Disclosing Materials at the Nanoscale [1 ed.] 9783038130994, 9783908158073

Citation preview

DISCLOSING MATERIALS AT THE NANOSCALE

DISCLOSING MATERIALS AT THE NANOSCALE

Edited by

P. VINCENZINI World Academy of Ceramics and National Research Council, Italy G. MARLETTA University of Catania, Italy “Disclosing Materials at the Nanoscale”. Advances in Science and Technology, 51. Proceedings of the International Symposium “Disclosing Materials at the Nanoscale” of CIMTEC 2006 11th International Ceramics Congress and 4th Forum on New Materials, held in Acireale, Sicily, Italy on June 4-9, 2006

TRANS TECH PUBLICATIONS LTD Switzerland UK USA on behalf of TECHNA GROUP Faenza Italy

Copyright  2006 Trans Tech Publications Ltd, Switzerland Published by Trans Tech Publications Ltd., on behalf of Techna Group Srl, Italy

All rights reserved. No part of this book may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, recording, photocopying or otherwise, without the prior written permission of the Publisher. No responsibility is assumed by the publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein.

Trans Tech Publications Ltd Laubisrutistr. 24 CH-8712 Stafa-Zuerich Switzerland http://www.ttp.net ISBN 3-908158-07-9 ISBN-13 978-3908158-07-3 Volume 51 of Advances in Science and Technology ISSN 1661-819X Full text available online at http://www.scientific.net The listing of the other Volumes (1 to 44) of the Series "Advances in Science and Technology" is available at TECHNA GROUP website: http://www.technagroup.it

Distributed worldwide by

and in the Americas by

Trans Tech Publications Ltd Laubisrutistr. 24 CH-8712 Stafa-Zuerich Switzerland

Trans Tech Publications Inc. PO Box 699, May Street Enfield, NH 03748 USA

Fax: +41 (44) 922 10 33 e-mail: [email protected]

Phone: +1 (603) 632-7377 Fax: +1 (603) 632-5611 e-mail: [email protected]

CIMTEC 2006 th

11 INTERNATIONAL CERAMICS CONGRESS & 4th FORUM ON NEW MATERIALS

Chairman Pietro Vincenzini, World Academy of Ceramics, Emeritus Research Manager National Research Council, Italy Co-Chair International Ceramics Congress Robert Freer, International Ceramic Federation, University of Manchester and UMIST, UK Co-Chair Forum on New Materials Robert Nemanich, International Union of Materials Research Societies, North Carolina State University, USA

International Symposium “Disclosing Materials at the Nanoscale” Symposium Co-Chairs: Yoshio Bando, Japan D. Bimberg, Germany Richard W. Siegel, USA Programme Chair: Giovanni Marletta, Italy Members: James H. Adair, USA François Beguin, France Richard J. Blaikie, New Zealand Jürgen Brugger, Switzerland Enric Canadell, Spain Kee Joo Chang, Korea YongHo Choa, Korea Gan-Moog Chow, Singapore Alain Claverie, France Vicki Colvin, USA Antonio Correia, Spain M. Lucia Curri, Italy Ali Eftekhari, Iran Daisuke Fujita, Japan Lian Gao, China Ricardo Garcia, Spain Yury Gogotsi, USA D. Wayne Goodman, USA Horst Hahn, Germany J.G. Hou, China Michael Z. Hu, USA Sumio Iijima, Japan Philippe Lambin, Belgium Charles M. Lieber, USA Jacques Livage, France Meyya Meyyappan, USA Paolo Milani, Italy Seizo Morita, Japan Koichi Niihara, Japan Asao Oya, Japan Jong-Wan Park, Korea David Pettifor, UK Simon R. Phillpot, USA Riichiro Saito, Japan Massimo Sancrotti, Italy Zhigang Shuai, China Zhong Lin Wang, USA Andrew T.S. Wee, Singapore Mark E. Welland, UK Roland Wiesendanger, Germany Zhiyuan Zhu, China

PREFACE CIMTEC 2006-11th INTERNATIONAL CERAMICS CONGRESS & 4th FORUM ON NEW MATERIALS was held in Acireale, Sicily, Italy on June 4-9, 2006. This qualitative and comprehensive congressional event, similarly to the previous editions, has been designed to encompass and derive synergism from a broad interdisciplinarity network capable of offering opportunities for identifying and exploring new directions for research and production. The above based on the view that ongoing and future innovations require at an ever increasing extent a complex array of interconnections among scientific research, innovating technology and industrial infrastructure. CIMTEC 2006 consisted of two major, closely intertwined events: the 11th INTERNATIONAL CERAMICS CONGRESS and the 4th FORUM ON NEW MATERIALS. The World Academy of Ceramics and the International Ceramic Federation (ICF) acted as principal endorsers for the first one, and the International Union of Materials Research Societies (IUMRS) for the FORUM. The 11th INTERNATIONAL CERAMICS CONGRESS included 13 Sections (61 Sessions) which covered recent progress in all relevant fields of ceramics science and technology, including the emerging area of nanomaterials in which a Special Symposium has been devoted. The 4th FORUM ON NEW MATERIALS consisted of five parallel International Conferences (“Mass and Charge Transport in Inorganic Materials”; “Science and Engineering of Novel Superconductors”; “Diamond and New Carbon Materials”; “Materials in Clinical Applications”and “Advanced Inorganic Fibrous Composites for Structural Applications”) and of two Special Symposia (“Spin Injection and Transport in Magnetoelectronics” and “Biomedical Applications of Nano Technologies”). A balanced, high quality programme of invited and contributed papers resulted from the over one thousand scientific and technical contributions effectively presented during the five working days to a large international audience coming from fifty-three countries throughout the world. The 9 volumes which constitute the Official Proceedings of the CIMTEC 2006 contain a wide selection of the papers presented. Where appropriate, the chapters of each volume have been organized in such a way to follow the flowsheet of the sessions of the congress. The volume dedicated to the INTERNATIONAL CERAMICS CONGRESS hosts Invited and Contributed matter given at the thirteen Technical Sections, i.e.: Section A - Fundamentals of Structure, Property, Reaction and Unit Processes of Ceramic Systems, Section B - Corrosion and Tribology Behaviour of Ceramics, Section C - Ceramic Powders Synthesis and Processing, Section D - Sintering Science and Technology, Section E - Non Conventional Routes to Ceramics, Special Session E-11 - Self-propagating High-temperature Synthesis of Ceramics, Special Session E-12 - Layered and Functionally Graded Materials, Section F - Surface Engineering with Ceramics, Section G - Ceramic Composites, Section H Ceramic Joining, Section I - Structural Ceramics, Section J - Ceramics for Electrochemical, Chemical, Energy, Environmental and Refractory Applications, and Section K - Electrical, Magnetic and Optical Ceramics. Matter presented at the International Symposium on nanoceramics has been collected in a separate volume: "Disclosing Materials at the Nanoscale". Invited and contributed papers presented at the FORUM ON NEW MATERIALS have been collected in seven volumes. Volume 1: 3rd International Conference "Mass and Charge Transport in Inorganic Materials"; Volume 2: 5th International Conference "Science and Engineering of Novel Superconductors"; Volume 3: 4th International Conference "Diamond and Other New Carbon Materials"; Volume 4: 7th International Conference "Materials in Clinical Applications"; Volume 5: 5th International Conference "Advanced Inorganic Fibrous Composites for Structural Applications"; Volume 6: “Spin Injection and Transport in Magnetoelectronics”; Volume 7: “Biomedical Applications of Nano Technologies”. It is noteworthy pointing out how the attribution of papers to the various sections of the books may having been subject to some shortcoming and uncertainties deriving essentially from the same material or

compound possibly involving different functions and uses, or papers containing at the same time aspects linked to structure, processing techniques and properties and their relationships. Where possible the general criterium was adopted to account for the predominant function performed in the specific context in all those cases where the material by itself might be able to carry out different functions. Likewise, an attempt was made to determine the most appropriate location for those communications where complex relationships among processing, properties and structure are involved. It may be supposed that not all may be entirely satisfied with the solutions adopted, being the matter subjective to some extent. Nevertheless it is hoped that, in spite of the above limitations, also deriving from the very large number and variety of the matter dealt with, a satisfactory compromise may have been reached in making these proceedings volumes logically presented and easy to consult. Most of the papers were written by authors whose mother tongue is not English. Therefore, considerable revision of the original texts was often required. The partial reworking of several papers and sometimes even complete rewriting was necessary to make clear work valid as regards the technical content but difficult to understand because of lack of proficiency in the English language. Even so, in order to allow the scientific and technical community to have access to the proceedings volumes within a reasonable length of time, compromise was necessary in regard to the quality of writing, and papers containing language imperfections were considered acceptable provided that their technical content was adequate and easily understandable. The Editor, who also acted as the Chairman of CIMTEC 2006, would like to express his sincere appreciation to all the Institutions and Professional Organizations involved in the congress, to the members of the International Advisory Committees, the National Coordinating Committees, the Co-Chairs Prof. Robert Freer (UK) for the INTERNATIONAL CERAMICS CONGRESS and Prof. Robert Nemanich (USA) for the FORUM ON NEW MATERIALS, the Programme Chairs, the Lecturers, the technical staff of Techna Group, and to the many others who directly or indirectly contributed to the organization. Indeed it was mainly through the involvement of the above organizations and individuals, and the active participation of most qualified experts from major academic and government research institutes and industrial R&D centers of many countries that a very valuable scientific programme could be arranged. It is therefore expected that the Proceedings of CIMTEC 2006-11th INTERNATIONAL CERAMICS CONGRESS & 4th FORUM ON NEW MATERIALS will be accepted as an original and valuable contribution to the literature in the field.

P. VINCENZINI World Academy of Ceramics Emeritus Research Manager National Research Council of Italy

Table of Contents Committee Preface

Session 1 - Synthesis, Functionalization and Properties of Nanomaterials Direct Synthesis of Tungsten Oxide Nanowires on Microscope Cover Glass F.C. Cheong, Y.W. Zhu, B. Varghese, C.T. Lim and C.H. Sow Electrochemical Control of the Magnetic Properties of Co and CoCu/Co Nanowires L. Piraux, M. Darques and S. Michotte In Situ Observation of Quantized Growth of Titanium Silicide in Ultra High Vacuum Transmission Electron Microscope (UHV-TEM) C.L. Hsin, W.W. Wu, H.C. Hsu and L.J. Chen Nanocrystalline TiO2 for Solar Cells and Lithium Batteries L. Kavan Synthesis and Characterization of Mesostructured Silicas and Gold Frameworks as Active Matrices for Biomolecule Encapsulation L.E. Iton, A.J. Crisci, V. Vajdova, P.D. Laible, C.T. Burns and M.A. Firestone Synthesis and Cathodoluminescence Study of Well-Aligned Planar-Tip and Tapered-Tip ZnO Nanorods J.H. He, C.W. Wang, K.F. Liao and L.J. Chen Self-Assembled Low-Resistivity NiSi Nanowire Arrays on Epitaxial Si0.7Ge0.3 on (001)Si W.W. Wu and L.J. Chen Preparation and Cathodoluminescence Properties of Ga-Doped ZnS Nanowalls M.Y. Lu, H.C. Hsu and L.J. Chen Nanoparticles of La(1-x)SrxMnO3 (x = 0.33, 0.20) Assembled into Hollow Nanostructures for Solid Oxide Fuel Cells A.G. Leyva, J. Curiale, H.E. Troiani, M. Rosenbusch, P. Levy and R.D. Sánchez Preparation of SiC Nanofibers by Using the Polymer Blend Technique Z. Correa, H. Murata, T. Tomizawa, K. Tenmoku and A. Oya Investigation on the Reinforcement of Multi-Walled Carbon Nanotubes on Alumina Matrix J. Sun and L. Gao Template Synthesis of Nanostructured Carbons T. Kyotani and H. Orikasa Characterization of Fullerene / TriA-PI Composites Y. Shinoda, I. Shiota, Y. Ishida, T. Ogasawara and R. Yokota

1 7 14 20 30 38 42 48 54 60 64 68 75

Session 2 - Nanoscale Characterization and Techniques

Carbon-Based and Other Nanostructures Obtained via Cluster-Assembling: A View Combining Electron Spectroscopies and Nanospectroscopies L. Gavioli and M. Sancrotti Probing the Role of Nanoroughness in Contact Mechanics by Atomic Force Microscopy R. Buzio and U. Valbusa Structural Modification of Doped and Undoped Nanocrystalline TiO2 by TemperatureResolved XRPD F. Matteucci, G. Cruciani, M. Dondi, G. Baldi, M.C. Dalconi, A. Barzanti, G. Lorenzi and C. Meneghini

81 90

99

Session 3 - Nanomanufacturing and Tools

Patterned 2D and 3D Assemblies of Nanoparticles on Molecular Printboards J. Huskens Sub-Wavelength Texturing for Solar Cells Using Interferometric Lithography W.L. Chiu, M.M. Alkaisi, G. Kumaravelu, R.J. Blaikie, R.J. Reeves and A. Bittar Some Investigations on Gallium Arsenide MEMS. Simulation of Microstructure Shapes C.R. Tellier, G. Huve and T.G. Leblois

105 115 121

b

Disclosing Materials at the Nanoscale

Session 4 - Theory, Modelling and Simulation

Dependence of Adhesion and Reflection on Orientation in Nanocluster Deposition S.C. Hendy and A. Awasthi Many-Scale Simulation of ABS/PC Blends for the Automotive Industry M. Fermeglia, M. Ferrone, P. Cosoli, M.S. Paneni, R. Venica, S. Pricl, S. Sinesi, P. Posocco and L. Martinelli Molecular Dynamics Simulation of Organic Molecules Distorted Conformation in Zeolites E. Semprini, G. Perez, F. Stefani, P. Cafarelli, A. De Stefanis and A.A.G. Tomlinson

127 134 140

Session 5 - R&D Advances in Devices and Applications

Nanotubes Based Composites for Energy Storage in Supercapacitors E. Frackowiak Nanocrystals in High-k Dielectric Stacks for Non-Volatile Memory Applications M. Fanciulli, M. Perego, C. Bonafos, A. Mouti, S. Schamm and G. Benassayag Simulation of the Growth of Copper Films for Micro and Nano-Electronics M. Cobian, E. Machado, M. Kaczmarski, B. Braida, P. Ordejon, D. Garg, J. Norman and H. Cheng Industrial Ink-Jet Application of Nano-Sized Ceramic Inks M. Dondi, F. Matteucci, D. Gardini, M. Blosi, A.L. Costa, C. Galassi, G. Baldi, A. Barzanti and E. Cinotti Evolution of Pt Nanoclusters Morphology on PEMFC Electrode due to Methanol Oxidation Reaction Studied by Electron Microscopy and Synchrotron Grazing Incidence X-Ray Diffraction M. Alvisi, G. Galtieri, L. Giorgi, E. Serra, T. Di Luccio and R. Giorgi Nanostructured Films of Polyphthalocyanines for Sensor Applications V. Anisimov, A. Borisov, O. Ivanova, S. Krutovertsev, A. Sherle and E. Oleinik LGS and LGN Microresonators: Applications to High Temperature Nanobalances T.G. Leblois Humidity Sensors Based on Nanostructured Materials S. Krutovertsev, A. Tarasova, L. Krutovertseva, A. Zorin and O. Ivanova Quantum Dots in GaInP/GaInAs/GaAs for Infrared Sensing H. Lim, S. Tsao, M. Taguchi, W. Zhang, A.A. Quivy and M. Razeghi

145 156 167 174

181 187 191 197 201

Advances in Science and Technology Vol. 51 (2006) pp 1-6 © (2006) Trans Tech Publications, Switzerland doi:10.4028/www.scientific.net/AST.51.1

Direct Synthesis of Tungsten Oxide Nanowires on Microscope Cover Glass F.C. Cheong1,a, Y.W. Zhu1,3,b, B. Varghese1,3,c, C.T. Lim2,3,d , C.H. Sow1,3,e 1

Department of Physics, Blk S12, Faculty of Science, National University of Singapore, 2 Science Drive 3, Singapore 117542 2

Division of Bioengineering and Department of Mechanical Engineering, National University of Singapore, 9 Engineering Drive 1 Singapore 117576

3

National University of Singapore Nanoscience and Nanotechnology Initiative, 2 Science Drive 3, Singapore 117542

a

[email protected], [email protected], [email protected], [email protected], e [email protected]

Keywords: tungsten oxide, nanowire, thermal annealing

Abstract. A simple technique to synthesis crystalline Tungsten Oxide nanowires is presented. Using a standard thermal hotplate, a pure 99.9% tungsten foil is annealed to 484 ± 5 oC under ambient condition to generate vapor deposition of the heated materials on a piece of 150µm thick glass cover slide pressing on the tungsten foil. Tungsten oxide nanowires are found to deposit on the cover slide facing the heated tungsten foil. These tungsten oxide nanowires were characterized with SEM, TEM, EDX, micro-Raman and XRD. The crystalline nanowires were found to be straight and clean with a diameter of 10-300nm and a length of a few tens of micrometers. Introduction One dimensional metal oxide nanostructures have been studied extensively due to their unique chemical and physical properties. Tungsten oxide nanowire is one of such materials in the nano-regime. It is an important metal oxide semi-conductor material that offers a wide spectrum of useful properties [1-9]. Many different routes to synthesize WO3-X nanorods have been discussed before, they include high temperature thermal treatment [10–14], vapor deposition [15], wet chemical method [16, 17], etc. Liu et al. have grown tungsten oxide nanowires on a tungsten filament by heating it in vacuum with some room air leakage [10]. Gu et al. synthesized tungsten oxide nanowires on tungsten (W) tips/plates by thermal annealing at 700–750 ◦C in Ar ambient with oxygen leakage in the system [11]. Cho et al. prepared tungsten oxide whiskers by a short period pre-treatment of tungsten films in a steam of H2O before annealing them at above 1000◦C in vacuum [13]. In addition, Jin et al. synthesized tungsten oxide micro-bundles on W powders, foils, and wires under the vapour of H2O at 800–1000 ◦C [12]. A similar report by Pfeifer et al. used W-based solution for crystallization to synthesis large-scaleW18O49 crystals [14]. Liu et al. made use of current passing through a tungsten coil for thermal vapor deposition of oxidized tungsten to produce array of aligned tungsten oxide nanowires [15]. Using pre-synthesized mesolamellar precursor, Li et al. [17] controlled the removal of surfactant at high temperature to obtain tungsten oxide nanowires. For large-scale material synthesis, it is important that the process does not involve high temperature, multiple steps, and low pressure to reduce operation cost. And for applications such as photo-sensor, it is also important to grow WO3-x nanostructures on optical transparent substrates such as glass or quartz. Following a similar technique report by Yu et al. in using thermal hotplate in ambience to grow metal oxides nanostructures [18], we report a vapor phase deposition route to synthesize crystalline tungsten oxide nanowires assembled onto thin glass cover slide substrate by thermal annealing. However, the difference between current method compare to Yu et al. reported technique is the ability to directly deposit nanowires on other substrates rather than on the source material.

2

Disclosing Materials at the Nanoscale

Experimental Setup The tungsten foil used in this work is pure 99.9% W, (9.6g, 100mm x100mm x0.05mm) metal foil, from Aldrich Chemical Company, Inc. The microscope cover glasses are from Fisher Scientific, 35x50 mm2. Thickness is about 150µm. And the thermal hotplates used are Cimarec digital stirring hot plates. The temperature of the hot plates can be adjusted to a maximum of 484 ± 5 oC.

Fig. 1 (a) Schematic of the setup used in WO3-X synthesis with the thermal hotplate technique. (b, c) Electron micrographs of WO3-X found on the glass cover slides and a single nanorod found on the substrate.

