Nanotechnology and NanoInterface Controlled Electronic Devices
M. Iwamoto K. Kaneto S. Mashiko
Elsevier
Nanotechnology and Nano-Interface Controlled Electronic Devices
Nanotechnology and Nano-Interface Controlled Electronic Devices
Edited by
M. Iwamoto K. Kaneto S. Mashiko
2003 ELSEVIER Amsterdam – Boston – London – New York – Oxford – Paris San Diego – San Francisco – Singapore – Sydney – Tokyo
ELSEVIER SCIENCE B.V. Sara Burgerhartstraat 25 P.O. Box 211, 1000 AE Amsterdam, The Netherlands © 2003 Elsevier Science B.V. All rights reserved. This work is protected under copyright by Elsevier Science, and the following terms and conditions apply to its use: Photocopying Single photocopies of single chapters may be made for personal use as allowed by national copyright laws. Permission of the Publisher and payment of a fee is required for all other photocopying, including multiple or systematic copying, copying for advertising or promotional purposes, resale, and all forms of document delivery. Special rates are available for educational institutions that wish to make photocopies for non-profit educational classroom use. Permissions may be sought directly from Elsevier’s Science & Technology Rights Department in Oxford, UK: phone: (+44) 1865 843830, fax: (+44) 1865 853333, e-mail:
[email protected]. You may also complete your request on-line via the Elsevier Science homepage (http://www.elsevier.com), by selecting ‘Customer Support’ and then ‘Obtaining Permissions’. In the USA, users may clear permissions and make payments through the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA; phone: (+1) 978 7508400, fax: (+1) 978 7504744, and in the UK through the Copyright Licensing Agency Rapid Clearance Service (CLARCS), 90 Tottenham Court Road, London W1P 0LP, UK; phone: (+44) 207 631 5555, fax: (+44) 207 631 5500. Other countries may have a local reprographic rights agency for payments. Derivative Works Tables of contents may be reproduced for internal circulation, but permission of Elsevier Science is required for resale or distribution of such material. Permission of the Publisher is required for all other derivative works, including compilations and translations. Electronic Storage or Usage Permission of the Publisher is required to store or use electronically any material contained in this work, including any chapter or part of a chapter. Except as outlined above, no part of this work may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without prior written permission of the Publisher. Address permissions requests to: Elsevier’s Science & Technology Rights Department, at the phone, fax and e-mail addresses noted above. Notice 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. Because of rapid advances in the medical sciences, in particular, independent verification of diagnoses and drugs dosages should be made.
First edition 2003 Library of Congress Cataloging in Publication Data A catalog record from the Library of Congress has been applied for. British Library Cataloguing in Publication Data A catalogue record from the British Library has been applied for. ISBN: 0-444-51091-5 ∞ The paper used in this publication meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper). Printed in The Netherlands.
Contents Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
ix
List of Contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
xi
A. Single Molecular Electronics and Photonics 1.
2.
3.
4.
5.
Nanostructure fabrication using electron and ion beams Shinji Matsui (Himeji Institute of Technology) . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3
Information storage using a scanning probe Kiyoshi Takimoto (Canon Co. Ltd.) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
21
Single electron tunneling organic devices Tohru Kubota, Shiyoshi Yokoyama, Tatsuo Nakahama, Shinro Mashiko (Communications Research Laboratory), Yutaka Noguchi, and Mitsumasa Iwamoto (Tokyo Institute of Technology) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
31
Spatial light confinement and laser emission from a gain medium containing dendrimer Shiyoshi Yokoyama and Shinro Mashiko (Communications Research Laboratory) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
41
Control of molecular selective-assembling on metal surface Takashi Yokoyama (National Institute for Materials Science), Toshiya Kamikado, Shiyoshi Yokoyama, Yoshishige Okuno, and Shinro Mashiko (Communications Research Laboratory) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
49
B. NICE Devices 6.
7.
8.
Polymer optoelectronics – towards nanometer dimensions Olle Inganäs and Fengling Zhang (Linköping University) . . . . . . . . . . . . . . . . . . .
65
Control of charge transfer and interface structures in nano-structured dyesensitized solar cells Shozo Yanagida, Takayuki Kitamura, and Yuji Wada (Osaka University) . . . .
83
Materials and devices for ultrafast molecular photonics Toshihiko Nagamura (Shizuoka University) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
vi
Contents
9.
Carrier transport behavior in OLED Tatsuo Mori and Teruyoshi Mizutani (Nagoya University) . . . . . . . . . . . . . . . . . . 133
10.
Electrical characterization of organic semiconductor films by in situ fieldeffect measurements Kazuhiro Kudo (Chiba University) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
C. Smart Soft Materials 11.
Introducing ruber into the Langmuir–Blodgett technique H. Xu, R. Heger, F. Mallwitz, M. Blankenhagel, C. Peyratout, and Werner A. Goedel (University of Ulm and Max-Planck-Institut für Kolloid- & Grenzflächenforschung) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
12.
Design of functional interface between living systems and semiconductor nano-structures Motomu Tanaka (Technische Universität München) . . . . . . . . . . . . . . . . . . . . . . . . 191
13.
Structural color forming system composed of polypeptide-based LB films Takatoshi Kinoshita (Nagoya Institute of Technology), Shujiro Hayashi, Yoshiyuki Yokogawa (National Institute of Advanced Industrial Science and Technology), and Shintaro Washizu (Fuji Photo Film Co. Ltd.). . . . . . . . . . . . . . 233
14.
Generation of a strong dipole layer and its function by using helical peptide molecular assemblies Shunsaku Kimura, Tomoyuki Morita, and Kazuya Kitagawa (Kyoto University) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253
D. Interfacial Dynamic Technology 15.
Guided mode studies of liquid crystal layers Fuzi Yang (Tsinghua University) and J.R. Sambles (University of Exeter) . . . 271
16.
Explanation of the static and dynamic director orientation in thin nematic liquid crystal films using deuterium NMR spectroscopy Akihiko Sugimura (Osaka Sangyo University) and Geoffrey R. Luckhurst (University of Southampton) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313
17.
MDC–SHG spectroscopy of organic monolayer film Atsushi Tojima, Ryouhei Hiyoshi, Takaaki Manaka, Mitsumasa Iwamoto (Tokyo Institute of Technology), and Ou-Yang Zhongcan (Academia Sinica) 351
18.
Light-driven dynamic controls in nano-hybrid materials Takahiro Seki (Tokyo Institute of Technology) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 377
Contents
vii
E. Fabrication and Characterization Technology 19.
Solvent-induced morphology in nano-structures Bin Cheng, Hongtao Cui, Brian R. Stoner, and Edward T. Samulski (University of North Carolina at Chapel Hill) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 399
20.
Polarons in conjugated polymer and its composite with fullerene Kazuhiro Marumoto and Shin-ichi Kuroda (Nagoya University) . . . . . . . . . . . . . 411
21.
Characterization of semiconductor surfaces with noncontact atomic force microscopy Seizo Morita and Yasuhiro Sugawara (Osaka University) . . . . . . . . . . . . . . . . . . . 429
22.
Transport and photocarrier generation in poly(3-alkylthiophene) and metal junctions Keiichi Kaneto, Koichi Rikitake, and Wataru Takashima (Kyushu Institute of Technology) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 455
23.
Thermochromic behavior in novel conducting polymers at the solid–liquid phase transition Mitsuyoshi Onoda and Kazuya Tada (Himeji Institute of Technology) . . . . . . . 479
Subject Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 509
Preface The International Workshop on Nanotechnology and NICE Devices (IWNND) was held on March 19 and 20, 2002 at Nagoya Congress Center, Nagoya, Japan. This international workshop was organized as one of the important events of the Nagoya Nanotechnology International Forum 2002 (NNIF2002), from March 18 through 20, 2002. Needless to say, Nano-Technology and Materials, Bio-Technology, Information Technology and others are recognized as the important key technologies in the 21st century, and these techniques are believed to make a significant contribution to our daily and fruitful life. The purpose of the IWNND workshop was to provide opportunities to discuss the present status and trends of Nanotechnology and Organic Electronics. Paying attention to the role of nano-interfaces, the experiments and ideas to create NICE (Nano-Interface Controlled Electronic) devices have been presented by distinguished scientists from overseas and Japan, working in universities, laboratories and companies. All the participants could benefit from the presentations and discussions. This book is a collection of papers based on the invited talks at this workshop. As you know, many organic materials that are interesting in terms of electronics have been synthesized and discovered during the past few decades. We can see one of the most remarkable achievements in the Nobel Prize for Chemistry 2000 awarded to Heeger, MacDiarmid and Shirakawa, for their contribution to the discovery and development of conducting polymers. In the hope of observing novel and useful electrical and optical properties, many investigations have been carried out to build up organic devices. Plastic solar cells, flexible-type field effect transistors (FETs), electroluminescent (EL) devices and so on have been developed, along with the development of new organic materials. Insightful ideas have also been proposed to open-up new methods in electronics. However these are no longer sufficient. One must develop techniques to catch the specific properties of organic materials, molecules, biological materials and so on, and then to create a novel method to benefit from the specific functions of these materials in electronic devices. One way would be to use nano-interfaces and related nanometric interfacial phenomena, although our understanding of nano-interfacial phenomena is far way from the viewpoint of science and technology. In the IWNND workshop, five topics, i.e., Single Molecular Electronics and Photonics, NICE Devices, Smart Soft Materials, Interfacial Dynamic Technology, and Fabrication and Characterization Technology, are selected in association with nanointerfacial phenomena and their electronic applications. This book covers these five topics. It will be very much appreciated to hear comments, critiques, and suggestions from any readers for the benefit of Nano-Interface Controlled Electronic devices. Finally, we would like to express our sincere thanks to the organizing members of the IWNND, the NNIF2002, the staffs of Nano-Device Group, National Institute for
x
Preface
Materials Science (NIMS), and the staffs of Nanotechnology Section, Kansai Advanced Research Center (KARC), Communications Research Laboratory (CRL) for their kind help and enthusiastic effort. Thanks are also given to the City of Nagoya, NIMS and KARC for their support. M ITSUMASA I WAMOTO K EIICHI K ANETO S HIRO M ASHIKO Editors
List of Contributors M. Blankenhagel Max-Planck-Institut für Kolloid- & Grenzflächenforschung, Berlin, Germany Bin Cheng Department of Chemistry, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599-3290, USA Hongtao Cui Department of Physics and Astronomy, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599-3255, USA Werner A. Goedel Organic and Macromolecular Chemistry, OC3, University of Ulm, Albert-EinsteinAllee 11, D-89069 Ulm, Germany Shujiro Hayashi National Institute of Advanced Industrial Science and Technology, Hirate-cho 1-1, Kita-ku, Nagoya 462-8510, Japan R. Heger Max-Planck-Institut für Kolloid- & Grenzflächenforschung, Berlin, Germany Ryouhei Hiyoshi Department of Physical Electronics, Tokyo Institute of Technology, 2-12-1 O-okayama, Meguro-ku, Tokyo 152-8552, Japan Olle Inganäs Biomolecular and organic electronics, Department of Physics and Measurement Technology, Linköping University, SE - 581 83 Linköping, Sweden Mitsumasa Iwamoto Department of Physical Electronics, Tokyo Institute of Technology, 2-12-1 O-okayama, Meguro-ku, Tokyo, 152-8552, Japan Toshiya Kamikado Communications Research Laboratory, 588-1 Iwaoka, Nishi-ku, Kobe 651-2401, Japan
xii
List of Contributors
Keiichi Kaneto Life Science & Systems Engineering, Kyusyu Institute of Technology, Iizuka, Fukuoka, 820-8502, Japan Shunsaku Kimura Department of Material Chemistry, Graduate School of Engineering, Kyoto University, Yoshida Honmachi, Sakyo-ku, Kyoto, 606-8501, Japan Takatoshi Kinoshita Deparment of Materials Science & Engineering, Nagoya Institute of Technology, Gokiso-cho, Showa-ku, Nagoya, 466-8555, Japan Kazuya Kitagawa Department of Material Chemistry, Graduate School of Engineering, Kyoto University, Yoshida Honmachi, Sakyo-ku, Kyoto 606-8501, Japan Takayuki Kitamura Department of Material and Life Science, Graduate School of Engineering, Osaka University, Yamada-oka 2-1, Suita, Osaka 565-0871, Japan Tohru Kubota Kansai Advanced Research Center, Communications Research Laboratory, 588-2 Iwaoka, Nishi-ku, Kobe, Hyogo, 651-2492, Japan Kazuhiro Kudo Faculty of Engineering, Chiba University, 1-33 Yayoi-cho, Inake-ku, Chiba, Chiba 263-8522, Japan Shin-ichi Kuroda Department of Applied Physics, Graduate School of Engineering, Nagoya University, Furo-cho, Chikusa-ku, Nagoya, 464-8603, Japan Geoffrey R. Luckhurst Department of Chemistry and Southampton Liquid Crystal Institute, University of Southampton, Highfield, Southampton, SO17 1BJ, UK F. Mallwitz Organic and Macromolecular Chemistry, OC3, University of Ulm, Albert-EinsteinAllee 11, D-89069 Ulm, Germany Takaaki Manaka Department of Physical Electronics, Tokyo Institute of Technology, 2-12-1 O-okayama, Meguro-ku, Tokyo 152-8552, Japan Kazuhiro Marumoto Department of Applied Physics, Nagoya University, Chikusa-ku, Nagoya 464-8603, Japan
List of Contributors
xiii
Shinro Mashiko Kansai Advanced Research Center, Communications Research Laboratory, 588-2 Iwaoka, Nishi-ku, Kobe, Hyogo, 651-2492, Japan Shinji Matsui Laboratory of Advanced Science and Technology for Industry, Himeji Institute of Technology, 3-1-2 Koto, Kamigori, Ako, Hyogo, 678-1201, Japan Teruyoshi Mizutani Department of Electrical Engineering, Graduate School of Engineering, Nagoya University, Furo-cho, Chikusa-ku, Nagoya, 464-8603, Japan Tatsuo Mori Department of Electrical Engineering, Graduate School of Engineering, Nagoya University, Furo-cho, Chikusa-ku, Nagoya 464-8603, Japan Seizo Morita Department of Electronic Engineering, Graduate School of Engineering, Osaka University, Yamada-Oka 2-1, Suita, Osaka, 565-0871, Japan Tomoyuki Morita Department of Material Chemistry, Graduate School of Engineering, Kyoto University, Yoshida Honmachi, Sakyo-ku, Kyoto 606-8501, Japan Toshihiko Nagamura Molecular Photonics Laboratory, Research Institute of Electronics, Shizuoka University, 3-5-1 Johoku, Hamamatsu, 432-8011, Japan Tatsuo Nakahama Kansai Advanced Research Center, Communications Research Laboratory, 588-2 Iwaoka, Nishi-ku, Kobe, Hyogo, 651-2492, Japan Yutaka Noguchi Department of Physical Electronics, Tokyo Institute of Technology, 2-12-1 O-okayama, Meguro-ku, Tokyo 152-8552, Japan Yoshishige Okuno Communications Research Laboratory, 588-1 Iwaoka, Nishi-ku, Kobe 651-2401, Japan Mitsuyoshi Onoda Graduate School of Engineering, Himeji Institute of Technology, 2167 Shosha, Himeji, Hyogo, 671-2201, Japan C. Peyratout Max-Planck-Institut für Kolloid- & Grenzflächenforschung, Berlin, Germany
xiv
List of Contributors
Koichi Rikitake Life Science & Systems Engineering, Kyusyu Institute of Technology, Iizuka, Fukuoka, 820-8502, Japan J.R. Sambles Thin Film Photonics, School of Physics, University of Exeter, Exeter, EX4 4QL, UK Edward T. Samulski Department of Chemistry, University of North Carolina, Chapel Hill, NC 275993290, USA Takahiro Seki Photofunctional Chemistry Devision, Chemical Resources Laboratory, Tokyo Institute of Technology, 4259 Nagatsuta-cho, Midori-ku, Yokohama, 226-8503, Japan Brian R. Stoner Department of Physics and Astronomy, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599-3255, USA Yasuhiro Sugawara Department of Electronic Engineering, Graduate School of Engineering, Osaka University, Yamada-Oka 2-1, Suita, Osaka, Japan Akihiko Sugimura Department of Information Systems Engineering, Osaka Sangyo University, 3-1-1 Nakagaito, Daito, Osaka, 574-0013, Japan Kazuya Tada Graduate School of Engineering, Himeji Institute of Technology, 2167 Shosha, Himeji, Hyogo 671-2201, Japan Wataru Takashima Life Science & Systems Engineering, Kyusyu Institute of Technology, Iizuka, Fukuoka, 820-8502, Japan Kiyoshi Takimoto Advanced Devices Division, Canon Research Center, Canon Inc., 5-1, MorinosatoWakamiya, Atsugi, Kanagawa, 243-0193, Japan Motomu Tanaka Lehrstuhl fur Biophysik E22, Technische Universität München, D 85748, Garching, Germany Atsushi Tojima Department of Physical Electronics, Tokyo Institute of Technology, 2-12-1 O-okayama Meguro-ku, Tokyo 152-8552, Japan
List of Contributors
xv
H. Xu Organic and Macromolecular Chemistry, OC3, University of Ulm, Albert-EinsteinAllee 11, D-89069 Ulm, Germany Shozo Yanagida Department of Material and Life Science, Graduate School of Engineering, Osaka University, Yamada-oka 2-1, Suita, Osaka 565-0871, Japan Fuzi Yang Liquid Crystal Research Center, Department of Chemistry, Tsinghua University, Beijing 100084, China and School of Physics, University of Exeter, Exeter Ex4 4QL, UK Yoshiyuki Yokogawa National Institute of Advanced Industrial Science and Technology, Hirate-cho 1-1, Kita-ku, Nagoya 462-8510, Japan Shiyoshi Yokoyama Kansai Advanced Research Center, Communications Research Laboratory, 588-2 Iwaoka, Nishi-ku, Kobe, Hyogo, 651-2492, Japan Takashi Yokoyama Nanomaterial Laboratory, National Institute for Material Science, Shidami Human Science Park, 2268-1 Anagahora, shimo-Shidami, Moriyama-Ku, Nagoya, 463-0003, Japan Yuji Wada Department of Material and Life Science, Graduate School of Engineering, Osaka University, Yamada-oka 2-1, Suita, Osaka 565-0871, Japan Shintaro Washizu Fujinomiya Research Laboratories, Fuji Photo Film co., LTD., Fujinomiya Shizuoka 418-8666, Japan Fengling Zhang Department of Physics and Measurement Technology, Laboratory of Applied Physics, IFM Linköping University, S-581 83, Linköping, Sweden Ou-Yang Zhongcan Institute of Theoretical Physics, Academia Sinica, P.O. Box 2735, Beijing, 100080, China
Part A
Single Molecular Electronics and Photonics
Nanotechnology and Nano-Interface Controlled Electronic Devices Editors: M. Iwamoto, K. Kaneto and S. Mashiko © 2003 Elsevier Science B.V. All rights reserved
CHAPTER 1
Nanostructure fabrication using electron and ion beams Shinji Matsui Laboratory of Advanced Science and Technology for Industry, Himeji Institute of Technology, 3-1-2 Koto, Kamigori, Ako, Hyogo, 678-1205, Japan
1. 2. 3.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Electron beam nanolithography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Room temperature nanoimprint technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Desktop compact imprint apparatus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Room-temperature nanoimprint into HSQ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. Three-dimensional nanostructure fabrication by focused-ion-beam . . . . . . . . . . . . . . . . . 4.1. Fabrication process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2. Micro-system parts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3 4 6 8 10 12 13 16 19 19
1. Introduction Recent years have witnessed a number of investigations concerning nanostructure technology. The objective of research on nanostructure technology is to explore the basic physics, technology, and applications of ultra-small structures and devices with dimensions in the sub-100-nm regime. Today, the minimum size of Si and GaAs production devices is down to 0.15 µm or less. Nanostructure devices are now being fabricated in many laboratories to explore various effects, such as those created by downscaling existing devices, quantum effects in mesoscopic devices, or tunneling effects in superconductors, etc. In addition, new phenomena are being explored in an attempt to build switching devices with dimensions down to the molecular level. Fig. 1 summarizes the resolution capabilities of several lithography processes that use electrons, ions, and photons. It includes the narrowest line width of feature size obtained with each process. Microfabrication can be classified into three regimes: submicron (1000 to 100 nm), nano (100 to 1 nm) and atom (or Ångstrom, less than 1 nm). A 256-Mb dynamic random-access memory (DRAM) Si ULSI of 0.25-µm dimensions can be fabricated by using an i-line stepper with a phase shift mask, or an excimer laser stepper. An excimer laser can be applied to a 1-Gb DRAM with 0.15 µm feature size.
4
S. Matsui
Fig. 1: Microfabrication using electrons, ions, and photons.
Electron beam (EB) lithography is the most widely used and versatile lithography tool used in fabricating nanostructure devices. Because of the availability of high-quality electron sources and optics, EB can be focused to diameters of less than 10 nm. The minimum beam diameters of scanning electron microscopes (SEM) and scanning transmission microscopes (STEM) are 1.5 and 0.5 nm, respectively. While focused-ion beam (FIB) can be focused close to 5 nm. EB and FIB can be used to make nano-scale features in the 100- to 1-nm regime. Scanning tunneling microscopy (STM) is used for atomic technology in the region of 1 to 0.1 nm. Fig. 2 shows the resolution of various resists, which were confirmed by experiment, for electrons and ions. Minimum sizes of 8 nm for PMMA [1,2], 10 nm for ZEP (Nippon Zeopn Co.) positive resists [3], 20 nm for SAL601 (Shipley Co.) [4], and 10 nm for CALIXARENE negative resists [5] have been demonstrated using EB lithography. Nano-scale patterns have also been written in inorganic resists such as AlF3 , NaCl, and SiO2 using STEM [6,7] and SEM [8]. Furthermore, carbon contamination patterns of 8 nm have been fabricated with SEM [9], and 8-nm PMMA patterns have been demonstrated by using Ga+ FIB [10]. In this chapter, recent progress in nanofabrication using EB and FIB is described.
2. Electron beam nanolithography Nanodevice fabrication requires not only high resolution but also high overlay accuracy. High-speed exposure very effectively meets the requirements because overlay accuracy is improved due to less beam drift on the nanometer scale. Moreover, it enables the use of a highly sensitive resist such as ZEP520 [11], which has sufficient resolution
Nanostructure fabrication using electron and ion beams
5
Fig. 2: Resolution of various resists for electrons and ions.
Fig. 3: 10-nm line width ZEP patterns.
and high dry etching durability for nanolithography. A 10-nm-scale resist pattern was obtained using ZEP520 positive resist. The ZEP520 resist was spin-coated onto a Si wafer to a thickness of 50 nm, and prebaked at 200°C. After EB exposure, the ZEP520 was developed with hexyl acetate for 2 min and rinsed with 2-propanol. Fig. 3 shows
6
S. Matsui
a ZEP520 resist pattern in which the lines are 10 nm wide and have a pitch of 50 nm [3]. CALIXARENE is roughly a ring-shaped molecule with a diameter of about 1-nm and it works as an ultrahigh-resolution negative EB resist. CALIXARENE is a single molecule and thus is monodispersed with a molecular weight of 972. In contrast, other phenol-based resists have dispersive weights from 1000 to 100,000, which set a resolution limit. The molecular uniformity of CALIXARENE and its small molecular size is the origin of surface smoothness and the resulting ultrahigh resolution. Such characteristics seem to be convenient for a nanodevice fabrication process. The basic component of CALIXARENE is a phenol derivative which seems to have high durability and stability, originating from the strong chemical coupling of the benzene ring. The threshold of sensitivity was about 800 µC/cm2 , which is almost 20 times higher than that of PMMA. CALIXARENE negative resist exposure was carried out. A 30-nm-thick resist was coated on a bare Si wafer. After prebaking at 170°C for 30 min, EB exposure was carried out and then the resist was developed in xylene for 20 s and was rinsed in IPA for 1 min. The etching durability of CALIXARENE was tested using a DEM-451 (ANELVA Corp.) plasma dry-etching system with CF4 gas. The etching rate of CALIXARENE is almost comparable with that of Si, and the durability is about four times higher than that of PMMA. This durability seems to be sufficient to make a semiconductor or a metal nanostructure. Nanodot arrays are useful not only for quantum devices but also for studying exposure properties. In this experiment, the EB current was fixed to 100 pA at 50 kV accelerating voltage, for which the spot size is estimated to be about 5 nm. All the dot arrays were fabricated on Si substrates. And the typical exposure dose (spot dose) was about 1 × 105 electrons/dot. Fig. 4 shows typical dot array patterns having 15 nm diameter with 35 nm pitch. Germanium pattern transfer is shown in Fig. 5. The 20-nm-thick Ge layer requires at least a 5-nm-thick CALIXARENE layer to be etched down, and the resist thickness was 30 nm. Fig. 5(a) shows the line patterns of the resist on Ge film exposed at a line dose of 20 nC/cm. Delineation was done using the S-5000 (Hitachi Corp.) SEM with a beam current of 100 pA at a 30-kV acceleration voltage. A 10-nm line width and a smooth line edge were clearly observed. This smoothness is the key point in fabricating quantum nanowires by etching processes. Fig. 5(b) shows the transferred pattern treated by 1 min of overetching, followed by oxygen-plasma treatment to remove the resist residues. A Ge line of 7 nm width was clearly observed without short cutting. Narrowing by overetching is a standard technique to obtain a fine line, however, side-wall roughness limits the line width. The smoothness of the CALIXARENE side wall enables the line width to be narrowed below the 10-nm region by overetching.
3. Room temperature nanoimprint technology Nanoimprint-lithography (NIL) [12–18], in which resist patterns are fabricated by deforming the physical shape of the resist by embossing with a mold, is a very useful
Nanostructure fabrication using electron and ion beams
7
Fig. 4: CALIXARENE dot array patterns with 15 nm diameter and 35 nm pitch.
technique to make nanostructure devices; and various nanostructure devices, such as a quantized magnetic disk have, been demonstrated [19–22]. The technique has excellent features that are sub-10 nm in size over a large area and have high throughput and are low in cost. However, a conventional NIL process has to heat a resist above the glass transition temperature to deform the physical shape of the resist with a mold. Consequently, a conventional NIL process (system) must require a thermal cycle of the resist. This heating process causes serious problems in replicated pattern accuracy, and a reduction in throughput due to the thermal cycle of the resist. In order to overcome these problems, a room temperature nanoimprint-lithography (RT-NIL) has been proposed. This RT-NIL process does not require a thermal cycle of the resist in pressing a mold onto the resist. Fig. 6 shows the difference between conventional NIL and RT-NIL. The RT-NIL process steps without heating and cooling are shorter compared with those in conventional NIL as shown in Fig. 6(a) [12,13]. First, we have demonstrated RT-NIL using Spin-on-Glass (SOG), as the material replicated and obtained excellent replicated SOG patterns and Si etched patterns by transferring replicated SOG patterns using CF4 RIE [23]. However, SOG has the important technological shortcoming that SOG hardens gradually by reaction with water in the air at room temperature. To overcome the disadvantage, we have proposed RTNIL using a hydrogen silsequioxane (HSQ) as the replicated material. Further, we have carried out the step-and-repeat imprinting using HSQ, and evaluated the uniformity of the replicated HSQ patterns.
8
S. Matsui
Fig. 5: Pattern transfer to Ge. (a) 10-nm line width CALIXARENE pattern and (b) transferred 7-nm line width Ge pattern.
3.1. Desktop compact imprint apparatus The desktop compact imprint apparatus which uses a stepping motor as driving power that we have developed is shown in Fig. 7(a). The apparatus is 17 cm in width and 30 cm in height, and has 10 × 10 mm2 mold-mask holder and 2 inch wafer stage. The z-positioning accuracy of the stepping motor is 2 µm per pulse. A heater is buried in the wafer stage to heat resist coated on a wafer to above the glass temperature. Therefore, by using this system, conventional NIL and RT-NIL can be performed. A 2 in. wafer can be imprinted on the x–y step-and-repeat stage. The z-axis of the mold holder and the x- and y-axis of the wafer stage are controlled by three stepping motors receiving pulse signals from a computer. The wafer temperature can be varied from room temperature up to 200°C by heating the stage, while measuring temperature with a digital thermometer. A piezo component is buried in the z axis of the mold holder to be able to measure pressures when pressing a mold-mask into the resist on the substrate.
Nanostructure fabrication using electron and ion beams
9
Fig. 6: Schematic of a NIL process: (a) conventional NIL using PMMA, and (b) room temperature NIL using SOG or HSQ.
Fig. 7: (a) Desktop compact imprint system which uses a stepping motor as driving power. (b) Imprint system with a mold-mask holder and an imprint stage.
10
S. Matsui
Fig. 7(b) illustrates the system of the mold-mask holder and the imprint stage. A magnet is inserted between a mold-mask and a mold wafer, and a magnetic sheet and an iron plate are inserted between the imprint stage and a wafer. The purpose of using a magnet is to make hard contact between the mold-mask and the wafer, and can bring a mold-mask into contact with a wafer in parallel in printing. 3.2. Room-temperature nanoimprint into HSQ In this experiment, HSQ (Dow Cornig Co. FOX) which has as structure formula H2 (SiO)3 was used. Infrared (IR) spectra confirmed that there was no hydrocarbon in the HSQ material. Therefore, this HSQ resin contains no organic groups, such as vinyl groups. And then, HSQ has a favorable etching durability [24]. Since HSQ has a high viscosity without pre-baking, pre-baking must be performed, in contrast to SOG, before the imprinting process. The effect of pre-baking for the HSQ is to remove the solvent from the content of HSQ, and to make the viscosity of the HSQ moderate so that physical transformation of the HSQ with a mold pattern is possible. Fig. 8 shows the pre-baking dependence of the imprint depth when using HSQ. We measured the depth of replicated HSQ patterns with 4-µm line widths using a profiler. Imprint pressures were varied from 1.0 to 4.5 MPa. And pre-baking temperatures were varied from 50 to 200°C. The results indicate that the imprinting depth decreases suddenly around 150°C. It is suggested that the hardness of HSQ increases around 150°C. Therefore we used a pre-baking temperature in the range from 50°C to 100°C to obtain a suitable imprint depth of the replicated HSQ patterns. An RT-NIL process using HSQ is shown Fig. 6(b). First, 0.3-µm-thick HSQ was spin-coated on an Si substrate. Then a mold and an HSQ coated substrate with prebaking were pressed together for 1 min at a set press-pressure in the range from 2.5 to 4.5 MPa. After that, the mold was removed by the driving power of the stepping motor. Fig. 9(a), (b) and (c) show SEM photographs of a top view and at a tilt angle of 45 degrees of imprinted HSQ patterns with 0.8-µm line widths and 6-µm pitches after 1 min pressing using RT-NIL. In this experiment, an imprinting pressure of 2.0 MPa and
Fig. 8: Imprinting characteristics using HSQ.
Nanostructure fabrication using electron and ion beams
11
Fig. 9: SEM micrograph of (a) a top view and (b), (c) at a tilt angle of 45 degrees of imprinted HSQ patterns with 0.8-µm line widths and 6 µm pitches.
Fig. 10: SEM micrograph of (a) a top view and (b) at a tilt angle of 45 degrees of SiO2 /Si pillars mold patterns with 90 nm diameter, 600 nm period and which are 0.4 µm in height.
a pre-baking temperature of 50°C were used. It was confirmed that the imprinted depth was 150 nm and the residual depth was 150 nm from Fig. 9(c), which nearly corresponds to the imprinting characteristics using HSQ as shown in Fig. 8. Fig. 10(a) and (b) shows SEM photographs of a top view and at a tilt angle of 45 degrees of SiO2 /Si pillars mold patterns with 90 nm diameter, 600 nm period and which are 0.4 µm in height. By using the mold, holes with 90 nm diameter and 600 nm pitch were obtained after 1 min impress time by RT-NIL, as shown in Fig. 11(a) and (b). The pre-baking temperature and pressure were 60°C and 4.0 MPa. This indicates that the mold pillar patterns were
12
S. Matsui
Fig. 11: SEM micrographs of a top view of holes with 90 nm diameter and 600 nm pitch imprinted after 1 min press time by RT-NIL.
Fig. 12: SEM micrographs of patterns with 50 nm line width and 200 nm pitch replicated in HSQ using RT-NIL.
imprinted with high precision. Fig. 12 (a) and (b) show SEM photographs of patterns with 50 nm line width and 200 nm pitch replicated in HSQ using RT-NIL. These results demonstrate that RT-NIL using HSQ is a useful nanostructure fabrication technique.