For this experiment, a piece of tungsten foil is placed on top of a thermal hot plate with its heating plate maintained at a temperature of 484 ± 5 oC. A piece of Fisher microscopic glass cover slide was placed on top of the W foil and was heated together. Typical dimension of the W foil used is 1 cm x 1 cm. Two weights, usually standard microscopic glass slides, were placed at the edge of the cover slide to prevent it from shifting during the synthesis. A schematic of the whole experimental setup is as illustrated in Figure 1(a). The sample was heated for hours to days in ambience. After a few hours of heating the microscopic cover glass with W foil at 484 ± 5 oC, white patches were found on the glass surface facing the foil. The nature of the materials found on the glass surface was investigated with the Scanning Electron Microscope SEM (JEOL JSM-6400F) and transmission electron microscope TEM (JOEL JEM 3010 TEM with in build EDS). Fig 1(b) shows a SEM image of the surface of the glass surface scattered with nanowires. Fig 1(c) shows a TEM image of an isolated nanowire synthesized. The nanowires are mostly long and straight, growing randomly on the surface of the cover glass. The diameter of the nanowires ranges from 10 to 300 nm but for the same nanowire, the diameter is quite uniform along different parts of the nanowire. The synthesized nanowires were further characterized with Micro-Raman (Renishaw System2000), and XRD (Phillips PW 127).

Advances in Science and Technology Vol. 51

3

Results and Observations The diameter of the as grown nanowires ranges from tens of nanometers to a few hundreds nanometers. And the length of individual nanowires ranges from a few micrometers to hundreds of micrometers. Longer duration of heating naturally resulted in longer and denser nanowires growth. Figures 2(a), (b) and (c) show SEM images of the as grown nanowires on glass cover slide grown at different heating duration of 1, 2, and 3 days respectively.

Figure2 (a-c) Scanning Electron micrographs of as synthesized WO3-X on glass slide after 1, 2, and 3 days of heating. (Scale bar =10 micrometers) (d) A plot of the average length of nanowires versus the growth time. (e) A plot of average diameter of the nanowires versus the growth time.

From the analysis of the SEM images, the average length of the nanowires increases linearly with time, but the average width (diameter) of nanowires plateaus at about 200nm after about 2 days of synthesis as shown in Figure 2(d) and (e) respectively. On other hand the tungsten foil that was heated had little or no nanowires being observed on the surface. And at slightly lower temperature of 474 ± 5 oC, no nanowires were observed growing on the substrate after 12 hours of heating. The electron diffraction x-ray spectrum (EDX) analysis of a single nanowire, in Figure 3(a), suggested that the chemical composition of the nanowire is mainly tungsten and oxygen. The Cu peak was attributed to the TEM copper grid used as support structure for the nanowire. Hence, it appears that the nanowires are not doped with any foreign elements. The micro-Raman spectrum of the as-synthesized products is shown in Figure 3(b). According to the literature reports [17,19,20], the strong well-defined bands between 268 to 426 cm-1 correspond to the bending of O-W-O bonds and the modes between 636 to 850 cm-1 was ascribed to the stretching of W-O bonds.

4

Disclosing Materials at the Nanoscale

Figure 3 (a) EDX signal at one portion of WO3-X nanowire. (d) Micro-Raman spectrum of the as-synthesized product deposited on Silicon (c) XRD signal from the as synthesized tungsten oxide nanowires on cover glass substrate.

From the XRD in Figure 3(c), we can identify (010) peak and the (014) peaks of monoclinic W18O49, (010) from the broad amorphous glass substrate peak in background. On the basis of Transmission electron microscope, the as synthesized nanowires are usually long and straight, with diameters ranging from 10nm -300nm and length of the nanowires up to several tens of micrometers.

Advances in Science and Technology Vol. 51

5

Figure 4 Transmission electron micrograph of the WO3-x nanowire. The inset is the SAED image of this nanowire.

A high-resolution transmission electron microscope (HRTEM) image (Figure 4) was taken on a single nanowire with diameter of about 50nm, providing a clear insight into the structure lattice spacing of the tungsten oxide nanowires. The spacing of the lattice fringes was found to be about 0.377nm and 0.374nm respectively as shown in Figure 4. This has close resemblance to the monoclinic WO3-x [17], indexing in the {010} and {103} planes. The nanowires grow preferentially along the [010] due to the close-pack plane of the monoclinic crystal in {010}. The selected area of the electron diffraction (SAED) diagram shown in the inset of Figure 4 supported the periodic growth in [010]. We propose the following mechanism for the growth of the tungsten oxide nanowires. Even though the temperature employed in this hotplate technique is much lower than the melting point of bulk tungsten [21], surface melting and evaporation for tungsten is possible at a lower temperature. The surface melting will constantly provide some minute amount of tungsten vapor even at such low temperature to interact with the oxygen in the atmosphere to form WO3-x. The vapor will condense onto the cooler cover glass on top of the tungsten foil. During the condensation, these vapor aggregate for the formation of the nanowires. As the kinetics of crystal growth dominates in the {010} direction, one-dimensional nanowires form on the cover glass. Moreover, it should be noted that we did not specifically made use of any catalysts on the coverglass, thus it is possible that the formation of the nanowire is self-catalytic. Conclusion WO3-x nanowires with relatively large density were synthesized on a large scale on microscope cover glass using the hotplate technique. We can control the average length and density of the nanowires by the duration of synthesis. The ability to directly synthesize nanowires onto glass substrate will be useful for studying optical properties of such nanowires.

6

Disclosing Materials at the Nanoscale

Acknowledgements C.H. Sow acknowledges support from National University of Singapore Academic Research Fund. References [1] Lee K.H., Fang Y. K., Lee W.J., et al., Sens. Actuators, B Vol. 69, (2000), p. 96 [2] Llobet E., Molas G., Molinas P., el at., J. Electrochem. Soc. Vol. 147. (2000), p. 776 [3] Li F.B., Gu G.B., Li X.J., Wan H.F., Acta Phys.-Chim. Sinica, Vol. 16 (2000), p. 997 [4] Qu W.M., Wlodarski W., Sens. Actuators, B Vol. 64, (2000), p. 42. [5] Turyan I., Krasovec U.O., Orel B., Saraidorov T., el at., Adv. Mater. Vol. 12 (2000), p.330 [6] Sayama K. Mukasa K., Abe R., Abe Y., Arakawa H., Chem. Commun. Vol. 21 (2001), p2416 [7] Bock C., MacDougall B. Electrochim. Acata (2002), Vol. 47, p.3361 [8] Li Y, Bando Y and Golberg D, Adv. Mater. Vol. 15 (2003), p.1294 [9] Zhou J., Gong L., Shao Z.D., et al., Appl. Phys. Lett, Vol. 87 (2005) p.223108 [10] Liu K, Foord D T and Scipioni L, Nanotechnology Vol. 16 (2005), p. 10 [11] Gu G, Zheng B, Han W Q, Roth S and Liu J, Nano Lett. Vol. 2 (2002), p.849 [12] Jin Y Z, Zhu Y Q, Whitby R L D, Yao N, Ma R, Watts P C P, Kroto H W andWalton D R M, J. Phys. Chem. B Vol. 108 (2004), p. 15572 [13] Cho M H et al, J. Vac.Sci. Technol. B Vol. 22 (2004), p. 1084 [14] Pfeifer J, Badaljan E, Tekula-Buxbaum P, et al., Cryst. Growth Vol. 169 (1996), p. 727 [15] Liu J, Zhao Y and Zhang Z, J. Phys.: Condens. Matter Vol. 15 (2003), p. L453 [16] Lou X.W., Zeng H.C., Inorg. Chem. Vol. 42(20) (2003), p 6169 [17] Li X L, Liu J F and Li Y D, Inorg. Chem.(2003) Vol. 42(3), 921 [18] T. Yu, Y.W. Zhu, X.J Xu, et al., Small Vol. 2(1) (2006), p. 80 [19] Frey G.L., Rothschild A., Sloan J., Rosentsveig R., et. al., J. S. S. Chem. Vol 162 (2001), p. 300 [20] Boulova M., Rosman N., Bouvier P., and Lucanzeau G., J. Phys.: Condens. Matter, Vol. 14 (2002), p.5849–5863 [21] Langmuir I., Phys. Rev., Vol. 4(2), (1915), p.138-157

Advances in Science and Technology Vol. 51 (2006) pp 7-13 © (2006) Trans Tech Publications, Switzerland doi:10.4028/www.scientific.net/AST.51.7

Electrochemical Control of the Magnetic Properties of Co and CoCu/Co Nanowires L. Piraux, M. Darques and S. Michotte Unité de Physico-Chime et de Physique des Matériaux, Place Croix du Sud, 1, 1348-Louvain-la-Neuve, Belgium [email protected] Keywords: nanowires, nanomagnetism, electrodeposition, templating.

Abstract. Using appropriate electrodeposition conditions, it is shown that the structural and magnetic properties of arrays of Co and CoCu/Co nanowires can be controlled. The hcp c axis orientation can be oriented parallel or perpendicular to the wire axis simply by changing the pH of the electrolytic solution and/or deposition rate. This selected orientation of the c axis leads to a drastic change in overall magnetic anisotropy and is promising for the fabrication of spin valves structures by electrodeposition. Introduction Over the last decade, many groups have been exploring the concept of using the pores in nanoporous media as templates to prepare arrays of nanowires or nanotubes by various synthesis techniques, such as electrodeposition and chemical methods. This so-called “template method” makes possible to fabricate various nanowires that are difficult to form using conventional nanolithographic process, including noble metals, ferromagnets, superconductors, semimetals, alloys, oxides, carbon nanotubes, conducting polymers and different types of multilayers. Though there now exist a huge range of hosts, most studies in this area has entailed the use of two types of templates: track-etch polymer membranes and nanoporous aluminas. Under optimized fabrication conditions, these two types of templates exhibit well-defined nanoporosity with geometrical parameters controllable to a large extent [1-3]. The resulting nanowires have diameters ranging between a few nm and a few microns and the aspect ratio (length to diameter) can be controlled to a large extent up to 103. A variety of novel physical properties and potential applications have been identified in relation to the nanowire composition and to their nanoscopic dimensions [4]. The intent of this article is to present a convenient way of controlling the structural and magnetic properties of arrays of electrodeposited hcp Co nanowires and CoCu/Cu multilayered nanowires. Fabrication Arrays of pure cobalt and CoCu/Co nanowires were grown by electrodeposition into the pores of polycarbonate membranes with diameter 23-70 nm. All the membranes used in this study have a porosity P around 3.5% to ensure that for each nanowires sample, the dipolar interactions between wires has the same magnitude, whatever the diameter of the wires. A Cr(20 nm)/Au(300 nm) layer is beforehand evaporated on one face of the membrane to serve as a cathode. For pure Co nanowires synthesis, we use a concentrated 0.85-M Co electrolyte buffered with 0.48-M boric acid. The growth is done in galvanostatic mode with current densities ranging from 5 to 150 mA.cm-2. For the synthesis of CoCu/Cu multilayered nanowires we used a 250 g/l CoSO4 + 0.25 g/l CuSO4 + 30 g/l H3BO3 electrolyte and a pulse-plating method in which the magnetic and nonmagnetic layers are deposited from a single solution by switching between the deposition potentials of the two constituents. The Cu layers were deposited at -0.5V vs Ag/AgCl reference electrode and the magnetic ones at -0.9V or -1.2V. The thicknesses are CoCu(50nm)/Cu(5nm) for all samples and the Cu content in the magnetic layers is less than 2%. For both electrolytic

8

Disclosing Materials at the Nanoscale

solutions, the initial pH is around 4.0 and pH is increased or lowered by adding either diluted NaOH or H2SO4. FMR experiments Ferromagnetic resonance (FMR) measurements are used to probe the magnetic properties of the nanowires using a transmission line technique [5]. After filling of the pores, a 150µm width Cu(800nm)/Au(200 nm) transmission line is evaporated on the free surface of the membrane which serves as a waveguide for frequencies up to 50 GHz. The small wire diameter (compared to the skin depth > 0.5µm) and the insulating nature of the nanoporous templates make these magnetically-filled porous membranes particularly well adapted for properties investigation at GHz frequencies. The microwave signal propagating along the microstrip transmission line produces a microwave pumping field which is perpendicular to the nanowires and induces a precession of the magnetization around the static equilibrium position. At ferromagnetic resonance, maximum power is absorbed from the incident microwave signal and the corresponding minimum in the transmitted power is recorded by a network analyzer. The analysis of the FMR results is made in the saturation state where the magnetization inside the wires can be considered as single domain. The absorption of this microwave by the nanowires is measured at different excitation frequencies, while sweeping an external magnetic field parallel to the nanowires from 10 kOe down to zero field. More details about the experimental setup can be found elsewhere [5]. The resonance condition for an array of nanowires is obtained using simple magnetostatic arguments, after the formulation by Smit and Beljers [6]. For hexagonal close packing (hcp) Co nanowires, when the magnetic field is applied parallel to the wires, these conditions are : If the c-axis is parallel to the wire axis (HU >0) : f = H R + 2#Ms (1- 3P) + HU " (1)

If the c-axis is perpendicular to the wire axis (HU 2 Å to avoid too high energy interactions. The MD run at T = 200K (NVT ensemble), 100ps, time step 0.1ps, update charge equilibration every 1000 steps, shows that the twist-boat conformation is retained trough all run (Fig. 3).

Figure 3. Snap-shot of the trajectory file for cyclohexane sorption on a mordenite side channel at 200K

144

Disclosing Materials at the Nanoscale

Similar MD at T = 600K gave a trajectory file equivalent to the one at 200K (Fig. 4)

Figure 4. Snap-shot of the trajectory file for cyclohexane sorption on a mordenite side channel at 600K Conclusions The calculations show that it is possible that a molecule configuration, at higher energy content than another, can be stored in a particular void space of a zeolite in selected ranges of temperature and pressure. References [1] G. Vitale, C.F. Mellot, and A.K. Ceetham, J. Phys. Chem. B 101 (1997) 9886. [2] H.C. Andersen, J. Chem. Phys. 72(4) (1980) 2384. [3] D.W. Heerman Ed., Computer Simulation Methods in Theoretical Physics, (Springer-Verlag, UK 1990). [4] N. Metropolis, A.W. Rosenbluth, M.N. Rosenbluth, A.H. Teller, E. Teller, J Chem Phys., 21 (1953) 1087. [5] K.B. Wilkberg, V.A. Walters, and W.P. Dailey, J. Am. Chem. Soc. 107 (1985) 4860. [6] MSI, User Guide (MSI press, USA 1996). [7] N. Sivasankar and S. Vasudevan, Catalysis Letters, 97(1-2) (2004) 53.

Advances in Science and Technology Vol. 51 (2006) pp 145-155 © (2006) Trans Tech Publications, Switzerland doi:10.4028/www.scientific.net/AST.51.145

Nanotubes Based Composites for Energy Storage in Supercapacitors Elzbieta Frackowiak Poznan University of Technology Institute of Chemistry and Technical Electrochemistry 60-965 Poznan, Piotrowo 3, Poland e-mail: [email protected] Keywords: carbon nanotubes, conducting polymers, composites, manganese dioxide.

Abstract. Composites based on nanotubes with such active materials as conducting polymers (e.g. polyaniline, polypyrrole), transition metal oxides (manganese oxide) and carbons enriched in heteroatoms (e.g. nitrogen) have been considered as electrodes for supercapacitors. The open mesopores network formed by the entanglement of nanotubes permits the ions to diffuse easily to the active surface of the composite components, hence, a good charge propagation and high values of capacitance (100-350 F/g) have been obtained. Since nanotubular materials are characterized by a high resiliency, the composite electrodes can easily adapt to the volumetric changes during charge/discharge, that drastically improves the cycling performance of supercapacitors. Additionally, it has been proved that combining materials with pseudocapacitance properties in an asymmetric configuration is a very promising direction for developing a new generation of high performance supercapacitors. Introduction Supercapacitors are attractive energy devices with a high power especially adapted for transportation systems where a peak power is necessary for starting or acceleration whereas energy is recovered during braking. Performance of electrical double layer capacitor (EDLC) is generally based on electrostatic attraction of ions at the charged/discharged interfaces of high surface area materials. In this case capacitance is almost proportional to specific surface area according to Eq.1. The most often used materials for EDLC are activated carbons (Fig. 1).

Fig. 1 Performance of electrochemical capacitor (EDLC)

146

Disclosing Materials at the Nanoscale

Energy and power of capacitor is strongly determined by operating voltage being enhanced in organic electrolytic solutions according to Eqs. 2 and 3, however, the values of capacitance are smaller due to a definitively lower conductivity of organic medium versus aqueous electrolyte. Hence, there is a great interest to enhance operating voltage of capacitor in aqueous solutions. εS Cn = (1) d E = 12 CU 2 (2) U2 (3) 4 RS Enhancement of capacitor voltage can be realized by using materials different than activated carbons. Apart of electrostatic attraction of ions as in the case of activated carbons, a different type of capacitance can arise at electrodes of special type as for example conducting polymers, transition metal oxide (RuO2, MnO2) or N-riched carbons, when the extent of faradaically admitted charge depends linearly on the applied voltage [1-9]. Pseudocapacitance is an intermediate situation where faradaic reactions occur, but the potential changes as in a real capacitor [1]. High values of capacitance can be theoretically obtained with such materials which undergo quick pseudofaradaic reactions, as illustrated by polypyrrole (PPy) and ruthenium oxide (Eqs. 4, 5): P=

[PPy+ A-] + e- ↔ [PPy] + ARuOa(OH)b + δH+ + δe- ↔ RuOa-δ(OH)b+δ

(4) (5)

Interesting capacitance values could be obtained from pseudocapacitance materials, e.g. electrically conducting polymers (ECP), however, during cycling a degradation and consequently a loss of performance takes place because of swelling and shrinkage of ECPs. The doping of polymers requires the insertion/de-insertion of counter ions, which causes volumetric changes. The cycle-life of the polymer-based electrodes relates directly with the mechanical stress in the polymer film. A very interesting solution to overcome this drawback is to use carbon nanotubes (NTs) for improving the properties of the electrodes [2-5] [7-9]. Carbon nanotubes due to their unique morphology are characterized by exceptional conducting and mechanical properties which allow them to be used as three-dimensional support for active materials in supercapacitors. The percolation of the active material with nanotubes is more efficient than with traditional carbon blacks because of nanoscale distribution of electrode mass. Three types of composites based on manganese oxide, conducting polymers and N-rich carbon from polyacrylonitrile will be described. In all cases multiwalled carbon nanotubes (MWNTs) play a perfect role of a conducting component in composites. Experimental Amorphous manganese dioxide was obtained by potentiodynamic and galvanostatic method directly on nanotubess. MWNTs were prepared by decomposition of acetylene over a CoxMg(1-x)O solid solution catalyst [10]. The potentiodynamic deposition was realized on MWNTs from an electrolytic solution consisting of 0.5 mol.L-1 MnSO4 + 0.5 mol. L-1 H2SO4, with scan rate of 200 mV/s. The reference electrode was Hg/Hg2SO4 and Pt sheet served as auxiliary one. The deposition was carried out potentiodynamically between the potential limits of 0.3 and 1.3 V. In the case of the galvanostatic method, composites were prepared in 1 mol.L-1 MnSO4 + 1 mol. L-1 H2SO4 electrolytic solution with 3h deposition time and current density of 500 mA/g.

Advances in Science and Technology Vol. 51

147

Amorphous a-MnO2/MWNTs composites obtained by chemical deposition were prepared through precipitation of a-MnO2 from a KMnO4 + Mn(OAc)2.4H2O mixture which contains different and predetermined amounts of carbon nanotubes. The capacitor electrodes were formed as pellets consisting of 90% composite material and 10 % binder (PVDF, Kynar Flex 2801). In case of asymmetric system, active carbon NORIT (85% active material, 10% PVDF, 5% acetylene black) was used as material of negative electrode. Supercapacitors were operating in 1 mol.L-1 Na2SO4 electrolytic solution. To obtain ECP composites chemical and electrochemical polymerization of the monomer has been considered in order to get a polymer layer on the nanotubular materials. Chemical polymerisation supplied porous type of composites but composite homogeneity strongly depend on ECP type. For preparation of C/C composite based on nanotubes and polyacrylonitrile PAN we used PAN from Aldrich and multiwalled carbon nanotubes prepared by catalytic decomposition of acetylene at 600°C on a CoxMg(1-x)O solid solution [10]. Different proportion of blends with CNTs from 15-70 wt% and PAN were thoroughly mixed in an excess of acetone. After evaporation of acetone at room temperature, the blend was pressed to form pellets. The pellets were carbonised at 700-900°C for 30-420 min under nitrogen giving rise to C/C composite electrodes. In this case no binder was necessary. For physicochemical characterization of composites the nitrogen adsorption/desorption isotherms were analysed with ASAP 2010 (Micromeritics). The micropore volumes were determined by application of the Dubinin-Radushkevitch equation. The texture of pristine MWNTs and the composites was also studied by scanning electron microscopy (SEM, Hitachi S 4200) as well as transmission electron microscopy (TEM, Philips CM 20). The capacitance values were estimated in two-electrode Swagelok® system by galvanostatic charge/discharge, cyclic voltammetry with different scan rates and impedance spectroscopy techniques (from 100kHz to 1 mHz) using AUTOLAB potentiostat/galvanostat (ECOCHEMIE) and VMP II (Biologic-France). The capacitance values were calculated per mass of active material in one electrode. Results and Discussion Capacitance properties and electrochemical behaviour of all the composites were strongly determined by the presence of carbon nanotubes. A tight electrical contact of active material with nanotubes is crucial. The structure of carbon nanotubes is generally characterised by extended graphitic type layers which provide them a high electrical conductivity. It is believed that ballistic electronic transport can be also considered due to their perfection and one dimensional electronic structure. On the other hand, the high entanglement of the nanotubes creates an open network of mesopores. Hence, they are a perfect backbone for preparing nanocomposites with various electroactive materials. The open mesoporous network allows the ions to easily diffuse to the active surface of the composite components. The conducting properties of three dimensional composites and mechanical ability to adapt to their volumetric changes are essential to lower the equivalent series resistance (ESR) and consequently to increase the power of the device (Eq. 3). MnO2/CNTs composites An amorphous manganese oxide (a-MnO2·nH2O) belongs to cheap transition metal oxides but of poor electrical conductivity. A decrease of its resistivity is crucial for any practical application, especially for supercapacitor. It has been proved that with nanotubes, the percolation of the MnO2 active particles is more efficient than with the traditional carbon black which are generally used for the manufacture of electrodes [2].

148

Disclosing Materials at the Nanoscale

Two methods of MnO2 deposition on CNTs have been realized, i.e. electrochemical and chemical one. It is well known that the values of specific capacitance obtained for pristine CNTs is very low ca. 20-30 F/g [8]. Depending on the electrodeposition method and different parameters of composite preparation, it is possible to obtain MnO2/CNTs composite with various composition. The composite obtained by galvanostatic way consists of 30 % MnO2 and 70 % CNTs. The values of specific capacitance obtained for symmetric two-electrode cell depends on potential scan rates during voltammetry measurements and current density for galvanostatic charge/discharge. For example, at 2 mV/s scan rate, the capacitance is equal to 92 F/g and for 100 mA/g the capacitance is 68 F/g. Measurements of impedance spectroscopy were made from 100 kHz to 1 mHz. The value of specific capacitance (at 1 mHz) is 56 F/g. Experiments performed in a three electrode cell proved that MnO2/CNTs composite appears to be a perfect material as a positive electrode. Hence, building an asymmetric system in which positive electrode was from MnO2 composite and activated carbon served as negative one appears to give the most interesting features. As it is seen in Fig. 2 the operating voltage could be greatly enhanced until 2.0 V, that improves significantly a capacitor energy according to Eq. 2.

250 200 150 C [F/g]

100 50 0 -50 -100 -150 -200 0

0.5

1

1.5

2

2.5

U [V] 1V

1,5 V

2V

Figure 4. Voltametry characteristics (20 mV/s) in the following asymmetric configuration: (+)30 % MnO2/70 % CNTs//AC (-) The values of capacitance for the manganese dioxide/carbon nanotubes composite and activated carbon in asymmetric system are 172 F/g at 2 mV/s scan rate and 158 F/g for 500 mA/g current load. Examples of galvanostatic charge/discharge characteristics at two different loads are presented in Fig. 5 where very limited ohmic drop is observed at higher current. The application of composite consisting of CNTs and electrochemical manganese dioxide allows to reach capacitance values varying from 56 to 110 F/g in symmetric supercapacitor and from 110 to 216 F/g in the asymmetric configuration which additionally can operate in a wide potential range (2V in aqueous solution) increasing extremely energy of system. Carbon nanotubes were also used as a perfect support for amorphous manganese oxide obtained chemically. In both cases (chemical and electrochemical preparation) the same pseudocapacitance of hydrous oxides like a-MnO2·nH2O is attributed to redox transitions with exchange of protons and/or cations with the electrolyte following Eq. (6):

Advances in Science and Technology Vol. 51

149

MnOa(OH)b + nH+ + ne- ↔ MnOa-n(OH)b+n

(6)

where MnOa(OH)b and MnOa-n(OH)b+n indicate interfacial a-MnO2·nH2O in higher and lower oxidation states, respectively. Due to the low electrical conductivity of chemically prepared aMnO2·nH2O, a conducting additive is also required to realize a composite electrode for supercapacitors. 2.5

U [V]

2 1.5 1 0.5 0 -400

-200

0

200

400

t [s] 500 mA/g

1000 mA/g

Figure 5. Galvanostatic charge-discharge characteristics of asymmetric supercapacitor (+) MnO2(30%)/CNTs(70%) composite//activated carbon (-) at two different current densities. It is clearly demonstrated that adding carbon nanotubes improves the behavior of a-MnO2·nH2O as capacitor electrode. The specific capacitance values referred to the mass of a-MnO2·nH2O increases with the amount of CNTs. However, when the specific capacitance is referred to the total mass of the composite electrode material, it can be noticed that a carbon nanotubes loading higher than 10-15 wt% does not improve the electrodes performance. Therefore, 10-15 wt% of CNTs as conductive additive are enough to increase the capacitance of the a-MnO2·nH2O based electrodes to ca. 140 F/g. The microscopy observation for an a-MnO2/CNTs composite containing 15 wt% of nanotubes shows a remarkable template effect of the entangled nanotubes framework. Consequently, the composite electrodes have a good resiliency, and their porosity is high enough to favor the access of ions to the bulk of the active material. The other interesting property of this composite is an extremely good adhesion of the coating layer on the carbon nanotubes, that is reflected by the low values of electrical resistance (loss of resistivity from 2000 Ω cm2 to a few Ω cm2). Using carbon nanotubes instead of carbon black produces a remarkable improvement of the capacitive behavior of the a-MnO2·nH2O based electrode. The shape of the voltammetry curve demonstrates that the resistance of the electrodes with CNTs is much lower than in the case of using the conventional carbon black finding a performance closer to an ideal capacitor. An additional highly valuable effect of CNTs is the possibility to extract more energy from the a-MnO2·nH2O electrodes, obtaining specific capacitance values of 140 F/g in comparison with only 70 F/g when carbon black is used. Similar profitable effects of CNTs as backbones for nanocomposite electrodes can be expected for other transition metal oxide composite components.

Conducting polymers/CNTs composites

150

Disclosing Materials at the Nanoscale

Conducting polymers are very interesting materials for getting high capacitance values due to their pseudocapacitive character. Electrically conducting polymers (ECP) as polypyrrole (PPy), polyaniline (PANI) or polythiophene (PTh), can store and release charge through redox processes (eq. 4). When oxidation occurs (also referred as “doping”), ions from the electrolyte are transferred to the polymer backbone, and upon reduction (“dedoping”) the ions are released back into the solution. The doping/dedoping process takes place throughout the bulk of the electrodes, offering the opportunity of achieving high values of specific capacitance. However, degradation of ECPs during cycling leads to a loss of performance. It occurs because the doping of polymers requires the insertion/de-insertion of counter ions, which cause volumetric changes, in turn, a loss of interconnectivity. The cycle-life of the polymer-based electrodes relates directly with the mechanical stress in the polymer film. A very interesting solution to overcome this drawback is to use carbon materials for improving the mechanical properties of the electrodes [3-5] [7-8]. Chemical and electrochemical polymerisation of the monomer has been considered in order to get a homogenous layer of the conducting polymer on the nanotubular materials. Electrochemical deposition of polypyrrole (PPy) supplied a more uniform polymer coating than chemical deposition, but it does not affect so much electrochemical behaviour. SEM images of PANI/CNTs composite electrodes with 20 wt% of MWNTs compared with 100 wt% PANI pellet electrodes proved that the composite material is porous, keeping the advantage of the entangled network of the nanotubes, that allows a good access of the electrolyte to the active polymer. There is no doubt that such a texture of the capacitor electrode is optimal for a fast ionic diffusion and migration in the polymer so that the electrode performance should be improved. By contrast, the electrodes which are pressed from pure ECP, were very compact and not porous. Real supercapacitors have been built from such CNTs/ECP composite electrodes.The detailed electrochemical characterization was performed for all the type of composites. The square shape of the voltammogram demonstrates a typical capacitor behaviour, even if pure ECP usually gives more irregular characteristics, confirming a good synergy between ECP and CNTs. This kind of capacitive behavior is confirmed by the linear discharge on the galvanostatic curve. The maximum value of capacitance which was reached using a two electrode construction is 190 F/g for the PPy/CNTs composite and 360 F/g for PANI/CNTs. In fact, it has been demonstrated that the high values claimed in literature for very thin layers of pure conducting polymers in three electrode cells are valid only in a given potential range. For example, the PPy/CNTs composite investigated in a three electrode cell gives 250 F/g for the positive range of potential from 0.2 V to –0.3 V and 903 F/g for the negative range from -0.3 V to –0.6 V vs Hg/Hg2SO4. Taking into account these values, and the fact that the overall capacitance for a two electrode cell is inversely proportional to capacitance of both electrodes in series, the theoretical value cannot be higher than 196 F/g, that fits well with the maximum experimental value of 190 F/g. Electrochemical characterization of the electrochemically obtained CNTs/polypyrrole (PPy) and CNTs/PANI composites shows that the values of capacitance reaches ca. 170 F/g and 220 F/g with a good cyclic performance over 3000 cycles. Capacitance frequency dependence for composite with PANI shows only a moderate aggravation with cycling (Fig. 6). The high values of capacitance found with the CNTs/ECPs composites are due to the unique property of the entangled nanotubes which supply a perfect three-dimensional volumetric charge distribution and a well accessible electrode/electrolyte interface. It can be concluded that the bulk of the thin ECP layer coating the nanotubes is fully involved for quick pseudofaradaic processes, due to the open network of mesopores formed by the nanotubes, which allows a full doping of the polymer.

Advances in Science and Technology Vol. 51

151

250 before after cycling

C (F g-1)

200

150

100

50

0 1.E-04 1.E-03 1.E-02 1.E-01 1.E+00 1.E+01 1.E+02 1.E+03 1.E+04 1.E+05 f (Hz)

Fig. 6 Capacitance vs frequency for the PANI/CNTs composite before and after 3000 cycles at 300 mA g-1 current load using a voltage range from 0 to 0.6 V. Two electrode cell. Since the nanotubular materials are characterized by a high resiliency, the composite ECP electrodes can easily adapt to the volumetric changes during charge and discharge, that improves drastically the cycling performance. For all these reasons, composites incorporating a nanotubular backbone coated by conducting polymers with pseudo-capacitive properties represent an interesting breakthrough for developing a new generation of supercapacitors. N-riched carbons/CNTs composites A self standing C/C composite electrode for supercapacitor has been prepared by pressing a carbon nanotubes/polyacrylonitrile blend, followed by one-step pyrolysis of the pellet under neutral atmosphere. Although their specific surface area is very low, e.g. 200 m2/g, the composites demonstrate high values of capacitance, up to 100 F/g. These remarkable values are due both to the mesoporous nanotexture of the composite and to the pseudo-faradaic properties of the in-frame incorporated nitrogen functionality. Polyacrylonitrile (PAN) was selected for obtaining an electrochemically active carbon matrix containing in-frame incorporated nitrogen, because of its high carbonisation yield and high residual nitrogen content in the char. The pellets after pyrolysis of the CNTs/PAN blend keep their original shape without any noticeable cracks or defects, indicating that CNTs act as reinforcing backbone preventing from dimensional changes and shrinkage during the C/C composite formation. However, when the CNTs content is less than 15 wt% in the original composite, the weight loss increases dramatically during carbonisation and shrinkage appears. The optimal composition was obtained by carbonisation of the CNTs/PAN (30/70 wt%) blend at 700°C under nitrogen. SEM image of such composite is shown in Fig. 7. It is possible to distinguish zones where individual or interconnected aggregates of carbonised PAN are present. At such a high content of PAN (70%), most of the nanotubes are embedded in carbon from PAN, therefore they are not easily visible in the SEM pictures. Macropores due to the evolution of decomposition gases are also visible; this induced porosity of the pellet electrodes should be profitable for the electrolyte penetration.

152

Disclosing Materials at the Nanoscale

Fig. 7 SEM image of CNTs/PAN (30%/70%) composite after carbonization at 7000C From elemental analysis C=86.7%, N=9.2%, O=4.1% whereas C=89.5%, N=7.3%, O=3.1% from XPS, showing comparable values in the bulk and on the surface. The final C/C composite is rich in nitrogen, demonstrating that PAN is an efficient nitrogen carrier. The amount of oxygen in the composite is quite high, due to its incorporation by addition to the dangling bonds when the C/C composite is exposed to air after its formation. of CNTs composite with PAN blend [9]. The values of capacitance were determined in 1 mol.L-1 H2SO4 for C/C composites obtained from different blend compositions and carbonisation conditions, at temperatures ranging from 700°C to 900°C. The remarkable fact is that the values of capacitance are definitively larger for the composites than for pristine CNTs or carbonised PAN, showing an interesting effect after pyrolysis of the PAN/CNTs blends. Voltammograms obtained at different scan rates with the C/C composite formed by pyrolysis of the CNTs/PAN (30/70 wt%) blend at 700°C presented the typical “box-like” shape expected for an ideal capacitor. At higher scan rate, e.g. 100 mV/s the shape of the curve is still satisfactory, that indicates a quick dynamic of charge propagation with this kind of composite. The galvanostatic charge/discharge curves even at 500 mA/g loading current are close to linearity. It confirms that capacitors based on this composite can be loaded at a high regime, while keeping good capacitive behaviour. The next noticeable fact on the galvanostatic curves is a very small ohmic drop when the sign of current is reversed, that reflects a low value of equivalent series resistance RS of the cell. The impedance spectra (Nyquist plot) confirmed a value of RS below 0.7 Ω cm2. The almost vertical dependence of the imaginary part demonstrates a good capacitive behaviour without diffusion limitations. The estimated time constant does not exceed 0.08 s, that proves a high dynamic of charge propagation. Consequently, according to eq. 3, this new kind of material, without conductivity additive and binder, is quite appropriate for high power capacitors. As already known for activated carbons with nitrogen functionality, capacitance can be enhanced by pseudo-faradaic charge transfer [6]. It has been also shown that the amount of nitrogen and specific capacitance decrease almost linearly as heat-treatment temperature increases. In our case, the contribution of this effect due to nitrogen will be higher for the composites prepared at 700°C, because they have a higher nitrogen content than those obtained at 900°C. Assuming that 180 min and 700°C are the optimal carbonisation time and temperature, the comparison of the results obtained with three CNTs/PAN ratios, 15:85, 30:70 and 50:50, shows that the highest capacitance value is for the CNTs/PAN (30/70 wt%) blend. Hence, we can assume that a well adjusted balance of PAN and CNTs amounts is required for obtaining an optimal performance. The content of PAN must be enough high to favour a large gas evolution which

Advances in Science and Technology Vol. 51

153

develops porosity, but also to get the largest amount of residual nitrogen which contributes to the pseudocapacitance effect. The content of CNTs must be also quite high to prevent the composite shrinkage during carbonisation and consequently to assist the pores formation. In order to observe the electrochemical properties of the optimal C/C composite (CNTs/PAN (30/70 wt%) blend carbonised at 700°C during 180 min) in the full potential range (negative and positive), we used a three-electrode cell where the C/C composite served both for the working and auxiliary electrodes. Specific capacitance values of 120 and 76 F/g of pellet are obtained in the negative and positive ranges of potential, respectively, showing that the C/C composite is more efficient when used as negative electrode in the full capacitor. Such a difference of values between the two electrodes in aqueous acid medium could be related to different charge transfer reactions involving nitrogen functionalities, as in the examples shown in the following equations: >C=NH + 2 e- + 2 H+ ↔ >CH-NH2 (7) >CH-NHOH + 2 e- + 2 H+ ↔ >CH-NH2 + H2O (8) where >C stands for the carbon network. Since the specific performance of the C/C composites could be as well attributed to a sufficiently developed surface area, the porous texture data determined by nitrogen adsorption at 77 K for the different composites prepared at 700°C were compared to the characteristics of the single components (CNTs and carbonised PAN). The isotherms of pristine nanotubes and C/C composites are of type IV, characteristic of mesoporous carbons with a small amount of micropores. Both isotherms of the composite and of the pristine nanotubes are almost similar, showing that despite their minor proportion, the nanotubes act as a template determining the nanotextural properties of the composites. The low amount of gas adsorbed at low relative pressure indicates that the carbon from PAN does not contribute to any enhancement of the micropore volume. Before reaching the carbonisation temperature, PAN transiently melts and it has a tendency to strongly adhere to the nanotubes. Interestingly, although the C/C composites have a lower (or comparable) specific surface area than CNTs, their capacitance values are always higher, showing a profitable contribution of the carbon from PAN. The capacitance values of the three composites obtained after 180 minutes of pyrolysis, and with different proportions of CNTs and PAN, are not correlated with the specific surface area. For the composite from the CNTs/PAN (15/85 wt%) blend, the proportion of nanotubes is not high enough to provide a good backbone preventing from shrinkage. The specific surface area is rather low (164 m2/g), giving rise to a moderate value of capacitance (58 F/g). Although the specific surface area of the composite from the CNTs/PAN (30/70 wt%) blend is lower than that of the composite from CNTs/PAN (50/50 wt%), the former gives a noticeably higher value of capacitance (100 F/g), suggesting that its higher amount of nitrogen contributes more to the pseudo-faradaic charge transfer reactions. In order to confirm the beneficial effect of nitrogen on the capacitance properties, two C/C composites with different nitrogen content have been prepared from the CNTs/PAN (30/70 wt%) blend by pyrolysis at 700°C and 900°C. The specific capacitance decreases from 100 F/g for the 700°C composite to 48 F/g for the 900°C one, whereas the nitrogen values determined by XPS are 7.3 at % and 3.8 at %, respectively. It may be assumed that pyridinic nitrogen, which is the dominant contribution, may play a prominent role in the pseudo-faradaic properties. It is noteworthy to remark that C/C composites from CNTs/PAN were able to tolerate high current loads until 10 A/g (Fig. 8) that can have an important practical application. To summarize, high capacitance properties were demonstrated by low surface area C/C composite electrodes prepared by one-step carbonisation of CNTs/PAN blends. During the thermal decomposition of PAN, the layer which adheres to the nanotubes shrinks, leaving pores which reflect directly the nanotexture of the pristine nanotubular framework.