4. Three-dimensional nanostructure fabrication by focused-ion-beam Two-dimensional nanostructure fabrication using electron-beam (EB) and focused-ionbeam (FIB) has been achieved, and has been applied to make various nanostructure
Nanostructure fabrication using electron and ion beams
13
devices such as single-electron transistors and MOS transistors with nanometer gatelength. Ten-nm structures can be formed by using a commercially available EB or FIB system with 5–10 nm beam diameter and a high-resolution resist [25]. Because of this, the technique of two-dimensional nanostructure fabrication is considered as established. As outlook on three-dimensional fabrication, there are three techniques using laser, EB, and FIB Chemical Vapor Deposition (CVD). Compared with three-dimensional fabrication using laser-CVD, FIB- and EB-CVD are superior [26] in the points of spatial resolution and beam-scan control. Koops et al. demonstrated some applications such as an AFM tip and field emitter by using EB-CVD [27]. Blauner et al. demonstrated pillars and walls with high aspect ratios by using FIB-CVD [28,29]. The deposition rate of FIB-CVD is much higher than that of EB-CVD due to factors such as the difference of mass between electrons and ions. Furthermore, a smaller penetration-depth of ions compared to electrons allows to make complicated threedimensional nanostructures. For example, when we make a coil nanostructure with 100 nm line width, electrons with 10–50 keV pass the ring of the coil and reach the substrate because of the large electron range (over a few µm), so it may be difficult to make a coil nanostructure by EB-CVD. On the other hand, as the ion range is less than a few ten nm, ions stop inside the ring. So far complicated nanostructures using FIB-CVD have not been reported. This paper presents a description of the fabrication of a complicated three-dimensional nanostructure using FIB-CVD [30,31]. 4.1. Fabrication process All experiments were carried out with a commercially available FIB system (SIM9200: Seiko Instrument Inc.) utilizing a beam of 30 keV Ga+ ions. The beam is focused to a spot size of 7 nm at 0.4 pA beam-current, and is incident perpendicularly to the surface. Phenanthrene precursor gas is evaporated from a heated container and is injected into the vacuum chamber by means of a nozzle, which is located at a height of 500 µm above the sample surface, at an angle of about 45 degrees with respect to the sample surface. The nozzle system serves to create a local high-pressure region over the surface. The base pressure of sample chamber is 2 × 10−5 Pa and the chamber pressure after introducing the source gas is 5 × 10−5 Pa. The FIB is controlled by a computer to write the desired pattern and the ion dose is adjusted to deposit a film of the desired thickness. The experiments were carried out at room temperature on a silicon substrate. The characterization of the deposited film was performed by observation of transmission electron microscope (TEM) and measuring of Raman spectra. A carbon thin film with 200 nm thickness was deposited on a silicon substrate by 30 keV Ga+ FIB using a phenanthrene precursor gas. The cross-section structures and electron diffraction patterns were observed by using a 300-kV TEM. The result was that there were no crystal structures in the TEM images and diffraction patterns. It is concluded that the deposited film is amorphous carbon (a-C). Raman spectra of a-C films were measured at room temperature with the 514.5 nm line of an argon ion laser. The Raman spectra were recorded by a monochromator equipped with a CCD multi-channel detector. Raman spectra were measured at 0.1–1.0 mW to avoid thermal decomposition of the samples. Fig. 13 shows Raman spectra of an a-C film deposited on a silicon substrate. A relatively
14
S. Matsui
Fig. 13: Raman spectra of a diamond-like amorphous carbon film obtained at 0.1 mW by 514.4 nm excitation wavelength. The decomposed bands are shown as the G (graphite: 1550 cm−1 ) and D (diamond: 1400 cm−1 ) solid lines.
sharper Raman band at 1550 cm−1 and a broad shoulder band at around 1400 cm−1 are observed in the spectra excited by a 514.5 nm line. Two Raman bands were plotted after Gaussian line shape analysis. The Raman bands at 1550 cm−1 and 1400 cm−1 originate from the trigonal (sp2 ) bonding structure of graphite and tetrahedral (sp3 ) bond structure of diamond. This result indicates that a-C film deposited by FIB-CVD is diamond-like amorphous carbon, which has attracted attention because of the hardness, chemical inertness, and optical transparency. The three-dimensional structure fabrication process by FIB-CVD is illustrated in Fig. 14. In FIB-CVD processes, a beam-scan is done in digital mode. First, a pillar is formed on the substrate by fixing a beam-position (position 1). After that, the beam-position is moved within the diameter of the pillar (position 2) and is then fixed until the deposited terrace thickness exceeds the ion range which is a few ten nm. This process is repeated to make three-dimensional structures. The key point to make three-dimensional structures is to adjust the beam-scan speed so that the ion-beam remains within the deposited terrace, which means that the terrace thickness exceeds an ion range. The growth conditions in the x and y directions are controlled by the beam-deflectors. The growth in the z direction is determined by the deposition rate, that is, the height of a structure is proportional to the irradiation time when the deposition rate is constant. Figure 15 shows a branch structure made by 30 keV Ga+ FIB-CVD. First, a pillar was made by fixing the beam-position for 120 s at 16 pA beam-current. And then, the growth of a branch was carried out using the process explained above. The diameter of the branch is 0.08 µm. The exposure time was 270 s at 0.4 pA beam-current. The experimental result indicates that arbitrary three-dimensional nanostructures can be fabricated by FIB-CVD.
Nanostructure fabrication using electron and ion beams
Fig. 14: Fabrication process for three-dimensional nanostructures by FIB-CVD.
Fig. 15: Branch structure with 0.08 µm diameter.
15
16
S. Matsui
Fig. 16: Micro-beakers with 1.0 and 1.5 µm diameter, and 1.0 µm height.
4.2. Micro-system parts Various micro-system parts were fabricated by FIB-CVD. Fig. 16 shows micro-beakers with 1.0 µm height, and 1.0 and 1.5 µm diameters. Total exposure-time of two beakers was 600 s at 16 pA beam-current. Some applications are considered, such as the study of micro-crystal growth or micro-chemical reactions by filling a beaker with the examined material.
Fig. 17: Micro-coil with 0.6 µm coil-diameter, 0.7 µm coil-pitch, and 0.08 µm line width.
Nanostructure fabrication using electron and ion beams
17
Fig. 18: Micro-drill with 0.25 µm diameter, 0.20 µm pitch, and 3.8 µm height.
Fig. 17 shows a micro-coil with 0.6 µm coil-diameter, 0.7 µm coil-pitch, and 0.08 µm line width. Exposure time was 40 s at 0.4 pA beam current. The coil-pitch can easily be changed by controlling the growth speed. Reducing the diameter of the micro-coil, a micro-drill was formed, as shown in Fig. 18. The diameter, pitch, and height of the micro-coil are 0.25, 0.20, and 3.8 µm, respectively. Exposure time was 60 s at 0.4 pA beam-current. A bellows is one of the important parts in a mechanical system, just like a coil and a drill. Fig. 19 shows a micro-bellows with 0.8 µm pitch, 0.1 µm thickness, 2.75 µm external diameter, and 6.1 µm height. Exposure time was 300 s at 16 pA beam-current. The results show that FIB-CVD is one of the promising techniques to make parts of microsystems, although the mechanical performances of these parts have to be measured. We intend to open up microstructure plastic arts as a new field using FIB-CVD. To demonstrate the possibility, a micro-wineglass was created as a work of microstructure plastic arts. A micro-wineglass with 2.75 µm external diameter and 12 µm height was formed on Si substrate and a human hair, as shown in Fig. 20(a) and (b). Fabrication time was 600 s at 16 pA beam-current. The beautiful micro-wineglass gives us expectations of opening up microstructure plastic arts.
18
S. Matsui
Fig. 19: Micro-bellows with 0.8 µm pitch, 0.1 µm thickness, 2.75 µm external diameter, and 6.1 µm height.
Fig. 20: Micro-wineglass with 2.75 µm external diameter and 12 µm height on (a) Si substrate and (b) human hair.
Nanostructure fabrication using electron and ion beams
19
5. Summary Ten-nanometer structure fabrication has been achieved by using a commercial available electron beam (EB) apparatus and resists. Nanoimprint lithography (NIL) is a very useful technique to overcome the low throughput and high cost in EB lithography. Room temperature NIL using HSQ as the replicated material has been proposed to achieve highly precise replication and to perform step-and-repeat imprinting. As a result, we demonstrated 50 nm line width and 90 nm in diameter hole replicated HSQ patterns, and the possibility of step-and-repeat imprinting in HSQ to 1.5 in. wafers. The results reveal that the room-temperature NIL process using HSQ as the replicated material is a very useful technique to achieve a highly precise nanoimprinting process and this technique can be applied to make nanostructure devices. Three-dimensional structures with spatial resolution in the nanometer range can be generated with the focused-ion-beam chemical vapor deposition (FIB-CVD) technique. Three-dimensional nanostructure fabrication has been demonstrated by 30 keV Ga+ FIB-CVD using a phenanthrene precursor. It is confirmed by TEM and Raman spectra that the deposited film is a diamond-like amorphous carbon. Micro-coil, drill, and bellows with 0.1 µm dimension were fabricated as parts of micro-systems. We propose microstructure plastic arts as a new field using micro-beam technology. A microwineglass with 2.75 µm external diameter and 12 µm height was created as one work. It is concluded from the experimental results that FIB-CVD direct-write processes may become interesting tools for the generation of micro- and nano-systems in the field of electronics, mechanics, optics and biology.
References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16.
F. Emoto, K. Gamo, S. Namba, N. Samoto, and R. Shimizu, Jpn. J. Appl. Phys. 24, L809 (1985). W. Chen and H. Ahmed, Appl. Phys. Lett. 63, 1116 (1993). K. Kurihara, K. Iwadate, H. Namatsu, M. Nagase, H. Takenaka, and K. Murase, Jpn. J. Appl. Phys. 34, 6940 (1995). T. Yoshimura, Y. Nakayama, and S. Okazaki, J. Vac. Sci. Technol. B10, 2615 (1992). J. Fujita, Y. Ohnishi, Y. Ochiai, and S. Matsui, Appl. Phys. Lett. 68, 1297 (1996). M. Isaacson and A. Murray, J. Vac. Sci. Technol. 19, 1117 (1981). D.R. Allee and A.N. Broers, Appl. Phys. Lett. 57, 2271 (1990). J. Fujita, H. Watanabe, Y. Ochiai, S. Manako, J.S. Tsai, and S. Matsui, Appl. Phys. Lett. 66, 3065 (1995). A.N. Broers, W.W. Molzen, J.J. Cuomo, and N.D. Wittles, Appl. Phys. Lett. 29, 596 (1976). R.L. Kubena, J.W. Ward, F.P. Stratton, R.J. Joyce, and G.M. Atkinson, J. Vac. Sci. Technol. B9, 3079 (1991). T. Nishida, M. Notomi, R. Iga, and T. Tamamura, Jpn. J. Appl. Phys. 31, 4508 (1992). S.Y. Chou, P.R. Krauss, and P.J. Renstrom, Appl. Phys. Lett. 67, 3114 (1995); Science 272, 85 (1996). S.Y. Chou, P.R. Krauss, W. Zhang, L. Guo, and L. Zhuang, J. Vac. Sci. Technol. B15, 2897 (1997). X. Sun, L. Zhuang, W. Zhang, and S.Y. Chou, J. Vac. Sci. Technol. B16, 3922 (1998). B. Heidari, I. Maximov, and L. Montelius, J. Vac. Sci. Technol. B18, 3557 (2000). H. Schift, R.W. Jaszewski, C. David, and J. Gobrecht, Microelectronic Eng. 46, 121 (1999).
20
S. Matsui
17.
T. Bailey, B.J. Choi, M. Colburn, M. Meissl, S. Shaya, J.G. Ekerdt, S.V. Sreenivasan, and C.G. Willson, J. Vac. Sci. Technol. B18, 3572 (2000). M. Komuro, J. Taniguchi, S. Inoue, N. Kimura, Y. Tokano, H. Hiroshima, and S. Matsui, Jpn. J. Appl. Phys. 39, 7075 (2000). W. Wu, B.C.X. Sun, W. Zhang, L. Zhuang, L. Kong, and S.Y. Chou, J. Vac. Sci. Technol. B16, 3825 (1998). L. Zhuang, L. Guo, and S.Y. Chou, Appl. Phys. Lett. 72, 1205 (1998). J. Wang, A. Schablitsky, Z. Yu, W. Wu, and S.Y. Chou, J. vac. Sci. Technol. B17, 2957 (1999). I. Martini, S. Kuhn, M. Kamp, L. Worschech, A. Forchel, D. Eisert, J. Koeth, and R. Sijbesma, J. Vac. Sci. Technol. B18, 3561 (2000). S. Matsui, Y. Igaku, H. Ishigaki, J. Fujita, M. Ishida, Y. Ochiai, M. Komuro, and H. Hiroshima, J. Vac. Sci. Technol. B19, 2801 (2001). H. Namatsu, Y. Takahashi, K. Yamazaki, T. Yamaguchi, M. Nagase, and K. Kurihara, J. Vac. Sci. Technol. B16, 69 (1998). S. Matsui, Proc. IEEE 85, 629 (1997). O. Lehmann, F. Foulon, and M. Stuke, NATO ASI Ser. E: Appl. Sci., 265, 91–102 (1994). H.W. Koops, Jpn. J. Appl. Phys. 33, 7099 (1994). A. Wargner, J.P. Levin, J.L. Mauer, P.G. Blauner, S.J. Kirch, and P. Long, J. Vac. Sci. Technol. B8, 1557 (1990). P.G. Blauner, Proc. 1991 Int. MicroProcess Conf., p. 309 (1991). S. Matsui, T. Kaito, J. Fujita, M. Komuro, K. Kanda, and Y. Haruyama, J. Vac. Sci. Technol. B18, 3181 (2000). J. Fujita, M. Ishida, T. Sakamoto, Y. ochiai, T. Kaito, and S. Matsui, J. Vac. Sci. Technol. B19, 2834 (2001).
18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31.
Nanotechnology and Nano-Interface Controlled Electronic Devices Editors: M. Iwamoto, K. Kaneto and S. Mashiko © 2003 Elsevier Science B.V. All rights reserved
CHAPTER 2
Information storage using a scanning probe Kiyoshi Takimoto Canon Research Center, Canon Inc., 5-1 Morinosato-Wakamiya, Atsugi, Kanagawa 243-0193, Japan, E-mail:
[email protected]
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. Restriction in transfer rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. Formation of an ideal metal–insulator–metal junction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. Size of data bits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5. Rate of reading and writing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6. Error rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
21 22 22 24 25 27 27 28
1. Introduction Surface modification with a scanning probe microscope (SPM) is very attractive, because the dimension of the modification is ranging from sub-micron to sub-nanometer, which is much smaller than the limit of conventional photolithography [1]. Even an atomic manipulation has been achieved [2,3]. A surface modification according to a given pattern and its observation with a scanning probe can also be seen as a writing and reading procedure in an information storage device. Then the pattern produced by the surface modification is regarded as a set of data bits. The recording density estimated from the typical size of each modification becomes much higher than that of any present storage device. If the size is 10 nm, for example, the recording density can be roughly estimated to be 1 Tbit/cm2 (= 1012 bit/cm2). Therefore the SPM-technology is thought to make it possible to establish a high-density and huge-capacity information storage system. There have been many reports concerning information storage using an SPM [4–12]. However, the recording density is only one of the performances characterizing an information storage device. Another important performance is the transfer rate, especially the reading rate. At least it is necessary to read and write a set of data bits at a rate comparable to that of the present storage devices in order to realize a practical system.
22
K. Takimoto
2. Restriction in transfer rate In the SPM, the surface image is based on some interaction between a probe and the surface of a sample when they are placed in close proximity. The surface topography is observed as the trajectory of the probe motion when the spacing between the probe and the surface is precisely controlled so as to maintain the interaction between them to be constant during scanning of the probe. The operation speed in the SPM is limited by the characteristic rate of the feedback loop, and it is generally low. The highest operation speed in the SPM is achieved in an atomic force microscope (AFM) operated in contact mode, where the tip of a probe is in contact with the surface of a sample. Then, the operation speed is limited to the mechanical resonance frequency of the cantilever for supporting the probe, which is limited to several mega-Hertz in practice. When the data bits are formed with topographical modification of the recording medium, the reading rate of information storage with a scanning probe is restricted by the operation speed in the SPM. One of the ways to overcome this restriction is the use of a cantilever with high mechanical resonance frequency. In this case, precise fabrication of a cantilever with a small size and a small weight is necessary. Another way is the use of a recording medium in which data bits can be written without topographical modification of this recording medium. In this procedure, a scanning probe must be used for simultaneously detecting two kinds of interactions, independent of each other. The spacing between the probe and the medium is controlled based on one interaction between them, and the data bits are written and read based on the other interaction. Then the former interaction does not influence the reading rate directly, since it is not affected by the presence of the data bits. The reading rate is determined by the rate for detection of the data bits and the rate for scanning of the probe on the medium. For high-speed scanning of the probe, reducing the surface roughness of the medium and controlling the spacing between the probe and the medium with an AFM based apparatus are effective.
3. Formation of an ideal metal–insulator–metal junction To overcome the restriction of the rate for reading and writing, another recording procedure was introduced which is not based on direct surface modification of a recording medium with a scanning probe. In this procedure, a scanning probe is used as a tool in order to form an ideal metal–insulator–metal (MIM) junction. That is, one electrode in the MIM junction is replaced by the scanning probe. For this purpose, an AFM with an electrically conducting probe is used. In practice, the conducting probe is obtained by coating a conventional cantilever for AFM with a metal film. The probe can be in contact with the sample surface using a weak force controlled by AFM so as to keep it constant, and an ideal MIM junction may be formed. In this scheme, the electronic properties of a film used as an insulating layer can be characterized with sufficient resolution as shown in Fig. 1. If the electronic properties of the film can be modified by application of a pulse voltage to the MIM junction formed with the probe, it can be considered as another procedure to form data bits. The area with modified electronic properties can be observed using the AFM based apparatus as
Information storage using a scanning probe
23
Fig. 1: Schematic drawing of an AFM based apparatus with an electrically conducting probe. Surface topography and electronic properties of a recording medium are simultaneously characterized using this apparatus.
shown in Fig. 1. If the size of the modified area is small enough, it can be regarded as a bit in high-density information storage. One expects to realize high-speed reading and writing if the modification is not associated with a change in surface morphology. One also expects to establish a system capable of rewriting if the modification is reversible. One of the actual recording media to realize the above mentioned new recording procedure is a polyimide Langmuir–Blodgett (LB) film. Sakai et al. reported a switching and memory phenomenon in a MIM junction with LB film as insulating layer [13]. It shows a reversible transition between high and low conductance states by application of pulse voltages. Each state has a corresponding threshold, and the state is maintained for application of voltage below the threshold, even 0 V. That is, this MIM junction shows a non-volatile memory effect. This phenomenon shows a clear material dependence, and does not depend on junction area [14]. Conventional polyimide is one of the typical materials showing this phenomenon. Formation of an MIM junction using a scanning probe and a polyimide LB film has been already achieved [15]. The polyimide LB film was deposited on Au(111) surface, and its thickness was 2.4 nm. Furthermore, formation of an ideal MIM junction using an AFM based apparatus has also been achieved [16,17]. The transition from low conductance state to high conductance state has been induced by application of pulse voltage to the polyimide LB film through the scanning probe. The area where the transition in conductance occurred has been observed as the conducting spot in the current image obtained using the conducting AFM probe. On the other hand, no change has been observed in the AFM image obtained simultaneously and independently. So, when a polyimide LB film is used as a recording medium in information storage with a scanning probe, a set of data bits composed of conducting spots can be formed without change in surface morphology. The reading rate depends on the rate in current detection
24
K. Takimoto
Fig. 2: Schematic drawing of an information storage system using an AFM based apparatus and a polyimide LB film. A local conducting area as a recording bit is not accompanied by a morphological change.
and the rate in probe scanning, as far as the probe traces the sample surface in contact. If the sample surface is very flat, the reading rate can be expected to be much higher than and not restricted by the mechanical resonance frequency of the cantilever. The size of the conducting spot is about 10 nm in diameter. The recording density can be estimated to be 1 Tbit/cm2 from this spot size when the conducting spots are considered to be data bits. Fig. 2 shows a schematic drawing of the information storage system using the AFM based apparatus and the polyimide LB film as the recording medium. The probe is scanned on the recording medium in contact. In the recording procedure, the array of pulse voltages according to a set of binary data is applied to the medium through the probe during scanning, and high-conductance regions are formed in the medium. After recording, current flow through the medium is detected with the probe during scanning and the obtained scan profile of current is converted to the set of binary data. In such a configuration, a pattern of encoded binary data consisting of more than a thousand conducting spots could be formed in a 2 × 2 µm2 area. And then reading the information back could be accomplished by converting the line scan profiles of the current image to bit patterns [18].
4. Size of data bits The size of the conducting spots was about 10–20 nm in diameter, independent of various kinds of probes. The probes are coated with metal thin films, typically Pt film. Fig. 3(a) shows a scanning electron microscope (SEM) image of a tip of a probe coated with Pt. The Pt film has grain structures, which are several 10 nm in diameter. The tip
Information storage using a scanning probe
25
Fig. 3: Scanning electron microscope images of a tip of a probe coated with Pt (a) and a tip of a probe fabricated by the replicating method (b). Reproduced from Ref. [19] with permission (© 1997 IEEE).
of the probe is also composed of some Pt grains. The size of conducting spots may be determined by the size of the grains. This seems to be a reason why the size of a conducting spot does not depend on the probes. To form a smaller conducting spot, a curvature in the tip of the probe has to be formed less than that in a grain of the coated metal. Yagi et al. fabricated a probe with a quite sharp tip by replicating a Si mold with pyramidal etch pits [19]. Fig. 3(b) shows an SEM image of a tip of a probe fabricated by the replicating method. Its curvature is estimated to be around 15 nm in radius. Fig. 4 shows a current image (a) and an AFM image (b) observed simultaneously using the replica probe [19]. Conducting spots observed in the current image were induced using the same probe. The sizes of the conducting spots are 10 nm or less. In addition, a monoatomic step structure in the Au(111) surface can be seen in the AFM image, which shows an increase in resolution of the AFM due to the decrease in the curvature in the tip of the probe. This shows that conducting spots with small sizes of 10 nm or less can be formed stably using the replica probe, and also shows that no degradation of the tip occurs during forming the conducting spots by application of pulse voltages.
5. Rate of reading and writing It has been reported that a conducting spot could be formed by application of a voltage pulse with 2 µs width, and the bit could be read within about 10 µs [18]. The main problem in high-speed writing is attributed to the stray capacitance around the MIM junction with the scanning probe. Actually, the writing with a 2 µs pulse was
26
K. Takimoto
Fig. 4: Current image (a) and AFM image (b) observed simultaneously using the replica probe. Conducting spots in the current image were induced by applying 13 V, 0.2 µs, rectangular voltage pulses using the same probe. Reproduced from Ref. [19] with permission (© 1997 IEEE).
achieved by a reduction of the stray capacitance. Furthermore, the transient response of the current observed during the spot formation by applying a 2 µs voltage pulse indicated that further reduction of the stray capacitance would make it possible to form a conducting spot with a voltage pulse shorter than 1 µs. The increase of the current by the contribution of the transition to the high-conductance state was fast enough. Actually, it is possible to form the conducting spot with a pulse voltage shorter than a 2 µs pulse. The conducting spots observed in Fig. 4(a) are formed by application of 0.2 µs voltage pulses [19]. The formation of conducting spots could be carried out by the application of pulse voltages of 25 ns in width under further optimization in stray capacitance, though such a result is not shown here. This indicates that a reading rate of about 40 Mbps may be achieved. For reading at a fast rate, high-speed scanning of the probe and high-speed detection of the current are necessary. To detect a low current at high scanning rate, a current amplifier is designed which has a small input capacitance and little gain for low frequencies. Using the current amplifier, it was demonstrated that the edge of the conducting spot could be detected within about 10 µs during scanning at a rate of 8 µm/s. This indicated that achieving a reading rate of about 100 kbps could be expected if the probe could be scanned with sufficient speed, typically 2 mm/s for an array of 10 nm spots [18]. High-speed scanning at a rate of 2 mm/s has been actually performed without damage to the medium. In this procedure, the probe was scanned so that the trajectory of its tip drew a circle. Although the scanning rate of 2 mm/s is much higher compared with that in a conventional AFM observation, the surface roughness of the polyimide LB film deposited on Au(111) is so small that high-speed scanning of the probe on it
Information storage using a scanning probe
27
Fig. 5: Current image of a part of the recorded 1 Mbits. The area is about 3.5 × 0.7 µm2 . (Reproduced from Ref. [20] with permission.)
seems to be possible. When both the detection rate of 10 µs per bit and the scanning rate of 2 mm/s are achieved simultaneously, a reading rate will be achieved of 100 kbps. It should be noticed that the severe requirements on mechanical resonance frequency of the cantilever are not necessary.
6. Error rate Stable formation of about one thousand data bits without the degradation of the tip of the probe has already been achieved in an area of 2 × 2 µm2 as shown in Fig. 4(a). For practical use, stable writing of a larger number of data bits must be confirmed, and error rate estimation is required. Yano et al. demonstrated 1 Mbit recording in an area of 40 × 80 µm2 without tip degradation [20]. Fig. 5 shows a current image of a part of the recorded 1 Mbits. The area is about 3.5 × 0.7 µm2 . An image similar to Fig. 5 can be acquired at any area where the voltage pulses were applied. In 1 Mbit recording, a transient response in current was monitored for each pulse application. A transition to the high-conductance state was confirmed when the current exceeded a predetermined value within a predetermined period of voltage application. That is, the case that the current exceeding this value was not observed within the period was regarded as a failure in the formation of the bit. Thus, the error rate was estimated to be 1.7 × 10−4 for 1 Mbit recording [20].
7. Conclusion The concept of an information storage system was demonstrated based on the formation of ideal MIM junctions using an AFM with a conducting probe. In practice, polyimide LB film is used as an insulating layer in an MIM junction, and a local conducting region in the polyimide LB film induced by applying a pulse voltage through the probe is considered to a recording bit. The reduction of bit size, the possibilities of fast rate reading and writing, and further stable bit writing were also shown. In this system, the surface topography of the medium is maintained even after writing data bits, which makes possible to read and write data bits at fast rates without severe requirements for the mechanical resonance frequency of a cantilever. Then an extremely flat surface of the recording medium is required over a wide range. It is essential to fabricate a sufficiently flat and large medium. LB film seems to be a suitable recording medium,
28
K. Takimoto
since the thickness of the film can be precisely controlled on a molecular scale. So, it is important to fabricate a flat and large substrate. Today, the recording density in some present storage devices reaches several tens gigabit per square inch. And it is growing at a rate of 60% every year. If today’s growing rate has to be kept, it is predicted that the recording density will be equal to the atomic density of the solid surface within twenty years. Molecular memory will also be realistic. Then, the storage devices will have to read and write the data bits with atomic resolution at a rate further exceeding that of present storages. The scanning probe method has enough potential concerning the resolution. However, the higher the recording density, the lower the reading rate becomes actually. The use of multiple probes and the parallel operation of them is an effective way to achieve a faster read and write rate[21–24]. However, essential breakthroughs on the problem of the reading rate also seems to be necessary.
References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22.
C.F. Quate, NATO ASI Ser. E: Appl. Sci. 184, 281 (1990). D.M. Eigler and E.K. Schweizer, Nature 344, 524 (1990); Appl. Phys. Lett. 68, 34 (1996). I.W. Lyo and P. Avouris, Science, 253, 173 (1991). H.J. Mamin, P.H. Guethner, and D. Ruger, Phys. Rev. Lett. 65, 2418 (1990) . H.J. Mamin and D. Ruger, Appl. Phys. Lett. 61, 1003 (1992). S. Hosaka, T. Shintani. M. Miyamoto, A. Kikukawa, A. Hirotsune, M. Terao, M. Yoshida, K. Fujita, and S. Kammer, J. Appl. Phys. 79, 8082 (1996). E. Betzig, J.K. Trautman, R. Wolfe, E, M. Gyorgy, P.L. Finn, M.K. Kryder, and C.-H. Chang, Appl. Phys. Lett. 61, 142 (1992). R.C. Barrett and C.F. Quate, J. Appl. Phys. 70, 2725 (1993). H. Kado and T. Tohda, Appl. Phys. Lett. 66. 2961 (1995). B.W. Chui, H.J. Mamin, B.D. Terris, T.D. Stowe, D. Rugar, and T.W. Kenny, Appl. Phys. Lett. 69, 2767 (1996). E.B. Cooper, S.R. Manalis, H. Fang, H. Dai, K. Matsumoto, S.C. Minne, T. Hunt, and C.F. Quate, Appl. Phys. Lett. 75, 3566 (1999). G. Binnig, M. Despont, U. Drechsler, W. Häberle, M. Lutwyche, P. Vettiger, H.J. Mamin, B.W. Chui, and T.W. Kenny, Appl. Phys. Lett. 74, 1329 (1999). K. Sakai, H. Matsuda, H. Kawada, K. Eguchi, and T. Nakagiri, Appl. Phys, Lett. 53, 1274 (1988). K. Takimoto, K. Yano, K. Hatanaka, K. Eguchi, and T. Nakagiri, Oyo Butsuri 63, 470 (1994) K. Takimoto, H. Kawade, E. Kishi, K. Yano, K. Sakai, K. Hatanaka, K. Eguchi, and T. Nakagiri, Appl. Phys. Lett. 61, 3032 (1992). K. Yano, M. Kyogaku, R. Kuroda, Y. Shimada, S. Shido, H. Matsuda, K. Takimoto, O. Albrecht, K. Eguchi, and T. Nakagiri, Appl. Phys. Lett. 68, 188 (1996). K. Yano, R. Kuroda, Y. Shimada, S. Shido, M. Kyogaku, H. Matsuda, K. Takimoto, K. Eguchi, and T. Nakagiri, J. Vac. Sci. Technol. B 14, 1353 (1996). K. Takimoto, R. Kuroda, S. Shido, S. Yasuda, H. Matsuda, K. Eguchi, and T. Nakagiri, J. Vac. Sci. Technol. B 15, 1429 (1997). T. Yagi, Y. Shimada, T. Ikeda, O. Takamatsu, H. Matsuda, K. Takimoto and Y. Hirai, Proc. Tenth Annual International Workshop on Micro-Electro-Mechanical Systems, p. 129 (1997). K. Yano and T. Ikeda, Appl. Phys. Lett. 80, 1067 (2002). M.I. Lutwyche, M. Despont, U. Drechsler, U. Dürig, W. Häberle, H. Rothuizen, R. Stutz, R. Widmer, G.K. Binnig, and P. Vettiger, Appl. Phys. Lett. 77, 3299 (2000). S.C. Minne, G. Yaralioglu, S.R. Manalis, J.D. Adams, J. Zesch, A. Atalar, and C.F. Quate, Appl.
Information storage using a scanning probe
23. 24.
29
Phys. Lett. 72, 2340 (1998). S.C. Minne, J.D. Adams, G. Yaralioglu, S.R. Manalis, A. Atalar, and C.F. Quate, Appl. Phys. Lett. 73, 1742 (1998). Y. Shimada, T. Yagi, T. Yamazaki, S. Shido, H. Matsuda, and K. Takimoto, Technical Digest of the 16th Sensor Symposium, p. 273 (1998).
Nanotechnology and Nano-Interface Controlled Electronic Devices Editors: M. Iwamoto, K. Kaneto and S. Mashiko © 2003 Elsevier Science B.V. All rights reserved
CHAPTER 3
Single electron tunneling organic devices Tohru Kubota a,* , Shiyoshi Yokoyama a , Tatsuo Nakahama a , Shinro Mashiko a , Yutaka Noguchi b , and Mitsumasa Iwamoto b a Kansai
Advanced Research Center, Communications Research Laboratory, 588-2 Iwaoka, Nishi-ku, Kobe 651-2492, Japan b Department of Physical Electronics, Tokyo Institute of Technology, 2-12-1 O-okayama, Meguro-ku, Tokyo 152-8552, Japan * E-mail:
[email protected]
1. 2.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Molecules and samples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Molecules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. Sample structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3. I –V measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. Current–voltage characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Single electron tunneling characteristic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Organic molecule as Coulomb island . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3. Light-irradiated single electron tunneling characteristics . . . . . . . . . . . . . . . . . . . 4. Future prospect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
31 32 32 32 33 34 34 37 38 39 39
1. Introduction In recent years, many research studies have been done along with the rapid progress of nano-fabrication technology in the field of nano-electronics [1]. Until now, much effort has been done on the fabrication of novel devices based on physics of quantum mechanics, principally using the nano-fabrication technique developed in semiconductor device technology [2]. As a result, single electron tunneling (SET) devices using small particles in their systems have been successfully prepared. Nano-fabrication technology developed in the field of semiconductor device technology may lead to a new way to electronics and many novel electronic devices such as high-density memory devices, high-speed low-power switching devices, high-sensitive electrometer devices and others will be produced in the near future. As such, it is needless to say that the research along with this trend is important. However, this is not sufficient. The study of observing
32
T. Kubota et al.
specific functions of organic molecules and applying these functions to electronic and optical molecular devices is of crucial help, because one can realize novel functional devices only by using organic molecules, without using the maturing nano-fabrication technology in the field of semiconductor device technology. In this chapter, single electron tunneling (SET) devices using organic molecules prepared by the Langmuir– Blodgett (LB) technique is briefly introduced. For the realization of SET devices, it is necessary to design the device system so that the one-electron charging energy of e2 /2C is greater than the thermal energy kT [2]. In this sense, so-called small particles, whose size is less than several nm, must be introduced into the molecular systems to be used. However, there are many difficulties to do this, e.g., synthesis and others. Overcoming the difficulties, an attempt to use an organic mono-molecule as a so-called Coulomb island has been successfully made by the present authors [3–6]. In the following, single electron tunneling devices prepared using organic molecules is described.
2. Molecules and samples 2.1. Molecules As the molecule used for fabricating the device, dendrimer molecules (Rh-G2), which have called much attention in the field of organic synthesis, were used as the Coulomb island. Polyimide LB film, which shows excellent insulating properties [7], was used as tunneling barrier in the device. The chemical structures of Rh-G2 and PI are shown in Fig. 1. The polyimide LB film functions as electron tunneling barrier, where the monomolecular film thickness of 0.4 nm can be controlled by the LB method. The dendrimer molecule Rh-G2 used as the Coulomb island has a spherical shell molecular structure, i.e. electrically insulating CH chains enclose rhodamine dye molecules [8]. That is, the dye molecules located at the center are electrically isolated from their surroundings. The use of such organic molecules facilitates the preparation of single electron tunnel devices. 2.2. Sample structure [3,4] Fig. 2 shows the sample structure of a SET device prepared using rhodamine dendrimer (Rh-G2) molecules, where an Rh-G2 molecule is assumed to function as a Coulomb island. Au was evaporated on glass substrate with a thickness of about 100 nm, and the evaporated electrode was used as a bottom electrode of the SET device. Onto this evaporated electrode, 13 to 31 layers of polyimide LB film were deposited by the LB technique to form an electron tunneling layer, in the same manner as that in our previous study [7]. Then, polyimide LB film mixed with rhodamine dendrite molecules was deposited by mono-layer deposition to form a single layer working as a Coulomb island. The mixing ratio of PI : Rh-G2 was 500 : 1 in molar ratio. From the experimental surface pressure–area (F–A) isotherm of the mixture monolayer film, a number of about 1000 molecules of Rh-G2 was estimated to be present within an area of about
Single electron tunneling organic devices
33
Fig. 1: Chemical structure of (a) polyimide (PI) and (b) dendrite polymer (Rh-G2).
1 µm2 . After the deposition of the mixing monolayer, 20–30 layers of polyimide LB films were again deposited as an upper electron tunneling layer. Finally, an Au electrode was evaporated with a thickness of 50–100 nm to form a top electrode. The working area of the resulting junction was about 50 × 100 µm (see Fig. 2). Furthermore, for the measurement of I –V characteristics under photoillumination, junctions with transparent indium–tin-oxide (ITO) electrodes were also fabricated. 2.3. I–V measurement The electrical resistance of the prepared devices was several hundreds of M to several tens of G. Thus, current–voltage (I –V ) measurement was performed using a 2-terminal method by applying a step voltage under constant temperature in a cryostat (Cryokelvin CG308SCPR; Nagase Electronics). The measurement temperature was between room temperature and 5 K.
34
T. Kubota et al.
Fig. 2: Sample structure of a molecular single electron tunneling device.