154

Disclosing Materials at the Nanoscale

120 100

C [F/g]

80 60 40 20 0 0

2000

4000

6000

8000

10000

12000

I [mA/g]

Fig. 8 Capacitance of composite from CNTs/PAN carbonised blend. versus current load A minimum amount of nanotubes, evaluated to ca. 30 wt%, is necessary to keep an optimised mesoporous texture which allows the active mass, i.e. the carbon from PAN, to be easily reached by the electrolyte ions. The remarkable capacitance properties of these new composites are due to a synergy between the template effect of CNTs on nanotexture and the pseudo-faradaic properties of the nitrogen functionality of carbonised PAN. Whereas nanotubes play an essential role on the mechanical point of view, they are also very profitable for a good charge propagation especially at high current loads.. Such C/C composites prepared without any activation treatment are very promising for high volumetric energy density capacitors, where high apparent density carbon materials are requested. Conclusions Carbon nanotubes appear to be a perfect component for composite materials with pseudocapacitance properties such as transition metal oxide (MnO2), conducting polymers and Nriched carbons. Such composites are promising alternative for the development of high performance supercapacitors. It is noteworthy that often a moderate porosity is sufficient for the perfect capacitor performance, hence, a useful volumetric capacity can be significantly improved.The developments in progress of hybrid systems, with two different electrodes working in their optimal potential range, are very promising. In these systems, nanotubes are essential for the electrodes realization, not only for conductive and mechanical purpose, but also to allow a good access of ions to the active surface. Hybrid capacitors, with an activated carbon as negative electrode and CNTs/a-MnO2 or a CNTs/conducting polymer composite as positive electrode, are actually demonstrating high capacitance values while operating at a voltage of 2 V in aqueous medium that greatly enhances capacitor energy. Such systems allow to reach electrochemical performance comparable to organic electrolytes, without the environmental drawbacks of these media. Acknowledgements The author acknowledge a partial financial support from grant DS 31-110/2006.

References

Advances in Science and Technology Vol. 51

155

[1] B.E. Conway Electrochemical supercapacitors – scientific fundamentals and technological applications Kluwer Academic/Plenum (1999). [2] E. Raymundo-Pinero, V. Khomenko, E. Frackowiak, F. Béguin, J. Electrochem. Soc.,152 (2005) A229 [3] K. Jurewicz, S. Delpeux, V. Bertagna, F. Béguin, E. Frackowiak, Chem. Phys. Lett. 347 (2001) 36 [4] V. Khomenko, E. Frackowiak, F. Béguin. Electrochim. Acta, 50 (2005) 2499 [5] K. Lota, V. Khomenko, E. Frackowiak, J. Phys. Chem. Solids, 65 (2004) 295 [6] G. Lota, B. Grzyb, H. Machnikowska, J. Machnikowski, E. Frackowiak., Chem. Phys.Lett. 404 (2005) 53 [7] E. Frackowiak, in; J. Schwarz et al. (Eds.) Encyclopedia of Nanoscience and Nanotechnology, Marcel Dekker, Inc., New York, 2004, 537. [8] E. Frackowiak, V. Khomenko, K. Jurewicz, K. Lota, F. Béguin. J. Power Sourc. 153 (2006) 153. [9] F. Béguin, K. Szostak, G. Lota, E. Frackowiak, Adv. Mater. 17 (2005) 238. [10] S. Delpeux, K. Szostak, E. Frackowiak, S. Bonnamy, F. Béguin, J. Nanosci. Nanotech., 2 (2002) 481.

Advances in Science and Technology Vol. 51 (2006) pp 156-166 © (2006) Trans Tech Publications, Switzerland doi:10.4028/www.scientific.net/AST.51.156

Nanocrystals in high-k dielectric stacks for non-volatile memory applications M. Fanciulli1,a, M. Perego1,b, C. Bonafos2,c, A. Mouti2,d, S Schamm2,e and G. Benassayag2,f 1CNR-INFM MDM National Laboratory, Via Olivetti 2, 20041 Agrate Brianza (MI), Italy 2 nMat Group, CEMES-CNRS, 29 rue J. Marvig 31055, Toulouse, France [email protected], b [email protected], c [email protected], d [email protected], e [email protected], f [email protected] Keywords: Silicon, nanocrystals, non volatile memory, high-k dielectrics

Abstract. The possibility to use semiconducting or metallic nanocrystals (ncs) embedded in a SiO2 matrix as charge storage elements in novel non volatile memory devices has been widely explored in the last ten years. The replacement of the continuous polysilicon layer of a conventional flash memory device by a 2-dimensional nanoparticle array presents several advantages but the fundamental trade-off between programming and data retention characteristics has not been overcome yet. The main problem is the limited retention time basically due to charge loss by leakage current through the ultra-thin SiO2 tunnelling dielectric. A longer retention time can be achieved by increasing the tunnel oxide thickness. This however implies higher operating voltages and consequently a reduced write/erase speed. Using high-k materials for tunnel and/or gate oxide it is in principle possible to achieve the goal of a low voltage non volatile memory device. The high dielectric constant of these materials allows using thicker tunnel oxide reducing leakage current. Several approaches have been explored to synthesise ordered arrays of ncs in SiO2 but the transfer of these methodologies to the synthesis of 2-d array of ncs in high-k materials is not trivial. In this work we address the material science issues related to the synthesis of metallic and semiconducting ncs in high-k materials using different techniques. A detailed review of the state of the art in the field is presented and further research strategies are suggested. Introduction In a conventional non-volatile memory (NVM) cell, the information is stored in a polysilicon floating gate embedded in the SiO2 matrix of a metal-oxide semiconductor field effect transistor (MOSFET). The floating gate can be charged or discharged either from the control gate or from the silicon substrate. Charge storage in the polysilicon layer results in a threshold voltage shift of the MOSFET. In order to fulfill the requirement of a retention time longer than ten years, relatively thick (8-10 nm) tunnel oxides are generally used in conventional floating gate NVM. As a consequence these devices are characterized by limited scalability (≥ 65 nm) and endurance (12 V) with slow programming cycles (>10 ms). Recently memory cell structures employing discrete storage nodes have been proposed to overcome these limitations. The use of semiconducting or metallic ncs embedded in a SiO2 matrix have been suggested as a solution to limit information loss due to leakage caused by localized defects in the oxide, thus improving the device retention characteristics [1,2]. As a consequence a more aggressive scaling of the tunnel oxide is allowed in these devices that exhibit therefore better characteristics compared to conventional NVM in terms of endurance, operation voltage, W/E speed and power consumption.

Advances in Science and Technology Vol. 51

The synthesis of ncs has been achieved by many different techniques, such as laser ablation [3], colloidal chemistry [4], sol-gel processes [5], chemical vapor deposition [6], molecular beam epitaxy [7], ion beam synthesis [8], evaporation and subsequent annealing of multilayered structures [9-10]. The main technological issues are related to the control on the position, density and size of the ncs and to the compatibility with conventional silicon planar technology. One of the most important parameter to control is the thickness of the tunnel oxide. Small variations in the distance between the ncs layer and the Si substrate strongly affect the charge transfer mechanism and induce dramatic changes of the retention time and W/E speed [11]. Using 2 nm thick tunnel oxides a direct tunneling mechanism can be used for programming and erasing. This allows to obtain a device operating at high speed (100 31.2 Electric conductivity µS·cm-1 10 1420 88

Black CoFe2O4 12 22 20-30 38.6 39.9 47.5 8

Results and discussion Chemico-physical properties of inks. The inks here considered are suspensions of actually nanometric pigments, mostly in between 10 and 70 nm, according to both DLS and TEM or FEG (Fig. 1). Therefore, the polyol method efficiency in synthesizing oxide or metal nanoparticles is confirmed regardless pigment composition [9-12]. Just the particles of the Magenta ink are slightly coarser, being to a large extent in the 50-70 nm range.

100 nm

MAGENTA TEM

CYAN FEG-SEM YELLOW TEM

50 nm Fig. 1. TEM and FEG-SEM micrographs of the nano-inks.

BLACK STEM

Disclosing Materials at the Nanoscale

RUTILE ANATASE

600°C 500°C

400°C 300°C

Mean Mobility Argument (°

176

80 60 40 20 0 Cyan Black

-20 -40 -60 0

2

4

6 8 10 12 14 16 18 Frequency [MHz]

Fig. 2. HT-XRD patterns of the Yellow ink Fig. 3. Dynamic mobility spectra of the Cyan and with anatase-to-rutile transformation. Black inks.

Phase Composition. The nano-inks consist of spinel-like cobalt ferrite (Black), gold (Magenta), anatase (Yellow) that will transform to Cr-Sb doped-rutile around 550 °C (Fig. 2). The phase composition of the Cyan pigment is not well characterized yet. Viscosity. The nano-inks have a Newtonian behaviour with viscosities ranging from about 40 to 200 mPa s at 25 °C (Table 1). These values are relatively high with respect to typical viscosities for ink-jet printing [4] due to the dispersing medium having a viscosity of about 24 mPa s. The Cyan, Magenta and Black inks have similarly low viscosities (around 40 mPa s) despite their different pigment loading, while the Yellow one has a quite high value (around 190 mPa s) hence out of the range of ink-jet applicability. The viscosity requirement is easily fulfilled by increasing the ink-jet printing temperature of the Yellow ink to approximately 70 °C. Surface tension. The equilibrium surface tension of a colloidal dispersion is not achieved instantaneously, because the nano-particles diffuse to the solid-liquid interface and a contamination by the environment occurs, causing a change with droplet ageing [13]. In order to take the most reliable surface tension data for ink-jet printing, where drop emission from printhead is very fast, the values are referred to the first instant after drop formation out of the needle (Table 1). All the tested samples fulfil the ink-jet requirement (i.e. 35-45 mN m-1) but the Yellow ink (~47 mN m-1 at 25 °C) which however reaches the desired value at the printing temperature (~45 mN m-1 at 65 °C). Zeta potential. These nanometric suspensions are stable for long time (i.e. at least several months) with no clue of pigment separation or settling. Electroacoustic measurements on the nanoinks give zeta potentials from 4 to over 100 mV (Table 1). The positive sign of zeta potential indicates a proton transfer between the glycol and the basic surface sites of pigment oxides. The absolute values of zeta potential depend also on the presence of other species that may come from the synthesis (e.g. water, surfactants). The Yellow and Black inks, and particularly the Magenta one, present high zeta potentials, implying a prevailing electrostatic mechanism of suspension stabilization. On the other hand, the analysis of the dynamic mobility spectra reveals an unexpected positive trend of the Cyan ink (Fig. 3) that should not be physically possible unless to hypothesize the presence of a thick elastic layer of polymer adsorbed onto the surface [14]. Such polymers, present in the reaction environment, give rise to a predominantly steric stabilization. Electrical conductivity. The electrical conductivity values are in the 1-100 µS cm-1 range, but for the Magenta one, whose high value is typical of a metal with a particle size over 10 nm [15].

Advances in Science and Technology Vol. 51

177

Table 2. Penetration kinetics of nano-inks into green porcelain stonewares. Parameter Unit Forming pressure MPa Open porosity % vol. Bulk density g cm-3 Mean pore size µm Penetration index adim. Cyan ink Magenta ink Yellow ink Black ink

10 33.5 1.71 0.39

Porcelain stoneware A 30 40 50 27.9 27.5 25.5 1.86 1.82 1.92 0.37 0.28 0.26

n.d. 0.32 0.23 0.31

n.d. 0.25 0.22 0.32

n.d. 0.23 0.20 0.25

0.50 0.19 0.13 0.25

10 33.9 1.76 0.43

Porcelain stoneware B 30 40 50 29.7 26.4 25.4 1.80 1.88 1.92 0.36 0.31 0.25

n.d. 0.42 0.28 0.26

n.d. n.d. 0.13 0.20

n.d. n.d. 0.10 0.11

0.82 n.d. 0.11 0.13

(n.d. = not determined)

Technological behaviour of inks. The penetration behaviour of the nano-inks into a ceramic greenware depends on both the physical properties of inks and the characteristics of substrates. The penetrability of the nano-inks was assessed on several substrates differing for composition, porosity and pore size distribution, controlled by different forming pressures, ranging from 10 to 50 MPa (Table 2). Although the kind of substrate slightly affects the penetration kinetics, the forming pressure does influence the degree of penetration. In fact, increasing the forming pressure, the pore size distribution shifts towards lower values, so the ink penetration through the porous media becomes more difficult because a higher pressure gradient is needed to overcome the surface tension. Ink penetration in ceramic substrates. The penetration kinetics of the four nano-inks is compared in Figure 4. Attention must be paid at the scale factor: the droplet volume used in this test (4 µL) is five orders of magnitude bigger than in the real ink-jet printing conditions (~80 pL). So, indices below 0.5 shall correspond to a full penetration in actual operating conditions. The penetration kinetics becomes slower along with decreasing pore amount and size, or increasing the forming pressure (Table 2). The Cyan ink is peculiar, being its penetration index much higher than those of the others. This behaviour could be connected with its different stabilization mechanism. In the hypothesis of a steric stabilization –such as that of the Cyan ink –it confirms that is beneficial for the ink penetration in green ceramic substrates. 30 Cyan

25

0.4 0.3

Contact angle (°)

penetration index (-)

0.5

Black

0.2

Yellow

0.1

20

Porc. stoneware Stoneware glaze Stoneware glaze Porc. stoneware Glazed wall tile

15 10 5

Magenta

time (s)

0

time (s) 0

0

1

2

3

Fig. 4. Penetration kinetics of nano-inks into a green porcelain stoneware (A, 50 MPa)

0

20

40

60

Fig. 5. Wettability of ceramic tile surfaces by the Black nano-ink.

178

Disclosing Materials at the Nanoscale

Table 3. Colour of nano-inks applied on glazes for stoneware tiles (firing temperature 1180 °C) and double-fired wall tiles (firing temperature 1120 °C). CIE-Lab coordinates: L* = white(+), black(-); a* = red(+), green (-); b* = yellow (+), blue (-). Nano-ink Cyan Magenta Yellow Black

Glaze for stoneware tiles L* a* b* mean s.d. mean s.d. mean 63.3 1.1 -3.3 0.1 -11.9 56.1 0.5 17.7 0.3 2.5 80.7 0.9 1.8 0.7 35.5 35.6 0.8 0.4 0.1 0.8

s.d. 0.7 0.1 0.8 0.1

Glaze for double-fired wall tiles L* a* b* mean s.d. mean s.d. mean 52.5 0.8 10.3 1.0 -35.1 37.0 1.1 43.4 0.2 4.4 79.9 2.1 1.8 1.2 47.2 14.2 0.9 3.1 0.1 -6.4

s.d. 2.0 1.1 1.8 1.0

Ink-jet printing. Preliminary testing of the nano-inks behaviour with laboratory-scale ink-jet printers was performed on different substrates, including glazes and bodies of industriallymanufactured ceramic tiles (e.g. double-fired and single-fired wall tiles, stoneware and porcelain stoneware floor tiles). The nano-inks spread uniformly on all these surfaces, wetting adequately both glazes and bodies (Fig. 5). The more difficult penetration is into the various substrates of unglazed porcelain stoneware, because of their low porosity and tiny pore size due to high forming pressures. The nanometric pigments resulted to be stable in a wide range of industrial tile firing temperatures (1000-1200 °C) giving rise to bright colours – comparable to those obtained with conventional ceramic pigments dispersed in the same ceramic matrices or even better –approaching quite satisfactorily the hues required by the quadrichromy printing process (Table 3). Conclusions Highly stable ceramic inks have been prepared by a modified polyol method, achieving pigments with actually nanometric sizes (i.e. 10-70 nm) based on metallic gold, spinel or anatase stoichiometries. These pigments develop vivid colours approaching the yellow, cyan, magenta and black required for the quadrichromy printing process. A wide-ranging characterization, including viscosity, surface tension, and zeta potential, proved that these inks do fulfil the requirements for the drop-on-demand ink-jet printing. Some nonoptimal behaviours have been corrected by increasing the ink temperature prior ejection. The physico-chemical properties of inks depend on both the composition of pigments and the occurrence of synthesis by-products (e.g. water). Ink penetration into the green ceramic substrate is affected by volume and size of pores, being promoted by lower forming pressures. Sterically-stabilized suspensions seem to penetrate faster than electrostatically-stabilized ones. Preliminary testing demonstrates that these nanosized inks are suitable for the four-colours inkjet printing in a wide range of ceramic applications, including low and high temperature firings (e.g. gloss fire of wall tiles at ~1000 °C and porcelain stoneware at ~1200°C, respectively). References [1]

H.R. Kang, J. Imaging Sci. Technol., 35 (1991) 179-188.

[2]

A. Atkinson, J. Doorbar, A. Hudd, D.L. Segal, P.J. White, J. Sol-Gel Tech., 8 (1997) 10931097.

[3]

H.P. Lee, J. Imaging Sci. Technol., 42 (1998) 49-62.

Advances in Science and Technology Vol. 51

179

[4]

P. Calvert, Chem. Mater., 13 (2001) 3299-3305.

[5]

G.J. Small, InterCeram, 44 (1995) 180-185.

[6]

R.A. Harvey, J.G. Sainza, Proc. Int. Conf. Digital Printing Technologies (2000) 516-518.

[7]

D. Gardini, F. Matteucci, M. Blosi, M. Dondi, A.L. Costa, C. Galassi, M. Raimondo, G. Baldi, E. Cinotti, Proc. QUALICER 2006, P.BC 397-408.

[8]

G. Baldi, M. Bitossi, A. Barzanti, Patent WO 03/076521 A1 (2003).

[9]

M. Fenandez-Garcia, A. Martinez-Arias, J.C. Hanson, J.A. Rodriguez, Chem. Rev., 104 (2004) 4063-4104.

[10] C.N.R. Rao, A. Müller, A.K. Cheetham, The Chemistry of Nanomaterials. Synthesis, Properties and Applications, Wiley-VCH, vols. 1-3 (2005). [11] C. Feldmann, H.-O. Jungk, Angew. Chem. Int. Ed., 40 (2001) 359-362. [12] L. Poul, S. Ammar, N. Jouini, F. Fiévet, F. Villain, J. Sol-Gel Sci. Tech., 26 (2003) 261-265. [13] L. Dong, D. Johnson, Langmuir, 19 (2003) 10205-10209. [14] M.L. Carasso, W.N. Rowlands, R.W. O’ Brien, J. Colloids Interface Sci., 193 (1997) 200-214. [15] C.N.R. Rao, G.U. Kulkarni, P.J. Thomas, P.P. Edwards, Chem. Eur. J., 8 (2002) 28-35.

Advances in Science and Technology Vol. 51 (2006) pp 181-186 © (2006) Trans Tech Publications, Switzerland doi:10.4028/www.scientific.net/AST.51.181

Evolution of Pt nanoclusters morphology on PEMFC electrode due to methanol oxidation reaction studied by electron microscopy and synchrotron grazing incidence x-ray diffraction M.Alvisi1, a, G.Galtieri1,b, L.Giorgi2,c, E.Serra2,d, T.Di Luccio1,e and R.Giorgi2,f 1

ENEA, Centro Ricerche Brindisi SS 7 Appia Km 706, 72100 Brindisi

2

ENEA, Centro Ricerche Casaccia,via Anguillarese, 00060 S.Maria di Galeria, Roma, Italy a

b

c

[email protected], [email protected], [email protected], [email protected], e [email protected], [email protected]

d

Keywords: fuel cell, catalyst, grazing synchrotron radiation, sputtering, platinum

Abstract. The proton exchange membrane fuel cells (PEMFC) have been developed mainly as a power source for vehicles, power generation and consumer electronics since they combine high energy conversion efficiency at relatively low temperatures without pollutants emission in the environment. An electrode for a PEMFC is a layered structure composed by a catalyst layer deposited on a porous carbon substrate. The substrate is usually covered by a diffusion layer that enhances the gas and water flow. Platinum nanoparticles supported by carbon microparticles are commonly employed as catalyst layer. In this work an extreme ultra-low loading of Pt catalyst (< 0.02 mg/cm2) has been deposited by magnetron sputtering on gas diffusion electrodes, with different carbon supports (Vulcan and SuperP), in order to enhance the activity of PEM fuel cells. The morphology (shape and grain size) and microstructure have been studied combining field emission scanning electron microscopy (FEG-SEM), grazing incidence synchrotron x-ray diffraction (GIXRD) and x-ray photoelectron spectroscopy (XPS). The results presented here concern the evolution of the cluster size and shape after the ageing, induced by cyclic voltammetry for methanol oxidation reaction. Introduction In fuel cell research much effort is devoted to increase the catalytic activity vs the load of the precious metal catalyst in the electrodes due to the day by day increase of metal bare costs. The chemical and electrochemical deposition methods are widely used to deposit platinum clusters on the gas diffusion electrodes (GDEs) and considerable progresses were achieved decreasing the platinum loading per unit area as shown in several articles [1 and ref. therein]. Recently also conventional physical vacuum deposition methods have been used to the same purpose [2,3] achieving ultra-low platinum loading and surprising elevated catalytic activity [4,5]. Earlier studies [6,7] have shown that PMFCs operating under constant potential operation for thousands of hours gradually lose surface area because of the growth of the cluster grains. Recent tests indicates that potential cycling may greatly accelerate the rate of surface area loss. Characterization of the crystallite particle size, crystal size distribution, amorphous material fraction and phase purity may be conveniently measured using x-ray scattering. Unfortunately nanometer sized electrocatalysts produce x-ray scattering with broad peaks often difficult to analyze because superimposed to high scattering signal from the used substrates. Moreover, the catalyst ultra-low loading gives very poor intensity signal. The use of synchrotron radiation is sometimes necessary to reveal appreciable signal from the low amount of the catalyst crystals thanks to its elevated brilliance and the versatility of the measurement conditions (e.g. different positions of the sample respect to the incident x-ray beam, tunable beam energy). Conventional x-ray diffraction has been successfully used down to platinum loading of 0.5 mg/cm-2. For lower loading only a work [8]

182

Disclosing Materials at the Nanoscale

reports data concerning platinum clusters with 0.03 mg/cm-2 platinum loading without showing the x-ray curves of low loading catalyzed electrodes. Following our previous work [9] on sputtered Pt catalyst deposition some questions have been opened. Which is the morphological and compositional difference between the Super P and Vulcan carbon gas diffusion electrodes? How do they influence the catalyst deposition and morphology? How is the compositional and morphological evolution of the clusters after methanol oxidation reactions? And how can these give rise to different electrochemical behavior? In this work we investigate the morphology and composition of two GDEs based on Super P and Vulcan carbon powders. We deposited sputtered platinum clusters (with an equivalent film thickness of 3 nm that gave us the better results in term of mass specific activity) in order to study the morphological and compositional modifications vs methanol oxidation reaction. Experimentals Gas diffusion electrode (GDE) preparation and catalyst deposition. The two layers (substrate + diffusion layer) which constitute the gas diffusion electrodes (GDEs) were prepared by a spraying procedure on carbon paper substrate (Toray TGPH-090). A homogeneous suspension composed of poly-tetrafluoroethylene (PTFE Teflon T-30, Du Pont) and carbon powder (Vulcan XC-72, Cabot Inc., USA) was sprayed onto the carbon paper to form the diffusion layer (thickness about 100 µm) of the electrode named CV. By using a Super P carbon (Timcal S.A., Belgium), instead of Vulcan carbon, and the same procedure we obtained the CS electrode. The composite structure was then dried in air at 120°C for 1 h, followed by thermal treatment at 280°C for 30 min to remove the dispersion agent contained in PTFE, and finally sintered in air at 350°C for 30 min. The amount of PTFE in the diffusion layer was 30 %wt. The Pt clusters (sample NF9) were deposited on GDE substrates (CS and CV) at room temperature by using a commercial RF magnetron sputtering system (SISTEC LS500). Before sputtering deposition, the Pt target (4” diameter, purity 99,95%) was sputtering cleaned in pure Ar. The Ar working pressure (2.8x10-1 Pa), the power supply (100W), the deposition rate were kept constant throughout these investigations. Electrochemical measurements. The electrochemical tests for methanol oxidation reaction (MOR) in CH3OH 0.5 M in H2SO4 1 M were carried out in a three-electrodes Pyrex cell cycling the samples 100 times. The working electrode was a disc inserted in a Teflon holder with an exposing area of 1 cm2, the counter electrode was a high purity (99.9999 %) platinum sheet placed in front of the working electrode, the reference electrode was a saturated Hg/Hg2Cl2 electrode (SCE). The instrument was a potentiostat-galvanostat EG&G PAR mod.273A controlled by a dedicated software CorrWare (Scribner Inc., USA). SEM-FEG, XPS and XRD measurements. The morphology of the electrode was examined by means of a Field Emission Gun SEM Leo mod.1530 equipped with a high-resolution secondary electron detector (in-lens detector). XPS measurements were performed with a V.G.ESCALAB MKII spectrometer. The excitation line was non-monocromatized MgK
, at 1253.4 eV. Survey spectra were collected at 50eV analyzer energy; detailed spectra at 20eV. The XRD experiments were performed at the beamline ID01 at the ESRF in Grenoble (France). The measurements consisted of grazing incidence x-ray diffraction (GIXRD) at grazing angle αi = αf = 0.5° by rotating the detector around the sample surface normal. The beam energy was fixed at 11.557 KeV (λ = 1.0728 Å). The diffracted intensity was recorded by a multichannel PSD detector with 2.5° opening.

Advances in Science and Technology Vol. 51

183

Results and discussions Gas diffusion electrode: The GDEs made with the two different carbon powders (Super P and Vulcan) 0.02 F1s

(upper curve) Vulcan (lower curve) Super P

NF9CV

FKLL

CKLL

C1s O1s

0

200

400

i / A cm-2

Intensity (a.u.)

0.01 NF9CS 0

-0.01

600

800

1000

Binding Energy (eV)

Fig.1a. XPS of CS and CV GDEs.

-0.02 -0.4

-0.2

0

0.2

0.4 0.6 E / V vs. SCE

0.8

1.0

1.2

1.4

Fig.1b Cyclic voltammetry in H2SO4 1 M of CV and CS GDEs.

present an equivalent microstructure (the x-ray spectra, not shown here, are equals) and an equivalent chemical composition visible in the XPS survey (fig.1a). Otherwise the surface morphology observed by using the scanning electron microscope presents some differences: a better carbon particles homogeneity is observed in CS electrode with respect to CV electrode where the dispersion of the particle dimensions is higher (as clearly visible in the figures 2a and 2b). The cyclic voltammetry curves of the two GDEs (fig. 1b) clearly show a different capacitive behaviour due to the different surface properties of the two substrates. Moreover, the NF9CV substrate shows a redox feature in the range 0.2-0.5 V, which can be assigned to the quinone/hydroquinone couple, indicating a different nature of the carbon powder. As deposited platinum clusters. The deposited platinum catalyst on the CV (NF9CV) and CS (NF9CS) GDEs are shown in the SEM images (figures 2a and 2b). The platinum clusters are the little (few nm) white spots covering the bigger (some tens of nm) carbon particles. The Pt clusters cover uniformly the carbon particles and have a medium size around 3 nm.

Fig.2a. SEM picture of the Pt clusters (NF9CS) on CS.

Fig.2b. SEM picture of the Pt clusters (NF9CV) on CV.

184

Disclosing Materials at the Nanoscale

The GIXRD measurements (fig. 3) refer to the as-deposited NF9CS and NF9CV samples. The Bragg reflections from metallic Pt are indicated, the rest of the spectrum is due to the GDE. The GIXRD confirms that the platinum was deposited in form of clusters of 2.5 - 3.0 nm (crystalline domain size) with cubic phase (space group Fm-3m, a = 3.92 Å [ref. 2001 JCPDS-ICDD) on CS and CV GDEs. Due to the ultra-low loading and cluster size the signal of platinum, although it is clearly visible, is not very intense also by using synchrotron radiation due to the extreme low size of the clusters. 0.020

Pt (111)

Intensity (a.u.)

NF9CS NF9CV

0.015

NF9CV

i / A cm-2

1

Pt (220)

0.010

0.005 NF9CS 30

40

50

60

2θ (deg)

0 0

Fig.3. GIXRD of the platinum clusters deposited on the CV and CS GDEs.

0.2

0.4 0.6 E / V vs. SCE

0.8

1.0

Fig.4. Cyclic voltammetry of CV and CS GDEs for methanol oxidation reaction (MOR).

In fig. 4 a zoom view of the voltammetry curves of the as-deposited NF9CS and NF9CV samples is shown. The methanol oxidation reaction (MOR) takes place in the range 0.55-0.80 V. The CS GDE presents two peaks (forward and back oxidation), while the CV GDE has only the forward oxidation. Besides, for the CV GDE it is still clear the quinone peak around 0.4 V. The presence of this specie probably inhibits the methanol back-oxidation. Platinum clusters after MOR. After 100 voltammetry cycles the NF9CS and NF9CV samples have been observed by SEM, XPS and XRD in order to quantify the change of clusters morphology, composition and microstructure. (upper curve) NF9CS as prepared (lower curve) NF9CS after MOR

Pt4d

O1s

OKLL F1s

Intensity (a.u.)

Intensity (a.u.)

Pt4f

Pt 4f peak CKLL

C1s

NF9/Super P as prepared NF9/Super P after MOR 0

200

400

600

Binding Energy (eV)

800

1000

70

75

80

85

90

Binding Energy

Fig. 5a. XPS survey of NF9CS before and Fig.5b. XPS platinum peaks of NF9CS after MOR. before and after MOR The XPS survey curves of the two samples are shown in fig. 5a revealing no significant changes in the composition of the GDEs after MOR. Only a close look on the characteristics Pt 4f peaks,

Advances in Science and Technology Vol. 51

185

(fig.5b) shown that after the MOR cycles the XPS Pt 4f doublet peaks become narrower and their separation is better defined. The peak width of the as-deposited platinum clusters indicates that a surface oxidation should be considered, whose reduction after MOR is revealed by the prevailing of the metallic character.

Fig. 6a. SEM of NF9CS after MOR

Fig.6b. SEM of NF9CV after MOR

The SEM pictures (fig. 6a and b) show a change of the clusters after the MOR with an increase of the cluster size and a different morphology, in fact the oxidation reaction of methanol on Pt induces a coalescence of the grains. This can be correlated to the results obtained from the grazing incidence x-ray diffraction spectra shown in figure 7. The x-ray confirmed that the Pt clusters on CV GDE experience a grain growth from around 2.5 nm to 5.5 nm in diameter (x-ray coherent domain size). The Pt clusters on CS GDE show similar behavior (from 3 to 7 nm). Due to the increase of the clusters size also the Pt (311) diffraction peak becomes visible. 0.06

= 2.5nm

0.02 0.00

1,5 -0.02

Intensity (a.u.)





Pt (220)

          

Pt (111)

0.04 Intensity (a.u.)

NF9CV NF9CVMOR

2,0

-0.04

= 5.5nm 40

42

44 46 2θ (deg)

48

50

1,0

Pt (220) 0,5





 






0,0

30

40 2θ (deg)

50

60




Pt (311)

















   

Fig. 7. GIXRD of NF9CV before and after MOR

Fig. 8. Calculated MSA of NF9CV and NF9CS for all cycles

The MOR activity was calculated from the charge QMOR measured by integrating the current density of the methanol oxidation anodic peak. The activity was calculated as mass specific activity (MSA) to take into account the platinum loadingLPt (mg cm-2): Q (1) MSA = MOR mC mg -1 LPt

(

)

186

Disclosing Materials at the Nanoscale

The electrocatalytic activity and its time stability (fig. 8) was evaluated by measuring it over 100 MOR cycles and was reported as mass specific activity (MSA). It is worth noting that the MSA of the CS GDE increases, while that of the NF9CV GDE decreases. The behavior of NF9CS vs MOR cycles correlated to the SEM, XPS and XRD characterization suggests that the increase in activity of the catalyst comes from the reduction of the adsorbed oxide products on the cluster surface; this phenomenon seems to prevail on the surface area reduction due to the doubling of the grain and domain size.

Summary Finally we can draw the following conclusions: a) the sputtering of platinum catalyst on GDE greatly enhances the activity of the catalyst on both CV and CS GDE, confirming our previous results; b) by using CS GDE we obtain an increase and a successive stability of the MOR activity that is normally expected to diminish as come out using a CV GDE; c) x-ray diffraction using synchrotron radiation allows us to study quantitatively the evolution of coherent domain size and microstructure of nanometer catalyst clusters also for extremely low metal loading (0.006 mg/cm2); d) the characterization of the Pt clusters before and after MOR shows that different carbon substrates induce different coalescence rate.

Acknowledgments The authors would like to thanks Mr. Domenico Dimaio for the technical assistance on the deposition apparatus and Dr. Dina Carbone and Dr. H.Metzger for assistance and very helpful discussions at ESRF Beamline. This work was financially supported in the framework of the National Project FISR “Fuel Cell”.

References [1] Liu, H., Song, C., Zhang, L., Zhang, J., Wang, H., Wilkinson, D.P. Journal of Power Sources 155 (2), (2006) pp. 95-110 [2] , A.F., Saha, M.S., Allen, R.J., Mukerjee, S., Electrochemical and Solid-State Letters 8 (10), (2005) pp. A504-A508 [3] , A.F., Saha, M.S., Allen, R.J., Mukerjee, S., Journal of the Electrochemical Society 153 (2), (2006) pp. A366-A371 [4] Alvisi M., Galtieri G., Giorgi L., Giorgi R., Serra E., Signore M.A., Surf. Coat. Tech. 200 (5-6), (2005) pp. 1325-1329 [5] Gruber, D., Ponath, N., ller, J., Lindstaedt, F., (2005) Journal of Power Sources 150 (1-2), pp. 67-72 [6] Wilson, Mahlon S., Garzon, Fernando H., Sickafus, Kurt E., Gottesfeld, Shimshon, Journal of the Electrochemical Society 140 (10), (1993) pp. 2872-2877 [7] Ascarelli, P., Contini, V., Giorgi, R., Journal of Applied Physics 91 (7), (2002) p. 4556 [8] Adora, S., Simon, J.P., Soldo-Olivier, Y., Faure, R.,  net, E., Durand, R., ChemPhysChem 5 (8), (2004) pp. 1178-1184

Advances in Science and Technology Vol. 51 (2006) pp 187-190 © (2006) Trans Tech Publications, Switzerland doi:10.4028/www.scientific.net/AST.51.187

Nanostructured Films of Polyphthalocyanines for Sensor Applications V. Anisimov, A. Borisov, O. Ivanova, S. Krutovertsev, A. Sherle1, E. Oleinik1 JSC “Practic-NC”, P/B 13, Zelenograd, Moscow, 124460, Russia 1

Institute of Chemical Physics of RAS,ul. Kosygina 4, Moscow, 117977, Russia [email protected]

Keywords: poly- oligophthalocyanines, film, sensor sensitivity

Abstract: The physicochemical properties and optical characteristics changes of novel synthesized polyphthalocyanines (PPc), containing Zn, Cu, Mn, Fe, Pb were investigated in different gas medium. The paper studies the characteristics changes of PPc nanostructured films, which were deposited on the interdigitated test structures and on the outside of U-shaped multimode cylindrical waveguide. Sensitivity of the films to different gases, such as NH3, NOx, H2S, O2 was investigated. Introduction The properties and applications of phthalocyanine (Pc) and its metall derivatives are studied extensively [1-3]. In the recent years special interest attracts to phthalocyanine films in regards with their application in gas sensors. This fact is due to the outstanding features distinguishing these substances from the multitude of other groups of macrocyclic complexes in that they are aromatic macrocycles with a unique conjugated π-system, which determines their chemical and thermal resistance, semiconductor properties and biological activity. It is known that the surface of phthalocyanine films can adsorb molecules of gases to change of physicochemical and optical properties of Pcs. The physicochemical properties (conductivity) and optical characteristics (absorption spectrum) of novel synthesized polyphthalocyanines (PPc), containing Zn, Cu, Mn, Fe, Pb were investigated. Experimental The conductivity of the films, their resistance-temperature relationship, sensor properties relative to gases as well as effect of the conditions of film formation on their characteristics were taken into account upon choice of method signal registration and construction developing of the PPc based sensors. The properties of the sensor were examined on sample gas mixtures using a dynamic blender “Environics-4000” (Environics, USA); the Dräger test ampoules (Dräger, Germany) were employed. All measurements were made at the same humidity and different temperature. The substances (fig.1) were synthesized by polycyclotetramerization of tetranitrile of pyromellitic acid in bulk and in polar solvents at 180-300oC for 5-30 hours in the presence of 0-5 mol% carbamide. The films of PPc were sputtered in vacuum or precipitated from the solutions. The films thermal sputtering were obtained in vacuum using UVN-71P3 plant at evaporation temperature of 300-1100оC and pressure in chamber 1,33x10-8 bar. The evaporator-to-base distance was 100 mm. To stabilize the film properties those were annealed in air at 100-250oC.

188

Disclosing Materials at the Nanoscale

M=H2, Zn, Fe, Cu, Co, Ni, Mn Fig.1 Typical structure of polyphthalocyanine (oligophthalocyanine - inside profile AA-BB).

1

2

3

Fig.2 Test structure for measurements: 1 - phthalocyanine film; 2-interdigitated electrodes; 3-heater. The films were deposited on tested structures and glass substrates. Test structures were formed on dielectric (sitall) substrate, with dimensions 15,0х5,0 mm. Two comb-like electrodes, made of nickel, were located on one side of the substrate; the opposite side contained nickel film heater. The active area with the comb-like electrodes and the heater on the opposite side was 4×8 mm in size (fig.2). The investigations were carried out under the sensor thermal stabilization conditions in the range 50 – 250оC. Sensor heating, conductivity and resistance measurements were maintained by two 16-channel measurement units and IBM PC program that operates in real time. Up to 32 sensors are able to be placed in the measurement chamber. The device has function to maintain a stable sensor temperature in the range 30-500oC with accuracy 1oC. The developed measurement unit allows to conduct simultaneously measurements of resistance of gas-sensitive layers on 32 channels in the range 500 Ohm – 1,5 GOhm with accuracy 2,5%. Besides that, the measurement unit regulates and controls sensor heating processes during the measurements. 11

1

10 3

12 4 7

2

8

6

5

9

Fig.3 The design of a flow-through optical sensor: 1 – the measuring block ; 2 – the reference photo receiver; 3 – the LED; 4 – the measuring photo receiver; 5 – the flow-through measurement cell; 6 – the film; 7 – the Ushaped cylindrical waveguide with film; 8, 9 – the input and output tubes; 10 – the beam-splitting system; 11 – the feedback channel; 12 – the channels of measuring; 13 – the current source.

Advances in Science and Technology Vol. 51

189

An optical sensor consists of sensitive element and measuring block (fig. 3). The physical principle of the sensor sensitive element action is based on the dependence of light loss from environment refraction index. The surface of sensitive element was covered with phthalocyanine films by aid of Langmuir – Blodgett method which allows obtaining very thin “molecular-smooth” films of a given thickness. The sensitive element is placed in a flow-through measuring cell (fig.1) and represents a cylindrical optical waveguide made of silicon. It is 2.5 mm in diameter and arched to 180 degrees. The arch radius is 2.5 mm (7). A light transmitter which divides a light beam according to ratio 1:1, a light source and a reference photo receiver (a photo diode) (2) are placed at the input end of the waveguide. A LED with the wavelength of 660 nm as the light source (3) was used for this experiment, because there is absorption maximum for phthalocyanine at this wavelength. The reference photo receiver is athwart to the axis of the light source. A measuring photo receiver is fixed at the output end of the waveguide. The measuring block contains a specially designed microprocessor unit which was calculated results of measurements. The signal magnitude of the sensor was determined by the equation: U=

U i − U i ,b U 0 − U 0,b

(1)

,

where U i ,U 0 - value of the signal in measuring channel and in reference channel correspondingly, U i ,b ,U 0,b – average value of the background signal in measuring channel and in reference channel correspondingly. The sensor relative signal was determined by the equation: Ur =

U − U air , U air

(2)

where U air - signal of the sensor in air. Results The typical characteristics of the PPc films at the influence of ammonia are shown on fig.4. The film resistance in the analyzed medium is seen to increase monotonically with the NH3 concentration. However the sensitivity range for FePPc is narrow and corresponds to gas micro concentrations. At the same time (fig.5) resistance of the PPc films decrease at the presence NO in air. This fact is interesting for realization of sensors based on PPc films as it permits to increase the selectivity of determination. 25000

800000

600000

1

17000

R, kOhm

R, kOhm

21000

13000 9000

200000

2

5000 0

20

40

60

80

400000

100

Concentration NH3 , ppm

3

0 0

2 55

110

1 220

440

Concentration NO, ppb

Fig.4 Typical NH3 concentration dependence of resistance for PPc films (100oC): 1-FePPc; 2- (Zn-Cu)PPc.