3. Current–voltage characteristics 3.1. Single electron tunneling characteristic Fig. 3(a) shows the I –V characteristics of an Au/PI25/PI+Rh-G2/PI30/Au device at a temperature of 5.2 K. The voltage step with constant spacing is clearly seen. This is the characteristic of SET devices, and the step spacing is given by e/C. Here C is the capacitance between the Coulomb island and the electrode [2]. The step spacing e/C is about 100 mV. In order to further clarify the step structure observed in the I –V characteristic, the dV /dI –V characteristic was plotted in Fig. 3(b). For both positive and negative voltage, peaks of dV /dI are seen at 50, 150 and 250 mV with an equal step voltage spacing of 100 mV.
Single electron tunneling organic devices
35
Fig. 3: (a) Typical I –V characteristics of metal/organic SET layer/metal junctions. (b) Typical dV /dI –V characteristics of metal/organic SET layer/metal junctions.
36
T. Kubota et al.
Fig. 4: Arrhenius plots (I –1/ T charcteristic) of an organic SET device.
Fig. 4 shows the temperature dependence of current flowing through the junction. The current flow decreases with the decrease of temperature in the region from room temperature to 50 K. This indicates that a thermally activated conduction type current flows through the junction. The thermal activation energy at a temperature higher than 50 K is estimated as less than 30 meV. On the other hand, the electrical current density is nearly constant at a temperature lower than 50 K. This indicates that a tunneling-conduction type current flows through the junction in this temperature region. These results reveal that both thermally activated conduction and tunnelingconduction currents are allowed to flow in this device. It is also suggested that the thermally activated conduction type current dominates at temperatures higher than room temperature, whereas this type of current decreases as temperature is getting lower and lower. Eventually, the tunneling-conduction type current becomes the main contributor. Of interest is that a single electron tunneling characteristic is found at lower temperature as shown in Fig. 3. Fig. 5 shows the I –V characteristics at various temperatures. At a temperature lower than 50 K, tunneling-conduction type current is observed, and the step voltage can be seen. The voltage step width is the same as that observed at 5.2 K. The position of the step voltage is not dependent on the temperature. As described above, in the junctions using Rh-G2 dendrite molecules, single electron tunneling behavior is observed. From the theoretical side, the charging energy e2 /2C associated with one electron tunneling must be greater than the thermal energy kT [2]. A voltage step due to single electron
Single electron tunneling organic devices
37
Fig. 5: Typical I –V characteristics of an organic SET device at various temperatures.
tunneling is about 50 meV for the device prepared here. We may expect that junctions showing a single electron tunneling characteristic is produced in the near future. 3.2. Organic molecule as Coulomb island It is interesting here to discuss whether the Rh-G2 molecule introduced in the junction actually functions as a Coulomb island or not. For this purpose, the size of the Coulomb island is estimated. The size of the Coulomb island is a dominant factor to explain the I –V characteristic. The size can be estimated from the spacing of step voltage of e/C in the I –V characteristic. The capacitance C of a spherical conductor with radius r that is separated by a distance d from a planar electrode is given by C = 4πεr ε0 F
with
F ≈ (1/r − 1/2d)−1 ,
(1)
under the assumption r d. Therefore, assuming the observed voltage step spacing is given by V , the radius r of the Coulomb island can be expressed as 4πεr ε0 1 −1 . (2) r= V + e 2d Using this equation, the size r of the Coulomb island is estimated as 3.8 nm for V = 100 mV, εr = 3 (polyimide), and d = 10 nm (25 layers of PI LB film). The size of an Rh-G2 molecule is speculated to be about 1–2 nm in radius from computer
38
T. Kubota et al.
Fig. 6: Typical I –V characteristics of a SET device under photoillumination.
simulation and F–A isotherm measurement. There is a discrepancy between the size estimated from I –V characteristics and that from the computer simulation, possibly due to the assumptions made in the calculation. This discrepancy is within the constraints of our estimation. As mentioned above, using dendrite molecules designed by molecular ensemble, it is possible to prepare a SET device. 3.3. Light-irradiated single electron tunneling characteristics When the fabricated samples are irradiated by light, electrons may be excited from the dye molecules and the single electron tunnel conduction mechanism may be changed. Further, the tunnel barrier height may be changed by the charge induced in the space charge layer. In other words, it is possible to control the single electron tunneling process by light illumination. Fig. 6 shows the I –V characteristics of a SET device under white light irradiation. As shown in the figure, the current decreases, but no change is observed in the position and the spacing of step voltages. These are specific characteristics seen in the junctions using dendrimer molecules. These results suggest that the changes induced by light irradiation did not originate from the Rh-G2 molecules working as the Coulomb island but from the PI tunnel barrier [9].
Single electron tunneling organic devices
39
Fig. 7: Image of a double tunneling type molecular SET device.
4. Future prospect By the use of core type dendrimer molecules created by molecular ensemble as Coulomb island, it is possible to fabricate SET devices. As has already been described, when organic molecules are used as the Coulomb island, the specific properties of the molecule will be added to the single electron tunnel characteristics. By choosing molecules, double type SET devices and others that can be used in e.g. detection of electromagnetic waves, light detection, etc. [10] will be produced. Furthermore, arranging the above double tunneling type device in another organic molecular matrix as shown in Fig. 7, the tunneling characteristics will be controllable by the field of the matrix with external stimuli, such as optical light or electrical field. As described above, by utilizing a molecular ensemble, building up organic electronics and molecular field control type electronics will be possible.
References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.
K.K. Likharev, Proc. IEEE 87, 606 (1999). J.H. Fendler, Nanoparticles and Nanostructured Fulms, (WILEY-VCH, Weinheim) Ch. 15 (1998). Y. Noguch, Y. Majima, M. Iwamoto, T. Kubota, S. Yokoyama, T. Nakahama, and S. Mashiko, IEICE Trans. Electron. E83-C, 1076 (2000). T. Kubota, S. Yokoyama, T. Nakahama, S. Mashiko, Y. Noguch, Y. Majima, and M. Iwamoto, Thin Solid Films 393, 379 (2001). Y. Noguchi, Y. Majima, and M. Iwamoto, J. Appl. Phys. 90, 1368 (2001). Y. Noguchi, M. Iwamoto, T. Kubota, and S. Mashiko, J. Appl. Phys. 92, 1174 (2002). M. Iwamoto, T. Kubota, and M. Sekine, J. Phys. D 23, 575 (1990). S. Yokoyama, T. Nakahama, A. Otomo, and S. Mashiko, Chem. Lett. 11, 1137 (1997). E. Itoh, Y. Niwa, and M. Iwamoto, Thin Solid Films 284, 545 (1996). T. Fujisawa and S. Tarucha, Jpn. J. Appl. Phys. 36, 4000 (1997).
Nanotechnology and Nano-Interface Controlled Electronic Devices Editors: M. Iwamoto, K. Kaneto and S. Mashiko © 2003 Elsevier Science B.V. All rights reserved
CHAPTER 4
Spatial light confinement and laser emission from a gain medium containing dendrimer Shiyoshi Yokoyama a,b and Shinro Mashiko a a
Communications Research Laboratory and b PRESTO, Japan Science and Technology Corporation (JST), 588-2, Iwaoka, Nishi-ku, Kobe 651-2492, Japan
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
41 43 43 47 48
1. Introduction Since the first reports on organic and polymeric materials that showed high optical gain and stimulated emission properties, there has been growing interest in exploiting laser applications. This growing interest is because of their wide wavelength tunability and processing flexibility in solutions [1–3], films [4–7], and fibers [8,9]. Dendritic macromolecules, called dendrimers, are a new category of hyper-structured material [10]. Their long branching chain and the high degree of control over regular molecular weight created a three-dimensional structure that is roughly spherical or globular [11]. In optical applications, radiative action from the high-gain medium, containing small particles such as dendrimers, may be altered significantly under coherent optical excitation [12]. In the study reported here, we found that using a homogeneous gain medium containing dendrimers increases the stimulated emission efficiency and facilitates fine-tuning the laser modes. We also identified an optical input–output threshold behavior above which laser emission with a linewidth of less than 0.1 nm was observed, even though our optical system lacked a real optical cavity. The input threshold energy from the gain medium was much smaller than the energy from the pure dye solution. Organic laser dyes typically show a large fluorescence yield ranging from about 0.6 to near the optimum 1.0. In spite of this large yield, the dye concentration in the dye
Fig. 1: Details of materials and optical experiment. (A) Chemical structure of dendrimers, 1 and 2, and DCM. (B) Schematic of stimulated emission experiment setup. (C) Photograph of laser emission from DCM/dendrimer solution showing the red output beam radiating some distance from the cuvette facet. Inset, interference pattern of output beam after passing through cross-diffraction grating.
42 S. Yokoyama and S. Mashiko
Spatial light confinement and laser emission from a gain medium containing dendrimer
43
laser medium must be kept low to achieve highly efficient spontaneous emission. At higher concentrations, molecular aggregation, which forms dimers or higher aggregates, almost completely suppresses the fluorescence [13]. This is in contrast with the general tendency of π-electron-conjugated chromophores, which easily aggregate to form complex structures. Therefore, a dye concentration of less than 10−3 mol/l is generally used in laser operations.
2. Experiment In order to obtain a higher gain medium for stimulated emission, we used a dendrimer, which encapsulates the laser dye inside and increases the dye concentration with little fluorescence quenching. The dendrimer used in this study was poly(amidoamine) with 64 hydroxyl-terminated groups (Starburst® PAMAM-OH dendrimer, Dendritech, Inc.) 1 [14], and the laser dye used was 4-(dicyanomethylene)-2methyl-6-(4-dimethylaminostyril)-4H-pyrane (DCM) (Fig. 1A). The DCM-doped dendrimer was obtained by mixing DCM and a dendrimer in a methanol solution. The DCM concentration was varied between 2.0 and 12.0 mM, whereas the DCM/dendrimer ratio was kept constant. As long as the DCM/dendrimer ratio was kept at 2.0, the fluorescence intensity increased as the level of the DCM concentration increased. However, the saturation concentration of DCM in methanol was less than 1.0 mM. The dendrimer was thus a good host for the DCM, increasing its solubility and yielding high emission efficiency. We used a nitrogen laser (337 nm, pulse duration 4.0 ns, repetition rate 10 Hz) as the excitation source for stimulated emission experiments. The excitation intensities were varied between 0 and 20 µJ/pulse. A cylindrical lens focused the excitation beam into a stripe, 200 µm × 5 mm on a quartz cuvette, which contained either the DCM and dendrimer mixture or a pure DCM solution in methanol (Fig. 1B). The emissions guided along the excitation stripe were collected from the side of the cuvette using a round lens, and were then spectrally analyzed using a spectrometer and a charge coupling device (CCD).
3. Results The emission spectra from the DCM/dendrimer ([DCM] = 2.0 mM) solution as a function of the excitation intensity gradually narrowed from relatively low excitation intensities up to 15 µJ/pulse (Fig. 2). The emission intensity, Ise , grew exponentially with I (Fig. 2, inset), which is consistent with amplified spontaneous emission (ASE) [15]. The gain guiding ASE process in an excitation-stripe is characterized by Ise = β(e(γ −α)L − 1), where β is a constant that depends on the excitation geometry, L is the excitation-stripe length, and γ and α are the optical gain and loss coefficients, respectively. Because γ is linear with excitation intensity in the simple approximation, ln(Ise ) ∝ I for the ASE process is in agreement with the experimental results fitted in Fig. 2, inset. When the excitation intensity became high, the emission spectrum
44
S. Yokoyama and S. Mashiko
Fig. 2: Emission spectra from DCM/dendrimer (= 2/1, [DCM] = 2.0 mM) solution in methanol at increasing excitation intensities. Spectrum a was magnified by 20. Spectrum b was magnified by 3. Inset: the line through emission intensities against excitation intensities was fit using ASE.
collapsed into multiple narrow peaks (Fig. 3), and the emission intensity increased linearly as the excitation intensity increased. A clear threshold behavior in the Ise vs. I plot (Fig. 4A) and a second decrease in the linewidth at higher excitation intensity (Fig. 4B) indicated the onset of laser action. The strongly modulated spectrum, with numerous peaks that were evenly spaced, clearly indicated the resonant cavity modes. The resonant peaks had a linewidth that was less than 0.1 nm. The laser beam was highly polarized in a longitudinal direction. The polarization ratio P = Ise,⊥ /Ise, was about 150, where Ise,⊥ and Ise, are emission intensities with polarization in longitudinal and lateral directions. The output beam was easily visible, as shown in Fig. 1C. The interference pattern is clearly projected after passing through a diffraction grating. This indicates that the output beam was coherent, though our optical setup lacked a real optical cavity. The distance between the resonance peaks can be given by λ = λ2 /(2n L), where n is the refractive index and L is the optical length of the resonator or cavity [16]. Using the measured peak separation of λ = 0.85 nm in Fig. 3, we estimated L to be 142 µm. Laser emission requires optical feedback, e.g., reflections from the cavity edges, and the cuvette sides may have become a reflecting mirror. However, such a reflection (10 mm separation) was inconsistent with the optical length estimated from the emission spectrum. We attributed the resonant mode of the output beam to the spatial confinement of the emitted light in the slab laser. To clarify this spatial confinement, the near-field
Spatial light confinement and laser emission from a gain medium containing dendrimer
45
Fig. 3: Spectrum of output laser beam from DCM/dendrimer (= 2/1, [DCM] = 2.0 mM) solution in methanol at excitation intensity above the threshold. Inset: beam-intensity dependence of near-field pattern measured at facet of cuvette in lateral a direction.
pattern of the laser beam was measured using side-imaging spectroscopy. We placed a 10 µm pinhole near the face of the cuvette to monitor the emission intensity at a given position (Fig. 3, inset). The laser intensities were concentrated on the face of the cuvette approximately 140 µm in the lateral section. Assuming that the excitation stripe can act as a waveguide on the cuvette, forming a 140 µm-long slab laser, this optical length is in excellent agreement with the resonant mode separation of L = 142 µm estimated from the spectrum. This indicates that the emitted light was confined by gain guiding within the stripe, resulting in laser feedback. The dendrimer is a good host to encapsulate DCM; its concentration is increased up to 12 mmol/l. Since the DCM/dendrimer has a high emission efficiency at various concentrations, laser emission intensities increased as the DCM/dendrimer concentration was increased as shown in Fig. 4. More importantly, the lasing threshold intensity became much lower as the concentration of the DCM/dendrimer was increased. DCM/dendrimer was found to be a high-gain medium for laser emission. The explanation of how the dendrimer acts as a small particle for the laser feedback is not obvious. However, optical gain in the homogeneous medium may, in part, have occurred when dendrimer behaves as a small particle. The surface unit of dendrimer 2 was modified with a hydrogen-bonding unit, such as a N-tert-butoxycarbonyl-Lphenylalanine. Because of this modification, dendrimer 2 achieved a highly dense hydrogen-bond shell with solid-state characteristics [17], becoming a molecular particle with a diameter of about 5 nm. Lasing actions from the DCM solution with or without
46
S. Yokoyama and S. Mashiko
Fig. 4: (A) Dependence of emission intensity on excitation intensity for DCM/dendrimer (= 2/1) solutions at various concentrations. Concentrations of DCM: 2.2 mM (), 4.4 mM ( ), and 13 mM (•). (B) Dependence of emission linewidth on excitation intensity.
dendrimer were compared in Fig. 5. In this case, dendrimer does not encapsulate DCM inside because of its hard shell structure. The threshold intensity of lasing became very small when the DCM solution contained dendrimer. It seems that dendrimer behaves as a scattering center, increasing the optical gain. In fact, a demonstration of stimulated emission from an inhomogeneous scattering medium, such as a microcavity
Spatial light confinement and laser emission from a gain medium containing dendrimer
47
Fig. 5: Dependence of emission intensity on excitation intensity for (•) DCM (1.0 mM) and dendrimer (2) mixture and ( ) pure DCM (1.0 mM) solution.
or mirrorless laser [18], is an excellent example of how photochemical and optical technologies can be used for emitting materials to make the optical devices simple and small [2,19]. These systems focused on the random laser, where the feedback mechanism of the laser emission is attributed to multiple scattering by the particles which keeps the light inside the scattering media for an extended period [19]. In these studies, though a submicrometer particle became a strong optical scattering center, it may also have induced a large optical loss. Since, in our experiment, the diameter of the dendrimer was much smaller than the optical wavelength, the optical gain was large in the DCM/dendrimer media, while the optical loss due to the passive scattering was inhibited by gain guiding. The significantly lower threshold of the laser action from the DCM and dendrimer mixture, compared with that from the pure DCM solution, provides clear evidence that emitted light can spend a great amount of time inside the gain medium. It is obvious that dendrimer produced multiple light scattering in the homogeneous gain medium. We are tempted to attribute the phenomena to photon localization providing an optical feedback for the high-gain laser dye.
4. Conclusion In conclusion, we have described the spatial light confinement in DCM/dendrimer media, which generated a resonant mode in lasing action. The laser emission was characterized by (i) the appearance of resonance peaks with lines less than 0.1 nm; (ii) a clear threshold excitation intensity for lasing; (iii) a high degree of polarization above the threshold; and (iv) increased coherency. Encapsulating the laser dye into
48
S. Yokoyama and S. Mashiko
the dendrimer increased the optical gain of the emitting media. More importantly, the emission process by the resonant modes is applicable to laser-like emission. The results showed that a random optical system consisting of emitting materials, optical excitation, and cavity modes, can be used to fine-tune mirrorless optical devices, even in small-device applications.
References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.
11. 12. 13. 14. 15. 16.
17. 18.
19.
S. Qian, J.B. Snow, H. Tzeng, and R.K. Chang, Science 231, 486 (1986). N.M. Lawandy, R.M. Balachandran, A.S.L Gomes, and E. Sauvain, Nature 368, 436 (1994). S.V. Frolov, Z.V. Vardeny, and K. Yoshino, Phys. Rev. B 57, 9141 (1988). F. Hide, M.A.Díaz-García, B.J. Schwartz, M.R. Andersson, Q. Pei, and A. J. Heeger, Science 273, 1833 (1996). V.G.K. Fozlov, V. Bulovi´c, P.E. Burrows, and S.R. Forrest, Nature 389, 362 (1997). N. Tessler, G.J. Denton, and H. Frend, Nature 382, 695 (1996). J.H. Schon, Ch. Kloc, A. Dodabalapur, and B. Batlogg, Science 289, 599 (2000). S.V. Frolov, A. Fujii, D. Chinn, Z.V. Vardeny, K. Yoshino, and R.V. Gregory, Appl. Phys. Lett. 72, 2811 (1998). F. Marlow, M.D. Mcgehee, D. Zhao, B.F. Chmlka, and G.D. Stucky, Adv. Mater. 11, 632 (1999). D.A. Tomalia, A.M. Naylor, and W.A. Goddard III, Angew. Chem. Int. Ed. 29, 138 (1990); in Advances in Dendrite Macromolecules, edited by: G.E. Newkome (JAI Press, Greenwich, CT, 1994); J.M. Fréchet, Science 263, 1710 (1994). M.L. Mansfield and L.I. Klushin, Macromolecules 26, 4262 (1993); C.L. Jackson, H.D. Chenzy, F.P. Booy, B.J. Drake, D.A. Tomalia, B.J. Bauer, and E.J. Amis, Macromolecules 31, 6259 (1998). A. Otomo, S. Yokoyama, T. Nakahama, and S. Mashiko, Appl. Phys. Lett. 77, 3881 (2000). G. Jones II, in Dye Laser Principle, edited by: F.J. Duarte and L.W. Hillman (Academic Press, San Diego, 1990) Ch. 7. D.A. Tomalia, V. Berry, M. Hall, and D.M. Hedstrand, Macromolecules 20, 1164 (1987). A.E. Siegman, in Laser (Univ. Sci. Book, CA, 1986) Ch. 13. R.C.H.P. Weber and R. Ulrich, Appl. Phys. Lett. 19, 38 (1971); R.C. Polson, G. Levina, and Z. Vardeny, Appl. Phys. Lett. 76, 3858 (2000); E. Siegman, in Laser (Univ. Sci. Book, CA, 1986) Ch. 11. J.F.G.A. Jansen, E.M.M. de Brabander-van den Berg, and E.W. Meijer, Science 266, 1226 (1994). D. Wiersma, Nature 406, 132 (2000); H. Cao, Y.G. Zhao, H.C. Ong, S.T. Ho, J.Y. Dai, J.Y. Wu, and R.P.H. Chang, Appl. Phys. Lett. 73, 3656 (1998); H. Cao, J.Y. Xu, E.W. Seelig, and R.P.H. Chang, Phys. Rev. Lett. 76, 2997 (2000). R.M. Balachandran, N.M. Lawandy, and J.A. Moon, Opt. Lett. 22, 319 (1997); D.S. Wiersma, P. Bartolini, A. Lagendijk, and R. Righini, Nature 390, 671 (1997); W. Kim, V.P. Safonov, V.M. Shalaev, and R.L. Armstrong, Phys. Rev. Lett. 82, 4811 (1999).
Nanotechnology and Nano-Interface Controlled Electronic Devices Editors: M. Iwamoto, K. Kaneto and S. Mashiko © 2003 Elsevier Science B.V. All rights reserved
CHAPTER 5
Control of molecular selective-assembling on metal surface Takashi Yokoyama a , Toshiya Kamikado b, Shiyoshi Yokoyama b , Yoshishige Okuno b , and Shinro Mashiko b a National
Institute for Materials Science, 2268-1 Shimo-shidami, Moriyama-ku, Nagoya 463-0003, Japan b Communications Research Laboratory, 588-1 Iwaoka, Nishi-ku, Kobe 651-2401, Japan
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. H2 -TBPP and Au(111) as basic molecule and substrate . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. Nonplanar conformation and orientational ordering of H2 -TBPP on Au(111) . . . . . . . 4. Selective aggregation of cyanophenyl-substituted porphyrins on Au(111) . . . . . . . . . . . 5. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
49 50 50 55 60 60
1. Introduction The realization of molecular nanodevices with advanced functions requires the development of new approaches to construct desired molecular nanostructures [1,2]. Supramolecular approach starting from molecular building blocks can lead to controlled structures [3], which is achieved by selective and directional intermolecular interactions. When non-covalent intermolecular interactions such as hydrogen bonding are introduced into functional molecules, the selective intermolecular interaction results in the controlled formation of molecular nanostructures, which have yielded exclusively crystals or dissolved structures [3]. To adapt the functional supramolecular structures to nanodevices, it should be necessary for the supramolecular structures to be supported on suitable substrates and at suitable positions. On substrate surfaces, atomic-scale investigation of adsorbed molecules has been enabled by using scanning probe microscopy, particularly scanning tunneling microscopy (STM) [4]. In particular, recent advance of high-resolution STM imaging allows one to directly determine their arrangement, configuration, and conformation of individual largish molecules on surfaces [5–9]. Based on the submolecular-resolution STM studies
50
T. Yokoyama et al.
and on theoretical calculations, several groups have been reported specific interactions of adsorbed molecules on metal surfaces [10–12], revealing self-assembled aggregation on surfaces. Whereas these surface-supported supramolecular structures have been directly observed using STM, the further control of their size and shape on surfaces should become a next step for realizing molecular nanodevices. In this work, we demonstrate selective assembly of supramolecular aggregates with controlled size and shape on a gold surface by modifying substituent structures.
2. H2 -TBPP and Au(111) as basic molecule and substrate The basic molecule used in this study is 5,10,15,20-tetrakis-(3,5-di-tertiarybutylphenyl) porphyrin (H2 -TBPP), which has a free-base porphyrin core and four di-tertiarybutylphenyl (tBP) substituents, as shown in Fig. 1(a). An ideal shape of H2 -TBPP exhibits a planar macrocyclic conformation of the central porphyrin through 60°–90° rotation of the phenyl rings with respect to the phenyl mean plane, as shown in Fig. 1(b), obtained from semi-empirical molecular orbital calculations. By the bulky tBP substituents, the aromatic π system of the central porphyrin should be sterically decoupled to the substrate surface even after adsorption, fulfilling the requirements for the molecular nanoelectronic or optoelectronic devices [5,6,13]. To understand the conformation and arrangement of the H2 -TBPP molecules, all experiments were performed in an ultrahighvacuum (UHV) chamber with a low-temperature STM. A Au(111) surface was used as a substrate because of its inertness and its properties of reconstruction. The atomically clean surface of Au(111), which was formed by deposition of Au on mica in UHV, was prepared by repeated cycles of Ar+ sputtering and annealing at 700 K. Fig. 2 shows an STM image of the reconstructed Au(111) at 63 K. The reconstruction of the Au(111) surface results from alternating face centered cubic (fcc) and hexagonal close packed (hcp) stacking of the surface atoms with respect to the bulk lattice, and long-range “herringbone” patterns are formed by periodic rotations of the uniaxial domains [14,15]. In addition, each elbow of this pattern contains a dislocation of the surface lattice, and the preferential nucleation of adsorbates at the elbows has been observed in various systems [12,16].
3. Nonplanar conformation and orientational ordering of H2 -TBPP on Au(111) The H2 -TBPP molecules were deposited onto the Au(111) surface at room temperature by sublimation from a Knudsen cell in UHV. The sample was then subsequently transferred to the cooled STM stage for direct observation. Fig. 3(a) shows an STM image of the Au(111) surface at 63 K after a small amount of the H2 -TBPP deposition. In this image, most of the molecules are located at the elbows of the surface reconstructed patterns, revealing regular arrays of isolated single molecules. In this system, it should be noted that the selective molecular positioning allows direct imaging of isolated single molecules without intermolecular interactions. Fig. 3(b) shows the high-resolution STM
Control of molecular selective-assembling on metal surface
51
Fig. 1: (a) A structure of H2 -TBPP, which includes the central free-base porphyrin and four tBP substituents. (b) Calculated conformation of H2 -TBPP, obtained from the semi-empirical molecular orbital method. The about 65° rotations of phenyl rings results from steric hindrance between the porphyrin and phenyl rings.
image of a single H2 -TBPP molecule on the Au(111) surface at 63 K, which exhibits four paired lobes surrounding two oblong protrusions. From the molecular dimension, we assign each lobe as one of the tertiary-butyl substituents, and the appearance of
52
T. Yokoyama et al.
Fig. 2: STM image (70 nm × 53 nm) at 63 K of an Au(111) surface, in which bright stripes are associated with domain walls between fcc and hcp stackings, as indicated by dashed lines, and the herringbone patterns are formed by periodic rotations of the uniaxial domains.
the paired lobes suggests that the phenyl rings are oriented close to the porphyrin macrocyclic plane, different from the ideal conformation of H2 -TBPP shown in Fig. 1(b). In this high-resolution image, each of the paired lobes consists of brighter and darker ones, and the lateral distance is estimated to be about 4.5 Å. By comparing the STM results with the molecular model, we have derived the dihedral angle between the porphyrin and phenyl rings to be about 20°, where the four phenyl rings are alternately rotated with respect to the porphyrin mean plane. The rotations of the phenyl–porphyrin bonds have been reported for adsorbed Cu-TBPP molecules on several metal surfaces, which depend on the substrate structures [6,9]. These results indicate that the rotational flexibility of the phenyl–porphyrin bonds allows the tertiary-butyl substituents to fit into the surface geometry, leading to the conformational changes. In the STM image of Fig. 3(b), the most distinguishing feature is that the internal structure of the central porphyrin has been resolved in the STM image, which is composed of the two oblong protrusions. We observed that the STM images were independent of the bias polarity, suggesting that the atomic structure was mainly contributed in the STM image, compared with the electronic structure. Thus, the oblong protrusions should be associated with the nonplanar deformation of the central porphyrin, induced by the rotations of the phenyl-based substituents. To confirm the nonplanar macrocyclic conformation, we performed the semi-empirical molecular orbital calculations with the AM1 hamiltonian [18]. Fig. 3(c) shows the calculated conformation of H2 -TBPP with fixed 20° rotations of the four phenyl rings with respect to the central porphyrin. The relaxed structure shows that the 20° alternate rotations
Control of molecular selective-assembling on metal surface
53
Fig. 3: (a) STM image (20 nm × 20 nm) of the Au(111) surface at 63 K after a small amount of the H2 -TBPP deposition. A regular array of single molecules results from preferential adsorption at the elbows of the herringbone patterns. (b) High-resolution STM image (2.1 nm × 2.1 nm) at 63 K of a single H2 -TBPP molecule, which is composed of four paired lobes surrounding two oblong protrusions. (c) Top view of calculated macrocyclic conformation of the H2 -TBPP molecule with 20° alternate rotations of the phenyl rings, obtained using semi-empirical molecular orbital calculations. A saddle-shaped nonplanar deformation of the central porphyrin is induced by the steric interactions with the rotated phenyl rings.
of the phenyl rings induce nonplanar deformation of the porphyrin macrocycle, while the ground state conformation is formed through about 65° rotations as shown in Fig. 1(b). In this nonplanar conformation, the saddle-shaped deformation of the central
54
T. Yokoyama et al.
porphyrin is characterized by alternately tilting of pyrrole rings above and below the mean plane, which should be induced by steric hindrance with the rotated phenyl rings. The maximum deviation of the porphyrin macrocycle from the mean plane is estimated to be about 0.95 Å, which is roughly consistent with the STM corrugations of 0.6 Å. In the high-resolution STM image of Fig. 3(b), the two oblong protrusions should be associated with the tilt-up pyrrole rings of the nonplanar porphyrin macrocycle, whereas the tilt-down pyrrole rings weakly appear as bridges between two oblong protrusions. In addition, we have obtained similar STM images for Cu-TBPP molecules on the Au(111) surface at 63 K. Due to the symmetric structure of the central Cu-porphyrin, this result should exclude a possibility that the two oblong protrusions are related to the electronic asymmetry of the central H2 -porphyrin. With increasing coverage of H2 -TBPP onto the Au(111) surface, two-dimensional islands are formed through self-assembled aggregation. As shown in Fig. 4(a), we have observed larger islands even without thermal annealing, suggesting a low diffusion barrier of the adsorbed molecules on the surface. In the islands, the nonplanar macrocyclic conformation remains, and the molecules exhibit a close-packed arrangement on the surface [Fig.4(b)]. A detailed analysis of the STM images indicates that the molecular √ arrangement exhibits an 11 × 5 3 superstructure, commensurate with the underlying substrate lattice of Au(111). Fig. 4(c) shows the model of the H2 -TBPP island formed on the Au(111)-1 × 1 structure, where intermolecular interactions should be governed by the van der Waals force between the tBP substituents. Due to the saddle-shaped deformation of the central porphyrin with mirror (C2v ) symmetry, the molecular orientations can be determined from directions of a dark line (symmetric axis) in the STM images. As shown in Fig. 5(a) and (c), two different orientations of the nonplanar porphyrins are randomly distributed within the island. We find that an orientational ordering is obtained via a thermal activation process. Fig. 5(b) shows the STM image of the supramolecular island after short thermal annealing (for 1 min at about 470 K). Inside the island, the orientations of neighboring molecules are the same in the [1 1 2] direction and rotated by 90° in the [2 1 1] direction, as illustrated in Fig. 5(d). It should be noted that the orientational ordering is accompanied without a change of the molecular arrangement, and such the orientations have been observed over the surface. Because it appears that the degree of the orientational ordering depends on the annealing temperature and time, the transformation should be related to rapid molecular diffusion on the surface, promoting breaking and rearrangement of the molecular islands. This kinetic process might allow the molecular islands to be rearranged in a stable manner. The orientational ordering should be associated with steric intermolecular interactions between the tBP substituents. Because of the 20° rotations of the phenyl rings, two possible steric interactions are expected between tBP substituents; cross- and paralleltype interactions of the tBP substituents between neighboring molecules in both the [1 1 0] and the [1 1 2] direction. The ordered orientations indicate that the cross-type interaction (up–down connections of each tertiary-butyl substituent) should be more stable than the parallel-type one (up–up and down–down connections), although the energy difference might be extremely small.
Control of molecular selective-assembling on metal surface
55
Fig. 4: (a) STM image (65 nm × 50 nm) at 63 K of a H2 -TBPP island formed on the Au(111) surface. (b) and (c)√ High-resolution STM image (5.3 nm × 5.3 nm) and its model of the H2 -TBPP island, exhibiting an 11 × 5 3 superstructure.
4. Selective aggregation of cyanophenyl-substituted porphyrins on Au(111) The orientational ordering of H2 -TBPP should be due to the weak steric intermolecular interactions between tBP groups. This result indicates that the molecular assembly formed on a surface can be controlled by changing substituents. In supramolecular chemistry, a large number of different selective and directional intermolecular interactions has been developed to control molecular aggregation, although it has been focused mainly on dissolved structures [3]. In this work, we have used a cyanophenyl substituent
56
T. Yokoyama et al.
Fig. 5: STM images (9.5 nm × 9.5 nm) of the H2 -TBPP island before (a) and after (b) thermal annealing at about 470 K for 1 min. (c) and (d) Schematic illustrations of the molecular arrangement and orientations within the island [17].
to control the molecular aggregation [19,20], because it has a simple and symmetric structure as well as an asymmetric charge distribution at the cyano group that should introduce dipole–dipole interactions between neighboring cyanophenyl substituents. An illustration of this interaction is given in Fig. 6, which shows the optimized arrangement of a cyanobenzen dimer and trimer obtained in ab initio molecular orbital calculations at the MP2/6-31G* level [20,21]. In the dimer, the cyano groups have an antiparallel configuration, whereas the trimer structure results from a cyclic arrangement of the cyano groups. The length of the CH. . . NC contacts is about 2.39 Å and 2.65 Å for the dimer and trimer, respectively, and thus shorter than the van der Waals distance of about 2.7 Å. The interaction energies are estimated to be −7.12 kcal/mol and −12.40 kcal/mol for the dimer and trimer structures, respectively, and thus comparable to the energy of hydrogen-bonding interactions. The relative orientation of molecules in these aggregates is therefore likely to be influenced by long-range dipole–dipole interaction, with hydrogen-bonding interactions further stabilizing the structure. To introduce the characteristic interaction of the cyano substituents, we have synthesized 5-(4-cyanophenyl)-10,15,20-tris(3,5-di-tertiarybutylphenyl) porphyrin (CTBPP), where a tBP group of H2 -TBPP was replaced with a cyanophenyl substituent, as shown in Fig. 7(a). At low coverage, most CTBPP molecules assembled into triangular clusters
Control of molecular selective-assembling on metal surface
57
Fig. 6: Calculated molecular aggregations of (a) a cyanobenzen dimer and (b) trimer, which are obtained from ab initio molecular orbital calculations [19].
on the Au(111) surface. As shown in Fig. 8(a), identical clusters are located separately at the elbows of the herringbone patterns. From the molecular structure of CTBPP, three paired lobes are expected as a single molecule in the STM image, because a paired lobe corresponds to one di-tertiarybutyl substituent. The high-resolution STM
58
T. Yokoyama et al.
Fig. 7: Structural formula of the cyanophenyl-substituted porphyrins (a) CTBPP, (b) cis-BCTBPP, and (c) trans-BCTBPP [19].