Fig.5 Typical NO concentration dependence of resistance for PPc films (100oC): 1-(Zn-Cu)PPc; 2- CoPPc; 3- FePPc.

190

Disclosing Materials at the Nanoscale

Change of electro-physical characteristics of CuPPc at the interaction with NO2 micro concentrations is accompanied by optical changes of substance. The typical behaviour of PPc films in the presence NO2 in the measuring air is shown on fig.6. The advantage of the optical method (fig.7) consists in possibility of measurement at room temperature as all electro-physical changes are implemented at increased temperature. Moreover films have high resistance and it makes difficulties for measurement.

Air

NO2 Air 1,89 1,885

10

NO2

sensor signal

R,MΩ

8

6

1,88 1,875 1,87

1 1,865

4

2 1,86

2

0

1000

2000

3000

time, s

0

6

12

18

24

30

36

Fig. 6 Resistance change of PPc at 1,9 ppb NO2 and 180oC: 1- zinc oligophthalocyanine; 2- copper polyphthalocyanine.

38

τ, min

Fig. 7 Optical signal of copper PPc at 1,9 ppb NO2 and room temperature.

The measurements were run for the various thicknesses of films and temperature. Optical sensor signal was determined on 400 and 600 nm. It was found, that PPc of Cu has high sensitivity to NO2 micro concentrations and H2S. The sensitivity increases with growth of Cu content in substance. It was shown, that PPc of Mn can be utilized as sensitive material to detect O2 concentration change. The investigated materials have satisfactory sensitivity to different gases: PPc of Fe and Pb can be used as sensitive layers in gas NO2 and H2S sensors. Besides, PPc of Fe demonstrated reversible signal change to NH3 in range 0-100ppm. It is important, that the character of resistive sensor signal change is determined by PPc conductivity type for all examined materials and corresponds with optical sensor signal in most cases. References [1] W. Hu, Y. Liu, Y. Xu, S. Liu, S. Zhou, P. Zeng, D.B. Zhu: Sensors and Actuators, B 56 (1999), p.228 [2] Kuo-Chuan Ho, Yi-Ham Tsou: Sensors and Actuators, B 77 (2001), p.253 [3] E. van Faassen, H. Kerp: Sensors and Actuators, B 88 (2003), p.329

Advances in Science and Technology Vol. 51 (2006) pp 191-196 © (2006) Trans Tech Publications, Switzerland doi:10.4028/www.scientific.net/AST.51.191

LGS and LGN microresonators: Applications to high temperature nanobalances T.G. LEBLOIS Institute FEMTO-ST, Dpt LCEP, 26 Chemin de l’Epitaphe, 25030 Besançon cedex FRANCE [email protected] Keywords: resonators, piezoelectric materials, high temperature

Abstract. New materials such as langasite and langanate crystals present some interesting perspectives for acoustic wave devices due to strong piezoelectric coefficients and due to the opportunity to work through a wide range of temperature. Theoretical formulations for temperature coefficients and electromechanical coupling coefficients of resonators are given in the case of thickness modes of vibration. Results are computed using two data sets of elastic, piezoelectric and dielectric constants. Orientations of BAW microresonators which give zero frequency shift versus temperature at 600°C are determined to serve as high temperature nanobalances. Introduction Langasite (LGS) and langanate (LGN) are new piezoelectric crystals which belong to the trigonal class 32. They constitute promising materials for applications in high temperature transducers because their melting points are located around 1400°C whereas the piezoelectric form of quartz crystal is limited to 573°C. An interesting application consists in the conception of microresonators with zero frequency shift versus temperature for mass measurements. To develop these new transducers, based on BAW vibrating plates operating in thickness modes, we have to investigate the orientations of cuts that offer high electromechanical coupling coefficients and very low temperature coefficients. These parameters are obtained using Lagrangian description in order to take account of the variation of size and orientation with temperature. Theoretical formulation. In order to evaluate the metrological performances of LGS and LGN resonators, the resonant structure is assimilated with an infinite plate perpendicular to the doubly rotated x’’2 axis, which constitutes a good approximation for the propagation of waves in the resonator. Consequently, the model of the BAW resonator vibrating in thickness modes becomes unidimensional and the equations only depend on x’’2 coordinates. Besides, the evolution of temperature is supposed to be slow and homogenous. To solve the equations of propagation and determinate the resonance frequency and the electromechanical coupling coefficient, we adopt the Lagrangian model [1]. In this model, we use the coordinates of a material point in the reference state that is to say before the evolution of temperature and we define effective elastic (G), piezoelectric (R) and dielectric (N) coefficients. The equations of piezoelectricity are given below:  G ' ' 2α 2β u ' 'β, 22 + R ' ' 2α 2 φ' ', 22 = ρ 0 &u&' ' α .   R ' ' 2α 2 u ' ' α,22 + N' ' 22 φ' ',22 = 0

(1)

These equations are established in the general case of doubly rotated cuts, which is indicated by (‘’). In these equations only effective constants vary with temperature. As an example, the elastic components of the tensor G are given in the vicinity of the reference temperature T0 by: n

GLmNo (T) = GLmNo (T0) [1+ ∑ TnGLmNo (T-T0)n]. n =1

(2)

192

Disclosing Materials at the Nanoscale

1 ∂ n G LmNo ( ) is the temperature coefficient of GLmNo of order n. Similar n! ∂T n expressions occur for N and R. In Eq.1, ρ0, uα and Φ are the mass density in the reference state, a component of the displacement and the electrical potential of waves respectively. The expressions of uα and Φ are given below: where TnGLmNo=

  x ,,   u α = u (αµ ) exp  jω  t − (2µ )    V    

  x ,,   φ α = φ (µ ) exp  jω  t − (2µ )   .  V    

(3)

The superscript (µ) indicates the mode of vibration (A, B or C), V(µ) the associate velocity of wave and ω the pulsation. Eq. 1 leads to the Christoffel system in terms of the Lagrangian effective coefficients: G ' ' 2α 2β u ' 'β, 22 +

R ' ' 2α 2 R ' ' 2 γ 2 u ' ' γ , 22 N' ' 22

= ρ 0 &u&' 'α .

(4)

The solutions V(µ) and u(µ) are respectively the eigenvalues and the eigenvectors of the Christoffel system. These solutions are thermally dependent by the elastic tensor G’’. As usual, the resonance frequency fR associated to each mode (µ) is deduced from the dispersion equation [2].

( 2 πf R )

=

2

n 2π 2 4 h 02

(V )

(µ) 2

(µ) 2  1 − 8 k  n 2π 2 

 .  

(5)

where 2h0 is the thickness of the endless plate and n the overtone number. The electromechanical coupling coefficient k(µ) is given by:

k

(µ)

=

R' '2α2 u (nµα)

(ρ V 0

(µ)2

N' '22

)

1

2

.

(6)

In this last expression, unα(µ) are the normalized eigenvectors components. Eq. 5 and Eq. 6 allow us to follow the evolution of resonant frequency and electromechanical coupling factor with temperature for any orientation of cuts. Results

The material constants of LGS and LGN used in this section are given by Malocha et al [4]. The values reported in the literature by several authors [5, 6] agree well enough to establish the correct order of magnitude for the properties of greatest significance to piezoelectric devices such as piezoelectric coupling coefficients and the existence of temperature compensated cuts. We compute results according to the theoretical formulation presented below. We present results on BAW resonators vibrating in thickness shear modes that are usually called b and c modes. C mode is the transverse mode which propagates with the lower velocity. A systematic study on the thermal sensitivity of the resonance frequency has been made in the case of doubly rotated cuts defined by the angles (ϕ0, θ0) for temperature of 300°C - 700°C. The angles of cuts are given by the IEEE standard on piezoelectricity [7]. The resonance frequencies are calculated in the case of the third overtone mode of vibration for a 150 µm thick resonator. Case of langasite. Let us start with the c mode of propagation. Three angles of cuts are selected to provide a temperature compensation at 600°C (Tinv = 600°C):

Advances in Science and Technology Vol. 51

193

(ϕ0, θ0)1=(0, 37°), (ϕ0, θ0)2=(40°, 39.3°) and (ϕ0, θ0)3=(60°, 37.5°). Fig. 1 displays plots of the relative frequency shift versus temperature, for several angles of cuts. Fig. 1a, 1b and 1c show the evolution of the frequency-temperature characteristics with θ when ϕ0=0, ϕ0=40° and ϕ0=60° respectively. Fig. 1d exhibits the changes of the frequency temperature curves with ϕ for θ0=40°. In order to plot the following curves, the reference temperature To is taken at 250°C and the resonance frequency at To is called fr0. 0,008

0,000

-0,001

a

0,006

b

0,004 0,002

(∆fr/fr0)

(∆fr/fr0)

-0,002

-0,003

0,000

-0,002

-0,004

-0,005

(0, 42°)c (0, 40°)c (0, 38°)c (0, 36°)c

-0,004

(40°, 42°)c

(40°, 40°)c

-0,006

(40°, 38°)c

(40°, 36°)c

(40°, 34°)c

(40°, 32°)c

-0,008

(40°, 30°)c -0,006 300

350

400

450

500

550

600

650

-0,010 300

700

350

400

450

0,002

c

550

600

650

700

d

0,000

(∆fr/fr0)

0,003

(∆fr/fr0)

600

0,001

0,004

-0,001

0,002

-0,002

0,001

(60°, 30°)c (60°, 32°)c (60°, 34°)c (60°, 36°)c (60°, 38°)c

0,000 -0,001 -0,002 -0,003 300

550

0,003

0,006 0,005

500

T(°C)

T(°C)

350

400

-0,003 -0,004 -0,005

450

500

550

600

650

-0,006 300

700

(0, 40°)c (4°, 40°)c (8°, 40°)c 350

(2°, 40°)c (6°, 40°)c (10°, 40°)c 400

450

500

650

700

T(°C)

T(°C)

Figure 1: Changes in frequency-temperature curves with θ ((a), (b) and (c)) and ϕ ((d)) for the third overtone c mode of LGS resonators. (a) case of ϕ0=0° (b) case of ϕ0=40° (c) case of ϕ0=60° (d) case of θ0=40° We observe that: (i) For ϕ0=0, the curves passe through minima whereas for ϕ0=40° and ϕ0=60° the characteristics exhibit maxima. Consequently, for ϕ0=0, the first temperature coefficient appearing in the polynomial function fr(T) is negative whereas it is positive for ϕ0=40° and ϕ0=60°. It is of interest to evaluate the deviation dfr of the relative frequency shift (∆fr/fr) with θ or (ii) ϕ at the given temperature T=600°C. According to Fig.1a, 1b, 1c and 1d, we obtain the results in Table 1. (ϕ0, θ0) dfr (ppm/°) dfr (ppm/’)

ϕ0=0 333 5.55

ϕ0=40° 950 12.5

ϕ0=60° 650 10.9

θ0=40° 450 7.5

Table 1: deviation of the relative variation of frequency at T=600°C The sensitivity of the resonance frequency with the angle θ is more accentuated for ϕ0=40°. If we perform the same calculation using Fig. 1d, we obtain dfr=450 ppm/°. After the fabrication process of resonators (cutting, lapping, polishing…), we can reasonably obtain plates with a precision of +/- 1’ on the angles (ϕ0, θ0). So, the changes in the relative variation of resonance frequencies do not exceed +/- 12.5 ppm. We have to keep in mind this value which corresponds to a deviation of +/- 200 Hz on a 16 MHz resonance frequency. Fig. 2 a and 2b illustrate the influence of the angles of cuts on the turnover temperature Tinv .

194

Disclosing Materials at the Nanoscale 800

610

ϕ=0° ϕ=40° ϕ=60°

700

600

Tinv (°C)

Tinv(°C)

600

θ=37° ϕ=0 θ=37.5° ϕ=60° θ=39.3° ϕ=40°

500

400

590

300

a

b

200

580 30

32

34

36

38

40

42

-8

-6

θ (°)

-4

-2

0

2

∆ϕ (°)

Figure 2: Changes in Tinv with θ (a) or with ϕ (b) for selected cuts. Carefull examination of these figures call for some remarks: (i) The evolution of Tinv with θ (Fig. 2a) in the range [ 30°, 42°] is more marked than the evolution of Tinv with ϕ (Fig. 2b) when ∆ϕ belongs to the range [-8°, 2°]. The lowest sensitivity dTinv/dθ occurs for the orientation (0, 37°): (ii) dTinv/dθ=17.5°C/° The previous remarks lead to retain the orientation (0, 37°). An other fundamental parameter for piezoelectric resonators is the electromechanical coupling factor k. Table 2 gives the values of k associated to the selected cuts for three different temperature values. (ϕ0, θ0) (0, 37°) (40°, 39.3°) (60°, 37.5°)

T=580°C 4.88313 6.170 10.3933

T=600°C 4.88324 6.161 10.3931

T=620°C 4.8832 6.152 10.3933

Table 2: Electromechanical coupling coefficients k (%) for selected cuts at three temperature This table reveals that: (i) the temperature does not influence significantly the coupling coefficient. (ii) According to the coupling factor the best orientation is (60°, 37.5°). Among the selected cuts the lowest value of k is obtain for (ϕ0, θ0)1=(0°, 37°). The conception of microresonators for microbalances applications passes through the determination of the mass sensitivity. So, we search the frequency shift fr-f’r induced by an increase of mass ∆m equal to 1 µg. The results on mass sensitivity Sm = (fr-f’r)/ ∆m.fr are given in Table 3. fr (kHz) f’r(kHz) fr-f’r (kHz) Sm (ppm/µg)

(0, 37°) 15971.36198 15971.26918 0.09280 5.81

(40°, 39.3°) 13057.01219 13056.93632 0.0759 5.81

(60°, 37.5°) 12166.20536 12166.13467 0.0707 5.81

Table 3: Changes in resonance frequencies and mass sensitivity for the three selected orientations As seen on table 3, all the orientations give the same mass sensitivity. It means that an increase of mass ∆m=1µg induces a shift of 75 Hz on the resonance frequency fr=13 MHz. These results must be compared with the values shown in table 1 and 2 and with results plotted on Fig.2. According to this comparison, no orientation gives simultaneously satisfaction to the four criteria: 1) slow evolution of Tinv with ϕ and θ 2) low deviation dfr of the relative variation of resonance frequency with ϕ and θ at 600°C 3) high electromechanical coupling coefficient k

Advances in Science and Technology Vol. 51

195

4) high mass sensitivity Sm. The analysis of results leads to some remarks: (i) Criteria 1), 2) and 4) are well satisfied for the orientation (0°, 37°) . But the weak coupling factor does not allow to use this cut for nanobalances. (ii) (ϕ0, θ0)2=(40°, 39.3°) is not a satisfactory orientation because the resonance frequency is too much sensitive with θ. The mass sensitivity is twice as low as the orientation sensitivity. (iii) Orientation (60°, 37.5°) is worth retaining because of its high piezoelectric performances. The orientation sensitivity is slightly greater than the mass sensitivity. Moreover, Tinv varies rapidly with ϕ and θ. As a conclusion, according to the above analysis, for nanobalances applications we retain the orientation (ϕ0, θ0)2=(60°, 37.5°) provided that the angles were very precisely known. In the same way, let us deal with the b mode of vibration. Determination of orientations for which an extremum in the frequency temperature plot occurs at 600°C gives no result. All the curves present minima but Tinv always remains below 400°C. Moreover, the coupling coefficient never exceeds 5.2% at 600°C. For these reasons no orientation of resonator vibrating on b mode are retained for nanobalances applications. Case of langanate. In this second part, we propose to follow the same method as for LGS material to determine temperature compensated cuts at 600°C. Fig. 3 displays for c mode the frequency temperature changes with θ and ϕ in the case of three selected orientations whose characteristics exhibit an extremum at 600°C: (ϕ0, θ0)1=(10°, 0°), (ϕ0, θ0)2=(19.5°, 8°) and (ϕ0, θ0)3=(50°, 0°). 0,04

0,07

a

0,05

0,01

0,04

0

(10, 0)c (10, 2)c (10, 4)c (10, 6)c (10, -2)c (10, -4)c

-0,01 -0,02 -0,03

b

0,06

0,02

(∆fr/fr0)

(∆fr/fr0)

0,03

0,03 0,02 0,01 0

-0,04

(50, 4)c

(50, 6)c

(50, 8)c

(50, 10)c

(50, 12)c

-0,01 300

350

400

450

500

550

600

650

700

300

350

400

450

T(°C)

500

550

600

650

700

650

700

T(°C) 0,07

0,07

c

0,06

0,06

0,05

0,05

0,04

0,04

(∆fr/fr0)

(∆fr/fr0)

(50, 2)c

0,03 0,02 0,01

d

0,03 0,02 0,01

0 -0,01

(10, 0)c

(8, 0)c

(12, 0)c

(14, 0)c

(16, 0)c

(18, 0)c

-0,02 300

350

400

450

500

T (°C)

550

600

(15,5, 8)c (19,5, 8)c (23,5, 8)c

0 -0,01 650

700

-0,02 300

350

400

450

500

550

(17,5, 8)c (21,5, 8)c (25,5, 8)c 600

T (°C)

Figure 3: Changes in frequency-temperature curves with θ ((a) and (b)) and ϕ ((c) and (d)) for the third overtone c mode of LGN resonators. Cases of ϕ0=10°(a) and ϕ0=50°(b). Cases of θ0=0°(c) and θ0=8° (d) As for LGS, no resonator presents a turnover temperature at 600°C on b mode. If we compare Fig.3 with Fig. 1 we can note that the variation of the frequency with temperature is sensibly more marked in the case of LGT material. Fig. 4 shows the evolution of Tinv with the angles of cuts. If we compare Fig. 2a and Fig. 4a, we observe that the evolution of Tinv with ∆θ is more slight in the case of LGT.

196

Disclosing Materials at the Nanoscale

680

640

θ=0° ϕ=10° θ=8° ϕ=19,5° θ=0° ϕ=50°

ϕ=10° θ=0° ϕ=50° θ=0°

640

620

600

Tinv (°C)

Tinv (°C)

θ=8° ϕ=19,5°

a

560

600

b

580 520

480

560 -4

-2

0

2

4

6

8

-4

-2

0

2

4

6

8

∆θ(°) ∆ϕ(°) Figure 4: Changes in Tinv with ∆θ (a) or with ∆ϕ (b) for selected cuts.

The metrological performances of resonators are given in table 4 in order to select good orientation. The orientation sensitivity given in the table is the maximum value between θ and ϕ sensitivity. (ϕ0, θ0) dfr (ppm/’) kc (%) Sm (ppm/µg)

(10°,0°) 90 14.34 5.532

(19.5°, 8°) 86.9 8.14 5.492

(50°, 0°) 137.5 14.34 5.532

Table 4: Metrological performances of resonators at 600°C All the orientations give a similar mass sensitivity. The higher electromechanical coupling factor is obtained for (ϕ0, θ0)=(10°,0°) or (50°,0°). We find the same value for kc and Sm when (ϕ0, θ0)=(10°,0°) and (50°,0°); this is due to the symmetry of LGN crystal. Differences appear for the sensitivity of fr with θ because we are then concerned with doubly rotated cuts. According to the results in the previous table, we choice the orientation (ϕ0, θ0)=(10°,0°). But, once again, we have to take care of the orientation sensitivity which is 16 times higher than the mass sensitivity. Conclusion

Frequency temperature characteristics, electromechanical coupling coefficients and mass sensitivities are computed using Malocha et al sets of constants. Two temperature compensated cuts at 600°C are extracted and offer good metrological performances for nanobalances applications. Indeed the changes in frequency reach 75 and 72 Hz for a 13 MHz resonator LGS and LGT respectively when we load the balance with 1 µg. LGS nanobalance seems rather more performant because it is less sensible at a disorientation of the substrate. References

[1] B. Dulmet, R. Bourquin, E. Bigler and S. Ballandras: J.A.S.A., vol 110(4) (2001), p1800-1807 [2] E. Dieulesaint and D. Royer: Ondes élastiques dans les solides, Masson and Cie Eds. Paris, France (1991) [3] R. Bechmann, A.D. Ballato and T.J. Lkaszek: Proc. IRE vol 50 (1962), pp 1812-1822 [4] D.C. Malocha, M. Pereira Da Cunha, E. Adler, R.C. Smythe, S. Fredrick, M.Chou, R. Helbold and Y.S. Zhou: Proc. IEEE IFCS (2000), pp 200-205 [5] Y.S. Pisarevsky, P.A. Senushenkov, P.A. Popov and B.V. Mill: Proc. IEEE IFCS (1995), pp 653-656 [6] R.C. Smythe and G.E. Hague: Proc IFCS (2000), pp 191-194 [7] Standards on Piezoelectricity, Convention IEEE, IER 37 (1978), p15

Advances in Science and Technology Vol. 51 (2006) pp 197-200 © (2006) Trans Tech Publications, Switzerland doi:10.4028/www.scientific.net/AST.51.197

Humidity Sensors Based on Nanostructured Materials S. Krutovertsev, A. Tarasova, L. Krutovertseva , A. Zorin, O. Ivanova JSC “Practic-NC”, P/B 13, Zelenograd, Moscow, 124460, Russia [email protected] Keywords: micro- relative humidity, SiOx film, polyimide film

Abstract: Constraction and characteristics of sorption polymeric relative humidity sensors, sorption SiOx microhumidity sensors, and condensation type (dew point) sensors are compared. The characteristics of the sensors are examined in dew point range from -80 to +20oC. An integrated multifunctional humidity sensor, for measurements in wide humidity range at different conditions is developed. The sensors are intended for use in various branches of industry, and in scientific researches. Introduction In modern engineering many problems are encountered, when it is necessary to measure the humidity of a gas in a wide range and under various conditions. Currents for this purpose capacitive and resistive humidity sensors are widely used. However the majority of existing sensors are intended for specialized measurement in two ranges - microhumidity (usually this level is less than 1000 ppm) or relative humidity under certain conditions (temperature, pressure, volume and so on). Usually for humidity measurements sorption capacitive type sensors based on inorganic sorbents (Al2O3 or SiO2) are used [1, 2]. For relative humidity measurement capacitors based on organic sorbents (polyimide, polysulfone, etc.) are widely used [3, 4]. For specific measurements (for example, measuring in a small volumes such as an integrated circuit package or in technological gases) condensation type sensors showing a decrease in resistance with the amount of condensed moisture can be used [5]. In this article, the characteristics of these sensors (of the indicated types) and their comparison are considered. The design and characteristics of an integrated sensor combining all the above enumerated types is described. Experimental During the realization of this work two base variants of a humidity sensor design were used (fig. 1). In all cases the sizes of the substrate made was 4x8 mm. The substrate was made from silica or alumina. Nickel with a sublayer of vanadium as electrode material is used in all cases. The thickness of the electrodes is between 0,2 - 0,3 mm. An interdigital structure was made with the width of finger pairs of electrodes and distance between them 20 - 100 micron (for various variants).The sensor' s structure is shown in fig. 1a, 1b represents a cross-sechon of the device. The advantage of such a design, in comparison with the traditional approach (i.e. "sandwich type") is the absence of an external contact with the very thin top gold electrode. In microhumidity sensors, SiOx can be used as the sensitive layers. The thickness of the SiOx layer was made 0,2 - 0,3 microns. A film of SiOx is produced by a method of hydrolytic polycondensation from solutions based on tetraethoxysilane (sol-gel method) [6]. The initial solution consists of: tetraethoxysilane, water and ethanol (as a solvent) in respective proportion, and catalysts of a hydrolyse process. A sensitive film is put on the surface of the substrate by centrifugal coating. The films were subsequently subjected to a two-step heat treatment at the temperature up to 140oC. A layer of gold with thickness of 0,01 - 0,015 micron was sputtered on the sorbent surface after disclosing the contact platforms by methods of photolitography and chemical etching. At such thickness of gold water molecules can easily pass through the layer . After dividing the plate into chips, external leads well attached to the contact platforms.

198

Disclosing Materials at the Nanoscale

c)

a)

4

b)

3

5 4 2

2

3

1

Fig. 1. The base sensors designs: a) capacitive type humidity sensor and its cross section (b); c) interdigital resistive type sensor. 1 - substrate; 2, 3 - electrodes; 4 - sensitive layer; 5 - gold. In sensors of relative humidity polyimide can be used as the sorbent. A special technology of heat treatment and stabilization was developed and used. The thickness of the polyimide film was between 0,3 -1,2 microns. The design and technological process of the relative humidity sensor' s is the same as described above for the microhumidity sensor. The sensors'response to relative humidity was measured by a two pressure humidity generator. Gas microhumidity measureents a reference generator "Polus-3,"which enabled humidity to be set in range from - 90 up to + 20 oC dew point was used. The results of the polyimide film sensors'characteristics are shown in fig. 3. From the drawing in follows, that the sensor' s characteristic is practically linear in the relative humidity range from 10 up to 98 % . It can also be seen, that the threshold of sensitivity for these sensors is 0,5% of relative humidity or -- 40oC on dew point. The temperature factor of the sensor characteristic does not exceed 0,2%/ oC.The sensor is stable in the temperature range from -100 up to 250oC. The developed technology of multistep temperature treatment and stabilisation of the polyimide film properties has allowed a high long-term stability of the sensors'characteristics including the ability to withstand humidity at long-term (80 - 98 %) . The hysteresis does not exceed ±1 % of relative humidity. The calibration characteristics of the SiOx film based sensor in the dew point range from -80 up to + 20oC are shown in fig. 2. Researches have shown, that sensor has a high sensitivity over the investigated range of humidity and stable characteristics. In the "high" humidity area (from + 10 up to + 20oC dew point) the hysteresis does not exceed ±2 %. However, rather high dependence on temperature is observed. Therefore at measurements in the microhumidity range of it is necessary to take into account the gas temperature, or to stabilise the temperature of the sensor surface . One of the most widely used methods of humidity measurement is the condensation method (or method of dew point). In this method the temperature is covered untile condensation occurs on the surface. The condensation can be detected in various ways, including a change of conductivity or capacitance [5]. The cooling of a sensor' s surface is made with the help of thermoelectrical cooler. The temperature is measured by thermoresistors or thermocouples. For realization of a sensor based on condensation method a design has been developed as shown in fig. 1c. As the thermosensitive element a nickel thermoresistor films is used. The sensors'characteristics with interdigital nickel electrodes without a cover and with a SiOx film, including the edditive of various doped hygrophilic additives (LiCl, P2O5) was investigated. It was found, that for sensors covered by doped SiOx film,

Advances in Science and Technology Vol. 51

199

5 .5 5 4 .5 4 3 .5 3 2 .5 2 1.5 1 0 .5 - 10 0

220

Capacitance, !F

Capacitance, nF

the change both in conductivity and capacitance was 3 - 8 times grater, than for the uncovered sensor. 2 10 200 19 0 18 0 17 0 16 0 15 0 14 0

-50

0

0

50

H u m i d i ty , d e w p o i n t

10 2 0 3 0 4 0

50 6 0

70 8 0 9 0

R e l a ti v e H u m i d i ty , %

Fig2. Humidity dependence of SiOx sensor capacitance at 20 °C

Fig3. Relative humidity dependence of polyimide sensor capacitance at 20 °C

The comparison of the results of simultaneous measurements by the three types of sensors in the dew point range from -80 up to +20 oC has showed a good agrement of results for the condensation type sensor and sorption capacitor sensor in dew point range from -70 up to +15oC, and also for the sorption capacitor microhumidity sensor and relative humidity sensor in the dew point range from -30 up to +15oC . However it is necessary to note, that the best measurements accuracy is obrawed for sorption type sensors. In the range from 5 up to 99 % of relative humidity this is a capacitor sensor based on a polyimide film, and in dew point range from - 20oC (4 % of relative humidity at 20oC) up to -80oC this is a capacitor sensor with sensitive layer of undoped SiOx. At the same time simultaneous use of two independent methods of measurements (with the help of condensation and sorption types sensors) at the points where their measurements allows a highly reliable reference for the measurements. On the basis of the conducted research an integrated multifunctional wide range sensor for humidity, combining in it`s design all the considered variant sensors, was developed. (fig.4).

2

1

4

Fig. 4. Integrated multifunctional humidity sensor. 1 - SiOx capacitive sensor 2 - polyimide capacitive sensor 3 - condensation type interdigital sensor 4 - nickel thermoresistor

3

The size of the sensor is 4,0x4,0 mm. The sensor is formed on a dielectric substrate ceramic, alumina).The sensor electrodes are made from nickel, material having a good temperature resistance coefficient (4,0-4,8 . 10-2 1/oC- for film material). The capacitance values of sorption sensors of microhumidity and relative humidity are 0,5 - 0,6 nF and 20 -30 pF respectivity. A comparative simplicity of processing the sensors'signals is thus provided. The sensor is installed in a standard microcircuits package. The conducted researche of these sensor' s characteristics has shown, that they correspond to above those shown for the individial structures. The sensor,s measurement range is from -80 to +20 oC dew point. The developed sensors are intended for use in various branches of industry and in scientific investigations.

20 15 31.3.06 1.4.06

2.4.06

3.4.06

4.4.06

25

100

20

o

120

Temperature, C

25

Humidity, %

Humidity,%

30

o

29 28 27 26 25 24 23 22 21 20

35

C

Disclosing Materials at the Nanoscale

Temperature,

200

80

15

60 10

40

5

20 0 31.3.06

0 1.4.06

2.4.06

3.4.06

4.4.06

Time, day

Time, day

(a)

(b)

Fig.5 Relative humidity dependence of device based on polyimide sensor of air temperature: (a) in the room; (b) outside. For example, it is possible to control in time the humidity changes in different processes by device based on developed polyimide sensors (fig.5).

References [1] R.K. Nahar, V.K. Khanna: Sensors and Actuators, B 46 (1998), p.35 [2] Weon-Pil Tai, Jae-Hee Oh: Sensors and Actuators, B 85 (2002), p.154 [3] J.M. Ingram, M. Greb, J.A. Nicolson, A.W. Fountain: Sensors and Actuators, B 96 (2003), p.283 [4] V. Ducéré, A. Bernès, C. Lacabanne: Sensors and Actuators, B 106 (2005), p.331 [5] K.Stephenson: Solid State Technology, (1980), p.34 [6] C.J.Brinker, G.W.Sherer, Sol-Gel Science (Academic Press, San Diego 1990), pp. 839.

Advances in Science and Technology Vol. 51 (2006) pp 201-208 © (2006) Trans Tech Publications, Switzerland doi:10.4028/www.scientific.net/AST.51.201

Quantum Dots in GaInP/GaInAs/GaAs for Infrared Sensing H. Lim, S. Tsao, M. Taguchi, W. Zhang, A. A. Quivy, M. Razeghia Center for Quantum Devices,b Department of Electrical Engineering and Computer Science, Northwestern University, Evanston, IL 60208, USA. a

[email protected], bhttp://cqd.ece.northwestern.edu

Keywords: self assembly, Stranski-Krastanov, quantum dot, infrared, detector, MOCVD, focal plane array

Abstract. Nanotechnology is occurring simultaneously in almost every field with strong interdisciplinary applications which have unique and important characteristics for potential novel and high performance devices. Quantum dots grown by epitaxial self-assembly via StranskiKrastanov growth mode have many favorable properties for infrared sensing. Because of their very small size and three-dimensional confinement, the electronic energy levels are quantized and discrete. These quantum effects lead to a unique property, “phonon bottleneck”, which might enable the high operating temperature of infrared sensing which usually requires cryogenic cooling. Here we report a focal plane array (FPA) based on an epitaxial self-assembled quantum dot infrared detector (QDIP). The device structure containing self-assembled In0.68Ga0.32As quantum dots with a density around 3×1010 cm-2 was grown by low-pressure metalorganic chemical vapor deposition (LP-MOCVD). Using different structures, we successfully developed QDIPs with a peak photoresponse around 5 m and 9 m. High peak detectivities were achieved at 77 K from both QDIPs. By stacking both device structures, we demonstrated a two-color QDIP whose peak detection wavelength could be tuned from 5 m to 9 m by changing the bias. 256×256 detector arrays based on 5 m and 9 m-QDIPs were fabricated with standard photolithography, dry etching and hybridization to a read-out integrated circuit (ROIC). We demonstrated thermal imaging from our FPAs based on QDIPs.

1. Introduction Infrared sensing has been extensively studied for many purposes such as military, medical, and commercial applications.1 Quantum-well infrared photodetectors (QWIPs) have achieved good performance in the middle-wavelength infrared regime but suffer from drawbacks 2 such as the requirement of optical coupling schemes for the absorption of normally incident radiation and the need for cryogenic cooling for operation. Quantum-dot infrared photodetector (QDIP) technology has improved in recent years and emerges as a strong competitor 3 , 4 , 5 , 6 due to their inherent absorption of normally incident light, potential room-temperature operation and higher responsivity than QWIPs. These unique features are a direct consequence of the three-dimensional confinement potential achieved in quantum dots (QDs) that provides a discrete density of states. Most of the QDIPs reported in the literature are based on the InAs/GaAs system and were grown by molecular beam epitaxy (MBE).5,6 In the present work, we report on the growth and testing of an InGaAs/InGaP QDIP grown on a GaAs (001) substrate using low-pressure metal-organic chemicalvapor deposition (LP-MOCVD). We also demonstrated QDIP-based focal plane arrays which resulted in thermal imaging.

202

Disclosing Materials at the Nanoscale

2. Growth and characterizations of Quantum dots

Room temperature

0.4

PL intensity (arb. un.)

PL intensity (arb. un.)

For the QD growth, we use triethylgallium, trimethylindium and pure arsine and phosphine as precursors, as well as dilute silane (200 ppm) to provide the n-type doping of the QDs. For the first kind of QDs, an InGaAs layer was deposited at 440°C on an In0.49Ga0.51P layer which was lattice matched to GaAs. In order to determine the composition of the QDs, a thick layer of InGaAs material was grown on a GaAs substrate and its lattice mismatch to the GaAs substrate was estimated by high-resolution x-ray diffraction. From that measurement, the composition of the QDs in this study was found to be In 0.68Ga0.32As. The QD growth time was 4 seconds with a growth rate of 0.70 monolayer per second (ML/s) while the QD ripening time was 30 seconds under arsine flow. The V/III ratio for the InGaAs quantum dots was 480. The QDs were formed by selfassembly using the Stranski-Krastanov growth mode. They had a lens shape and their density was around 3×1010cm-2, as observed by atomic force microscopy in Figure 1 (left). The typical size of a QD is 20~30 nm in diameter and 3~4 nm in height. Dilute silane with a flow rate of 74 sccm was used during InGaAs deposition in order to n-type dope the QDs. The QD layer was covered with InGaP which was lattice matched to GaAs at 440°C in order to measure the photoluminescence (PL). The room temperature PL shows an interband transition around 1230 nm as seen in Figure 1 (center). We used this kind of QDs for the active region of our MWIR (Mid-Wave Infra Red)-QDIP whose peak intersubband transition was around 5 m (~250 meV).

InGaP

InGaP

0.2

0.0

1.0

Room temperature InGaP

0.8 GaAs

0.6

InGaP

0.4 0.2 0.0

900

1000

1100

1200

1300

Wavelength (nm)

1400

1000

1100

1200

1300

1400

Wavelength (nm)

Figure 1. (left) 1µm×1µm atomic force microscopy and scanning electron microscopy (SEM, inset) pictures of uncapped QDs. (center) Room-temperature PL of InGaAs QDs buried in InGaP. (right) Room-temperature PL of InGaAs QDs inserted in the center of a 5.4nm-wide GaAs/InGaP quantum well. We also grew InGaAs QDs in the center of a 5.4nm-wide GaAs/InGaP quantum well (as shown in the inset of Figure 1 (right)) in order to check the effect of the thin GaAs layers on the QD electronic structure. The density and size of the QDs were very similar to the ones of the QDs grown directly on InGaP. The room-temperature PL spectrum revealed two peaks around 1244 nm and 1170 nm. The main peak was caused by the interband transition inside the QDs, which was between the ground state of the conduction and valence bands. On the other hand, the second peak resulted from the interband transition from the first excited state of the conduction bands to the ground state of the valence band, as was confirmed by power-dependent PL measurements. From the difference between these two peaks, the electronic energy spacing between the ground state and first excited state in the quantum dots can be estimated to be 63 meV. This InGaAs QD layer in a GaAs/InGaP quantum well was used as the active region of the LWIR (long-wavelength infrared)QDIP whose peak intersubband transition was 9 m.

Advances in Science and Technology Vol. 51

203

3. MWIR-QDIP structure and characterizations The structure of the MWIR-QDIP considered here belongs to a class of devices3,7 studied at Northwestern University. The structure was grown on a semi-insulating GaAs(001) substrate and started with a 0.5µm-thick GaAs:Si (n=5×1017cm-3) bottom contact layer deposited at 480°C followed by a 100nm-thick In0.49Ga0.51P layer lattice matched to the GaAs bottom contact layer. The active region was grown at 440°C and consisted of ten In0.68Ga0.32As QD layers separated by 35nmthick In0.49Ga0.51P barriers. Finally, an InGaP barrier layer followed by a top GaAs:Si contact layer identical to the bottom ones were deposited to complete the structure as shown in Figure 2 (left). To test the QDIP’s performance, 400×400 µm2 mesas were fabricated using reactive ion etching to etch through the top contact and active region down to the bottom GaAs:Si layer. AuGe/Ni/Au bottom and top metal contacts were made via lift-off technique and alloyed at 400 ºC for 3 minutes. The sample was then mounted on a copper heatsink and attached to the cold finger of a liquid nitrogen cooled cryostat capable of varying the sample temperature between 77 K and room temperature. The spectral response of the QDIP was tested on a Fourier transform infrared spectrometer. The peak responsivity was determined using a 800 ºC blackbody source modulated at 400 Hz. A 2-12 m optical band-pass filter plus a ZnSe cryostat window suppressed near-infrared radiation from the blackbody. The noise current was measured with a fast Fourier transform spectrum analyzer and a low-noise current amplifier. The dark current was extracted using a semiconductor parameter analyzer. Figure 2 shows the photoresponse and the peak responsivity of the MWIR QDIP at two different temperatures. The signal has a maximum at 4.7 m and a cutoff at 5.5 m. The spectral width (Δ / peak) was 18 %, indicating that the intersubband transition probably occurred between two bound states in the QDs. The peak wavelength showed negligible change with increasing temperature, but the cutoff wavelength decreased and the peak narrowed to produce a spectral width of 7.8 % at 120 K. This narrowing can be attributed to either inhomogeneous doping or size distribution of the QDs in each of the 10 periods of InGaAs QD layers (due to growth fluctuations, indium segregation, inhomogeneous strain generated by the QDs). Therefore, with increasing temperature, carrier relaxation will dominate carrier escape in certain QDs, and those QDs will no longer contribute to the photoresponse spectrum.8 4

10

Photoresponse (arb. un.)

Peak responsivity (mA/W)

77K 120K

1.0 0.8 0.6 0.4 0.2

77K 100K 120K

3

10

2

10

1

10

0

10

-1

10

0.0 -2

3

4

5

6

7

Wavelength (µm)

8

9

10

10

-3

-2

-1

0

1

2

3

Bias voltage (V)

Figure 2. (left) MWIR-QDIP structure (center) Spectral photoresponse of the MWIR QDIP at 77 K and 120 K showing a peak at 4.7µm. The inset shows the MWIR-QDIP structure. (right) Peak responsivity of the MWIR-QDIP as a function of bias voltage at 77 K, 100 K, and 120 K.