Control of molecular selective-assembling on metal surface
59
Fig. 8: STM images at 63 K of CTBPP [(a) and (b)], cis-BCTBPP [(d) and (e)], and trans-BCTBPP [(g) and (h)]. The corresponding molecular model is for (c) CTBPP, (f) cis-BCTBPP, and (i) trans-BCTBPP, respectively.
image of Fig. 8(b) shows that the triangular cluster is a CTBPP trimer. The cyanophenyl substituents are assembled into a cyclic configuration in the trimer structure of Fig. 8(c), in agreement with the cyanobenzen trimer aggregation of Fig. 7(a). Compared to the characteristic aggregation of CTBPP, we have not observed such supramolecular clusters for phenyl-tris(3,5-di-tertiarybutylphenyl) porphyrins, where the cyano substituents were removed from the CTBPP molecule. In this case, a random arrangement of the molecules was formed on the surface, confirming that the supramolecular aggregation is dominated by the cyano groups.
60
T. Yokoyama et al.
To probe the supramolecular aggregation in our system further, we substituted one more cyanophenyl group to give bis(4-cyanophenyl)-bis(3,5-di-tertiarybutylphenyl) porphyrin (BCTBPP), which forms two types of isomers (cis and trans) with respect to the configuration of two cyanophenyl substituents [Fig. 7(b) and (c)]. Fig. 8(d) and (e) shows that the cis-BCTBPP molecules are aggregated into a supramolecular tetramer. In this structure, the antiparallel intermolecular connections of all the cyanophenyl substituents lead to a macrocyclic arrangement of the porphyrin molecules, forming a molecular ring [Fig. 8(f)]. In contrast to the macrocyclic clusters of CTBPP and cis-BCTBPP, sequential aggregation was achieved for the trans-BCTBPP molecules, where two cyanophenyl groups were substituted at the trans positions. Fig. 8(g) shows the STM image of the Au(111) surface after deposition of the trans-BCTBPP molecules. In this structure, the antiparallel configuration between the cyanophenyl substituents results in a linear arrangement of the trans-BCTBPP molecules, forming supramolecular wires [Fig. 8(h) and (i)]. In the STM images, most of the individual wires extended across the elbows of the herringbone patterns, because the molecules were initially nucleated at the elbows. Furthermore, it is remarkable that the maximum length of the straight wire was above 100 nm, although some branches are also formed due to the three-fold symmetry of Au(111).
5. Summary The controlled aggregation of porphyrins has succeeded on a gold surface, which was visualized by low-temperature STM. On this surface, monomer, trimer, tetramer, or wirelike arrangements were controlled by local substituent interactions, and these structures were spontaneously and selectively formed by changing substituents. We believe the selective aggregation approach should become a general strategy for the rational design and construction of desired molecular architectures on substrate surfaces.
References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.
Y. Wada, M. Tsukada, M. Fujihira, K. Matsushige, T. Ogawa, M. Haga, and S. Tanaka, Jpn. J. Appl. Phys. 39, 3825 (2000). C. Joachim, J. K. Gimzewski, and A. Aviram, Nature 408, 541 (2000). J.-M. Lehn, Supramolecular Chemistry: Concept and Perspectives (VCH, Weinheim, 1995). H.-J. Guntherodt and R. Wisendanger (editors), Scanning Tunneling Microscopy I (Springer-Verlag, 1994). T.A. Jung, R.R. Schlittler, J.K. Gimzewski, H. Tang, and C. Joachim, Science 271, 181 (1996). T.A. Jung, R.R. Schlittler, and J.K. Gimzewski, Nature 386, 696 (1997). G.P. Lopinski, D.J. Mofatt, D.D.M. Wayner, and R.A. Wolkow, Nature 392, 909 (1998). M.O. Lorenzo, C.J. Baddeley, C. Muryn, and R. Raval, Nature 404, 376 (2000). F. Moresco, G. Meyer, K.-H. Rieder, H. Tang, A. Gourdon, and C. Joachim, Phys. Rev. Lett. 86, 672 (2001). M. Frukawa, H. Tanaka, K. Sugiura, Y. Sakata, and T. Kawai, Surf. Sci. Lett. 445, L58 (2000).
Control of molecular selective-assembling on metal surface 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21.
61
J.V. Barth, J. Weckesser, P. Gunter, L. Burgi, O. Jeandepeux, and K. Kern, Angew. Chem. Int. Ed. 39, 1230 (2000). M. Bohringer, K. Morgenstern, W.-D. Schneider, R. Berndt, F. Mauri, A.D. Vita, and R. Car, Phys. Rev. Lett. 83, 324 (1999). K. Sugiura, K. Iwasaki, K. Umishita, S. Hino, H. Ogata, S. Miyajima, and T. Sakata, Chem. Lett. 841 (1999). U. Harten, A.M. Lahee, T. Peter, and Ch. Woll, Phys. Rev. Lett. 54, 2619 (1985). J.V. Barth, H. Brune, G. Etrl, and B.J. Behm, Phys. Rev. B 42, 9307 (1990). D.D. Chambliss, R.J. Wilson, and S. Chiang, Phys. Rev. Lett. 66, 1721 (1991). T. Yokoyama, S. Yokoyama, T. Kamikado, and S. Mashiko, J. Chem. Phys. 115, 3814 (2001). M.J.S. Dewar, E.G. Zoebisch, E.F. Healy, and J.J.P. Stewart, J. Am. Chem. Soc. 107, 3092 (1985). T. Yokoyama, S. Yokoyama, T. Kamikado, Y. Okuno, and S. Mashiko, Nature 413, 619 (2001). Y. Okuno, T. Yokoyama, S. Yokoyama, T. Kamikado, and S. Mashiko, J. Am. Chem. Soc. 124, 7218 (2002). M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb, J.R. Cheeseman, V.G. Zakrzewski, J.A. Montgomery, R.E. Stratmann, J.C. Burant, S. Dapprich, J.M. Millam, A.D. Daniels, K.N. Kudin, M.C. Strain, O. Farkas, J. Tomasi, V.A. Petersson, P.Y. Ayala, Q. Cui, K. Morokuma, D.K. Malick, A.D. Rabuck, K. Raghavachari, J.B. Foresman, J. Cioslowski, J.V. Oritiz, B.B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaromi, R. Gomperts, R.L. Martin, D.J. Fox, T. Keith, M.A. Al-Laham, C.Y. Peng, A. Nanayakkara, C. Gonzalez, M. Challacombe, P.M.W. Gill, B.G. Johnson, W. Chen, M.W. Wong, J.L. Andres, M. Heak-Godon, E.S. Replogle, and J.A. Pople, Gaussian 98, version A,7 (Gaussian Inc., Pittsburgh, PA, 1998).
Part B
NICE Devices
Nanotechnology and Nano-Interface Controlled Electronic Devices Editors: M. Iwamoto, K. Kaneto and S. Mashiko © 2003 Elsevier Science B.V. All rights reserved
CHAPTER 6
Polymer optoelectronics – towards nanometer dimensions Olle Inganäs and Fengling Zhang Biomolecular and organic electronics, Department of Physics and Measurement Technology, Linköping University, SE - 581 83 Linköping, Sweden
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. Excited states in polythiophenes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. Diffusion length of excited states . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. Stratified photodiodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5. Excitation transfer in photodiodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6. Models of charge generation in photodiodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7. Nanodimension of electrodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8. Nano-pattern application in photovoltaic devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
65 66 69 72 73 75 75 79 80 80
1. Introduction The unification of the natural sciences in a form now widely being called nanosciences is one of the themes of the late 20th century scientific enterprise, and may well become a dominant theme in the 21th century. This reunification of the natural sciences, divided by processes of specialisation in the late 19th century, is central to the development of the field of conjugated polymers at the junction of physics and chemistry. Here chemistry delivers materials in the forms of conjugated polymers and molecules, using well-established methods of synthetic organic chemistry and polymer chemistry. The objects of synthesis are defined on the Å and nm length scale, and are synthesised in samples carrying millimoles of objects. The objects are quasizero- or one-dimensional electronic systems with a high degree of excitation and charge confinement, but also with a high mobility on the nm length scale. Physics offer one description of the electronic structure of these systems, complementary to the quantum chemical description but all based in quantum mechanics; the language in which these objects are described may however differ between chemistry and physics. The description of the excited states, so crucial to the development of optoelectronic devices for emission of light by
66
O. Inganäs and F. Zhang
electronic injection of charge into molecules and polymers, or for creation of charges from excited states, is an area of controversy, and where physical and chemical models meet and collide. While these models may be fought over, using sophisticated models and spectroscopies, there is in the background a steady development of new molecules, materials and devices, which sometimes leads to results contradicting current dogma. A particular enigmatic example of these controversies is the issue of charge recombination in organic layers, where a simple argument from symmetry shows that electrons and holes recombining in a solid should lead to singlet and triplet excited states in a ratio of 1 : 3. More recent experimental determinations do not agree with this rule [1], and the ratio has been shown even to go beyond 1 : 1. As the ratio is now considered to be a material and device parameter, rather than a fundamental property of recombining charges, one of the ultimate limiting factors of light emitting diodes is no longer present. Likewise, in the field of organic photodiodes, a central issue is the binding energy of the excited state, as it has a clear impact on the formation of mobile charges in photodiodes. No clear consensus is yet found, and the range of binding energies reported from experimental studies is broad. Internal quantum efficiencies of excited state dissociation into charges can be very high in donor–acceptor systems, and the kinetics of these processes is extremely fast [2]. In pure homopolymer systems, routes towards photogeneration of charges are present [3], but much less efficient than what is found in donor–acceptor complexes where an electron acceptor receives an electron from the excited state on a conjugated polymer or molecule [4,5]. However, progress in organic photodiodes has now generated photodiodes with external quantum efficiency at low intensity monochromatic illumination of better than 60%, and solar cells with solar energy efficiencies of 2.5% at AM 1.5 conditions [6–9]. A steady development of new materials, new devices and new patterning methods contributes towards incremental progress in this field. The length appropriate for analysis and characterisation in this field spans from the Å and nm level, for chemical structure in donor–acceptor moieties, to 1–10 nm for the diffusion of excited states and charge carriers, to 10–100 nm for the description of multilayer devices where a photoactive solid is confined between electrodes separated by this thickness, and to 100–1000 nm, where electrodes and photoactive solids are patterned on this length scale for trapping and confining light. Looming in the background are found also the large area devices, which are deposited as thin films on flexible carriers of cm to dm dimensions, and which may one day be found as an organic solar cell for energy conversion. Here conditions for large or giant area electronics are crucial, and it is of great importance to develop green chemistry and large area deposition methods. This chapter focuses on some of our recent contributions to the field of organic photodiodes, with particular emphasis on the nanometer dimensions of materials, processes and devices.
2. Excited states in polythiophenes We have chosen polythiophenes (PT) as our main group of conjugated polymers for photovoltaics [10]. The molecular structures of polythiophenes (PT) are shown in Fig. 1. This choice is because this family of polymers is very versatile from the point of view
Polymer optoelectronics – towards nanometer dimensions
67
Fig. 1: The chemical structures of polythiophenes and a copolymer, some of which are used in photodiodes. The copolymer [22] demonstrates covalent linking of polymer donor and acceptor.
68
O. Inganäs and F. Zhang
of polymer modification by chemical substitution, and from the point of view of the higher photostability of polythiophenes as compared to poly(paraphenylene vinylenes), for example. It is thus quite easy to move the optical absorption over much of the visible spectrum, and the range of polythiophenes covers a sizable part of the solar spectrum. Another reason is the possibility of making high mobility polymers, as shown in recent years where poly(alkylthiophenes) have been induced to have high field effect mobility in polymer transistors, and even the possibility to field dope into the metallic and superconducting regime [11]. The nature of the excited state in PTs has been extensively studied through photoluminescence (PL) quantum yield and photoluminescence and photoinduced absorption kinetics [12,13]. The radiative lifetime of the excited state is typically 1–2 ns, and one source of decay is through internal conversion. That process is much enhanced in poly(alkylphenylthiophenes) designed for a non-planar geometry and high bandgap [12]. As these polymers do not give much absorption in the solar spectrum, they are less significant for photovoltaics. More important are the low bandgap PTs, where aggregates have been shown to lead to non-radiative recombination of excited states [13]. These non-radiative processes are not sufficiently fast to outcompete the process of photoinduced charge transfer in photodiodes. The possibility of intersystem crossing in PTs, induced by the spin–orbit coupling due to the presence of a sulphur atom in the carbon conjugated chain, is of little importance in photovoltaics, as it appears that both singlets and triplets may be dissociated by the presence of acceptors. The diffusion of the excited state is much influenced by the chemical structure of the polymer, as well as by the morphology of the polymer when aggregating into solid films. The typical behaviour is a drastic loss of PL quantum yield upon precipitation of PTs from good solvents. There are important alternatives; we have observed an enhanced PL yield of some soluble polythiophenes [14] when decreasing the solvent quality on the route to the solid, where once more a general reduction of PL yield is found. It may be that special aggregation forms may retain and even increase the quantum yield, but most forms of aggregation lead to an increase of the non-radiative decay and thus quench luminescence. The spectral migration of excited states reveals a diffusion of these, which will couple the excited state to some source of non-radiative recombination, which is never far away in a polymer film. Copolymers and oligomers incorporating chemically modified thiophene units have been shown to have a much enhanced PL quantum yield [15]. The geometry of the polymer chain in good solvents has been studied by photophysical methods, where a sub-picosecond pulse of (polarized) monochromatic light is used to excite polymer chains in solid films or in solutions; during the coming picoseconds the intensity or polarisation state of the rapidly decaying luminescence is observed. In these studies is revealed how transport of the excited state along the single polymer chain in a dilute solution in a good solvent occurs [16]. This study reveals a tendency of the low bandgap PTs to give a stiff, close to planar, geometry as expected from the point of theory. When now packing these chains into a polymer solid, a memory of the chain conformation in the solvent may persist, as argued in other studies.
Polymer optoelectronics – towards nanometer dimensions
69
3. Diffusion length of excited states In the solid state, the length of diffusion of the excited state is one of the important parameters controlling the design of photodiode materials. If the excited state is capable of finding a site for non-radiative recombination prior to finding a site for photoinduced charge transfer, potential photocurrent is lost. Determination of the exciton diffusion length is therefore an important item. We have used a combination of photoluminescence measurements and optical modelling to deduce the diffusion length of the excited state which can be quenched by the presence of C60 molecules [17]. In order to be able to do this experiment, we have chosen a polythiophene, which does not easily dissolve fullerene compounds, as to be able to obtain sharp bilayers of polythiophene/C60, where the fullerene is evaporated onto a thin spin-coated polymer film, carried on a quartz substrate. These methods of deposition give reasonably flat films, and these films can be used to obtain the full dielectric function of the materials. To do this we use spectroscopic ellipsometry, a non-invasive method of high precision, which will also measure the thickness of films. We also determine the absolute PL quantum yield of these bilayers, in integrating sphere measurements, and vary the polymer and C60 thickness. As the thickness of the polymer films increases, more excited states are created at a distance from the polymer/C60 junction, and higher PL is obtained. With these methods, and with a detailed optical model of the thin film optical physics valid in films of 10–100 nm thickness (Fig. 2), we are able to determine the diffusion length of the excited state. We find that to be 5 nm in the polymer studied. This is the distance in which an acceptor must be found in order for an excited state to be able to generate charge. The diffusion length is a materials property, and may also be influenced by morphology and packing of chains. It is therefore an object of design and synthesis. Fig. 2 shows the distribution of absorbed electromagnetic energy from an incident monochromatic wave at a polymer/C60 bilayer supported on a semi-infinite quartz layer. The internal reflection of waves inside the bilayer gives a different excitation distribution compared to that obtained from a simple-minded application of the Lambert–Beer law as often used. The excitation or exciton distribution is actually not completely identical to the distribution of excited states formed by absorption, as exciton diffusion may have an impact, in particular close to boundaries. Even more is this physics relevant in the emission of photons from excited states within the polymer layer, a distribution that must also be included in the complete analysis of the emission profile versus wavelength in the far field. We may also approach this problem of quenching of excited states in the presence of acceptors by constructing objects where the excited state should always find an acceptor within that distance. A realisation of such elements is found in the double-cable polymers, so named because a polymer is carrying both an electron acceptor and an electron donator [18–22]. These elements are also necessary for the further transport of photogenerated charges, as they will hop between occupied and unoccupied states on the electron acceptor moiety and electron donor moiety, respectively. The term double cable refers to this property, and by incorporation of these elements on each polymer chain means for excited state dissociation and for charge transport are found everywhere in the material.
70
O. Inganäs and F. Zhang
Fig. 2: The lower panel shows the calculated power excitation density g(z), of a 60 nm polymer film excited at 2.34 eV, normalised to the incoming intensity. The axes thus show the fraction of the incoming intensity absorbed per length unit. The different lines are the absorbed power density for a neat film (thin solid line) and a polymer film with a 4 nm film of C60 on top (dashed line). The excitation density assuming an exponential decay of the incoming radiation (e.g. I = I0 e−αz ) is shown (thick solid line). Also shown is the steady-state power density with an infinite sink at the polymer/C60 heterojunction and a diffusion length of 5.3 nm (broken dashed line). The upper panel shows the calculated transmission of the polymer emission from different points z in the polymer film, to a point outside the quartz substrate (see text for more details). The transmission is shown for the case of a neat polymer film (thin solid line) and a polymer film with 4 nm C60 on top (dashed line). (From Ref. [17].)
A recent example of double cable polymers is found in a soluble polythiophene–C60 copolymer, where copolymerisation of substituted thiophenes with a fraction of C60 carrying substituted thiophenes leads to the polymer. Photoinduced charge transfer is easily observed in these polymers, and is almost complete in the solid and partial in the
Polymer optoelectronics – towards nanometer dimensions
71
Fig. 3: PL spectra and their PL yield from pure POMeOPT film and copolymer POMeOPT with 10% C60 film show the PL quenching in copolymer film of POMeOPT.
solution of the polymer. Fig. 3 shows the PL quenching in one of these copolymers. Variation of the stoichiometry of the polymer can be used to enhance the quenching; unfortunately the solubility of the polymer is not retained when putting more than one C60 per 4 monomers of thiophene in the copolymer. The processing through polymer solutions is crucial to obtain thin film devices, which are required for good devices. We are, however, free to enhance the C60 concentration in these polymers by mixing
72
O. Inganäs and F. Zhang
C60 into a common solvent. This will lead to polymer films with enhanced photodiode performance [22], indicating that the density of C60 is not sufficient in the copolymer per se.
4. Stratified photodiodes The sequel to photoinduced charge separation in photodiodes is the transport of generated charges to the electrodes. This should preferably happen without the trapping or recombination of these charges, which are very plausible events in the materials concerned. Localisation of charges into traps is very likely due to the high degree of disorder; recombination is likely when a stream of charges passes through fixed space charges of the opposite polarity. It is therefore essential to organise the paths of the electron and hole so that they can be transported with minimal trapping and recombination. Blends of donor and acceptor do not always deliver this. The size of domains in which donors and acceptors are aggregated is influenced both by kinetics and statics of blending, and it is expected and observed that the mode of preparation of these blends, and layers from blends in solutions, should have a major impact on the performance of devices. The process of removing solvent from a blend may be fast – as in spincoating – or slow, such as in dropcasting or doctor blading. The phase structure of the resulting film is probably very different under these different circumstances. We note that the highest solar cell efficiencies are reported [9] for films prepared from particular blends of solvents, where the change of composition during evaporation of the different solvents at different rates is yet another mechanism that will influence the kinetics of thin film formation and molecular organisation/morphology in the film. By influencing the degree of phase separation by purely chemical means, as for example when using a “immobilised solvent” as a side chain of a polythiophene, molecular miscibility of polymer/C60 is obtained, and the resulting phase structure does not show up above 20–30 nm [23]. To gain some small means of control of the 10–100 nm organisation of materials in the vertical dimension, we have developed stratified photodiodes [24,25]. These are assembled by using a soluble fullerene derivative to be spin-coated on top of a polymer film, previously deposited by spin-coating on a substrate. The structure of the stratified active layer in photodiodes is shown in Fig. 4. The second act of spin-coating could of course be expected to dissolve the first layer, unless solvents and conditions are carefully chosen so as not to dissolve the first layer when depositing the second. This has been possible for bilayers of high molecular weight PPV and soluble methanofullerene compounds, which form graded junctions [25]. The diffuse junctions of the polymer phase at the bottom, with a methanofullerene phase on top, is much more effective in generating photocurrent. This is very visible in the photoluminescence of the polymer layer, which is almost completely quenched in the presence of the methanofullerene, this indicating a very effective dissociation of excited states. The increase of photocurrent, due to the higher contact area between donor and acceptor molecules, now must be transferred through layers, which eventually
Polymer optoelectronics – towards nanometer dimensions
73
Fig. 4: A cross section of the stratified active layer of polymer and C60 in stratified photodiodes.
end up in polymer only contacts at the anode side, and methanofullerene only contacts at the cathode side.
5. Excitation transfer in photodiodes The collection of optical energy is in green plants very much helped by the construction of the photosynthetic apparatus. Here antenna pigments in a primary step absorb the incoming photons. That energy is transferred to a reaction center chlorophyll in reaction centers PSI and PSII, wherein the primary photochemistry occurs. The nature of these antenna pigments varies, and they may be chlorophylls but they may also be distinctly different, for instance in the form of xanthenes. The collection of energy from antenna pigments to reaction center chlorophyll occurs through excitation transfer, also known as Förster transfer. One advantage of this system is that more of the photon flow may be absorbed, through the use of pigments of different absorption spectra. The same advantage is something we would like to utilise in photodiodes, to sensitize some part of the green and blue part of the solar spectrum. This requires the use of several pigments, and we have implemented this in the form of polymer blends. In a study of three different red polythiophenes – absorbing in the green part of the spectrum – in combination with a polyparaphenylene vinylene absorbing in the blue-green, we have demonstrated that this principle may be used to enhance the action spectrum of the photodiodes [26]. These devices were built as bilayer photodiodes, where an evaporated C60 layer is the acceptor located on top of a polymer blend layer. We compare the performance of photodiodes incorporating blends with photodiodes where the polymer layer is a homopolymer. The high bandgap polymer shows an emission spectrum largely overlapping with the absorption spectra of the lower bandgap polythiophenes. By studying the photoluminescence spectrum of homopolymers and blend, we establish that almost complete excitation transfer occurs in the polymer blends of 1 : 1 (by weight). This is very visible in the complete suppression of emission from the PPV high bandgap polymer, and in the considerable enhancement of PL from the polythiophenes used. This enhancement is very visible as the emission intensity from the three different polythiophenes is not high. Photoluminescence and excitation transfer in photodiodes is
74
O. Inganäs and F. Zhang
Fig. 5: PL spectra of polymeric films of BEHP-PPV, PBOPT and the blend BEHP-PPV : PBOPT in the ratio 1 : 1.
shown in Fig. 5. Therefore we conclude that geometries suitable for excitation transfer are obtained in these blends obtained by spin-coating from a common solvent. We corroborate this interpretation by imaging the blend layer surface in scanning force microscopy, with the tapping mode of interaction between tip and surface. We are unable to resolve any fine-structure in these blends, which may be due to structures being smaller than the limits of resolution of the method, say 20–30 nm, or by a complete phase separation into a pancake geometry, where one of the polymers would be residing on top of the other. This does not appear to be a likely geometry, but is not excluded by our studies. The total absorption of the film is such as to make implausible that the degree of excitation transfer that is observed could be originating from a double layer, as the individual layers would have to be too thin to generate the optical absorption observed. It is therefore plausible that the dimensions of phase separation are small, and that most PPV chains are found close enough to a PT chain to allow excitation transfer. In photodiodes incorporating polymer blends we find consistently higher photocurrents and lower photovoltages. The photovoltages observed are similar to those that would be found in homopolythiophene/C60 layers, and we therefore conclude that photocurrent generation occurs through excitations found on the polythiophene chains, but to some degree delivered there by excitation transfer from the PPV. An enhancement
Polymer optoelectronics – towards nanometer dimensions
75
of photocurrent and an extension of the action spectrum is the result. We may in this case argue that the characteristic length of Förster transfer is smaller or comparable to the phase separation dimensions. Later and more extensive studies of polymer blends in other polymer families have revealed a multitude of length scales in the phase separation between two polymers [27,28].
6. Models of charge generation in photodiodes The same thin film optical phenomena found valuable in the analysis of photoluminescence quenching in polymer/C60 bilayers is also of importance in analysing the performance of photodiodes constructed by sandwiching this bilayer in between two electrode layers, one metallic and highly reflective and one conducting but almost transparent [29,30]. For all the materials in this photodiode, dielectric functions have been measured with the help of spectroscopic ellipsometry. With help of a detailed optical model of all internal reflections and transmission at internal boundaries, we are able to calculate the distribution of optical energy of the incoming (and partly reflected) wave. Fig. 6 shows the distribution of optical energy in photodiodes with different thicknesses of C60. The distribution of optical absorption is thus known, and we can couple this to the generation of photocurrent in the device by including this distribution as a source term in a diffusion equation appropriate for neutral excited states. With fixed thickness and optical parameters, due to the spectroscopic ellipsometrical data, we have only one free variable in fitting our real data of the external quantum efficiency of the devices; that is the diffusion length of the excited state contribution to the formation of photocurrent. We implicitly assume many things, in particular that the photoinduced charge transfer occurs only at the interface between polymer and C60. This assumption is identical in PL quenching modelling experiments, and both types of experiment end up with an exciton diffusion length of 5 nm. Considering this short distance, we note that we must position all polymer chains within a distance of less than 5 nm in order to transfer the excitation energy. This may not be a common situation in polymer blends.
7. Nanodimension of electrodes In developing nanoelectronic devices based on polymers, the electrodes must be patterned at a sub-micron scale; the vertical dimension of the device is almost always less than a few hundreds of nm. In order to pattern at a nano scale, in addition to the traditional photolithography processing, soft lithography was developed and is nowadays used as a suitable method for patterning polymers [31,32]. Micro- or nanopatterning of polymers is not only necessary for electrical addressing and wiring of the circuits, but also has other functions, such as trapping more light to improve the performance of polymer photodiodes [33] or increasing the efficiency, and controlling light output from microstructured LEDs [34,35]. The conducting polymer poly(3,4-ethylenedioxythiophene) doped with poly(4-styrenesulfonate) (PEDOT-PSS) is often used as a metal in polymer electronic devices,
76
O. Inganäs and F. Zhang
Fig. 6: Calculated distribution of the normalised modulus of the optical electric field |E|2 inside a photovoltaic device: glass (1 mm)/ITO (120 nm)/PEDOT (11O nm)/PEOPT (40 nm)/C60/Al with a C60 layer thickness of (a) 35 nm and (b) 80 nm for a wavelength of 460 nm.
as a modified anode on top of indium tin oxide (ITO), in photodiodes of enhanced performance [36] and for improvement of the rectification ratio in polymer diodes PEDOT-PSS can be modified with glycerol or sorbitol to increase the conductivity by two orders of magnitude. This makes it possible to use this material as a flexible electrode for application in optoelectronic devices. The PEDOT-PSS has double functions in electronic devices: to enhance the electrode performance by adjusting the work function of an electrode [37] and for use as a flexible anode [38]. We here introduce our experimental results on nano-patterning of conducting polymer PEDOT-PSS, both on glass and Si wafer, patterned by a soft lithography technique, MIMIC and liquid printing, which we believe are new approaches for obtaining polymer nanostructures. These could have potential application in biology and optoelectronics. Nanowires of PEDOT-PSS was fabricated from its aqueous solution on glass or Si wafers, by using capillary action in MIMIC structures. Two different elastomer submicron patterns were made by casting polydimethylsiloxane (PDMS) onto commercially available diffraction gratings. We put the elastomer stamp in conformal contact with a piece of cleaned glass or an Si wafer, and then applied a drop of PEDOT-PSS (Baytron P, Bayer AG, concentration 1.3%) aqueous solution in front of the capillary openings of
Polymer optoelectronics – towards nanometer dimensions
77
Fig 7: SFM image of polymer PEDOT-PSS nanowires and chemical structure of PEDOT-PSS. The period of the nanowires was 270 nm and the height ≈25 nm. The cross section picture shows the profiles of the wires and their height.
the stamp. The sample was then left for several hours, for the solution to migrate into the capillaries and subsequently dry out. After the solution had dried we carefully peeled off the stamp leaving the polymer nanowires standing on the surface. In this way the grating pattern was transferred from the stamp to the polymer layer. All the processes were performed in ambient condition. A scanning force microscope (SFM-Nanoscope III, Digital Instruments) was used in tapping mode to image the polymer nanowires. Fig. 7 shows SFM images of polymer PEDOT-PSS nanowires (3600 lines per millimeter) with the same period as the stamp
78
O. Inganäs and F. Zhang
Fig. 8: SFM image of nanowires with a period 833 nm molded with a compressed stamp, showing a height of the nanowires of ∼8 nm and a width of ∼600 nm with a separation of 200 nm.
(280 nm), but with a lower height (∼25 nm) after drying than the depth of the original stamp (∼55 nm). In order to make polymer nanowires of lesser height, we desired to reduce the size of the capillary channel. This we did by placing a weight (≈100 g), resulting in a pressure of ∼10 kPa on top of the stamp to compress the channel during molding. The pressure we used to deform the stamp is similar to that previously analysed [39]. We also diluted the PEDOT-PSS solution with distilled water (1 : 1). Si wafer was used instead of glass to decrease the roughness of the substrate. The SFM images show that the height of the lines resulting in the MIMIC process could be decreased, depending on the pressure exerted on the top of the stamp. For example, the height of the nanowires for the 1200 grating under zero load is 70 nm, which decreased to around 8 nm during loading with ∼10 kPa (see Fig. 8). The top of the wires was deformed under these conditions. The nanowires in Fig. 8 are separate from each other, the distance between lines is ∼200 nm and the width of the wires is around 650 nm. The lengths of the nanowires can reach 1 cm.
Polymer optoelectronics – towards nanometer dimensions
79
8. Nano-pattern application in photovoltaic devices In polymer photovoltaic devices, the thickness of the diodes is crucial. To absorb more light a thick film is needed, but due to the low mobility of charge carriers, this is associated with increasing series resistance. In bilayer diodes built from the combination of an electron donor and acceptor, the excited state dissociates at the interface between electron donor layer and acceptor layer. The larger interface area thus causes more efficient charge separation. These two aspects of device thickness motivate efforts to combine donor and acceptor in micro and nano-patterned diodes, to control the geometry and thus influence photon propagation in the device as well as to control the interface area between donor and acceptor. We have used soft embossing to pattern active polymer layer to enhance the external quantum efficiency (EQE) of photodiodes [33]. Here we combine two soft lithography methods, liquid printing and soft embossing, to fabricate the devices with two periodic patterns to increase the performance. Metallic polymer anode PEDOT-PSS was micro patterned in 600 lines/mm and polythiophene POMeOPT layer was nano-patterned in 3600 lines/mm followed by vacuum deposition of C60 as electron acceptor to keep a sharp interface. The bilayer diodes were fabricated as ITO/PEDOT-PSS/POMeOPT/C60/Al. Our results show that the EQE of these diodes with patterned anode and polymer was enhanced compared with non-patterned planar diodes (see Fig. 9). These results demonstrate the possibility of using sub-micro and nano-pattern by soft lithography to implant structure in polymer photodiodes to improve light trapping in thin active layers, so that the performance of these devices may be enhanced.
Fig. 9: The EQE of photodiodes with pattern PEDOT-PSS in 600 lines/mm, pattern polymer (POMeOPT) in 3600 lines/mm (open triangles) and planar PEDOT-PSS, planar POMeOPT (filled squares).
80
O. Inganäs and F. Zhang
In summary, we have demonstrated how processes and structures with nanometer extension are crucial to the operation of polymer photodiodes. Structuring of materials on the nanometer length scale is thus an important tool for obtaining higher detection and conversion efficiencies in these devices, always relying on parallel progress in chemical synthesis, materials formulation and device assembly.
Acknowledgements We wish to acknowledge all the coworkers and students contributing to the studies here reported, in particular Lucimara Stolz Roman, Lichun Chen, Mathias Theander, Leif Petterson, Yohannes Teketel and many more. Joint European projects with S. Sariciftci, C. Brabec, J. Hummelen, R. Janssen, M. Maggini, M.R. Andersson and M. Prato have been instrumental in delivering materials. Fundings from the Göran Gustafsson foundation and the Carl Tryggers foundation have been instrumental to achieve these results, as well as the Engineering Research Board (TFR).
References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19.
Y. Cao, I.D. Parker, G. Yu, C. Zhang, and A.J. Heege, Nature 397, 414 (1999). C.J. Brabec, G. Zerza, G. Cerullo, S. De Silvestri, S. Luzzati, et al., Chem. Phys. Lett. 340, 232 (2001). A. Ruseckas, M. Theander, M.R. Andersson, M. Svensson, M. Prato, et al., Chem. Phys. Lett. 322, 136 (2000). N.S. Sariciftci, L. Smilowitz, A.J. Heeger, and F. Wudl, Science 258, 1474 (1992). G. Yu, J. Gao, J.C. Hummelen, F. Wudl, and A.J. Heeger, Science 270, 1789 (1995). C.J. Brabec, N.S. Sariciftci, and J.C. Hummelen, Adv. Func. Mater. 11, 15 (2001). C.J. Brabec and S.N. Sariciftci, Monatsh. Chem. 132, 421 (2001). O. Inganäs, L.S. Roman, F.L. Zhang, D.M. Johansson, M.R. Andersson, and J.C. Hummelen, Synth. Metals 121, 1525 (2001). S.E. Shaheen, C.J. Brabec, N.S. Sariciftci, F. Padinger, T. Fromherz, and J.C. Hummelen, Appl. Phys. Lett. 78, 841 (2001). M.R. Andersson, O. Thomas, W. Mammo, M. Svensson, M. Theander, and O. Inganäs, J. Mater. Chem. 9, 1933 (1999). J.H. Schon, A. Dodabalapur, Z. Bao, C. Kloc, O. Schenker, and B. Batlogg, Nature 410, 189 (2001). M. Theander, O. Inganäs, W. Mammo, T. Olinga, M. Svensson, and M.R. Andersson, J. Phys. Chem. B 103, 7771 (1999). A. Ruseckas, E.B. Namdas, T. Ganguly, M. Theander, M. Svensson, et al., J. Phys. Chem. B 105, 7624 (2001). M. Theander, M. Svensson, A. Ruseckas, D. Zigmantas, V. Sundstrom, et al., Chem. Phys. Lett. 337, 277 (2001). G. Gigli, O. Inganäs, M. Anni, M. De Vittorio, R. Cingolani, et al., Appl. Phys. Lett. 78, 1493 (2001). M.M.L. Grage, T. Pullerits, A. Ruseckas, M. Theander, O. Inganäs, and V. Sundstrom, Chem. Phys. Lett. 339, 96 (2001). M. Theander, A. Yartsev, D. Zigmantas, V. Sundstrom, W. Mammo, et al., Phys. Rev. B 61, 12957 (2000). A. Cravino, G. Zerza, M. Maggini, S. Bucella, M. Svensson, et al., Chem. Commun. 24, 2487 (2000). A. Cravino, G. Zerza, H. Neugebauer, S. Bucella, M. Maggini, et al., Synth. Met. 121, 1555 (2001).
Polymer optoelectronics – towards nanometer dimensions 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39.
81
A. Cravino, G. Zerza, H. Neugebauer, M. Maggini, S. Bucella, et al., J. Phys. Chem. B 106, 70 (2002). G. Zerza, A. Cravino, H. Neugebauer, N.S. Sariciftci, R. Gomez, et al., J. Phys. Chem. A 105, 4172 (2001). F.L. Zhang, M. Svensson, M.R. Andersson, M. Maggini, S. Bucella, et al., Adv. Mater. 13, 1871 (2001). l.S. Roman, M.R. Andersson, T. Yohannes, and O. Inganäs, Adv. Mater. 9, 1164 (1997). D. Godovsky, L.C. Chen, L. Pettersson, O. Inganäs, M.R. Andersson, and J.C. Hummelen, Adv. Mater. Optics Electr. 10, 47 (2000). L.C. Chen, D. Godovsky, O. Inganäs, J.C. Hummelen, R.A.J. Janssens, et al., Adv. Mater. 12, 1367 (2000). L.C. Chen, L.S. Roman, D.M. Johansson, M. Svensson, M.R. Andersson, et al., Adv. Mater. 12, 1110 (2000). A.C. Arias, J.D. MacKenzie, R. Stevenson, J.J.M. Halls, M. Inbasekaran, et al., Macromolecules 34, 6005 (2001). A.C. Arias, N. Corcoran, M. Banach, R.H Friend, J.D. MacKenzie, and W.T.S. Huck, Appl. Phys. Lett. 80, 1695 (2002). L.A.A. Pettersson, L.S. Roman, and O. Inganäs, J. App. Phys. 86, 487 (1999). L.A.A. Pettersson, L.S. Roman, and O. Inganäs, Synth. Met. 102, 1107 (1999). Y.N. Xia and G.M. Whitesides, Angew. Chem. Int. Ed. 37, 551 (1998). E. Kim, Y.N. Xia, and G.M. Whitesides, Nature 376, 581 (1995). L.S. Roman, O. Inganäs, T. Granlund, T. Nyberg, M. Svensson, et al., Adv. Mater. 12, 189 (2000). J.A.E. Wasey, A. Safonov, I.D.W. Samuel, and W.L. Barnes, Phys. Rev. B 6420, art. no.-205201 (2001). B.J. Matterson, J.M. Lupton, A.F. Safonov, M.G. Salt, W.L. Barnes, and I.D.W. Samuel, Adv. Mater. 13, 123 (2001). L.S. Roman, W. Mammo, L.A.A. Pettersson, M.R. Andersson, and O. Inganäs, Adv. Mater. 10, 774 (1998). T. Kugler, W.R. Salaneck, H. Rost, and A.B. Holmes, Chem. Phys. Lett. 310, 391 (1999). F. Zhang, M. Johansson, M.R. Andersson, J.C. Hummelen, and O. Inganäs, Adv. Mater. 14, 662 (2002). A. Bietsch and B. Michel, J. Appl. Phys. 88, 4310 (2000).
Nanotechnology and Nano-Interface Controlled Electronic Devices Editors: M. Iwamoto, K. Kaneto and S. Mashiko © 2003 Elsevier Science B.V. All rights reserved
CHAPTER 7
Control of charge transfer and interface structures in nano-structured dye-sensitized solar cells Shozo Yanagida, Takayuki Kitamura, and Yuji Wada Department of Material and Life Science, Graduate School of Engineering, Osaka University, Yamada-oka 2-1, Suita, Osaka 565-0871, Japan
1. 2. 3.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Light harvesting efficiency based on dynamics of dye-sensitization . . . . . . . . . . . . . . . . . Electron transport in nano-structured TiO2 electrodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Ambipolar electron diffusion mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Evaluation method of electron diffusion in plain nano-structured TiO2 layers 3.3. Electron transport in nano-structured TiO2 layers . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4. Effects of electrolytes on the electron transport . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5. Effect of surface states in nano-structured TiO2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. Charge transport in iodide/polyiodide electrolytes of DSCs . . . . . . . . . . . . . . . . . . . . . . . . 4.1. Mechanistic studies using quasi-solid-state electrolytes produced by lowmolecular-weight gelators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2. Quasi-solid stated DSCs using imidazolium molten iodides as electrolytes . . 5. Importance of interface control in DSC fabrication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1. Interface structures affecting efficiencies, ηei and ηhi . . . . . . . . . . . . . . . . . . . . . . . 5.2. Important role of interfaces affecting efficiencies, ηec and ηhc . . . . . . . . . . . . . . 6. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
83 85 86 86 88 88 91 95 96 96 97 100 100 100 101 102 102
1. Introduction Almost 10 years have passed since dye-sensitized solar cells (DSCs) were innovated by Grätzel’s group [1–3]. DSCs can be constructed by Ruthenium complex [for example, N3 = cis-RuII (dcbpy)2 (SCN− )2 (dcbpy = 2,2 -bipyridine-4,4 -dicarboxylic acid); Scheme 1] dye-anchored nano-structured TiO2 films and redox electrolyte made from iodine and iodide salts, and is currently under large, intense investigation, because
Control of charge transfer and interface structures in nano-structured DSCs
85
In a DSC, the seven factors must be well balanced to achieve the high efficiency. With regard to ηlh of the N3 dye, incident photon to current efficiency (IPCE) is almost 100% in the wide absorption band, which is comparable to amorphous silicon solar cells. This is due to the high absorption coefficient of N3 dye molecules and large ratio (>1000) of the nano-structured TiO2 surface (mesoporous surface) area to the projected one defined as roughness factor. Current research topics to improve DSCs are electron transfer dynamics that determine ηei and ηhi , charge transport mechanism in mesoporous nano-crystalline TiO2 phase referring to ηet , and mastering the interfaces that control ηec and ηhc . In this article, we will report how to improve the DSC efficiency in view of the above-mentioned factors. The charge transport efficiencies of both nano-structured TiO2 and electrolyte solution, and electron transport at interfaces of window electrode and counter electrode will be discussed. On the basis of mechanistic points of view, we will propose some concepts for solidification of hole transporting phases.
2. Light harvesting efficiency based on dynamics of dye-sensitization The dynamics of the interfacial electron-transfer from excited state of dye, N3, to TiO2 was examined precisely by laser-induced ultra-fast transient spectroscopy as summarized in Fig. 1 [4–9]. These kinetics are largely affected by the composition of the electrolyte. For example, since decrease of surface pH makes the flat band potential positive, the presence of Li+ , which acts like H+ , leads it to the positive side, while the
Fig. 1: Electron transfer dynamics at the interfaces of nano-structured TiO2 /N3 and N3/iodide electrolyte.
86
S. Yanagida et al.
presence of basic pyridine derivatives shifts it to the negative side [10]. Concentrations − − of I− and I− 3 formed by equilibrium of I + I2 ↔ I3 is also noted to affect charge transfer kinetics. At the efficient DSC conditions, however, the two forward electrontransfer steps are much faster than the corresponding reverse electron-transfer steps (charge recombination), by a factor of 106 to 109 . The high conversion efficiency well explains effective and vectorial electron transfer in a DSC with N3, suggesting high values of ηei and ηhi . 3. Electron transport in nano-structured TiO2 electrodes 3.1. Ambipolar electron diffusion mechanism The porous TiO2 films of 10 µm-thickness formed by sintering av. 20 nm-sized TiO2 particles have more than 500 boundaries across the film. The resulting nano-structured TiO2 films have a large volume of pores inside with a porosity of 45 to 60%, giving a large surface area for dye-adsorption. Accordingly, the high efficiency of the DSC suggests that the nano-structured dye-coated TiO2 films should exhibit high ηet , i.e., anomalous electron transport properties. Interestingly, such electron transporting properties of the nano-structured TiO2 films should appear only when the films are filled with highly ionic electrolyte. Electron transport in nano-structured TiO2 films in a liquid electrolyte has been described in terms of carrier diffusion due to a lack of large electric field gradient in the film [11–26]. The diffusion coefficients of the dye-coated TiO2 and the plain TiO2 electrodes were reported to depend on light intensities, ranging from 10−8 to 10−4 cm2 s−1 . A trapping model [13,21,24–26] can well explain such slow diffusion of electrons, where electrons spend a large fraction of transient time in traps, as shown in Fig. 2. According to the model, the number of traps, their energy distribution, and the steady state population of the trapped electrons affect the diffusion coefficients. For the nano-structured electrodes, the effect of traps is significant since the number of traps is extremely large due to the high surface area and boundaries where traps likely exist. It is interesting to note that other factors, which affect measured diffusion coefficients, are the concentration and diffusion coefficient of ions in an electrolyte. The effects of electrolyte ion concentrations on measured diffusion coefficient were reported by Solbrand et al. [19]. The diffusion coefficient and total amount of charges decreased as the electrolyte concentration decreased. Kopidakis et al. proposed an ambipolar diffusion mechanism to interpret the effect of electrolyte on the electron transport in a mesoporous TiO2 –electrolyte system [23]. Photoinjected electrons in TiO2 were surrounded by an electrolyte consisting of various kinds of ionic species. The ambipolar diffusion coefficient is expressed by Damb =
(n + p) (n/D p ) + ( p/Dn )
(5)
where n and p are the density of electrons and cations, and Dn and D p are the diffusion coefficients of electrons and cations, respectively. The diffusion coefficient of
Fig. 2: Schematic view of electron transport mechanism in nano-structured TiO2 electrode soaked in high ionic strength electrolyte.
Control of charge transfer and interface structures in nano-structured DSCs 87
88
S. Yanagida et al.
electrons depends not only on the density of electrons but also on the concentration of ions in electrolytes such as lithium perchlorate (LiClO4 ). The density of electrons is proportional to light intensity. 3.2. Evaluation method of electron diffusion in plain nano-structured TiO2 layers Electron diffusion in mesoporous TiO2 films soaked in electrolyte was determined by laser-induced transient current measurements, which is a kind of time-of-flight technique [14,27–29]. When the electrode is illuminated by UV-light from the TiO2 – electrolyte interface side, only the shallow region of electrode surface is excited, since the absorption coefficient of TiO2 in the UV region is large (∼10−7 m−1 ). The charge separation in the TiO2 electrode after excitation is completed in the sub-nanosecond range. The hole at the valence band of TiO2 is rapidly removed by supplying an electron from quenchers in solvent, suppressing initial electron–hole pair recombination inside the TiO2 . When a laser is employed as the excitation light source, electrons are generated at the shallow region of the electrolyte side of the TiO2 electrode, and then travel to the substrate side by a diffusion process because of thermal fluctuations in the system. Therefore, with increasing thickness of the electrode the carriers travel a longer distance. The electron diffusion coefficient of the nano-structured TiO2 electrode can be determined by analyzing the photocurrent transient response using Fick’s diffusion model. Neglecting a current due to electrostatic repulsion in the solution of the timedependent diffusion equation, the time for the current maximum, tpeak , appears when tpeak = W 2 /2D
(6)
where D is the electron diffusion coefficient, W is film thickness, and tpeak is the time of photocurrent maximum [27]. The nano-structured TiO2 electrodes were prepared on transparent conducting glass (F-doped SnO2 ) and annealed at 450°C for 30 min in air before the measurements. The electrodes were immersed in an ethanolic or acetonitrile electrolyte solution composed of LiClO4 , 0.7 M and using a platinum wire as a counter electrode. Short duration of excitation light was obtained using a 10 Hz Nd-YAG laser (The Quanta-Ray INDI Series Pulsed Nd : YAG Lasers, pulse width 7 ns, wavelength 355 nm) and time transient photocurrent was monitored by a digital oscilloscope (Tektronix TDS 3052, 500 MHz). The schematic picture is shown in Fig. 3. Filters were employed to prevent the effects of the 2-fold and 4-fold wavelength light. The density of photo-formed electrons in TiO2 film, i.e., light intensity was controlled by using ND filters. The geometric area of an electrode was fixed as 0.093 cm−2 by an aperture. All the measurements were performed in air with at least 3-min intervals between measurements. Thickness of the films was measured by a Dectak profilometer and the surface roughness was about 5%. 3.3. Electron transport in nano-structured TiO2 layers Transient curves of the typical photocurrent observed for films having different thicknesses are shown in Fig. 4 [28]. The positions of the peaks in the current rise were shifted to longer times with increasing film thickness, indicating that the electrons have
Control of charge transfer and interface structures in nano-structured DSCs
89
Fig. 3: The setup for measurement of the diffusion coefficient of electrons in nano-structured TiO2 .
Fig. 4: Typical transient photocurrent observed for A1 films with different thicknesses: Nd : YAG (3ω = 355 nm), 7 ns, 0.98 mJ cm−2 , 0.093 cm2 .
90
S. Yanagida et al.
Fig. 5: The current peak vs. the square of each different nano-structured TiO2 film thickness sintered at 450°C. A1 (•), A2 ( ), A3 (), A4 (), A5 (), R1 (◦), R2 ( ), A5 containing 20 wt% of larger particles (), A2 sintered at 550°C () and R1 treated with TiCl4 (♦).
to travel a longer distance to reach the conducting glass layer. The decay of the current was slower for the samples with larger thickness. Therefore, the experimental observations were qualitatively in agreement with those expected for the diffusive process of electrons in the films. The generated charge in our setup was evaluated as 1.2 µC for the electrode with 5 µm in thickness by integrating the transient curve between the times 0 and 100 ms and was almost constant for the films with <12 µm thickness, but decreased to 0.5 µC for those 15 µm in thickness. This is probably due to the back electron transfer to electron acceptors in the electrolyte solution such as oxidized ethanol, water, or dissolved oxygen. We plotted W 2 vs. the peak time tpeak for the electrodes made with the TiO2 samples according to the relationship expressed by Eq. 6 and obtained Fig. 5. A straight line was obtained for all of the examined nano-structured TiO2 films, which indicates that the transport of electrons in the TiO2 electrodes will follow the diffusion model. The diffusion coefficients of electrons in nano-structured TiO2 (D) were determined for comparing the electron transport properties of the TiO2 films made with different TiO2 nanoparticles [28] as follows. A1 was anatase TiO2 having spherical shape with an average size of 12 nm, which was prepared from titanium isopropoxide, containing a small amount of brookite crystal structure. A2 was anatase TiO2 having a higher crystallinity than A1. A3 was phase-pure anatase TiO2 having rod-like shape with an average size of 13 × 4 nm (supplied by Catalysis & Chemicals Industry (denoted as CCI, hereafter) HPW-25R). A4 was phase-pure anatase having cubic shape with an
Control of charge transfer and interface structures in nano-structured DSCs
91
Table 1 Characteristics of TiO2 nanoparticles and the electron diffusion coefficients of the electrodes
A1 A1TiCl4 c A2 A2550 d A3 (CCI) A4 (CCI) A5 (P25) A5Large e R1 R1TiCl4 c R2 (CCI)
Structure a
Shape
Size (nm) b
D (×10−5 cm2 s−1 )
A/Amor A/Amor A A A A A/R A/R R R R
Spheric Spheric Cubic Cubic Rod-like Cubic Spheric Spheric Spheric Spheric Rod-like
12 12 12 12 13/34 11 21 21 27 27 23/73
2.2 2.2 0.3 2.0 4.0 4.1 4.0 4.0 0.1 0.4 0.3
A = anatase structure, Amor = amorphous phase, R = rutile structure. Average size. c Treated with aq. TiCl4 solution. d Annealed at 550°C. e With 20 wt% of large TiO2 (Fluka). a
b
average size of 11 nm (CCI, HPW-10R). A5 was nano-structured TiO2 (P25) having mixed crystal structure (anatase : rutile = 8 : 2), spherical shape and an average size of 21 nm (supplied by Nippon Aerosil). R1 was rutile TiO2 having spherical shape with average size 27 nm, which was prepared from titanium isopropoxide. R2 was rutile crystal structure, rod-like shape and average size of 23 × 73 nm (CCI, HPR-16). The D values determined for all the electrodes made from the TiO2 samples are listed in Table 1. In general comparison, the electron diffusion coefficient of the rutile electrodes (DR1 and DR2 ) was almost one order smaller than that of the anatase electrodes (DA1 , DA3 , DA4 and DA5 ). Similar diffusion coefficients were obtained for A3 and A4 which were prepared by the common synthesis method, but possessed different particle sizes and shapes. The A2 having the anatase structure but higher crystallinity than A1 gave a smaller D value, but one which was highly increased by annealing at 550°C, being closed to that of DA1 . The detailed comparison of the diffusion coefficient depending on the morphology of nano-structured TiO2 film well indicates the importance of surface structure of anatase TiO2 for fabrication of good electron transport layers of DSCs. 3.4. Effects of electrolytes on the electron transport Ambipolar diffusion coefficients of electrons in nano-structured anatase TiO2 –electrolyte systems were determined in the presence of a wide range of concentrations of Li+ , Na+ , Mg2+ , tetrabutylammonium cation (TBA+ ) or dimethylhexylimidazolium cation (DMHI+ ) in electrolytes [29]. An imidazolium salt was examined as an electrolyte, because imidazolium cations are essential components to achieve high performance of dye-sensitized solar cells [30–32].
92
S. Yanagida et al.
Fig. 6: Photocurrent transients for 7.2 µm-thick nano-structured TiO2 electrodes in acetonitrile with various 17 cm−3 . From below 0.3, 0.7, 1.0, concentrations of TBA+ ClO− 4 at average electron density n = 2.1 × 10 2.0, 5.0, 10, 20, 40, 80, 100, 200, 500 and 1000 mM of the cation concentration. Inset shows longer time scale at low concentrations of TBA+ from below 0.3, 0.7, 1.0 and 2.0 mM.
Fig. 6 shows the photocurrent transients in acetonitrile-electrolytes consisting of various densities of TBA+ at a photoelectron density of n = 2.1 × 1017 cm−3 . The shape of photocurrent transients changed in the series and the peak of the photocurrent increased with an increase of TBA+ density at a comparable photogenerated electron density. Fig. 7 shows the change in diffusion coefficient (DAmb ) as a function of TBA+ density at different electron densities (n). A solid line obtained by fitting DAmb according to Eq. 5 is shown in Fig. 7. The fitted curve of DAmb agreed closely with the measured DAmb at Dn = 2.1 × 10−4 cm2 s−1 and D p = 2.8 × 10−6 cm2 s−1 . The D p obtained by the curve fitting is close to the limiting cation diffusion coefficient −6 cm2 s−1 derived from the Nernst equation using the limiting molar D∞ p = 2.32 × 10 conductance Λ0 = 61.63 S cm2 mol−1 [33]. The diffusion coefficients (DAmb ) increase with an increasing cation density. Different n values give different DAmb , and DAmb was larger at higher n. This difference supports not only an ambipolar diffusion
Control of charge transfer and interface structures in nano-structured DSCs
93
Fig. 7: Diffusion coefficients of electrons in nano-structured TiO2 as a function of TBA+ ClO− 4 density in acetonitrile at different electron densities n, 2.1 × 1017 cm−3 (•) and 5.0 × 1016 cm−3 ( ). A line is calculated from Eq. 5 where Dn = 2.1 × 10−4 cm2 s−1 and Dp = 2.8 × 10−6 cm2 s−1 .
mechanism but also the trap-filling effect of electrons on electron transport in a DSC [12,13,22,34]. DAmb as a function of the cation density was also determined for the case of Li+ , + Na , Mg2+ and DMHI+ , as shown in Fig. 8. The DAmb depends on kind of the cation and decreases in the order DMHI+ > TBA+ > Na+ > Li+ > Mg2+ . Interestingly, the diffusion coefficients increased drastically in the presence of highly concentrated Li+ or DMHI+ without fitting well with the ambipolar diffusion mechanism. These behaviors were in contrast with that in the case of TBA+ . We have reported previously that the measured D was well interpreted with the ambipolar diffusion mechanism for a wide range of Li+ densities in ethanolic electrolyte [27]. The difference observed in acetonitrile and in ethanol could be explained by the difference in Li+ adsorption behavior on TiO2 . The adsorption of Li+ in the TiO2 electrodes increases the local cation density at the TiO2 surface. And the results suggest that the adsorption might form favorable trap states in the surface TiO2 and influences the electron transport. The behavior of DAmb as a function of the cation density in the case of DMHI+ was similar to that in the case of Li+ rather than TBA+ , although DMHI+ is a quaternary ammonium cation. Taking into account the similarity of DAmb in the case of DMHI+ and Li+ , DMHI+ adsorption on the TiO2 was expected. While the adsorption of TBA+ on TiO2 (P25) surface was negligible, the estimated amount of the adsorbed DMHI+ on the TiO2 surface was ca. 100 molecules per nm2 , that is too large for a monolayer of DMHI+ on the TiO2 , suggesting multi-layered adsorption of DMHI+ on TiO2 . The proximity of multi-layered DMHI+ to all particles on the surface causes screening of photoinjected electrons [20,35], leading to their enhanced transport.
94
S. Yanagida et al.
Fig. 8: Diffusion coefficients of electrons in nano-structured TiO2 as a function of cation density of LiClO4 17 (a), NaClO4 (b), Mg(ClO4 )2 (c) and DMHI+ ClO− cm−3 (Li+ ), 1.6 × 1017 cm−3 4 (d) when n = 2.1 × 10 (Na+ ), 2.1 × 1017 cm−3 (Mg2+ ) and 1.8 × 1017 cm−3 (DMHI+ ) in acetonitrile. Lines are calculated from Eq. 5 at lower cation density.
Control of charge transfer and interface structures in nano-structured DSCs
95
3.5. Effect of surface states in nano-structured TiO2 Positive introduction of oxygen vacancies or surface states as electron trapping site was expected to increase electron diffusion coefficients and to contribute to an increase of ηet of the TiO2 films when they are formed just below the conduction band. The oxygen vacancies should originate from the 3d-orbital of Ti3+ but that is unstable in the presence of oxygen and water molecules. An alternative way to produce Ti3+ in the TiO2 films is fluorine doping, i.e., the chemical replacement of O2− with F− . A high physical stability can be expected in the fluorine-doped TiO2 films (TiO2 /F), because O2− sites come to be occupied with the similar sized F− [36,37]. Thus we applied a fluorine doped TiO2 (TiO2 /F) to DSCs. Fluorine-doped TiO2 was synthesized by hydrolysis of Ti(Oi Pr)4 followed by autoclaving in the presence of HF. The content of fluorine in the doped particles was estimated as F/Ti = 0.0011 by applying the fundamental parameter method to the observed F-Kα signal on X-ray fluorescent analysis (Rigaku, ZSX100e; RX35 analyzing crystal; F-PC detector). The films (TiO2 /F) prepared in the similar procedure had the similar morphology to that of fluorine-free TiO2 . For comparison, TiO2 films with excess oxygen vacancies were prepared by calcination under N2 atmosphere, and the formation of oxygen vacancies was proved by absorption of UV-VIS spectroscopy at longer wavelength. Laser pulse induced photocurrent measurements were performed under comparable conditions (in ethanolic solution of LiClO4 ) for the TiO2 /F film and fluorine-free TiO2 films with and without introduction of oxygen vacancies. Fig. 9 shows that the photocurrent transient is larger for the TiO2 /F film. The diffusion coefficients of the TiO2 /F film and the fluorine-free TiO2 film were determined to be 1.5 × 10−4 cm2 s−1 and 1.3 × 10−4 cm2 s−1 , respectively. TiO2 films with oxygen vacancies showed too poor current to determine the diffusion coefficient. Fig. 10 shows I –V curves of the cells fabricated by using these films with ∼4 µm thickness. Introduction of oxygen vacancies reduced photocurrent and photovoltage drastically but fluorine doping induced
Fig. 9: Photocurrent transient induced by UV pulsed laser for 4.0 µm-thick TiO2 (thin line) and TiO2 /F (bold line) in 0.7 M LiClO4 ethanolic solution: Nd : YAG (3ω = 355 nm), 6 ns, 15 µJ/0.02 cm2 .
96
S. Yanagida et al.
Fig. 10: Photocurrent density–voltage curve of dye-sensitized solar cells with 4 µm-thick nano-structured TiO2 film under AM 1.5 irradiation: TiO2 (thin line), TiO2 /F (bold line) and TiO2 with excess oxygen vacancies (dashed line).
the increase of photovoltage [38]. These results suggest that while the oxygen vacancies may produce deep trap states as back electron transfer site, the doped fluorine states may serve as shallow trap sites just below the conduction band that might serve effective electron-transporting sites with a slight shift of the Fermi level to the negative.
4. Charge transport in iodide/polyiodide electrolytes of DSCs 4.1. Mechanistic studies using quasi-solid-state electrolytes produced by low-molecular-weight gelators Quasi-solid-state DSCs were fabricated using low-molecular-weight gelators (Scheme 2) [39,40]. They showed comparable photoenergy conversion efficiencies to the liquid cell at high illumination intensity up to AM 1.5 (one sun). This fact implies the specific charge transport of the electrolytes of the iodine/iodide redox couple. The employed electrolyte consisted of dimethylpropylimidazolium iodide (DMPImI), LiI, and I2 , and tert-butylpyridine (BP) as an additive and methoxypropylnitrile (MPN) as solvent. Conductivity measurements of the electrolyte phases revealed that the gelation does not affect largely the conductivity of the electrolyte and that the conductivity increased with an increase of iodine in both gel electrolytes and liquid electrolyte (Fig. 11). Raman spectroscopic measurements confirmed the formation of polyiodide ions − − (I− 2n+1 ) such as I3 and I5 by addition of iodine. The self-diffusion of iodide species in the gel electrolyte was about a quarter of that of I− in acetonitrile. Lindquist’s group previously reported that the diffusion of I− 2n+1 in nano-structured TiO2 space is one order
Control of charge transfer and interface structures in nano-structured DSCs
97
Scheme 2.
of magnitude lower than that in solution [41]. The less-mobile polyiodide ions in electrolyte did not influence the charge transport process, which is in contrast to the lower limiting molar conductivity of polyiodide species than monoiodide [33]. In addition, the rather high concentration (0.5 M) of polyiodide and iodide species in the electrolyte implies that such iodide species are located in the proximity of each other (∼0.8 nm). These facts suggest that the effective charge transport in the electrolyte phase should be rationalized as due to electron hopping or iodine exchange (Grotthuss-type) mechanism caused by the rather packed polyiodide species in the electrolytes. Fig. 11 depicts the electronic transport in the iodide and polyiodide species in electrolyte. The effective electron injection (see Fig. 1) from iodide species to oxidized dye may suggest chemical interaction of polyiodide species and thiocyanide groups of the ruthenium dye molecule as shown in Fig. 12. 4.2. Quasi-solid stated DSCs using imidazolium molten iodides as electrolytes In the measurement of the conductivity of the DSC’s I− /I− 3 redox electrolyte systems, the presence of imidazolium iodides instead of lithium iodide was found to increase the electron conductance of the redox electrolyte (Fig. 11) [40]. On the other hand, the presence of imidazolium cations accelerates the diffusion coefficient of electrons in the mesoporous TiO2 phase (Dn ) through their multi-layer adsorption [29]. In addition,
98
S. Yanagida et al.
Fig. 11: Effect of the concentration of Gelator 2 on conductivity for electrolytes containing 0.5 M of iodide salts (DMPImI: ◦, LiI: ) and electrolytes containing 0.5 M of iodide salts with 0.1 M of I2 (DMPImI: •, LiI: ).
Fig. 12: Schematic view of interactions at the interfaces of nano-structured TiO2 /N3 and N3/iodide electrolyte and electron transport mechanism in iodide/polyiodide electrolyte.
some imidazolium iodides are known as chemically inert room-temperature molten salts, i.e., ionic liquids. Currently, many ionic liquids have attracted much attention as electrolyte in electrochemistry because of their features such as high ionic conductivity, non-volatility, thermal stability and non-flammability [42–44]. In fact, such imidazolium salts were applied to DSCs as non-volatile electrolyte solvent [31,45]. However, the
Control of charge transfer and interface structures in nano-structured DSCs
99
Scheme 3.
conversion efficiency of the cells showed lower than that of the cells using liquid electrolyte containing organic solvent. When we applied 1-alkyl-3-methylimidazolium iodides (Scheme 3) to the electrolyte of DSCs by combining iodine but without using any volatile solvents, the electrolyte with 1-hexyl-3-methylimidazolium iodide gave respectable photoenergy conversion efficiency (η = 5.0%) when combined with 1/10 molarity of iodine. Further, solidification of the molten imidazolium salts using gelator (Scheme 2, Gelator 1) showed comparable conversion efficiency (5.0%) under AM 1.5 irradiation (Fig. 13) [46]. Under dry heat test, the quasi-solid-state imidazolium DSCs showed a higher stability than the quasi-solid-state DSCs fabricated using organic solvent with the same gelator [47]. It is interesting to note that the imidazolium DSC is non-flammable because of the high boiling point of the imidazolium salts as a room-temperature molten salt.
Fig. 13: Photocurrent–voltage curves of dye-sensitized solar cells with 1-hexyl-3-methylimidazolium iodide containing 8.7 wt% of I2 under AM 1.5 irradiation without (dotted curve) and with (solid curve) 1.4 wt% of Gelator 1.
100
S. Yanagida et al.
Scheme 4.
5. Importance of interface control in DSC fabrication 5.1. Interface structures affecting efficiencies, ηei and ηhi Dye adsorption on TiO2 is important for fast electron injection in a dye-sensitization mechanism. In the case of N3/TiO2 interface, the dye molecules are chemically adsorbed through one carboxylic group on each dcbpy ligand by an ether-like bond [48] or bidentate coordination [49] contributing to high efficiency of ηei . The importance of a covalent bond between dye molecules and the hole transport layer was also displayed when polypyrrole (PPy) as a hole transport layer and a Ru dye with pyrrole group, cis-RuII (dcbpy)2 (pmp)2 , (pmp = 3-(pyrrole-1-ylmethyl)-pyridine; Scheme 4) were employed [50–52]. As for TiO2 interface control, Huang et al. reported the suppression of back electron transfer to redox electrolytes by surface treatment with pyridine derivatives [53] or cholic acids [54,55]. Covering of solvent-exposed interface parts of OTE and TiO2 surfaces by insulating molecules such as poly(phenylene oxide-co-2-allylphenylene oxide) or poly(methylsiloxane) contributes to a decrease of the back electron transfer [56]. 5.2. Important role of interfaces affecting efficiencies, ηec and ηhc With regard to conducting window and counter electrodes, the conductivity relating to surface structures and morphology and the transparency especially at window OTE play some decisive roles. On the other hand, Willig pointed out the importance of the junction at window OTE (SnO2 /F) and nano-structured TiO2 as a driving force of electron transfer in DSCs [20]. Gregg et al. reported the difference in role of metal deposition on window OTE in view of the work function [57]. These facts and findings are closely related with interfacial control affecting ηec . Fullerenes are good electron acceptors due to the small reorganization energy in
Control of charge transfer and interface structures in nano-structured DSCs
101
Scheme 5.
electron transfer and are applied to a variety of electron mediators in electron transfer reactions [58–60]. We successfully attempted to apply a C60 derivative to acceleration of electron capture at OTE in DSCs. Since the HOMO level of C60 is almost comparable to the conduction band of TiO2 , two systems, OTE/C60 /TiO2 /N3 and OTE/TiO2/C60 /N3, were examined to know the role of C60 at interfaces of DSCs. As a C60 derivative, 1,2methanato[60]-fullerene-61,61-dicarboxylic acid [61,62] was adsorbed on an OTE film or on OTE/TiO2 film from toluene solution. The resulting OTE/C60 film was coated with nano-structured TiO2 and sintered at 400°C for 30 min, and then N3 dye was adsorbed on the electrodes. The resulting OTE/C60 /TiO2 /N3 and OTE/TiO2/C60 /N3 systems can be depicted as shown in the Scheme 5. In the system of OTE/TiO2/C60 /N3, the amount of the adsorbed dye molecules was decreased to about 40% of that of the OTE/TiO2/N3 system fabricated as a reference system. On the other hand, the OTE/C60 /TiO2 /N3 system gave 80% dyeadsorption when compared to the OTE/TiO2/N3 system. While negligible response in measurement of IPCE was observed in OTE/TiO2 /C60 /N3, the OTE/C60 /TiO2 /N3 system gave a response comparable to that of the OTE/TiO2/N3 system in spite of the decrease of dye adsorption (Fig. 14). This result would suggest that C60 has a potential to mediate electron capture from TiO2 to SnO2 /F, i.e., increase of ηec . Platinum or carbon deposition on counter OTE (SnO2 /F) electrode is known as a requisite in DSC fabrication in view of efficiency in hole capture at counter electrodes. In fabrication of solid state DSCs using polypyrrole as a hole transport layer and a Ru dye with pyrrole group, cis-RuII (dcbpy)2 (pmp)2 , carbon-based counter electrode improved cell performance compared to the cell with gold or platinum counter electrode [52]. A good electric contact of the hole transport layer of polypyrrole with the carbon electrode will enhance the efficiency ηhc . 6. Conclusions The respectable conversion efficiency of Grätzel solar cells is now rationalized by 7 high efficiencies, i.e., light harvesting efficiency, electron injection, transport and capture efficiencies, and hole injection, transport and capture efficiencies. Interfacial structures
102
S. Yanagida et al.
Fig. 14: IPCE spectra of dye-sensitized solar cells with 8.6 µm-thick TiO2 (P25) film and N3 under AM 1.5 irradiation: OTE/TiO2 /N3 (thin line), OTE/C60 /TiO2 /N3 (bold line) and OTE/TiO2 /C60 /N3 (dashed line).
between nano-structured TiO2 , dye molecules, iodide/polyiodide couple, and window OTE and counter electrodes have influence on the efficiency for each step. For keeping good electron diffusion and charge transport both in TiO2 and in iodide/polyiodide electrolyte, the presence of highly concentrated imidazolium species is favorable for DSC systems. Employment of a series of imidazolium iodides as ionic liquid electrolyte led to successful fabrication of DSCs with respectable efficiency and thermal stability when followed by solidification using a low-molecular-weight gelator. Control of the interfaces to reduce the back electron transfer by chemically modifying the surfaces of TiO2 and OTE materials is also an important subject to improve DSC performance. In the solidification of DSCs, we must maintain vectorial charge flow with keeping each efficiency optimal.