The peak responsivity of the device versus bias voltage at different temperatures is shown in Figure 2 (right). At low bias, the signal varies by several orders of magnitude, which is consistent with a bound-to-bound transition and with the extraction of the photoexcited carriers out of the QDs by

204

Disclosing Materials at the Nanoscale

voltage-assisted tunneling.8 At 77 K, a maximum responsivity of 5.5 A/W was measured around –2 V, while the magnitude of the signal was still larger than 2 A/W at 120 K for the same bias voltage. The noise current was extremely low, as illustrated in Figure 3. At 77 K, a noise floor around 3.5×10-14 A was measured in the –0.9 V to 2.5 V range and was determined by the instrumental limitation of the setup. The level of the noise floor remained constant but narrowed with increasing temperature. The specific peak detectivity was calculated from the noise current (in) and peak responsivity (Rp) measurements using the expression. R p AΔ f (1) D *p = in

where A is the detector area and Δf (=1Hz) the noise bandwidth. The results are shown in figure 3 and reveal that the specific peak detectivity has a maximum value of 1.1×1012 cmHz1/2W-1 at a bias of –0.9 V. It is worth noting that, in this device, the maximum D*p value always corresponds to the onset of the noise floor (flat region) because it will be at that point that the ratio of the peak responsivity over the noise current is maximized. The quantum efficiency η could be obtained using the relation η=

hcR p

(2)

qgλ

-8

1x10

-9

1x10

-10

10

-11

10

-12

10

-13

10

-14

-3

77K 100K 120K

-2

-1

0

1

Bias voltage (V)

2

1/2

1x10

Specific peak detectivity (cm Hz / W)

Noise current (A)

where h is Plank’s constant, c is the speed of light, q is the elementary charge of the electron, Rp is the peak responsivity, g is the photoconductive gain, and is the peak detection wavelength. Its value was determined to be 1.0% at 77K and decreased to 0.2% at 120K. The quantum efficiency of a QDIP is known to be low, but it is generally compensated by a very high photoconductive gain that reached a value of 200 at the peak detection bias. Such low quantum efficiency is generally attributed to the small fill factor (low QD density) and to the low oscillator strength of the principal intersubband transition involved in the optical process. However, it can be considerably improved by incorporating the QDIP structure into a resonant cavity. Preliminary calculations show that 30% quantum efficiency can be expected in a good cavity design. By comparing temperature dependent dark-current measurements3 with the photocurrent generated by a 300K background, the background-limited performance (BLIP) temperature of the QDIP was found to be 200K under a field of view of 45°.

3

10

13

10

12

10

11

10

10

10

9

10

8

10

7

-3

77K 100K 120K

-2

-1

0

1

2

3

Bias voltage (V)

Figure 3. (left) Noise current as a function of bias voltage at 77 K, 100 K, and 120 K. At low bias, the signal is limited by the experimental setup around 3.5×10-14 A. (right) Specific peak detectivity of the QDIP as a function of bias voltage at 77 K, 100 K, and 120 K.

Advances in Science and Technology Vol. 51

205

4. LWIR-QDIP and two-color QDIP structures and characterization 0.5 µm n-GaAs top contact n=5x1017cm-3

x10

32.3 31 nm 32.3 nmGaInP GaInPbarrier barrier 100 nm GaInP spacer 0.5 µm n-GaAs bottom contact n=5x1017cm-3

Relative photoresponse (a.u.)

100 nm GaInP spacer 2.7 nmGaAs

T=77 K Bias=-1.0 V

2.0

1.5

1.0

0.5

0.0 3

6

9

12

15

18

21

24

Wavelength (µm)

SI GaAs substrate

Figure 4. (left) LWIR-QDIP structure and (right) its spectral photoresponse at 77 K, showing a peak at 9 m. The LWIR-QDIP structure was also grown on a semi-insulating GaAs (001) substrate by LP-MOCVD and its schematic is shown in Figure 4. The active region consisted of ten In0.68Ga0.32As QD layers sandwiched by 2.7 nm-thick GaAs layers and surrounded by 32.3 nm-thick In0.49Ga0.51P barriers. 3.6 monolayers of In0.68Ga0.32As self-assembled via the Stranski-Krastanov growth mode and formed lens-shaped InGaAs QDs with a density around 2.8×1010 cm-2. The photoresponse, measured using Fourier transform infrared (FTIR) spectroscopy, had a 9 µm peak as shown in Figure 4 (right). A peak responsivity of 1.76 A/W was obtained at 77 K and a bias of -3 V with a calibrated 800 °C blackbody source (Figure 5 (left)). A high peak detectivity of 5×1010 cmHz1/2/W was achieved at 77 K and a bias of -0.7 V (Figure 5 (right)). The quantum efficiency of the QDIP was estimated to be 3 % assuming the noise gain was the same as the photoconductive gain. 1E11

077K 100K

1/2

1000

Detectivity (mAHz /W)

Responsivity (mA/W)

10000

077K 100K

100

10

-4

-2

0

Bias (V)

2

4

1E10

1E9

1E8 -6

-4

-2

0

2

4

Bias (V)

Figure 5. (left) Peak responsivity of the LWIR QDIP as a function of bias voltage at 77 K and 100 K; (right) Peak detectivity of the QDIP as a function of bias voltage at 77 K and 100 K. In order to have a two-color QDIP operating around 5 and 9µm, we grew a structure where the two different QD layers discussed previously were stacked as shown in Figure 6 (left). The photoresponse of that new device was measured as a function of bias voltage. Two main peaks were observed, as expected, around 4.8 and 9 m. The relative intensity of the photoresponse depended upon the applied bias. Below -1.4 V, the photoresponse of the 9 m peak was stronger than that of the 5 m peak, whereas above that value the photoresponse from the 5 m signal was stronger.

206

Disclosing Materials at the Nanoscale

3.5

9 µm

35 nm GaInP barrier

x5 32.3 31 nm GaInP barrier 100 nm GaInP spacer 0.5 µm n -GaAs bottom contact n=5x10 17 cm -3

5 µm structure 9 µm structure

Relative photoresponse (a.u.)

100 nm GaInP spacer

x5

2.0

6.6 µm

T=77 K 5 Gain=10

1.2

-0.8 V -1.0 V -1.2 V -1.4 V

2.5

1.5

1.0

0.5

1.0

-1.5 V -1.6 V -1.8 V -2.0 V

0.8

0.6

0.4

0.2

0.0

0.0 3

SI GaAs substrate

T=77 K 6 Gain=10

4.8 µm 3.0

Relative photoresponse (a.u.)

0.5 µm n -GaAs top contact n=5x10 17 cm -3

4

5

6

7

8

9

10

11

12

13

14

15

3

Wavelength (µm)

4

5

6

7

8

9

10

11

12

13

14

15

Wavelength (µm)

Figure 6. (left) Two-color QDIP structure. (center and right) Spectral photoresponse of the twocolor QDIP at 77 K, showing the evolution of the peak intensity as a function of the applied bias voltage.

6. Focal plane arrays based on QDIPs

Figure 7. (left) SEM image of the pixel array with AuGe/Ni/Au ohmic contacts (right) SEM image of the indium bumps on individual pixels After the full characterization of our QDIPs using large mesas, 256×256 detector arrays were fabricated from the same wafer. The pitch and pixel size of the array were 30 µm and 25 µm respectively, which gave a fill factor of about 70%. To achieve higher uniformity, array pixels were fabricated with standard photolithography and electron cyclotron resonance reactive ion etching (ECR-RIE) technique. BCl3 gas was used for the etching. The dry etching was stopped in the middle of the GaAs bottom contact layer. Due to the inherent normal incident light absorption of quantum dots, no light coupling structure was required for our QDIP FPAs. After dry etching, top and bottom metal contacts (AuGe/Ni/Au = 700Å/350Å/1300Å) were fabricated with lift-off via “image reversal” lithography technique. 9 Indium bumps are usually used to interconnect each detector pixel to each readout circuit unit cell for signal readout in infrared FPA fabrication. In this work, indium bumps were made on the array pixel via lift-off, which was based on a special positive lithography process. In that process, the necessary undercut profile was formed by combining a develop-enhanced bottom resist layer with a develop-inhibited top resist layer. The details of this process were described elsewhere.9 Figure 7 shows a SEM picture of the QDIP array with ohmic contacts (left) and indium bumps (right) on top of each pixel. After indium bump evaporation, the QDIP FPA wafer was diced. A single QDIP FPA die was chosen and hybridized to a 256×256 Litton CMOS ROIC via flip chip bonding technique. The gap between the QDIP FPA and the Si ROIC was backfilled with a DP0110 underfill, 10 which provides the necessary mechanical strength to the FPA hybrid prior to substrate thinning or removal. Substrate thinning can significantly enhance the reliability of a hybrid FPA and eliminate pixel-to-pixel crosstalk.

Advances in Science and Technology Vol. 51

207

Finally, the FPA hybrid was bonded to a 68 pin leadless ceramic chip carrier (LCCC) by vacuum grease for subsequent testing. The FPA was tested on a CamIRa™ infrared FPA evaluation system made by SE-IR Corp. The FPA hybrid was mounted on the cold finger in the dewar which was cooled down to 70 K by pumping on liquid nitrogen. A 3-5 µm Ge lens (f/2.3) was attached with the dewar lid in which a 312 µm broadband Ge window was installed. In the following test, the background temperature was 300 K, the area of each detector pixel was (25µm)2 and the frame rate was 23.5 Hz. Our LWIRInGaAs/GaAs/InGaP QDIP FPA gave very good imaging with 99 % of the pixels operational as shown in Figure 8 (left and center). The most important parameter for an infrared FPA is the noise equivalent temperature difference (NEΔT). The NEΔT can be estimated as: 11 i kTλ NEΔT = n ⋅ ( ) ⋅T (3) I P hc where is the peak response wavelength and T is the blackbody temperature. With the peak photoresponse at 9 µm, the mean NEΔT was 168 mK at an operating temperature of 77K and bias of -2 V. A NEΔT histogram of the pixels in the 256×256 array is shown in Figure 8 (right).

Figure 8. (Left and center) images of a graduate student taken by the LWIR-QDIP FPA at 68 K. The veins of the hand were clearly visible in the image. (Right) NEΔT histogram of the pixels in the 256×256 array 7. Discussion

QDIPs are supposed to exhibit very good performance at high temperature, as predicted theoretically. Although the detectivity of the present MWIR-QDIP is still 8.3×1010cmHz1/2/W at 120K, as the temperature goes up further the detectivity decreases rapidly due to the increase of the dark current (and consequently of the noise) and to the decrease of the photocurrent (responsivity). The reason why the photocurrent decreases with increasing temperature is the following one. Due to the relatively large volume and flat lens-like shape of the quantum dots obtained by self assembly, there are many electronic energy levels in the conduction band and therefore their spacing can be very small. If the energy spacing is not so different form an integer number times the LO phonon energy in the quantum dots, the relaxation rate to lower energy states can be dramatically increased by multiphonon emission as the temperature goes up. This relaxation rate competes against the escape rate of the photoexcited electrons that generate the photocurrent. Above a certain temperature, the fast relaxation process starts to dominate the escape process and leads to the dramatic decrease of the photocurrent. In order to achieve the high-temperature operation predicted for QDIPs, the dark current should be suppressed at high temperature or (and) the photocurrent should be much less sensitive to the temperature of the detector. A good approach to reduce the dark current at high temperature is to use a resonant-tunneling filter around the quantum-dot layers.12

208

Disclosing Materials at the Nanoscale

8. Summary

In summary, we reported MWIR(4.7 µm) and LWIR (9 µm) InGaAs/GaAs/InGaP quantumdot infrared photodetector grown by LP-MOCVD. We also demonstrated a two-color QDIP whose peak detection wavelengths could be tuned from 4.8 m to 9 m by changing the bias. 256×256 detector arrays based on MWIR and LWIR-QDIPs were fabricated with standard photolithography, dry etching and hybridization to a read-out integrated circuit. We demonstrated good thermal imaging from our FPAs based on QDIPs.

References 1

A. Rogalski, Progr. Quant. Electron. 27, 59 (2003).

2

K.K. Choi, The Physics of Quantum Well Infrared Photodetectors (World Scientific, New Jersey, 1997).

3

J. Jiang, S. Tsao, T. O’Sullivan, W. Zhang, H. Lim, T. Sills, K. Mi, M. Razeghi, G.J. Brown, and M.Z.Tidrow, Appl.

Phys. Lett. 84, 2166 (2004). 4

W. Zhang, H. Lim, M. Taguchi, S. Tsao, B. Movaghar, M. Razeghi, Appl. Phys. Lett. 86 191103 (2005).

5

Eui-Tae Kim, Anupam Madhukar, Zhengmao Ye and Joe C. Campbell, Appl. Phys. Lett. 84, 3277 (2004).

6

S. Chakrabarti, A.D. Stiff-Roberts, X.H. Su, P. Bhattacharya, G. Ariyawansa, A.G.U. Perera, J. Phys. D 38, 2135

(2005). 7

S. Kim, H. Mohseni, M. Erdtmann, E. Mitchel, C. Jelen, and M. Razeghi, Appl. Phys. Lett. 73, 963 (1998).

8

M. Razeghi, H. Lim, S. Tsao, J. Szafraniec, W. Zhang, K. Mi, and B. Movaghar, Nanotechnology 16, 219 (2005).

9

J.Jiang, S.Tsao, T.O’Sullivan, M.Razeghi, G.Grown, accepted for publication in Infrared Physics & Technology, 2003.

10

11

12

J.H.Lau, Flip Chip Technologies., New York: McGraw-Hill, 1996 A.Zussman, B.F.Levine, J.M.Kuo, and J.de Jong, J.Appl.Phys. 70, 5101(1991). P. Bhattacharya, X. H. Su, S. Chakrabarti, G. Ariyawansa and A. G. U. Perera, Appl. Phys. Lett. 86, 191106 (2005).

Keywords Index A Aluminum (Al) Atomic Force Microscope (AFM)

Fullerene 64 90

B Biocomposite

30

C Carbon Nanoclusters Carbon Nanotube (CN) Catalyst Cathode Cathodoluminescence Ceramic Inks Chemical Vapor Deposition (CVD) Cluster Assembling Coefficient of Restitution Composite Conducting Polymer Contact Mechanics Cu(hfac)(tmvs) Cupraselect Cyclohexane Conformers

81 68, 145 181 54 38, 48 174 167 81 127 75, 145 145 90 167 167 140

D Dendrimers Density Function Theory (DFT) Detector Disproportionation Reaction Doping Dynamic

105 167 201 167 68 14

E Electro-Deposition Electronic Structure

7 81

F Film Film Electrode Finite Element (FE) Simulation Focal Plane Array Fuel Cell (FC)

187 68 134 201 181

75

G Ga-Doped GaAs Glass Transition Temperature Gold Frameworks Grazing Synchrotron Radiation

48 121 75 30 181

H High-k Dielectrics High-Temperature

156 191

I In Situ Ultrahigh Vacuum Transmission Electron Microscope Infrared Ink-Jet Printing Interferometric Lithography

14 201 174 115

L Lithium Ion Battery

20

M Manganese Dioxide Manganite Many-Scale Simulation Melt-Spinning Mesoscale Mesostructured Silica Metal Filling Micro-Electromechanical System (MEMS) Micro Machining Micro-Relative Humidity MOCVD Molecular Dynamic (MD) Molecular Printboards Multi-Walled Carbon Nanotube (MWCNT)

145 54 134 60 134 30 68 121 121 197 201 127, 140 105 64

N Nanocluster Deposition Nanocrystal

127 156

210 Nanofiber Nanomagnetism Nanomaterial Nanoparticle Nanorod Nanostructure Nanowall Nanowire NiSi Nanowires Non-Volatile Memory (NVM)

Disclosing Materials at the Nanoscale 60 7 99 105, 174 38 54 48 1, 7 42 156

P PC/ABS Piezoelectric Material Platinum Poly-Oligophthalocyanines Polyimide Film Polymer Blend Post-industrial Rejects Protein

134 191 181 187 197 60 134 30

201

R Reinforcement Resonator RIE Roughness

64 191 115 90

S Self-Assembly Sensor Sensitivity Si1-xGex/Si Heterostructures SIESTA Silicide Silicon Silicon Carbide (SiC) Simulation SiOx Film Soft Lithography Solar Cell Solid Oxide Fuel Cell (SOFC) Sputtering Step Bunching Stranski-Krastanov Surface Patterning Surface Texturing

99

T Ta Template Method Templating Thermal Annealing TiO2 Titanium (Ti) Titanium Nanoclusters TriA-PI Tungsten Oxide

167 68 7 1 20, 99 167 81 75 1

W W Water Dispersible

167 68

X X-Ray Powder Diffraction

99

Z

Q Quantum Dot (QD)

Synchrotron Radiation (XRD)

42, 201 187 42 167 14 156 60 121 197 105 20, 115 54 181 42 201 105 115

Zeolite Zinc Oxide (ZnO) ZnS

140 38 48

Authors Index A Alkaisi , M.M. Alvisi, M. Anisimov, V. Awasthi, A.

115 181 187 127

B Baldi, G. Barzanti, A. Benassayag, G. Bittar, A. Blaikie, R.J. Blosi, M. Bonafos, C. Borisov, A. Braida, B. Burns, C.T. Buzio, R.

99, 174 99, 174 156 115 115 174 156 187 167 30 90

C Cafarelli, P. Chen, L.J. Cheng, H. Cheong, F.C. Chiu, W.L. Cinotti, E. Cobian, M. Correa, Z. Cosoli, P. Costa, A.L. Crisci, A.J. Cruciani, G. Curiale, J.

140 14, 38, 42, 48 167 1 115 174 167 60 134 174 30 99 54

D Dalconi, M.C. Darques, M. De Stefanis, A. Di Luccio, T. Dondi, M.

99 7 140 181 99, 174

F Fanciulli, M.

156

Fermeglia, M. Ferrone, M. Firestone, M.A. Frackowiak, E.

134 134 30 145

G Galassi, C. Galtieri, G. Gao, L. Gardini, D. Garg, D. Gavioli, L. Giorgi, L. Giorgi, R.

174 181 64 174 167 81 181 181

H He, J.H. Hendy, S.C. Hsin, C.L. Hsu, H.C. Huskens, J. Huve, G.

38 127 14 14, 48 105 121

I Ishida, Y. Iton, L.E. Ivanova, O.

75 30 187, 197

K Kaczmarski, M. Kavan, L. Krutovertsev, S. Krutovertseva, L. Kumaravelu, G. Kyotani, T.

167 20 187, 197 197 115 68

L Laible, P.D. Leblois, T.G. Levy, P. Leyva, A.G. Liao, K.F. Lim, C.T. Lim, H.

30 121, 191 54 54 38 1 201

212 Lorenzi, G. Lu, M.Y.

Disclosing Materials at the Nanoscale 99 48

M Machado, E. Martinelli, L. Matteucci, F. Meneghini, C. Michotte, S. Mouti, A. Murata, H.

167 134 99, 174 99 7 156 60

N Norman, J.

167

75 187 167 68 60

P Paneni, M.S. Perego, M. Perez, G. Piraux, L. Posocco, P. Pricl, S.

134 156 140 7 134 134

Q Quivy, A.A.

201

R Razeghi, M. Reeves, R.J. Rosenbusch, M.

201 115 54

S Sánchez, R.D. Sancrotti, M. Schamm, S. Semprini, E. Serra, E. Sherle, A. Shinoda, Y. Shiota, I.

134 1 140 64

T Taguchi, M. Tarasova, A. Tellier, C.R. Tenmoku, K. Tomizawa, T. Tomlinson, A.A.G. Troiani, H.E. Tsao, S.

201 197 121 60 60 140 54 201

V

O Ogasawara, T. Oleinik, E. Ordejon, P. Orikasa, H. Oya, A.

Sinesi, S. Sow, C.H. Stefani, F. Sun, J.

54 81 156 140 181 187 75 75

Vajdova, V. Valbusa, U. Varghese, B. Venica, R.

30 90 1 134

W Wang, C.W. Wu, W.W.

38 14, 42

Y Yokota, R.

75

Z Zhang, W. Zhu, Y.W. Zorin, A.

201 1 197