Acknowledgements This work was partially supported by Grant-in-Aid for Scientific Research (A) (No. 11358006), and by Grant-in-Aid for the Development of Innovative Technology (No. 12310) from the Ministry of Education, Culture, Sports, Science and Technology of Japan.
References 1. 2. 3.
B. O’Regan and M. Grätzel, Nature (London) 353, 737 (1991). M.K. Nazeeruddin, A. Kay, I. Rodicio, R. Humphry-Baker, E. Müller, P. Liska, N. Vlachopoulos, and M. Grätzel, J. Am. Chem. Soc. 115, 6382 (1993). M. Grätzel, Nature (London) 414, 338 (2001).
Control of charge transfer and interface structures in nano-structured DSCs 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40.
103
Y. Tachibana, J.E. Moser, M. Grätzel, D.R. Klug, and J.R. Durrant, J. Phys. Chem. 100, 20056 (1996). T. Hannappel, B. Burfeindt, W. Storck, and F. Willig, J. Phys. Chem. B 101, 6799 (1997). J.S. Salafsky, W.H. Lubberhuizen, E. van Faassen, and R.E.I. Schropp, J. Phys. Chem. B 102, 766 (1998). S.A. Haque, Y. Tachibana, D.R. Klug, and J.R. Durrant, J. Phys. Chem. B 102, 1745 (1998). R.J. Ellingson, J.B. Asbury, S. Ferrere, H.N. Ghosh, J.R. Sprague, T. Lian, and A.J. Nozik, J. Phys. Chem. B 102, 6455 (1998). S.A. Haque, Y. Tachibana, R.L. Willis, J.E. Moser, M. Grätzel, D.R. Klug, and J.R. Durrant, J. Phys. Chem. B 104, 538 (2000). S.Y. Huang, G. Schlichthörl, A.J. Nozik, M. Grätzel, and A.J. Frank, J. Phys. Chem. B 101, 2576 (1997). S. Södergren, A. Hagfeldt, J. Olsson, and S.-E. Lindquist, J. Phys. Chem. 98, 5552 (1994). F. Cao, G. Oskam, G.J. Meyer, and P.C. Searson, J. Phys. Chem. 100, 17021 (1996). P.E. de Jongh and D. Vanmaekelbergh, Phys. Rev. Lett. 77, 3427 (1996). A. Solbrand, H. Lindström, H. Rensmo, A. Hagfeldt, S.-E. Lindquist, and S. Södergren, J. Phys. Chem. B 101, 2514 (1997). G. Schlichthörl, S.Y. Huang, J. Sprague, and A.J. Frank, J. Phys. Chem. B 101, 8141 (1997). L. Dloczik, O. Ileperuma, I. Lauermann, L.M. Peter, E.A. Ponomarev, G. Redmond, N.J. Shaw, and I. Uhlendorf, J. Phys. Chem. B 101, 10281 (1997). G. Franco, J. Gehring, L.M. Peter, E.A. Ponomarev, and I. Uhlendorf, J. Phys. Chem. B 103, 692 (1999). G. Schlichthörl, N.G. Park, and A.J. Frank, J. Phys. Chem. B 103, 782 (1999). A. Solbrand, A. Henningsson, S. Södergren, H. Lindström, A. Hagfeldt, and S.-E. Lindquist, J. Phys. Chem. B 103, 1078 (1999). K. Schwarzburg and F. Willig, J. Phys. Chem. B 103, 5743 (1999). J. Nelson, Phys. Rev. B 59, 15374 (1999). A.C. Fisher, L.M. Peter, E.A. Ponomarev, A.B. Walker, and K.G.U. Wijayantha, J. Phys. Chem. B 104, 949 (2000). N. Kopidakis, E.A. Schiff, N.-G. Park, J. van de Lagemaat, and A.J. Frank, J. Phys. Chem. B 104, 3930 (2000). J. van de Lagemaat and A.J. Frank, J. Phys. Chem. B 104, 4292 (2000). J. van de Lagemaat, N.-G. Park, and A. J. Frank, J. Phys. Chem. B 104, 2044 (2000). J. Nelson, S.A. Haque, D.R. Klug, and J.R. Durrant, Phys. Rev. B 63, 205321 (2001). S. Nakade, S. Kambe, T. Kitamura, Y. Wada, and S. Yanagida, J. Phys. Chem. B 105, 9150 (2001). S. Kambe, S. Nakade, Y. Wada, T. Kitamura, and S. Yanagida, J. Mater. Chem. 12, 723 (2002). S. Kambe, S. Nakade, T. Kitamura, Y. Wada, and S. Yanagida, J. Phys. Chem. B 106, 2967 (2002). C.J. Barbé, F. Arendse, P. Comte, M. Jirousek, F. Lenzmann, V. Shklover, and M. Grätzel, J. Am. Ceram. Soc. 80, 3157 (1997). N. Papageorgiou, Y. Athanassov, M. Armand, P. Banhôte, H. Pettersson, A. Azam, and M. Grätzel, J. Electrochem. Soc. 143, 3099 (1996). P. Banhôte, A.-P. Dias, M. Armand, N. Papageorgiou, K. Kalyanasundaram, and M. Grätzel, Inorg. Chem. 35, 1168 (1996). J. Berthel, L. Iberl, J. Rossmaier, H.J. Gores, and B. Kaukal, J. Solut. Chem. 19, 321 (1990). K. Schwarzburg and F. Willig, Appl. Phys. Lett. 58, 2520 (1991). D. Cahen, G. Hodes, M. Grätzel, J.F. Guillemoles, and I. Riess, J. Phys. Chem. B 104, 2053 (2000). S.N. Subbarao, T.H. Yun, R. Kershaw, K. Dwight, and A. Wold, Mat. Res. Bull. 13, 1461 (1978). S.N. Subbarao, T.H. Yun, R. Kershaw, K. Dwight, and A. Wold, Inorg. Chem. 18, 488 (1979). S. Yamagida, T. Kitamura, and M. Kohmoto, Electrochem. 70, 399 (2002). W. Kubo, K. Murakoshi, T. Kitamura, Y. Wada, K. Hanabusa, H. Shirai, and S. Yanagida, Chem. Lett., 1241 (1998). W. Kubo, K. Murakoshi, T. Kitamura, S. Yoshida, M. Haruki, K. Hanabusa, H. Shirai, Y. Wada, and S. Yanagida, J. Phys. Chem. B 105, 12809 (2001).
104 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62.
S. Yanagida et al. Z. Kebede and S.-E. Lindquist, Solar Energy Mater. Solar Cells 51, 291 (1998). R. Hagiwara and Y. Ito, J. Fluorine Chem. 105, 221 (2000). T. Welton, Chem. Rev. 99, 2071 (1999). A.B. McEwen, H.L. Ngo, K. LeCompte, and J.L. Goldman, J. Electrochem. Soc. 146, 1687 (1999). H. Matsumoto, T. Matsuda, T. Tsuda, R. Hagiwara, Y. Ito, and Y. Miyazaki, Chem. Lett., 26 (2001). W. Kubo, T. Kitamura, K. Hanabusa, Y. Wada, and S. Yanagida, Chem. Commun., 374 (2002). Japanese Industrial Standard, Environmental and endurance test methods for amorphous solar cell modules, JIS C 8938, (1995). K. Murakoshi, G. Kano, Y. Wada, S. Yanagida, H. Miyazaki, M. Matsumoto, and S. Murasawa, J. Electroanal. Chem. 396, 27 (1995). K.S. Finnie, J.R. Bartlett, and J.L. Woolfrey, Langmuir 14, 2744 (1998). K. Murakoshi, R. Kogure, Y. Wada, and S. Yanagida, Chem. Lett., 471 (1997). K. Murakoshi, R. Kogure, Y. Wada, and S. Yanagida, Solar Energy Mater. Solar Cells 55, 113 (1998). T. Kitamura, M. Maitani, M. Matsuda, Y. Wada, and S. Yanagida, Chem. Lett., 1054 (2001). S.Y. Huang, G. Schlichthörl, A.J. Nozik, M. Grätzel, and A.J. Frank, J. Phys. Chem. B 101, 2576 (1997). A. Kay and M. Grätzel, J. Phys. Chem. 97, 6272 (1993). K. Hara, H. Sugihara, Y. Tachibana, A. Islam, M. Yanagida, K. Sayama, H. Arakawa, G. Fujihashi, T. Horiguchi, and T. Kinoshita, Langmuir 17, 5992 (2001). B.A. Gregg, F. Pichot, S. Ferrere, and C.L. Fields, J. Phys. Chem. B 105, 1422 (2001). F. Pichot and B.A. Gregg, J. Phys. Chem. B 104, 6 (2000). H. Imahori, K. Hagiwara, M. Aoki, T. Akiyama, S. Taniguchi, T. Okada, M. Shirakawa, and Y. Sakata, J. Am. Chem. Soc. 118, 11771 (1996). M. Knupfer, Surface Sci. Rep. 42, 1 (2001). N.S. Sariciftci, Prog. Quant. Electr. 19, 131 (1995). X. Campas and A. Hirsch, J. Chem. Soc. Perkin Trans. 1, 1595 (1997). I. Lamparth, G. Schick, and A. Hirsch, Liebigs Ann./Recueil, 253 (1997).
84
S. Yanagida et al.
Scheme 1.
a DSC is noted to have the following advantages over silicon solar cells; respectable photo-conversion efficiency with high fill factor, low cost and earth-friendly materials (low material cost), economical production facilities like screen printing and ink jet printing (low labor cost), high open-circuit photovoltage at low light intensity that is favorable for indoor use, and less sensitiveness to angle of incidence of solar radiation which leads to an increase of total conversion of solar light per day. It is further interesting to note that a DSC permits the construction of transparent solar cell modules, so can be used in window and roof lighting. The photoconversion efficiency can be obtained by measuring short circuit photocurrent density (Jsc ), open-circuit photovoltage (Voc ) and fill factor (ff ) under one sun irradiation (Is ) conditions (Eq. 1). η = (Jsc × Voc × ff )/Is
(1)
The respectable photo-conversion efficiency can be explained by efficiency of light harvesting (ηlh ) of the sensitizing Ru dye and efficiencies of transport of photo-formed electrons and holes in a DSC, i.e., ηe and ηh , as expressed by Eq. 2. η = ηe × ηlh × ηh
(2)
The efficiency of transport of the photo-formed electrons can be subdivided in three factors, i.e., efficiency of electron injection from the excited dye molecule, ηei , efficiency of electron transport in nano-structured TiO2 phase, ηet , and efficiency of electron collection at the transparent electrode, ηec , as expressed by Eq. 3. ηe = ηec × ηet × ηei
(3)
The efficiency of transport of the photo-formed holes can be shown as a product of the efficiency of hole injection into the electrolyte as a hole-transport phase, ηhi , efficiency of charge (hole) transport in the electrolyte phase, ηht , and efficiency of hole collection at the counter electrode, ηhc (Eq. 4). ηh = ηhi × ηht × ηhc
(4)
Nanotechnology and Nano-Interface Controlled Electronic Devices Editors: M. Iwamoto, K. Kaneto and S. Mashiko © 2003 Elsevier Science B.V. All rights reserved
CHAPTER 8
Materials and devices for ultrafast molecular photonics Toshihiko Nagamura Molecular Photonics Laboratory, Research Institute of Electronics, Shizuoka University, 3-5-1 Johoku, Hamamatsu 432-8011, Japan E-mail:
[email protected]
1. 2. 3.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Materials and methods for ultrafast photoresponse measurements . . . . . . . . . . . . . . . . . . Absorption changes over wide ranges of wavelength and time by photoinduced electrochromism of ion-pair charge-transfer complexes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Photoinduced electrochromism in 4,4 -bipyridinium salts with various counter ions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Ultrafast dynamics of photoinduced electrochromism . . . . . . . . . . . . . . . . . . . . . . 3.3. Charge resonance band in the near infrared region and its ultrafast dynamics 4. Parallel all-optical processing in guided wave geometry containing organic dyes . . . . 4.1. Ultrafast spatial light modulation and parallel optical recording based on photoinduced complex refractive index changes upon nanosecond laser excitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2. Reversible reflectance control of fs white light in guided wave geometry containing photochromic compounds upon fs laser photoexcitation . . . . . . . . 5. Ultrafast nonlinear optical responses amplified by photoexcitation . . . . . . . . . . . . . . . . . . 6. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
105 107 109 109 111 113 116
116 124 127 129 129 129
1. Introduction Novel materials, devices and systems are required for much faster data processing, much higher recording density, or much more specific and efficient sensing. Ultrafast switching materials which work in less than 1 ps are essential for teraHerz (THz) optical communication. Several attempts have been made for this purpose, which include optical switching by tunneling bi-quantum well semiconductors or organic nonlinear optical materials [1]. Magnetic “hard” disks and heat-mode optical disks such as phase change
106
T. Nagamura
Fig. 1: Why and how molecular photonics?
or magneto-optical memories have rapidly been increasing their recording density owing to the development of new pick-up heads or blue semiconductor lasers. But there is a physical limit in such conventional memories recording data bit by bit in a serial way on the surface or in the thin surface layer of recording materials. Organic molecules have many useful optical and electronic functions that can be easily controlled by the structures, substituents, or external fields. Specific interactions or organization of molecules further can afford much higher functions than isolated or randomly distributed molecules. Photons have many superior properties such as wavelength, polarization, phase, ultrashort pulse, or parallel processibility. Through strong interactions of molecules or molecular assemblies with photons, many superior properties of photons can be directly converted to changes in physical properties of materials such as fluorescence, absorption, refractive index, conductivity, or optical nonlinearity. These interactions will be utilized as molecular photonics with ultra high speed, ultra high density or high performance sensing as schematically shown in Fig. 1. Excited state formation, photochromism, photoinduced electron transfer are some examples among them. Photon-mode recording or switching based on these changes of electronic states can therefore achieve ultrafast multiple or three-dimensional recording and parallel processing with ultimate spatial resolution at a molecular level. There will be no doubt that molecular photonics based on interactions of molecules and photons has many advantages as compared with electric or photoelectric switching, heat-mode or magnetic recording, switching based on liquid crystals or thermal phenomena. We have been making efforts to develop new molecular photonics materials and devices by making various organized molecular systems and by optically controlling their electronic states. So far we have achieved photoinduced electrochromism, which is color changes due only to the photoinduced electron transfer and reverse reactions, molecular control of the lifetime and the wavelength of colored species over extremely wide ranges, amplified fluorescence quenching in LB films, photon-mode super-resolution
Materials and devices for ultrafast molecular photonics
107
to exceed the diffraction limit of light in optical memory based on transitory photobleaching of phthalocyanine derivatives, ultrafast all-optical two-dimensional control of reflectance and parallel optical self-holding switch based on photoinduced complex refractive index changes [2,3]. In the present chapter, some of our recent achievements will be discussed on materials and devices for ultrafast molecular photonics.
2. Materials and methods for ultrafast photoresponse measurements The structures of typical compounds employed in our study are shown in Fig. 2. 1,2-Dimethoxyethane (DME) and methanol solutions of TFPB− or iodide (I− ) salts of polymeric 4,4 -bipyridinium (PV2+ ) were used together with polymer films cast or spin-coated from these solutions. The content of 4,4 -bipyridinium ions in a PV2+ polymer is 3.3 × 10−4 mol/g. Various types styrylpyridinium (NS+ , DCS+ , NS+ CnNS+ ) tetraphenylborate (TPB− ) salts in DME were also employed to study ultrafast absorption changes in the visible and near-infrared (NIR) region. Several derivatives of phthalocyanines including water soluble copper-phthalocyanine (CuPcS) and zinc-phthalocyanine (ZnPcS) were used for photon-mode spatial light modulation. Phthalocyanines were used in solutions or in poly(vinyl alcohol) (PVA) films. A photochromic spiropyran derivative, 1,3,3-trimethylindolino-6 -nitrobenzopyrylospiran (SP) was dispersed in polystyrene or Arton® (JSR Co. Ltd.), which became a colored photomerocyanine (PM) type upon UV irradiation. For ultrafast dynamics studies, these dyes were excited in air at room temperature by the second harmonics (400 nm) of a femtosecond (fs) Ti : sapphire laser with a regenerative Ti : sapphire amplifier and a double path amplifier pumped with the second harmonics (532 nm) of a Nd : YAG laser. The amplified Ti : sapphire laser delivered pulses with a FWHM of 200–250 fs, 10 Hz repetition, a maximum power of 6 mJ/pulse at 800 nm. The fs probe white light was obtained by focusing the residual 800 nm light into a cell containing D2 O/H2 O (2 : 1) mixture after passing through a BBO crystal to obtain the second harmonics. The transient absorption and the dynamics were observed with a Photonic Multi-channel Analyzer (PMA; Hamamatsu Photonics) system using a dual photodiode array (Hamamatsu Photonics C6140) for the UV–visible and an InGaAs multichannel detector (Hamamatsu Photonics C5890-256) for the NIR absorption using an optical delay system. The intensities of the probe light with and without the pump pulses were averaged by 20 times. The block diagram of the fs transient absorption measurement system is shown in Fig. 3. For measuring basic properties of reflection type spatial light modulation or parallel optical switching, the sample plate was index-matched with a BK7 prism, which was set on a computer-controlled rotating stage. The writing beam was a ns OPO laser at 670 nm or the third harmonic (355 nm) of Nd : YAG laser; each with 8 ns pulse width, 0.03–2 mJ/pulse, and ca. 0.2 cm2 beam area. He–Ne laser (543.5 nm) through a half-wave plate, a polarizer, and a chopper was used as a reading beam. The time dependence of a reflected intensity at a given incident angle upon ns laser excitation at different powers was detected with a photomultiplier and was recorded with a digital oscilloscope terminated with 50 ohm. The spatial resolution was evaluated by using a
108
T. Nagamura
Fig. 2: The structures and abbreviations of typical compounds employed.
USAF Test Target as a mask. In order to measure reflectance changes of fs white light as a reading light upon excitation by 400 nm fs pulses as a writing light, an Arton® film with spiropyran spin-coated on silver film was set with a prism on a rotating stage as schematically shown in Fig. 4.
Materials and devices for ultrafast molecular photonics
109
Fig. 3: The block diagram of a fs laser flash photolysis system.
3. Absorption changes over wide ranges of wavelength and time by photoinduced electrochromism of ion-pair charge-transfer complexes 3.1. Photoinduced electrochromism in 4,4 -bipyridinium salts with various counter ions Various photochromic systems employing polymeric thin films or Langmuir–Blodgett (LB) films have recently attracted much interest in view of their promising applicability to high-speed and high-density photon-mode optical memory [2–7]. The photochromism reported so far involves changes of chemical bonds such as heterolytic cleavage of a pyran ring in spiropyrans, ring opening and closing in diarylethenes and fulgides, or trans–cis isomerization in azobenzenes [4–7]. Recently we have reported novel photoinduced electrochromism as schematically shown in Fig. 5 [2,3,8–24]. It is the color change due to photoinduced electron transfer in ion-pair charge transfer (IPCT) complexes of 4,4 -bipyridinium salts with tetrakis[3,5-bis(trifluoromethyl)phenyl]borate [25] (abbreviated to TFPB− ) and thermal back electron transfer reactions. No changes of chemical structure were involved in photochromism. TFPB− and iodide (I− ) salts of 4,4 -bipyridinium ions showed pale yellow and red colors, respectively, though each ion is colorless. These new absorption spectra above 350 nm in solutions were attributed to the IPCT complexes with 4,4 -bipyridinium ion as an acceptor and TFPB− or I− as a donor. It was thus demonstrated that these ion pairs made electronic interactions at the ground state partially transferring electronic charges from a donor to an acceptor. No
Fig. 4: Schematic representation of measurement systems for the incident angle and the time dependences of reflected intensity in guided mode thin films by fs laser.
110 T. Nagamura
Materials and devices for ultrafast molecular photonics
111
Fig. 5: Schematic representation of photoinduced electrochromism in 4,4 -bipyridinium IPCT salts.
color changes were observed with I− salts by steady photolysis, which is in contrast to TFPB− salts. From steady and laser photolysis results it has been shown that 4,4 -bipyridinium radical cations, which escaped from the geminate reaction immediately after the photoinduced electron transfer in less than 1 ps [22] upon IPCT excitation, became metastable owing to the bulk and chemical stability of TFPB− , to the restriction of molecular motion by the microenvironment, and also probably to the very high exothermicity of the reverse reaction in the Marcus inverted region [26,27]. Highly sensitive detection of photoinduced electrochromism and transient absorption spectra in ultra-thin LB and polymer films have been achieved by the conventional and the white-light optical waveguide method [28–33]. Such photoinduced electrochromism may be applied to ultrafast photon-mode optical memory and to redox sensors. 3.2. Ultrafast dynamics of photoinduced electrochromism Immediately upon excitation of an IPCT band with a fs laser at 400 nm, transient absorption was observed for both salts in solutions with a peak at about 600 nm, characteristic of 4,4 -bipyridinium radical cations. Fig. 6 shows the transient absorption spectra of PV2+ (I− )2 in methanol solution. A marked increase in the absorbance of the 4,4 -bipyridinium radical cations took place with a rise time of about 0.3 ps upon excitation. 4,4 -Bipyridinium radical cations were thus formed in a fs time scale by the photoinduced electron transfer from a donor I− to an acceptor 4,4 -bipyridinium upon IPCT excitation [22]. The time profiles of transient absorption at 600 nm are shown in Fig. 7 for (a) PV2+ (I− )2 in a film cast from DME and (b) PV2+ (TFPB− )2 in DME solutions. Both of them showed a very rapid rise in about 0.3 ps, which was almost the same as the time resolution of our fs Ti : sapphire laser measurement system with a regenerative amplifier. Similar extremely rapid formation of 4,4 -bipyridinium radical cations was observed for PV2+ (I− )2 salts in methanol and dimethylsulfoxide solutions upon IPCT excitation, respectively. These results demonstrated that the charge separated 4,4 -bipyridinium radical cations were formed directly upon IPCT excitation because of
112
T. Nagamura
Fig. 6: Transient absorption spectra of PV2+ (I− )2 in methanol solution upon fs laser excitation at 400 nm.
Fig. 7: The time profiles of transient absorption at 600 nm for (a) PV2+ (I− )2 in a film cast from DME and (b) PV2+ (TFPB− )2 in DME solutions.
the nature of IPCT absorption bands that the electrons correlated with the IPCT band are transferred partially at the ground state and completely at the excited state. Such a situation is very different from usual photochromism, which is caused by various changes of chemical bonds mainly via the excited singlet state. No transient absorption was observed for PV2+ (I− )2 in DME solutions, which was most probably due to the
Materials and devices for ultrafast molecular photonics
113
decreased distance between ion pairs in such a less polar solvent and an appropriate thermodynamic driving force (−∆G 0 ) for reverse electron transfer reactions [26,27]. The decay behavior due to the reverse electron transfer was found to depend markedly on the microenvironment and the counter anion. The lifetime (τ ) and a fraction of a major component for PV2+ (I− )2 was 1.2 ps and 86% in films cast from DME as compared with 4.0 ps and 73% in methanol solutions [22]. Photogenerated 4,4 -bipyridinium radical cations disappeared completely during 10 Hz excitation, which corresponded well with the fact that no steady color changes were observed for PV2+ (I− )2 in solutions and in cast films upon IPCT excitation. No decay was observed on the same time scale, as shown in Fig. 7(b) for PV2+ (TFPB− )2 . About a half decayed with τ = 71 ps and the rest survived for an extremely long time corresponding to the reversible and persistent color changes observed by steady photolysis [20]. The lifetimes of photogenerated 4,4 -bipyridinium radical cations were thus controlled over a very broad range from about 1 ps to almost infinity by the −∆G 0 value, the polarity of solvents and microenvironments in solid films. The present result of color change in about 0.3 ps with IPCT complexes of PV2+ is the fastest response reported so far among materials which show steady photochromism. It will help a great deal to develop novel optical memory and also THz all-optical switching devices using visible light. 3.3. Charge resonance band in the near infrared region and its ultrafast dynamics Recently we have also reported, for the first time, the charge resonance (CR) band due to dimer radical cation formation as a broad absorption with a peak at 950–2000 nm upon steady photoexcitation of styrylpyridinium derivatives such as tetraphenylborate (TPB− ) salts of 1-hexadecyl-4-(4-dicyanovinylstyryl)pyridinium (DCS+ ), 1-hexadecyl-4-(4nitrostyryl)pyridinium (NS+ ) or 1,n-bis(4-nitrostyrylpyridinium)alkane (NS+ CnNS+ ), as shown in Fig. 2, in solutions at room temperature by steady photolysis [34–46]. The absorption spectra after irradiation (> 365 nm) for TPB− salts of DCS+ and NS+ in DME are shown in Fig. 8 with respect to those before irradiation. In addition to the absorption spectra in the visible region due to the radical formation, broad specific absorption spectra were observed in the NIR region. The peak wavelength and shape of the latter spectra depended on the substituents. They were assigned to a CR band as schematically shown in Fig. 9 in a dimer radical cation which was formed between a styrylpyridinium cation and a photogenerated styrylpyridinyl radical. We have also observed the CR band with a peak at 1500–1700 nm as a charge resonance band for intramolecular dimer radical cations, as shown in Fig. 10 for NS+ C3NS+ , NS+ C4NS+ , and NS+ in acetonitrile [38]. It is clearly shown that two very strong CR bands with peaks at 950 and 1700 nm were observed only in NS+ C3NS+ due to intramolecular dimer radical cations. The CR band energy is twice the stabilization energy of the dimer radical cations which is controlled by several factors such as the extent of overlap of two chromophores, their mutual distance, and/or the polarity of solvents. The CR bands at 950 and 1700 nm were assigned to fully and partially overlapped dimer radical cations, respectively. Very recently we also reported fairly strong and unusual absorption changes in 800–2200 nm due to the intramolecular CR band of styrylpyridinyl radicals, formed by steady photolysis of newly synthesized meso-2,4-bis(4-(4 -nitrostyryl)
114
T. Nagamura
Fig. 8: Difference absorption spectra of (a) DCS+ TPB− and (b) NS+ TPB− in DME after irradiation at > 365 nm in degassed conditions.
Fig. 9: Schematic representation of energy levels for monomer radical and dimer radical cations, ∆H is the stabilization energy of dimer radical cations.
pyridinium)pentane ditetraphenylborates, during storage in the dark in solutions at room temperature, clearly indicating the change in conformation and overlapping of two chromophores in dimer radical cations which was controlled by the molecular structure [45,46]. As shown in Fig. 11, the rise of absorption spectra at the visible region due to radical formation and at the near-IR region due to the CR band was observed in less than 1
Materials and devices for ultrafast molecular photonics
115
Fig. 10: Absorption spectra for TPB− salts of (a) NS+ C3NS+ , (b) NS+ C4NS+ , (c) NS+ irradiated in acetonitrile by a Xe–Hg lamp at λex > 365 nm at room temperature in degassed conditions.
Fig. 11: Transient absorption in (a) the visible and (b) near infrared region together with (c) time dependences at 580 and 900 nm upon fs laser excitation of NS+ TPB− in DME at room temperature.
ps upon a fs laser excitation at 400 nm. These results indicated that the dimer radical cations were formed immediately after the photoinduced electron transfer reaction. The CR band at 900 nm in NS+ TPB− decayed single-exponentially (τ = 3.3 ps) [39]. The transient absorption at 580 nm showed double exponential decay with lifetimes of 3.3 and 11.4 ps. Similar results were obtained for DCS+ TPB− ; the decay at 960 nm with τ = 3.2 ps, and that at 650 nm with lifetimes of 3.8 ps for a fast component and 17.4 ps for a slow one [39]. The difference in the decay behavior at the visible and the near-IR region was explained as follows. While the NIR region absorption is attributed to the
116
T. Nagamura
dimer radical cation alone as the CR band, both the monomer radical and the dimer radical cation contribute to the absorption in the visible region as a HOMO–LUMO transition as schematically shown in Fig. 9. The fast component in the visible region and that in the NIR absorption gave almost the same lifetime. They were attributed to the reverse electron transfer reactions from the dimer radical cation (SP+ · · · SP• ) to the TPB• radical. Then the slow component in the visible region was most probably due to the reverse electron transfer reaction from the monomer radical (SP• ) to the TPB• radical. The higher rate of the reverse electron transfer in the dimer radical cations than in the monomer radicals was explained by the classical Marcus theory as follows [26,27]. The −∆G 0 values for the reverse electron transfer from NS• and DCS• to TPB• were estimated as 1.69 and 1.61 eV from the redox potentials, respectively. The reduction potential of the dimer radical cation should be less negative by 0.65 and 0.59 V than that of the radical monomer radicals due to the stabilization energy. The −∆G 0 values for the reverse electron transfer from the dimer radical cation to TPB• were thus estimated to be 1.04 and 1.02 eV for (NS+ · · · NS• ) and (DCS+ · · · DCS• ), respectively [39]. The observed values of −∆G 0 in this study for the reverse electron transfer of monomer radicals and dimer radical cations are in the Marcus inverted region [26,27]. The rate constant of the reverse electron transfer reactions from the dimer radical cation to TPB• would become higher due to the smaller −∆G 0 in the inverted region. This is the reason why the observed decay was faster for dimer radical cations. These results strongly suggest the applicability of the present system to ultrafast optical switching in the NIR region if an appropriate combination of a donor anion and an acceptor cation is used.
4. Parallel all-optical processing in guided wave geometry containing organic dyes 4.1. Ultrafast spatial light modulation and parallel optical recording based on photoinduced complex refractive index changes upon nanosecond laser excitation All-optical data processing has recently attracted much interest especially in the fields of spatial light modulation and optical data storage. A spatial light modulator (SLM) is a device to two-dimensionally control the intensity or the phase of reading light by another (writing) light, which plays an essential role in a projection TV, wavefront correction and an optical correlator. However, no practical devices have been developed except some prototypes or SLMs based on liquid crystals (LCs). The response time of LC-SLM is controlled by the electric field induced motion of LC; about a few hundreds of microseconds (µs) for ferroelectric LC and a few tens of milliseconds (ms) for nematic LC. The spatial resolution of LC-SLM is also not so high, about 10–20 µm, because a photoconducting layer used to optically address LC limits it. Recently multiple-quantum-well (MQW) SLMs have been developed showing spatial resolutions of 5–7 µm and a response time of 1 µs or faster [47–49]. MQW-SLMs will need further improvements in properties as spatial resolution, contrast, or capability of handling large readout intensities.
Materials and devices for ultrafast molecular photonics
117
Several types of new devices for optical parallel data processing have been proposed based on surface plasmon resonance (SPR), guided wave mode and Fabry–Perot (FP) resonance [50–52]. However, all of these devices could not exceed the LCSLM especially in the response time. Okamoto et al. [50] reported an all-optical photoaddressed SLM using a dye-doped polymer film in a surface plasmon resonance (SPR) configuration. They demonstrated the SLM based on photothermal changes in the refractive index of the methyl orange-doped poly(vinyl alcohol) (PVA) film using an Ar laser of 1–6 W cm−2 as a writing beam. The rise and fall times at 6 W cm−2 laser power were about 10 s and 2 s, respectively [50]. Yacoubian and Aye proposed Fabry–Perot (FP) resonance shifting in attenuated-total-reflection (ATR) geometry using azo-dye polymers [51]. They reported that their ATR-FP device enhanced optical modulation speed and efficiency as compared with the conventional intensity modulation based on photoinduced birefringence of Disperse Red 1 dye-doped poly(methylmethacrylate) [51]. The response time, 50–200 ms, was still relatively slow, though it was improved as compared with the conventional modulation system [51]. Ho et al. proposed a polarization vectorial holographic recording based on birefringent polymeric materials containing a photochromic azo benzene dye. The response time was 80 µs with a 100 mW power (80 µJ/cm2 ) of Ar laser [52]. Fichou et al. [53] proposed an incoherent-tocoherent optical converter based on photoinduced absorption of sexithiophene film. No actual properties of such a device, including the response time, were reported. We have also proposed a novel all-optical SLM based on complex refractive index changes upon photoexcitation of an organic dye-doped polymer thin film [54–60]. Similar resonance shift in guided mode due to refractive index changes was first reported by Sekkat et al. [61,62] in photoisomerization of azobenzene derivatives in polymeric thin films by pumping at 546 nm. They studied photoisomerization and thermal relaxation by a shift of the reflectivity dips in the guided wave mode. The observed response time was 2–15 s, which was too slow to be used in spatial light modulation or optical memory [61,62]. The conformational changes (trans → cis) necessary in solid films and a smaller quantum yield of photoisomerization might contribute to such slow responses, though they did not show the value. Our system is very unique as compared to the previously proposed “all-optical” light modulation systems as mentioned above. In principle fs response can be achieved in this system, because we use resonance condition changes of the guided optical waves (guided mode) in the ATR geometry based on the changes in the imaginary or the real part of the refractive index due to transient absorption or its Kramers–Kronig transformation, as schematically shown in Fig. 12. The guided mode “resonance” pattern depends not only on the thickness of the dielectric layer but also on its complex refractive index composed of real and imaginary parts, in general. If the imaginary part increases due to transient absorption, for example, the reflectance increases from curve a to c as shown in Fig. 12. A change of the real part shifts the resonance from curve a to b as shown in Fig. 12 for the case of a decrease. The intensity of the probe beam can be two-dimensionally controlled by the pump (writing) beam through photoexcitation of a dye. The main advantages using the guided mode are (1) its high sensitivity to small changes in refractive index and thickness, and (2) its sensitivity to both p- and s-polarized light. So far we have achieved repeated light modulation using a pulsed ns
118
T. Nagamura
Fig. 12: Schematic representation of the all-optical parallel processing in guided mode geometry and the calculated reflectance for a polymer film (1600 nm) on a silver layer (50 nm). The complex refractive index of the polymer layer is (a) 1.60, (b) 1.58 and (c) 1.60 + 0.02i.
laser and CuPcS and ZnPcS in guided wave-mode geometry. The response time was controlled by the triplet lifetime of phthalocyanines, 30 ns for CuPcS and 0.55 ms for ZnPcS. We are making efforts to achieve much faster responses using a ps or fs laser and appropriate materials. We have also demonstrated self-held ultrafast parallel optical switching based on the same geometry and using photochromic compounds instead of phthalocyanines [55–58,60]. Absorption spectra of spiropyran derivative (SP) dispersed in polystyrene with a weight ratio of 1 : 10 are shown in Fig. 13 before and after UV irradiation for 5 s. Strong absorption in the visible region due to photomerocyanine (PM) can be held for a long time and be reverted to that of SP by visible irradiation. Spectra of extinction-coefficient and refractive-index changes (∆k and ∆n) of polystyrene thin film containing SP upon UV excitation for 5 s are shown in Fig. 14. The former is based on the observed difference absorption spectra before and after UV irradiation as shown in Fig. 13. The latter is calculated from the extinction coefficient ∆k by Kramers–Kronig transformation. Refractive-index changes with different signs can be seen near strong absorption changes. The extinction-coefficient and/or refractive-index changes over a wide wavelength range approximately from 400 to 800 nm can be utilized to operate a wide range all-optical switch. The guided mode structures are very important to
Materials and devices for ultrafast molecular photonics
119
Fig. 13: The structures of photochromic SP and PM, and absorption spectra of SP or PM dispersed in a polystyrene thin film with a weight ratio of 1 : 10 before and after UV irradiation.
Fig. 14: Spectra of (a) extinction coefficient and (b) refractive index of polystyrene thin film containing SP and PM. The former is based on the observed difference absorption spectra shown in Fig. 13 before and after UV irradiation. The latter was calculated from (a) by Kramers–Kronig transformation.
utilize ultrafast changes of molecular electronic state upon photoexcitation for practical photonics devices. The incident-angle dependences of measured reflectance of a probe beam at 543 nm are shown in Fig. 15 for a polystyrene film containing SP with a weight ratio of 10 : 1. Each dip shows the SPR for a silver film (a) and the guided TM wave mode for the composite thin film before (b) and after (c, d) excitation by a pulsed Nd–YAG laser
120
T. Nagamura
Fig. 15: Incident-angle dependences of observed reflectance of a glass slide covered with (a) a silver film (50 nm) alone, and an SP in polystyrene (1 : 10) film (264 nm thick), (b) before, and (c), (d) after ns laser excitation (1.25 and 2.5 mJ/pulse, respectively) at 355 nm.
at 355 nm. The power of the pulsed laser at 355 nm was set to 1.25 or 2.5 mJ/pulse. Photochromism induced by transformation from SP to PM increased the reflectance and slightly shifted the dip to lower incident angles from curve b to c or d as shown in Fig. 15. The simulation gave almost perfect reproduction of the observed results, from which the complex refractive indices before and after excitation were determined. From comparison between the measured and calculated dependences, the thickness of the silver film and the polymer film was evaluated as 50 nm and 264 nm, respectively. The reflectance at the incident angle of 50.76° was increased from 0.04 to 0.68 upon excitation as shown in Fig. 15. The reflectance increase and the shift were found to be due to the increase of extinction coefficient and the decrease of refractive index at 543.5 nm of the polystyrene thin film containing SP as shown in Fig. 13 by the formation of the PM form. The changes of refractive index and extinction coefficient were estimated to be −0.015 and +0.024 at 2.5 mJ/pulse from comparison between the measured and the calculated dependences. Transmittances of the probe beam before and after excitation at 2.5 mJ/pulse were calculated as comparison to the present reflectance changes by using the same value for the film thickness and the extinction coefficient changes. The estimated values were 0.99 and 0.86 at 543 nm before and after excitation, respectively. The dynamic range of reflection changes, 17.0, was thus demonstrated to be much better than that of transmission changes, 1.15. The reflectance at the incident angle of 50.76° increased by 5–10 times very rapidly with a rise time of about 20 ns upon pulsed laser excitation at 355 nm, depending on its power. This rise time corresponded to the response of a photomultiplier. Much better switching is expected if a picosecond laser and an experimental setup with much better
Materials and devices for ultrafast molecular photonics
121
Fig. 16: Image (×500) written through a USAF Test Target as a mask by a single shot of ns 355 nm laser (1 mJ/pulse) on a spiropyran-doped poly(styrene) film deposited on a silver film.
time resolution are used, since the rise time of transient absorption of a polystyrene film containing spiropyran was reported as about 200 ps upon excitation with a ps laser [63]. The reflectance after ns pulsed laser excitation at 355 nm was held at a high value without applied power. The switching OFF was also demonstrated to be very fast with a response time similar to that of switching ON, although its accurate estimation was difficult due to a smaller S/N ratio. These results indicate that this fast reflectance decrease was caused by the reverse photochromic reaction from PM to SP, and not by the thermal reaction. The observed smaller reflectance change as compared with switching ON was due to a lower quantum yield of ring closure of PM. Photochromic dyes which have a higher quantum yield for reverse photochromic reaction will switch from ON to OFF with a larger dynamic range and fast response. Wavelength dependences of the refractive index and the extinction coefficient changes evaluated from reflectance changes upon pulsed laser excitation in the polystyrene film containing SP corresponded well with the spectra of extinction-coefficient and refractive-index changes estimated from steady photolysis as shown in Fig. 14. These results confirmed the mechanism responsible for the reflectance changes in guided wave geometry and also demonstrated the wide range of operation wavelength of the present all-optical device. In addition to very fast photoresponses, it is essential to write and read a twodimensional image pattern for optical parallel data processing. Fig. 16 shows microscopic photographs (×500) of the image written through a USAF Test Target as a mask by a single shot of ns 355 nm laser (1 mJ/pulse) on a spiropyran-doped PS film
122
T. Nagamura
deposited on a silver film. At least 128 line pairs/mm was clearly seen, which corresponds to a spatial resolution of 3.9 µm. Hickel et al. reported that the lateral resolution of optical waveguide microscopy based on the resonance shift by using guided waves as illumination light source is better than 10 µm [64]. The spatial resolution in these imaging devices based on guided mode geometry is limited by the propagation length of the guided waves, which is reduced by coupling to surface plasmon states in the case of p-polarized light. As one of the applications of the present all-optical switch in the near future, the architectures of the optical parallel processing logic devices AND and OR were proposed [58]. Optical parallel AND and OR devices are composed of two present switches and one switch only, respectively. They are operated by two parallel data as two input signals, Input 1 and Input 2. An optical parallel NOT device will also be composed by a polymer film containing a photochromic dye which will be colored only during irradiation. Then, the excitation light as input signal causes an intensity decrease of a probe beam as output signal. A combination of the present all-optical switch and the photon-mode spatial light modulator will also contribute a great deal to construct an ultrafast parallel processing optical correlator. Several spatial optical logic devices can be composed simply and easily from the present all-optical switch which employs reflectance increase or decrease by irradiation. An example of possible all-optical parallel correlation based on the present device is schematically shown in Fig. 17.
Fig. 17: Schematic representation of a possible all-optical parallel correlation system based on the present device.
Materials and devices for ultrafast molecular photonics
123
Fig. 18: (A) Profiles of fs white light without and with a spin-coated Arton® film at an incident angle of 50.0°. (B) Relationship between the dip wavelength and incident angle for a spin-coated 360 nm thick film of spiropyran/Arton® (1 : 2) observed with fs white light.
124
T. Nagamura
4.2. Reversible reflectance control of fs white light in guided wave geometry containing photochromic compounds upon fs laser photoexcitation As mentioned in the previous section, our all-optical SLM or optical switching device based on complex refractive index changes upon photoexcitation of an organic dyedoped polymer thin film can make in principle fs response, because we use resonance condition changes of the guided optical waves (guided mode) in the ATR geometry based on photoinduced changes of the imaginary or the real part of the refractive index. In order to demonstrate it, we employed photochromic spiropyran in a fs pump-probe measurement system as shown in Fig. 4. Profiles of detected fs white light are shown in Fig. 18(A) at an incident angle θ = 50.0° without and with a 360 nm thick Arton® film containing SP. The dip found around 588 nm in Fig. 18(A) is due to the guided wave mode. The dip wavelength shifted to the longer side by decreasing the incident angle, as shown in Fig. 18(B), from about 500 nm at θ = 65° to about 760 nm at θ = 39°. This film showed highly efficient photochromism upon fs laser excitation at 400 nm, as shown in Fig. 19. Changes of reflected light intensity for SP/Arton® (1 : 2) film (360 nm) at θ = 53.0, 50.0, 42.6, and 39.6° upon fs laser excitation at 400 nm and He–Ne laser at 543.5 nm are shown in Figs. 20–23, respectively. All reflectance changes were highly reversible, corresponding to photochromism between SP and PM forms. The “direction” of changes of reflectance clearly depended on the incident angle or the dip wavelength. By photochromism from SP to PM, the reflectance increased with shifting the dip wavelength to the shorter side at θ = 53.0° and to the longer side at θ = 42.6°, or at almost the same wavelength at θ = 50.0°. At an incident angle of θ = 39.6°, the dip wavelength shifted to the longer side with almost no changes of reflectance. Upon CW He–Ne laser irradiation at 543.5 nm, all dips returned to the original position before
Fig. 19: Absorption spectra of SP/Arton® (1 : 2) film (360 nm thick) upon fs laser excitation at 400 nm with 20–160 shots.
Materials and devices for ultrafast molecular photonics
125
Fig. 20: Changes of reflected light intensity for SP/Arton® (1 : 2) film (360 nm thick) at an incident angle of 53.0° upon fs laser excitation at 400 nm and He–Ne laser at 543.5 nm.
fs laser excitation due to reverse photochromism from PM to SP. These results clearly indicate that fs white light can be used as a probe light in guided mode geometry and that the fs pump-probe method will be used in such geometry. Efforts are being
126
T. Nagamura
Fig. 21: Changes of reflected light intensity for SP/Arton® (1 : 2) film (360 nm) at an incident angle of 50.0° upon fs laser excitation at 400 nm and He–Ne laser at 543.5 nm.
made to construct fs SLM based on the guided mode geometry and appropriate ultrafast photoresponsive materials.
Materials and devices for ultrafast molecular photonics
127
Fig. 22: Changes of reflected light intensity for SP/Arton® (1 : 2) film (360 nm) at an incident angle of 42.6° upon fs laser excitation at 400 nm and He–Ne laser at 543.5 nm.
5. Ultrafast nonlinear optical responses amplified by photoexcitation Nonlinear optical responses are very important to achieve, for example, wavelength conversion, electro-optical or pure optical control of the refractive index, and all-optical
128
T. Nagamura
Fig. 23: Changes of reflected light intensity for SP/Arton® (1 : 2) film (360 nm) at an incident angle of 39.6° upon fs laser excitation at 400 nm and He–Ne laser at 543.5 nm.
logic. Many organic and inorganic materials have been developed. One of the main problems especially in organic compounds is the small nonlinear optical coefficient. We have also been making efforts to modulate or enhance the second- and the third-order nonlinear optical properties by changing the electronic state or the extent of electronic
Materials and devices for ultrafast molecular photonics
129
distribution upon photoexcitation [65–73]. We also observed considerable enhancement of the second- and the third-order optical nonlinearity upon photoexcitation [68–73].
6. Conclusion We have developed several molecules and organized molecular assemblies to control the lifetime of the colored state by photoinduced electrochromism, the complex refractive index, and the nonlinear optical responses by the interactions with photons. Application of such materials to novel optical devices based on photoinduced complex refractive index changes was proposed and was successfully demonstrated in photon-mode recording and reflection control. These results will contribute a great deal to realize parallel all-optical ultrafast data processing devices. Considering that our vision is initiated by a simple photoisomerization of 11-cis retinal and is processed in a parallel way to achieve extreme high functions, molecular photonics based on elegant combination of molecules, photons and appropriate devices is expected to be the very promising way of ultrafast information processing in the near future.
Acknowledgements The author would like to thank Dr. K. Sakai, Dr. H. Sakaguchi, Dr. H. Kawai, Dr. K. Sasaki, Dr. D. Matsunaga, Mr. S. Kashihara, Dr. S.H. Park, Mr. T. Adachi, and Mr. I. Yoshida for their great contributions. Partial supports by the Grant-in-Aids for Scientific Research on Priority Areas “Molecular Superstructures, Design and Creation” (No. 07241102), Monbusho International Scientific Research Program (Joint Research, No. 08044137, 10044144), “Creation of Novel Delocalized Electronic Systems”, (No. 10146219), and “Molecular Synchronization for Design of New Materials System” (No. 11167242, 13022230), from the Ministry of Education, Science, Sports and Culture, Japan are greatly acknowledged.
References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.
T. Kamiya, F. Saito, O. Wada, and H. Yajima (Eds.), Femtosecond Technology, Springer Series in Photonics 2, Series Eds. T. Kamiya, B. Monemar, and H. Venghaus (Springer-Verlag, Berlin, 1999). T. Nagamura, Ch. VI-4 in Ref. [1], (1999), p. 376. T. Nagamura, in Sensors and Optical Switching, Molecular and Supramolecular Photochemistry Series, Vol. 7, edited by V. Ramamurthy and K. S. Schanze (Marcel Dekker, New York, 2001) p. 387. H. Dürr and H. Bouas-Laurent, Photochromism Molecules and Systems (Elsevier, Amsterdam, 1990). M. Irie, Chem. Rev. 100 (5), 1685 (2000) and references cited therein. Y. Yokoyama, Chem. Rev. 100 (5), 1717 (2000) and references cited therein. G. Berkovic, V. Krongauz, and V. Weiss, Chem. Rev. 100 (5), 1741 (2000) and references cited therein. T. Nagamura and K. Sakai, J. Chem. Soc. Faraday Trans. 1 84, 3529 (1988). T. Nagamura and S. Muta, J. Photopolym. Sci. Technol. 4, 55 (1991). T. Nagamura and K. Sakai, J. Chem. Soc. Chem. Commun., 810 (1986) .
130 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59.
T. Nagamura T. Nagamura and K. Sakai, Ber. Bunsenges. Phys. Chem. 93, 1432 (1989). T. Nagamura, K. Sakai, and T. Ogawa, J. Chem. Soc. Chem. Commun., 1035 (1988). T. Nagamura and K. Sakai, Thin Solid Films 179, 375 (1989). T. Nagamura and Y. Isoda, J. Chem. Soc. Chem. Commun., 72 (1991). T. Nagamura, Polym. Int. 27, 125 (1992). T. Nagamura, S. Muta, and K. Sakai, J. Photopolym. Sci. Technol. 5, 561 (1992). T. Nagamura, Mol. Cryst. Liq. Cryst. 224, 75 (1993). T. Nagamura, H. Sakaguchi, S. Muta, and T. Ito, Appl. Phys. Lett. 63, 2762 (1993). T. Nagamura, H. Sakaguchi, T. Ito, and S. Muta, Mol. Cryst. Liq. Cryst. 247, 39 (1994). T. Nagamura, H. Sakaguichi, and S. Muta, Proc. SPIE 2514, 241 (1995). T. Nagamura, Pure Appl. Chem. 68 (7), 1449 (1996). H. Inoue, H. Sakaguchi, and T. Nagamura, Appl. Phys. Lett. 73, 10 (1998). T. Nagamura and K. Sakai, Chem. Phys. Lett. 141, 553 (1987). T. Nagamura and K. Sakai, Ber. Bunsenges. Phys. Chem. 92, 707 (1988). H. Nishida, N. Takada, M. Yoshimura, T. Sonoda, and H. Kobayashi, Bull. Chem. Soc. Jpn. 57, 2600 (1984). R.A. Marcus, J. Chem. Phys. 24, 966 (1956). J.R. Miller, L.T. Calcaterra, and G.L. Closs, J. Am. Chem. Soc. 106, 3047 (1984). T. Nagamura, H. Sakaguchi, K. Suzuki, C. Mochizuki, and K. Sasaki, J. Photopolym. Sci. Technol. 6, 133 (1993). T. Nagamura, H. Sakaguchi, K. Sasaki, C. Mochizuki, and K. Suzuki, Thin Solid Films 243, 660 (1994). T. Nagamura, D. Kuroyanagi, K. Sasaki, and H. Sakaguchi, Proc. SPIE 2547, 320 (1995). K. Sasaki and T. Nagamura, J. Photopolym. Sci. Technol. 9, 129 (1996). K. Sasaki and T. Nagamura, Mol. Cryst. Liq. Cryst. 294, 145 (1997). H. Kawai, K. Nakano, and T. Nagamura, Chem. Lett., 1300 (2001). T. Nagamura, A. Tanaka, H. Kawai, and H. Sakaguchi, J. Chem. Soc. Chem. Commun., 599 (1993). T. Nagamura, H. Kawai, T. Ichihara, and H. Sakaguchi, Synth. Metals 71, 2069 (1995). H. Kawai and T. Nagamura, Mol. Cryst. Liq. Cryst. 267, 235 (1995). T. Nagamura, T. Ichihara, and H. Kawai, J. Phys. Chem. 100, 9370 (1996). T. Nagamura, S. Kashihara, and H. Kawai, Chem. Phys. Lett. 294, 167 (1998). H. Kawai and T. Nagamura, J. Chem. Soc. Faraday Trans. 94, 3581 (1998). W.-S. Xia, H. Kawai, and T. Nagamura, J. Photochem. Photobiol. A: Chem. 136 (1–2), 35 (2000). H. Kawai and T. Nagamura, Mol. Cryst. Liq. Cryst. 344, 209 (2000). S.H. Park, H. Kawai, and T. Nagamura, J. Photopolym. Sci. Technol. 13 (2), 197 (2000). S.H. Park and T. Nagamura, Mol. Cryst. Liq. Cryst. 370, 249 (2001). S.H. Park and T. Nagamura, J. Photopolym. Sci. Technol. 14 (2), 227 (2001). S.H. Park and T. Nagamura, J. Chem. Soc. Chem. Commun., 2344 (2001). S.H. Park, H. Kawai, and T. Nagamura, J. Chem. Soc. Perkin Trans. 2 (3), 508 (2002). A. Partori, A.M. Glass, T.H. Chiu, and D.T.H. Liu, Opt. Lett. 18, 906 (1993). P. Tayebati, E. Canoglu, C. Hantzis, and R.N. Sacks, Appl. Phys. Lett. 71, 1610 (1997). S.R. Bowman, W.S. Rabinovich, G. Beadie, S.M. Kirkpatrick, D.S. Katzer, K. Ikossi-Anastasiou, and C.L. Adler, J. Opt. Soc. Am. B 15, 640 (1998). T. Okamoto, T. Kamiyama, and I. Yamaguchi, Opt. Lett. 18, 1570 (1993). A. Yacoubian and T.M. Aye, Appl. Opt. 32, 3073 (1993). Z.Z. Ho, G. Savant, J. Hirsh, and T. Jannson, Proc. SPIE 1773, 433 (1992). D. Fichou, J.-M. Nunzi, F. Charra, and N. Pfeffer, Adv. Mater. 6, 64 (1994). T. Nagamura and T. Hamada, Appl. Phys. Lett. 69, 1191 (1996). K. Sasaki and T. Nagamura, Appl. Phys. Lett. 71, 434 (1997). K. Sasaki and T. Nagamura, J. Appl. Phys. 83, 2894 (1998). T. Nagamura and K. Sasaki, Proc. SPIE 3466, 212 (1998). T. Nagamura and K. Sasaki, Mol. Cryst. Liq. Cryst. 344, 199 (2000). . T. Nagamura, T. Adachi, I. Yoshida, H. Inoue, H. Heckmann, and M. Hanack, Mol. Cryst. Liq. Cryst. 370, 97 (2001).
Materials and devices for ultrafast molecular photonics 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73.
131
T. Nagamura, K. Sasaki, F. Iizuka, T. Adachi, and I. Yoshida, Opt. Commun. 205, 107 (2002). Z. Sekkat and M. Dumont, Appl. Phys. B 53, 121 (1991). Z. Sekkat, D. Morichere, M. Dumont, R. Loucif-Saibi, and J.A. Delaire, J. Appl. Phys. 71, 1543 (1992). T. Ito, M. Hiramatsu, M. Hosoda, and Y. Tsuchiya, Rev. Sci. Instrum. 62, 1415 (1991). W. Hickel and W. Knoll, Appl. Phys. Lett. 57, 1286 (1990). H. Sakaguchi, T. Nagamura, and T. Matsuo, Jpn. J. Appl. Phys. 30, L377 (1991). T. Nagamura, H. Sakaguchi, and T. Matsuo, Thin Solid Films 210/211, 160 (1992). H. Sakaguchi, L.A. Gomez-Jahn, M. Prichard, T.L. Penner, D.G. Whitten, and T. Nagamura, J. Phys. Chem. 97, 1474 (1993). H. Sakaguchi and T. Nagamura, Nonlinear Optics 15, 73 (1996). H. Sakaguchi and T. Nagamura, in Ultrafast Phenomena X, edited by P.F. Barbara, J.G. Fujimoto, W.H. Knox and W. Zinth (Springer-Verlag, Berlin, 1996) pp. 62, 209. A. Harada and T. Nagamura, Mol. Cryst. Liq. Cryst. 316, 79 (1998). A. Harada and T. Nagamura, Nonlinear Optics 22, 169 (1999). H. Sakaguchi and T. Nagamura, Nonlinear Optics 22, 413 (1999). T. Nagamura, T. Umeda, and H. Sakaguchi, Mol. Cryst. Liq. Cryst. 370, 119 (2001).
Nanotechnology and Nano-Interface Controlled Electronic Devices Editors: M. Iwamoto, K. Kaneto and S. Mashiko © 2003 Elsevier Science B.V. All rights reserved
CHAPTER 9
Carrier transport behavior in OLED Tatsuo Mori and Teruyoshi Mizutani Department of Electrical Engineering, Graduate School of Engineering, Nagoya University, Furo-cho, Chikusa-ku, Nagoya 464-8603, Japan
1. 2. 3. 4. 5. 6. 7.
Conducting organic materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conduction in organic LED: experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conduction in organic LED: modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Band model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hopping and tunneling models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Carrier injection model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Space Charge Limited Current (SCLC) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.1. Theoretical introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2. Experimental verification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8. Simulation of carrier transport by directly calculated hopping model . . . . . . . . . . . . . . . 8.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.2. Model in detail . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.3. Carrier behaviors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.4. Transient response characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.5. Summary of the simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
133 134 135 136 137 139 140 140 142 144 144 145 149 151 153 154 154
1. Conducting organic materials Organic materials have been regarded as insulating before the appearance of conductive polymers. The concept of “organic semiconductor” was revealed from the studies of π-conjugated polymers, that polyacethylene was discovered, and that the doping method was developed [1,2]. In addition, research fields such as organic functional materials and organic electronics are growing through the application of photosensitive materials (photoconduction materials) to electrophotography. However, it is very unstable state for an essentially neutral organic molecule to ionize by negatively or positively discharging as shown in Fig. 1. Since the unstable state leads to a degeneration reaction (oxidation), it was thought to be one of interference factors for the practical use of organic materials. Organic light-emitting diodes (OLEDs) reported by Tang and VanSlyke in 1987 [3] can
134
T. Mori and T. Mizutani
Fig. 1: The conception diagram of charge transfer in organic molecules.
be operated in the high-current region of > 1 A/cm2 by means of the shielding of O2 and H2 O.
2. Conduction in organic LED: experiment The current–voltage (I –V ) characteristics in OLEDs show a non-linear behavior. For example, Figs. 2 and 3 show the current density–luminance–voltage and luminance– current density characteristics of ITO/TPD[50nm]/Alq3[50nm]/AlLi and ITO/CuPc [30nm]/NPD[50nm]/Alq3[50nm]/LiF[0.6nm]/Al, respectively. ITO is indium-tin-oxide and a typical transparent electrode. TPD is N,N -diphenyl-N,N -bis(3-methylphenyl)1,1 -diphenyl-4,4 -diamine and a famous but old-type hole transport material. Alq3 is 8-hydroxyquinoline aluminum (Alq3) and a most famous emitting material. CuPc is Phthalocyanine Copper as a hole injection layer. NPD is N,N -di(1-naphthyl)-N,N diphenyl-1,1 -diphenyl-4,4 -diamine and a famous and high Tg (glass transition point) hole transport material. The fabrication process is shown in the previous paper [4–6]. Although the total thickness of a trilayer OLED is thicker than that of a bilayer OLED, both current densities are almost the same without increasing operating voltage. That is, electroluminescence is observed in the trilayer OLED at lower electric field. After the current shows Ohmic behavior below a few volts, the current increases steeply and shows non-linear behavior. However, as soon as EL, in other words, electron–hole
Fig. 2: The current density–luminance–voltage characteristics of ITO/TPD[50nm]/Alq3[50nm]/AlLi.
Carrier transport behavior in OLED
135
Fig. 3: The current density–luminance–voltage characteristics of ITO/CuPc[30nm]/NPD[50nm]/Alq3 [50nm]/LiF[0.6nm]/Al.
recombination can be observed, the current increases loosely and is proportional to V 4 . Luminance is proportional to current density. We use these experimental data in the following discussion.
3. Conduction in organic LED: modeling If the carrier conduction in a material is unipolar, its current density can be described as J = qnµE,
(1)
where q is the charge, n is the carrier density, µ is the charge carrier mobility, and E is an electric field. However, if the current in a material is caused by many kinds of charged carriers (i.e. electron, hole, anions, cations), the current density must be described as J=
k
qi n i µi E,
(2)
i=1
where qi is the charge of the ith carrier species, n i is the carrier density of the ith carrier species, µi is the mobility of the ith carrier species, and E is an average electric field. Now we do not consider the modification of electric field in the layer. Since most polymeric LEDs (PLEDs) consist of an additional hole injection layer and an emitting layer, it is physically consistent to apply Eq. 2 to their conduction. However, since organic low-molecular LEDs have multi function-separated layers, their conduction mechanism is very complicated. Let us discuss the simple bilayer OLED with TPD as a hole transport layer and Alq3 as an emitting layer. TPD is well-known to be a hole transport material and then we consider only hole conduction in the TPD layer. In addition, the electron injection from Alq3 into TPD is strongly blocked because of the high barrier height between TPD and Alq3. On the other hand, Alq3 is a weak electron transport material because its electron
136
T. Mori and T. Mizutani
mobility is 100 times larger than its hole mobility [7]. Strictly speaking, the current of the TPD/Alq3 device must be written as J = JTPD = JAlq3 = Jh,TPD + Je,TPD = Jh,Alq3 + Je,Alq3 = epT µhT E T + en T µeT E T = epA µhA E A + en A µeA E A ,
(3)
where p and n are hole and electron densities, respectively. Subscripts T , A, e, and h mean TPD, Alq3, electron and hole, respectively. In a steady state, JTPD = JAlq3 (the law of continuity of current). Even if TPD thickness agrees with Alq3 thickness, the divided voltage of TPD layer is different from that of Alq3 because the former conductivity is lower than the latter. When the above experimental results are considered, Je,TPD can be neglected. However, although hole mobility is smaller than electron mobility in Alq3, the third term on the right-hand side cannot be neglected because of the hole density injected from TPD. Consequently, Eq. 3 becomes J = epT µhT E T = epA µhA E A + en A µeA E A .
(4)
The next problem is that carrier density and mobility in organic materials depend on electric field. That is, carrier density and mobility cannot be regarded as constant parameters. In addition, electric field is obtained as a function of position as well as each layer. As E T or E A is each average electric field in the TPD or Alq3 layer, this expression is ambiguous and it is right that the electric field should be described as E(x). Of course, although the current also depends on time after applying voltage, E(x) may be given as a distribution function of position since we treat the static state of the device. In addition, as high-performance OLEDs have a complicated multi-layer structure, one can understand that it is not easy to describe an analytical solution as the conduction model of OLEDs. In OLEDs, the fact that organic materials have low carrier mobilities is thought to lead to that the conduction mechanism in OLEDs is due to the space charge limited current (SCLC) model. In PLEDs, the SCLC model is comparably easy to be accepted because their layer structures are simpler than those of low-molecular LEDs. The conductive mechanism in OLEDs is categorized by two models: One is that the current in OLEDs is strongly controlled by injected carrier density since organic materials have low carrier concentration. The other is that it is strongly controlled by carrier mobility since organic materials have low carrier mobility. In the former example, some analyzed the current of OLED as Schottky current model controlled by hole or electron injection. However, the value of a physical parameter (a dielectric constant, the barrier height of carrier injection, etc.) determined from the approximate I –V curve is very different from that estimated from a direct measurement [8,9]. Although some interpretations for the conflict are suggested, they are not thought to be consistent with the physical phenomena. The carrier transport in OLEDs cannot be explained only by the unipolar carrier injection model. 4. Band model In general, the conduction behavior in OLEDs is often explained using an energy diagram on the basis of the band model. For example, the OLEDs in Figs. 2 and 3 can
Carrier transport behavior in OLED
137
Fig. 4: (a) The energy diagram of ITO/TPD/Alq3/AlLi. (b) The energy diagram of ITO/CuPc/NPD/ Alq3/LiF/Al.
be expressed by Fig. 4. When the energy diagram is made, the levels of the conduction band and valence band will be matched with LUMO (lowest unoccupied molecular orbital) and HOMO (highest occupied molecular orbital) levels, respectively. However, since the interaction between organic molecules is a van der Waals force, which is much weaker than covalent bond and metallic bond, the band width of the energy band becomes narrow even if an energy band may be formed in organic materials. The narrow band means low mobility for carrier transport in the band model. In nature, the carrier mobility in the band model is more than several hundreds cm2 /V s. The largest mobility in organic materials is at most 1 cm2 /V s, the carrier mobility in a pentacene crystal [10]. We must think that it is not appropriate to apply the band model to organic materials.
5. Hopping and tunneling models [11] Let us remember that the carrier transport in organic materials is caused by alternate ionization between ionized molecules and neutral molecules, as shown in Fig. 1. That is, M+(−) + M −→ M + M+(−) .
138
T. Mori and T. Mizutani
Fig. 5: The energy potential between molecules: a is molecular distance, U is potential energy.
The charge migration between molecules can be also explained by other, i.e. hopping and tunneling, processes. Now we regard the potential diagram of neighboring molecules in Fig. 5. U is the barrier height of the potential. a is the distance between two neighboring molecules. The hopping probability of thermally activated charge is given by U , (5) P = ν exp − kT where ν is trial frequency factor and k is Boltzmann constant. The mobility under electric field is given by the Einstein relation µ=
eD , kT
(6)
where D is the diffusion coefficient. In addition, using D = Pa 2 , we obtain the following relation, ea 2 ν U µ= exp − . (7) kT kT When temperature increases, the preexponential factor decreases inversely proportional to temperature but the exponential term increases steeply. Consequently, the thermally activated hopping process has a positive temperature dependence. On the other hand, the charge transfer between molecules may be caused by a tunneling process. The tunneling probability, PT depends on the number of carriers colliding with the potential barrier, N and the tunneling factor, T . PT can be described as the product of N and T , i.e. PT = N T . T can be written as √ 2w 2m(U − E) T = T0 exp − , (8) h where T0 is a constant, w is the barrier width, m is the electron mass, U is the barrier height of the potential barrier, E is the electron energy, and h is Planck’s constant. Although the tunneling transfer due to the quantum mechanical mechanism is not affected by temperature, it strongly depends on the distance between one molecule and the counter as well as the electric field. Usually the effective distance for tunneling transfer is said to be < 1 nm.
Carrier transport behavior in OLED
139
6. Carrier injection model Some researchers propose that the current in OLEDs can be simulated by the Schottky injection model [8,9]. The ground for their proposal is the temperature dependence of the current as shown in Fig. 6. If the current in OLEDs could be simulated by the tunneling injection model, it would not have shown a remarkable temperature dependence. However, let us remember Eq. 1. Since organic materials do not have intrinsic carrier density because they are essentially insulators, carrier density in bulk is due to carrier injection from the electrodes. Therefore, if current depended on only carrier density, we would consider the Schottky injection for the conduction in OLEDs since the current in OLEDs has temperature dependence. But we need to remember that the carrier transportation in organic materials is not caused by band conduction, but by such a discontinuous process as hopping conduction. The hopping conduction model has temperature dependence. The current due to Schottky injection is described as β E 1/2 − φ , (10) J = AT 2 exp kT where A is Richardson–Dushman’s constant, φ is the barrier height of carrier injection, and β is defined as e3 β= . (11) 4πε This current depends on the squared electric field as well as temperature. In order to judge whether Schottky current can be applied to the conduction current of a material or not a Schottky-plot, ln J : E 1/2 , is often used. When the dielectric constant estimated from the gradient of the graph agrees with the experimental value, it is possible that the conduction mechanism in the material may be due to Schottky emission current. If the dielectric constant does not agree with the experimental one, we think that the conduction mechanism should be treated carefully.
Fig. 6: The temperature dependence of the current density–electric field characteristics of ITO/TPD[50nm]/ Alq3[70nm]/Mg.
140
T. Mori and T. Mizutani
7. Space Charge Limited Current (SCLC) 7.1. Theoretical introduction Although SCLC is the conduction model controlling carrier injection, it is not an injection-controlled conduction but a bulk-controlled one. Some researchers believe that the current density, J , due to SCLC equation is 9 εµV 2 , (12) 8 d3 where e is a quantum of electricity, µ is the mobility of carriers, V is an applied voltage, and d is the sample thickness. However, this equation is a special solution obtained from an original Poisson equation and the boundary condition, E(0) = 0, V (0) = 0. In this section, we discuss the problem of the SCLC model. The SCLC model ought essentially to be applied to unipolar conduction. When voltage is applied to an insulator (organic material) interposed by two electrodes and the charged carriers injected from an electrode are not neutralized by the counter charged carriers injected from the counter electrode, the injected charged carriers form a space charge around the electrode. This space charge modifies the electric field between the electrodes in the case of low mobility. The homo space charge accumulated in front of an electrode reduces the electric field on the electrode. Therefore, the carrier injection after forming the space charge strongly depends on the modified electric field due to space charge. The necessary conditions that a conduction current becomes a SCLC are the following: 1. the current due to injected carriers has the same or higher value as the Ohmic current; J=
Fig. 7: Typical SCLC characteristics.
Carrier transport behavior in OLED
141
2. the dielectric relaxation time of the material is longer than the carrier drift time between the electrodes in the material. A typical SCLC example is the anode current–voltage characteristics of a twoelectrode vacuum tube. In general, it is not easy for organic materials to satisfy the necessary condition 1. Let us find Eq. 12 from Poisson’s equation. We regard a one-dimensional system containing the sample with two electrodes. The interface between cathode and sample is x = 0. Now we imagine that electrons are injected into the sample. Poisson’s equation is en d2 ϕ =− , (13) 2 dx ε where ϕ is the potential in the sample, n is injected electron density, and ε is the dielectric constant of the sample. The current density, J in the sample is described as J = enµE,
(14)
where µ is the electron mobility and E (= −dϕ/dx) is electric field in the sample. Deleting n using the two equations 13 and 14, J dE = . dx εµE
(15)
Integrating Eq. 15 with respect to x after separating variables, E 2 (x) =
2J x + C, εµ
Using E(0) = 0, C = 0, therefore: 2J 1/2 x . E(x) = ± εµ However, as the positive solution is not appropriate for this case, 2J 1/2 E(x) = − x . εµ V (x) is given by integrating E(x) with respect to x. x x 2J 1/2 E(x) dx = x dx V (x) = − εµ 0 0 8J 3/2 = x + C 9εµ
(16)
(17)
(18)
(19)
We can use V (0) = 0, C = 0. When the sample thickness is d and the applied voltage is V , the following equation can be given (Fig. 8), J=
9 εµV 2 . 8 d3
(12)
142
T. Mori and T. Mizutani
Fig. 8: The potential distribution between cathode and anode: thin solid line means average electric field. Thick solid line is the potential on the equilibrium condition.
7.2. Experimental verification [13] There are many papers using the SCLC model [14–18]. In our opinion, there are some problems for applying the SCLC model to organic LEDs. The system of the organic LED may not satisfy the above necessary conditions. Fig. 9 shows the experimental and calculated J –V characteristics of the OLEDs as shown in Figs. 2 and 3. The lines with marker are the experimental curves and the lines without marker are calculated curves. Estimated values are used for the parameters: ε = εr ε0 , the dielectric constant of organic material, εr is about 3 and ε0 is the permittivity of vacuum, µ is 10−3 –10−6 cm2 /V s. The thickness, d, is the sum of the CuPc and NPD layers in the dash-dotted line. d is the total thickness of organic layers for the other lines. The carrier mobility in both the dash-dotted line and the solid line is 10−3 cm2 /V s. That of dotted line, short-dashed one, and long-dashed one is 10−4 , 10−5 , and 10−6 cm2 /V s, respectively. In the low current region, the experimental current behavior does not agree with the calculated curves. In the high current region, the value of the former approaches the calculated one. A decrease of effective thickness contributes to an increase of current. Since the carrier mobility estimated by the TOF method is caused by the carrier transfer due to photoexcited carriers with high energy (> 3 eV), it is possible to overestimate the intrinsic mobility which will be excited by thermal activation (∼0.026 eV). Carrier mobility of organic materials needs to be discussed in detail. Fig. 10 shows the conductivity, dielectric relaxation time and drift time–voltage characteristics of ITO/TPD[50nm]/Alq3[50nm]/AlLi. Each parameter is calculated by the following. The apparent conductivity, σ is calculated by σ = J/E = J d/V . The
Carrier transport behavior in OLED
143
Fig. 9: The current density–voltage characteristics of ITO/TPD/Alq3/AlLi (a) and ITO/CuPc/NPD/Alq3/ LiF/Al (b). The lines with marker are the experimental curves and the lines without marker are calculated curves. TPD thickness is used as d in the dash-dotted line. d is the total thickness of organic layers in the other lines. The carrier mobility in both the dash-dotted line and solid line is 10−3 cm2 /V s. That of the dotted line, short-dashed one, and long-dashed one is 10−4 , 10−5 , and 10−6 cm2 /V s, respectively.
10
TPD/Alq3
-7
1
10
-9
-1
10
-11
-3
10
10 p
-13
τ [s]
σ [S/cm]
10
-5
10
10
-15
-7
10
10 0
4
8 Voltage [V]
12
Fig. 10: The apparent conductivity, dielectric relaxation time and drift time as function of voltage in ITO/TPD/Alq3/AlLi: closed circles mean apparent conductivity, open circles mean dielectric relaxation time, and solid line is calculated drift time.
144
T. Mori and T. Mizutani
dielectric relaxation time, τ , is estimated from τ = ε/σ . We used 3ε0 as the dielectric constant of organic materials: ε0 is the permittivity of vacuum. The drift time, td is calculated by td = d/v = d/µE. We used a constant mobility, 10−6 cm2 /V s, as average mobility. The hole mobility in TPD was estimated to be ∼ 10−3 cm2 /V s [19] and the hole and electron mobilities in Alq3 were estimated to be ∼ 10−5 and ∼ 10−7 cm2 /V s [7], respectively. Since these values were obtained by the time-of-flight method, we think that the carrier mobility obtained by the time-of-flight method is overestimated as the carrier mobility of organic material. In addition, the exact drift time in organic LEDs can be obtained by summing the hole drift time in TPD and the electron drift time in Alq3. However, such a calculation process is not consistent with the SCLC model. Therefore, we used a lower value as average mobility. The apparent conductivity of ITO/TPD/Alq3/AlLi is almost constant, ∼ 10−14 S/cm below ∼ 2 V. It increases steeply with starting carrier injection and achieves to ∼ 10−6 S/cm. Consequently the dielectric relaxation time is lower than the average drift time. Therefore, we conclude that the necessary condition 2 for the SCLC model is not satisfied in OLEDs. We have to consider the conduction mechanism in OLEDs on the basis of real charge transfer between molecules. 8. Simulation of carrier transport by directly calculated hopping model [20–26] 8.1. Introduction Although the structure of OLEDs in which organic layers are sandwiched between two electrodes is simple, the light-emitting mechanisms of the device are quite complicated. These mechanisms may be roughly divided into three processes: the carrier injection process from each electrode, the carrier transport process, and the emission process via excitons generated by electron–hole recombination. For example, many researchers tried to explain the carrier injection mechanism of OLED from the viewpoint of experimental current–voltage characteristics. However, such external information is insufficient to explain the injection mechanism. Clarification of each process will ease improvement of current performance of the device. When we improve on the device performance, it is important to discuss the balance between electron and hole injections. Rate of electron–hole recombination, electric field, and space-charge distributions in the OLED are also important. However, it is impossible to obtain and evaluate these parameters experimentally because these parameters are “internal” OLED parameters. In the present work, we assumed a simple model and attempted to calculate carrier behavior in OLED in order to clarify light-emitting mechanisms. Many groups have attempted to simulate I –V characteristics of devices [27–35]. Calculations were carried out using a “continuous model” in which conduction current density is explained by carrier drift and carrier diffusion. Considering recombination and Fowler–Nordheim injection, Khramtchekov et al. showed the distributions of electric field and current flows in a bilayer OLED [27]. Davids et al. assumed that initial hole distribution followed Maxwell–Boltzmann statistics as accompanied with Schottky and tunneling injection [28].
Carrier transport behavior in OLED
145
Crone et al. estimated the distributions of electric field, hole and electron currents, and recombination rate [29]. Although they explained the values by the conduction mechanism due to the SCLC model, they pointed out that space charge is not significant. Kawabe et al. analyzed the conduction characteristics in PLEDs on the basis of semiconductors [30]. They used Fowler–Nordheim and SCLC currents. Malliaras et al. used numerical methods to calculate the current and the efficiency of a single-layer organic LED, taking into account field-dependent mobilities, diffusion, and thermionic injection [31]. Staudigel et al. quantitatively simulated the conduction and EL mechanism in multi-layer OLEDs by a one-dimensional numerical model [32]. Of course, they compared the experimental results with their simulated data. Crone et al. gave the carrier mobility of a single-layer PLED the field dependence of the Pool–Frenkel form [33]. And they treated the conduction of PLEDs as a bipolar mechanism. They tried to explain the change of conduction in a single-layer PLED caused by the difference of cathode metal using their model. They claimed that their model successfully describes the I –V characteristics of a single-layer PLED. Crone et al. applied their conduction model to single-layer OLEDs [34]. They calculated the spatial variation of the carrier densities, electric field, and recombination rate. Tutis et al. proposed the discrete carrier injection model due to tunneling injection [35]. (However, some equations in this paper have errors!) Tsutsui et al. pointed out that the main factor of current in organic film is not always an equilibrium carrier density [36]. In general, space charge limited current (SCLC) is used to explain conduction of organic thin films such as OLEDs. In this model, the injection field becomes zero, so that carrier density is infinite at the interface. However, this density never becomes infinite since sites are limited in organic films. 8.2. Model in detail We proposed a one-dimensional discontinuous model for simulation as shown in Fig. 11. Simulation of carrier behavior in an insulator is based on the hopping model proposed by Iwamoto and Hino [37]. Although carrier density is not limited in continuous models, the carrier number accepted by a molecule is limited in our model. Because the carrier transport between organic molecules is regarded as an intermolecular oxidation– reduction, our model approximates carrier behavior more accurately than conventional continuous models. In continuous models, the carrier number accepted by a molecule is not limited. We assumed a bilayer OLED of ITO/TPD/Alq3/Al. Thickness of each organic layer is 50 nm. Since an Alq3 molecule is represented by a sphere of 0.8 nm diameter, we approximate that these molecules are arranged with average distance of 1.73 nm in an electric field. The number of sites is 30. Molecular stacking is not considered. Maximum carrier density per unit area is 1018 m−2 [(109 )2 ]. Most parameters obtained by experiments can be found in our previous papers [20– 26]. The carrier conduction process is assumed as follows: (I) a molecule is a hopping site, (II) a site can be occupied by an electron or a hole at most, (III) carriers move only to adjacent sites, and (IV) the hopping rate depends on not only to carrier density, but also to the rate of unoccupied adjacent sites. Conduction currents from the kth site to the
146
T. Mori and T. Mizutani
Fig. 11: Diagram of the one-dimensional hopping model.
adjacent k + 1th site for holes (Jp(k,k+1)) are represented as qa F(k,k+1) N − pk+1 − rk+1 exp Jp(k,k+1) = ν p qpk N 2kB T −qa F(k,k+1) N − p − r k k − ν p qpk+1 exp N 2kB T (k = 1, 2, . . . , m) ν p
= ν exp
−U p
kB T
,
and those for electrons (Jn(k,k+1) ) are qa F(k,k+1) N − n k+1 − rk+1 exp Jn(k,k+1) = νn qn k N 2kB T −qa F(k,k+1) N − n k − rk − νn qn k+1 exp N 2kB T (k = 1, 2, . . . , m − 1) νn = ν exp
(20)
−Un , kB T
(21)
(22) (23)
Carrier transport behavior in OLED
147
where N represents the maximum site density for a molecular layer (= 1018 m−2 ); pk , n k , and rk (m−2 ) are densities of the hole, electron, and exciton of the kth site, respectively; (k, k + 1) is an electric field between the kth and (k + 1)th site; and U p and Un (eV) are hopping barriers for holes and electrons, respectively. Also, ν (s−1 ) is the attempt-to-escape frequency; m shows site numbers for Alq3; and T , kB , and q are temperature, the Boltzmann constant, and elementary charge, respectively. The hopping distance a is assumed to be 1.73 nm, which is the average distance between the two centers of adjacent molecules. U p and Un were calculated using the equation for conventional hopping transport from experimental carrier mobility, µ p and µn [7]. Electron mobility is about 100 times higher than hole mobility in Alq3; hole mobility in the TPD bulk is about five orders of magnitude higher than that in the Alq3 bulk. We use U p (0.27 eV) as U p and Un (0.15 eV) as Un , respectively, in Alq3. At the TPD/Alq3 interface, U p and Un are U p + φbp and Un + φbn , where φbp (0.26 eV) and φbn (0.83 eV) are barrier heights for the hole and the electron, respectively. Since φbn is so high that electrons are almost blocked at the TPD/Alq3 interface, electron behavior can be ignored in the TPD bulk. We use both Schottky emission and Fowler–Nordheim emission for electron injection from the cathode. The electron current density passing between the Alq3 and the cathode interface, Jn(m,m+1) , is assumed as
N − n m − rm −φn q q F(m,m+1) 2 Jn(m,m+1) = An T exp exp N kB T kB T 4πεr ε0
2 q F(m,m+1) −8π 2m ∗ φn3 −qa F(m,m+1) +ρ exp , (24) − νn qn m exp 8πhφn 3qh F(m,m+1) 2kB T where φn (eV) is the barrier height for electron injection from the cathode to an Alq3 molecule and is estimated to be 0.67 eV, and An , ε0 , and εr are initial parameters based on the Richardson–Dushman constant for electrons, vacuum permittivity, and dielectric constant of Alq3 bulk, respectively. Hole injection from an anode is assumed to be due to Schottky emission. The hole current density passing through the TPD/Alq3 interface, Jp(0,1), is assumed to be the same at the ITO/TPD interface because the space charge density is negligible in the TPD bulk except for the site adjacent to the Alq3. Thus, the interface is assumed to be a hole reservoir as shown in Eq. 25. As holes are accumulated in the TPD site closest to the TPD/Alq3 interface, we can regard this site as a reservoir for holes. Hole density is represented as pres , that is, p0 = pres . Therefore, the hole conduction current passing through the TPD/Alq3 interface is obtained by substituting pres into Eq. 20:
q q F(0,1) N − pres −φ p 2 A p T exp exp Jp(0,1)) = N kB T kB T 4πεr ε0 −qa F(0,1) N − pres − ν p qp1 exp . (25) N 2kB T The barrier height, φbp , for hole injection from the TPD molecule to Alq3 is estimated to be 0.26 eV. Current density flowing in an external circuit consists of the
148
T. Mori and T. Mizutani
hole conduction component Eq. 26 and the electron conduction one Eq. 27, both of which are derived from the continuity equation under DC field, Jp is hole current density; Jp(0,1), Jp(m,m+1) , w Jp(0,1) and the sum of Jp(k,k+1) are hole current densities flowing in the TPD/Alq3 interface, the Alq3/Al interface, the TPD bulk, and the Alq3 bulk, respectively. m−1 1 {Jp(k,k+1)}. Jp = {Jp(0,1) + Jp(m,m+1) } + w Jp(0,1) + 2 k=1
(26)
In the equation above, Jn is electron current density; Jn(0,1), Jn(m,m+1) , and the sum of Jn(k,k+1) are flowing the TPD/Alq3 interface, the Alq3/Al interface, and the Alq3 bulk. Electron mobility in the TPD bulk is very low and electron current is negligible. m−1 1 {Jn(k,k+1) }. Jn = {Jn(0,1) + Jn(m,m+1) } + 2 k=1
(27)
Here, w is the number of sites in TPD. Time variation of hole density is shown in Eq. 28 and that of electron density is shown in Eq. 29. d pk 1 (28) = {−Jp(k,k+1) + Jp(k−1,k)} − Rn k pk , dt q dn k −1 (29) = {−Jn(k,k+1) + Jn(k−1,k) } − Rn k pk , dt q where R is the electron–hole recombination coefficient for Alq3 molecules. The fields are expressed as Eqs. 30–32, which are derived from the Poisson equation. k −qa 1 F(k,k+1) = ( ps − n s ) s− εr ε0 d s=1 2 m 1 qa Va ( ps − n s ) − , m −s + + (30) εr ε0 d s=k+1 2 d m 1 −qa Va ( ps − n s ) − , m −s + (31) F(0,1) = εr ε0 d s=k+1 2 d m 1 −2qa Va ( ps − n s ) − . m −s + (32) F(m,m+1) = εr ε0 d s=k+1 2 d In these equations, d and Va are thickness and applied voltage of the device. When L is the length of exciton diffusion and τ is the fluorescence lifetime in Alq3, the diffusion coefficient, D, is shown by L2 . τ Time variation of exciton density is shown by D=
(33)
d2 dn k d2rk rk = Rn k pk + D(N − pk − n k − rk ) 2 + Drk 2 (N − pk − n k − rk ) − . dt dk dk τ
(34)
Carrier transport behavior in OLED
149
Electroluminescence (EL) intensity is assumed to be proportional to the sum of Rn k pk in the Alq3 layer (k: from 1 to m), 1 Rn k pk . EL ∝ τ k=1 m
(35)
8.3. Carrier behaviors In this simulation, the carrier (electron and hole) distribution and field distribution as well as current density and EL intensity are calculated when a DC step voltage is applied. Distributions of hole density, electron density, and exciton generation density are shown in Figs. 12, 13, amd 14, respectively. In these calculations, a recombination rate R = 1.0 × 10−5 m2 /s is used to calculate the exciton generation distribution. Holes are accumulated near the TPD/Alq3 interface, as shown in Fig. 13. Hole density decreases with distance from the TPD/Alq3 interface. Holes are accumulated within 10 nm distance from the interface (Fig. 12) because of the low hole mobility in the Alq3 layer. In the emission layer (Alq3), electrons injected from a cathode move to the TPD/Alq3 interface. Electrons are comparatively uniformly distributed in Alq3 bulk (10 nm ≤ position ≤ 50 nm), and decrease near the TPD/Alq3 interface. Electron density near the TPD/Alq3 interface is lower than that near the cathode, as shown in Fig. 13. Distribution of hole density differs from that of electron density because the electron mobility is 100 times faster than the hole mobility in the Alq3 layer. Fig. 14 shows distribution of generated exciton density after 30, 100, and 250 ns. Exciton generation due to recombination occurs near the TPD/Alq3 interface. Distribution of exciton generation depends on the product of hole and electron densities. The electron density rapidly decreases near the interface because of the recombination of electrons and holes, resulting in generating excitons near the TPD/Alq3 interface.
Fig. 12: Distribution of hole density in Alq3 layer.
150
T. Mori and T. Mizutani
Fig. 13: Distribution of electron density in Alq3 layer.
Fig. 14: Distribution of generated exciton density in Alq3 layer.
Exciton generation density achieves a maximum value and it moves from the interface with time. Since electron density is lower than hole density, all electrons are considered to recombine with holes before reaching the TPD/Alq3 interface. Fig. 15 shows the field distribution in both organic layers at an average field of Fa = 140 MV/m. Field distortion in TPD bulk is little observed at Fa = 140 MV/m where an OLED shows strong luminance of over 600 cd/m2 . Our one-dimensional discontinuous calculation model suggests that conduction in OLEDs cannot be explained by a typical SCLC conduction model since field distortion is not observed near both cathodes in organic layers.
Carrier transport behavior in OLED
151
Fig. 15: Distribution of electric field at an average field of 140 MV/m.
8.4. Transient response characteristics Figs. 16 and 17 show calculated time dependence of current density and EL intensity at Fa = 140 MV/m. The hole current density, Jp , at 100 ns decreases until it reaches 90% at 30 ns. Since distribution of hole density spreads into the Alq3 bulk over time, as shown in Fig. 12, accumulation of electrons results in inducing field relaxation near the interface (Fig. 15). Also, the amount of injected electrons decreases. At 30 ns, the electron current density is 90% of that at 100 ns and saturated. Thus, electron current density appears to be saturated after 100 ns. It has a turning point at 30 ns; after which EL begins to increase. Amounts of injected electrons and recombining electrons
Fig. 16: Time dependence of current densities at an average field of 140 MV/m.
152
T. Mori and T. Mizutani
Fig. 17: Time dependence of EL intensity at an average field of 140 MV/m.
-2
Current density [Am ]
4
10
3
10
2
10
1
10
experiment calculation
0
10
60
80
100 120 140 -1
F [MVm ] Fig. 18: Comparison of calculated and experimental field dependence of current density.
equalize due to exciton generation near the TPD/Alq3 interface. When the applied electric field is small, (Fa = 100 MV/m), the delay time of EL (solid line) and 90% of EL value at 250 ns (dashed line) is longer than when a high electric field is applied. Fig. 18 shows the calculated current densities flowing in an external circuit. The calculated current density normalized by the current density at Fa = 100 MV/m are used to calculate those at other Fa . The calculated curves (solid line) agree to the experimental ones (dashed line), as shown in Fig. 18. In our previous work, we considered only Schottky emission as electron injection mechanism. EL intensity (Fig. 19) did not agree with experimental values at low electric field, although the calculated density agreed with experimental data. Considering both Fowler–Nordheim emission and Schottky emission into the electron injection mechanism, Fowler–Nordheim emission is dominant in high fields, as shown in Fig. 20.
153
-2
Luminance [cdm ]
4
10
10
experiment
3
10
10
22
21
2
10
calculation 10
20
1
10
10
19
EL Intensity [arb.units]
Carrier transport behavior in OLED
0
10
60
80 100 120 140 -1
F [MVm ] Fig. 19: Comparison of calculated and experimental field dependence of EL intensity.
3
-2
Jn(m+w,m+w+1) [Am ]
10
2
10
1
10
0
10
total ( i + ii ) i. Fowler Nordheim ii. Schottky
-1
10
60
80
100
120
140
160
-1
F [MVm ] Fig. 20: Field dependence of electron injection.
8.5. Summary of the simulation We assume a one-dimensional hopping conduction model for the OLED: each emitting molecule corresponds to a hopping site simulating actual charge transfer between adjacent molecules. Time dependence of carrier, exciton and EL intensity, and distributions of field and carrier density are calculated. Hole and electron densities decrease near the TPD/Alq3 interface. As a result, the density of exciton generation achieves its maximum within 10 nm from the TPD/Alq3 interface. Field distribution due to the space charge effect is not apparent in the TPD
154
T. Mori and T. Mizutani
bulk. These results suggest that the conduction mechanism in bilayer OLEDs cannot be explained by a typical SCLC conduction model. This model accommodates Fowler– Nordheim emission as an electron injection mechanism. As a result, behavior of current density and EL intensity agree with measured current density and luminance. From above results, a simple bilayer and discontinuous model is effective for investigating OLED carrier behavior.
9. Conclusion We showed that it is difficult to directly apply the SCLC model to OLEDs. If a researcher believes that the conduction mechanism of a material can be explained by a particular mechanism, that is, an equation, he can analyze the conduction current by the equation. And he may obtain the various information on the conduction mechanism of a material. However, if he does not verify his calculated parameters by means of other experimental results, his analysis will be almost nonsense in the special case with additional assumptions. Of course, our simulation is still incomplete and also needs to reflect the experimental results. Since OLEDs have multi-layer structure and a bipolar conduction mechanism, we have to treat the complicated conduction discretely.
References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20.
H. Shirakawa, T. Ito, and S. Ikeda, Polym. J. 2, 231 (1971). M. Hirooka and T. Doi, Synthetic Metals 17 372 (1987). C.W. Tang and S.A. Vanslyke, Appl. Phys. Lett. 51 913 (1987). T. Mori, K. Imaizumi, K. Yamashita, T. Mizutani, and H. Miyazaki, Synthetic Metals 111–112 79 (2000). T. Mori, K. Obata, and T. Mizutani, J. Phys. D 32 1198 (1999). T. Kato, T. Mori, and T. Mizutani, Thin Solid Films 393 109 (2001). R.G. Kepler, P.M. Besson, S.J. Jacobs, R.A. Anderson, M.B. Sinclair, V.S. Valencia, and P.A. Cahill, Appl. Phys. Lett. 66 3618 (1995). E.g., J. Kalinowski, M. Cocchi, V. Fattori, and P.D. Marco, J. Phys. D 34 2274 (2001). M. Matsuura, T. Akai, M. Saito, and T. Kimura, J. Appl. Phys. 79 264 (1996). S. Naka, H. Okada, and H. Onnagawa, Synthetic Metal 91 129 (1997). S.F. Nelson, Y.Y. Lin, D.J. Gundlach, and T.N. Jackson, Appl. Phys. Lett. 72 1854 (1998). K. Yamashita and A. Kitani, Dodensei-Yukihakumaku no Kinou to Sekkei (Function and Design of Conductive Organic Thin Films), (Kyoritsu Shuppan, Tokyo, 1998), pp. 18–25 [in Japanese]. T. Mori, T. Ogawa, D.C. Cho, and T. Mizutani, The 12th Int. Conf. on Solid Films and Surface, Marseille, France, 2002. P.W.M. Blom, M.J.M. de Jong, and J.J.M. Vleggaar, Appl. Phys. Lett. 68 3308 (1996). S. Karg, M. Meier, and W. Riess, J. Appl. Phys. 82 1951 (1997). A.J. Cambell, D.D.C. Bradley, and D.G. Lidzey, J. Appl. Phys. 82 6326 (1997). J.C. deMello, N. Tessler, S.C. Graham, and R.H. Friend, Phys. Rev. B 57 12951 (1998). U. Wolf, S. Barth, and H. Bässler, Appl. Phys. Lett. 75 2035 (1999). T. Mori, E. Sugimura, and T. Mizutani, J. Phys. D: Appl. Phys. 26 452 (1993). K. Imaizumi, K. Kaneko, T. Mori, and T. Mizutani, 1999 Autum Meeting, Materials Research Society, Boston, USA (1999).
Carrier transport behavior in OLED 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37.
155
K. Imaizumi, K. Kaneko, T. Mori, and T. Mizutani, Int. Symp. on Oranic Molecular Electronics Proc. Nagoya, Japan, 2000. K. Imaizumi, K. Kaneko, and T. Mori, T. Mizutani, The 10th Int. Work. on Inorganic & Organic. Electroluminescence, Hamamatsu, Japan, 2000. K. Imaizumi, K. Kaneko, T. Mori, and T. Mizutani, Trans. IEE Jpn. 121-A 332 (2001) [in Japanese]. K. Imaizumi, K. Kaneko, T. Mori, and T. Mizutani, Trans. IEE Jpn. 121-A 666 (2001) [in Japanese]. K. Imaizumi, K. Kaneko, T. Mori, and T. Mizutani, Jpn. J. Appl. Phys. 41 366 (2002). T. Ogawa, D.-C. Cho, K. Kaneko, T. Mori, and T. Mizutani, IEICE Trans. Electron. E85-C, 1239 (2002). D.V. Khramtchenkov, H. Bässler, and V.I. Arkhipov, J. Appl. Phys. 79 9283 (1996). P.S. Davids, I.H. Campbell, and D.L. Smith, J. Appl. Phys. 82 6319 (1997). B.K. Crone, P.S. Davids, I.H. Campbell, and D.L. Smith, J. Appl. Phys. 84 833 (1998). Y. Kawabe, M.M. Morrell, G.E. Jabbour, S.E. Shaheen, B. Kippelen, and N. Peyghambarian, J. Appl. Phys. 84 5306 (1998). G.G. Malliaras and J. Scott, J. Appl. Phys. 85 7426 (1999). J. Staudigel, M. Stöel, F. Steuber, and J. Simmerer, J. Appl. Phys. 86 3895 (1999). B.K. Crone, P.S. Davids, I.H. Campbell, D.L. Smith, C.J. Neef, and J. P. Ferraris, J. Appl. Phys. 86 5767 (1999). B.K. Crone, P.S. Davids, I.H. Campbell, and D.L. Smith, J. Appl. Phys. 87 1974 (2000). E. Tutis, M.N. Bussac, B. Masenelli, M. Carrad, and L. Zuppiroli, J. Appl. Phys. 89 430 (2001). T. Tsutsui, C.P. Lin, and S. Saito, Mol. Cryst. Liq. Cryst. 256 63 (1994). M. Iwamoto and T. Hino, Trans. IEE Jpn. A 100 291 (1980) [in Japanese].
Nanotechnology and Nano-Interface Controlled Electronic Devices Editors: M. Iwamoto, K. Kaneto and S. Mashiko © 2003 Elsevier Science B.V. All rights reserved
CHAPTER 10
Electrical characterization of organic semiconductor films by in situ field-effect measurements Kazuhiro Kudo Department of Electronics and Mechanical Engineering, Faculty of Engineering, Chiba University, 1-33 Yayoi-cho, Inage-ku, Chiba 263-8522, Japan
1. 2. 3.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Experimental details of in situ field-effect measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . Field-effect characteristics of organic films . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Merocyanine evaporated films . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Phthalocyanine evaporated films . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3. Hole transporting materials for EL devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4. Perylene evaporated films . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
157 158 160 160 162 168 173 178 179
1. Introduction Organic semiconductors have recently received increasing interest because of their potential applications in low-cost and large-area devices such as organic light emitting diodes (LEDs), organic field-effect transistors (FETs), and optoelectronic integrated circuits (OEIC). High carrier mobilities comparable to amorphous silicon have been obtained for several organic transistor materials. In particular, field-effect mobilities in the range from 1 to 5 cm2 /V s and on/off current ratios larger than 108 were reported in single crystal organic FETs [1–3]. Although many kinds of p-type organic semiconductors have been reported, there are few examples of n-type behavior. For practical device applications, both p-type and n-type semiconducting materials with high stability against air are highly desirable. From these points of view, it is important to investigate the intrinsic electrical properties of organic semiconductor films before and after exposing in atmospheric gasses, especially oxygen gas. In situ field-effect measurement is a promising method for the evaluation of conduction type (p or
158
K. Kudo
n), carrier mobility (µ), electrical conductivity (σ ), and carrier concentration (N ) of evaporated films [4,5]. Phthalocyanine (Pc) and merocyanine (MC) derivatives have potential for application to organic electronic devices, such as gas sensors [6,7] and optoelectronic devices [8,9] because of their p-type semiconducting properties and absorption bands which extend from the ultraviolet to the infrared region. In particular, the coordinated metals in Pc are important factors for the photoelectrical properties. It is necessary to investigate the relationship between their chemical structure and intrinsic electrical properties, especially without the influence of atmospheric gasses and impurities. The in situ fieldeffect measurement [4,5] is a promising method to evaluate the electrical parameters of organic thin films. In the present work, we have carried out the in situ field-effect measurement of several kinds of organic semiconductors expected for optoelectronic devices, and estimated the intrinsic electrical parameters, such as carrier mobility (µ), carrier concentration (N ) and electrical conductivity (σ ), and excluded the influence of atmospheric gasses and impurities with in situ field-effect measurements. The organic films were fabricated by a standard vacuum evaporation technique and the FET characteristics were investigated before and after breaking the vacuum by oxygen gas, and the thermal treatment. Furthermore, the conduction process is discussed with the experimental results.
2. Experimental details of in situ field-effect measurements A schematic of an in situ field-effect measurement system and the sample structure are shown in Figs. 1 and 2. The highly doped Si substrate which works as a gate electrode was covered with thermally grown SiO2 with a thickness of approximately 200 nm. The interdigital source and drain electrodes were formed on the substrate using standard vacuum evaporation and photolithographic techniques. The metal materials of the source and drain electrodes were chosen to make an ohmic contact to the organic
Fig. 1: Schematic of the in situ field-effect measurement.
Electrical characterization of organic semiconductor films
159
Fig. 2: Schematic of the sample structure of the in situ field-effect measurement.
materials, i.e., Au for p-type organic films and In for n-type films. The channel length and width were 0.1 and 56 mm, respectively. Substrates were thoroughly cleaned with organic solvent in an ultrasonic bath and prebaked at 373 K over 30 min in the vacuum chamber. Subsequently, organic semiconductor materials were evaporated as the active component of the FET. During the evaporation, substrate temperature, Tsub , was controlled depending on the organic materials from room temperature to 100°C (373 K). Typical thickness of the organic films was approximately 200 nm. Field-effect measurements were performed immediately after the evaporation of organic thin film, after the exposure to oxygen gas for 5 h, and after the thermal annealing of the sample at 373 K for 1 h in vacuum (10−5 Torr). All of the electrical measurements were carried out in the dark. The characteristics of the source–drain current (IDS ) vs source–drain voltage (VDS ) with applying gate voltage (VG ) were measured. IDS in the linear region for a standard TFT is given by [10] 2 ], IDS = −(W/L)COX [(VG − Vth )VDS − (1/2)VDS
(1)
where W is the channel width, L is the channel length, µ is the carrier mobility, COX is the capacitance of the SiO2 layer, and Vth is the threshold voltage. When IDS does not increase at higher VDS (saturation region), IDS is expressed by [10] IDS = −(W/2L)µCOX (VG − Vth )2 .
(2)
According to Eq. 2, µ can be obtained by plotting (VDS )1/2 against VG . The electrical conductivity (σ ) was obtained from the slope of the IDS –VDS plot at VG = 0. σ is also expressed by σ = q N µ,
(3)
where q is the elementary electric charge and N is the carrier concentration. N was obtained from Eq. 3.
160
K. Kudo
3. Field-effect characteristics of organic films 3.1. Merocyanine evaporated films Merocyanine (MC) derivatives are reported as p-type semiconducting materials and their photovoltaic effects were shown in Schottky diode or pn junction cells [6–8]. In this study, we have estimated their parameters such as field-effect mobility, carrier concentration, and electrical conductivity by field-effect measurements and discussed the transport mechanism of MC film from the temperature dependence of the FET characteristics. Fig. 3 shows the molecular structures of MC derivatives examined here. Typical characteristics of IDS vs VDS measured for MC(a) FET are shown in Fig. 4. IDS increases with negative VG and MC TFT operates in an enhancement mode. This result indicates that negative gate voltages form a hole accumulation layer and MC films show p-type semiconducting properties without influences of impurities and atmospheric gasses. From this procedure, the following electrical parameters of MC(a) film were estimated: the field-effect mobility was 1.4 × 10−6 cm2 /V s, the electrical conductivity was 2.6 × 10−10 S/cm, and the carrier concentration was 1.3 × 1015 cm−3 . The estimated value of µ depends on the thickness of the dye films. Fig. 5 shows µ variation as a function of film thickness of MC(a), MC(b), and MC(c). As the film thickness increases, µ increases and saturates at about 200 nm. These phenomena are mainly due to the effects of interface traps between SiO2 and organic film, and the carrier conduction of the discontinuous parts in thin films. Both effects appear strongly in thinner films and tend to be inconspicuous in thicker films. From this result, we have chosen as standard thickness of the organic films 200 nm. Table 1 shows the field-effect mobilities at the gate and drain bias voltages of 20 V,
Fig. 3: Molecular structures of MC derivatives examined here.
Electrical characterization of organic semiconductor films
161
Fig. 4: Typical characteristics of IDS vs VDS measured for MC(a) FET. Table 1 Field-effect mobility and photoelectric quantum efficiency of merocyanine films Dye
Field-effect mobility (cm2 /V s)
Photoelectric quantum efficiency (%)
(a) (b) (c)
1.5 × 10−5 5.0 × 10−6 1.0 × 10−7
1.0 0.3 0.1
and the photoelectric quantum efficiencies in the Schottky type cells using MC dyes (Fig. 3(a), (b) and (c)). The quantum efficiencies were roughly estimated using the illuminated monochromatic photon density and electrons produced as a short-circuit
162
K. Kudo
Fig. 5: Carrier mobility variation as a function of the film thickness of MC(a), MC(b), and MC(c).
photocurrent [8]. It should be noted that the photoelectric quantum efficiencies are closely related to the field-effect mobilities. The temperature dependence of electrical parameters obtained by in situ field-effect measurement is shown in Fig. 6. From these Arrhenius plots, the activation energies of conductivity and field-effect mobility in MC film were estimated to be 0.44 eV, and the activation energy of carrier concentration was very low (0.04 eV). These results indicate that the generation of thermally activated carriers from the shallow acceptor level is very low and there are few carriers directly excited from the band to band or HOMO (highest occupied molecular orbital) to LUMO (lowest unoccupied molecular orbital) level in this temperature region. Furthermore, the field-effect mobility is thermally activated, and hopping conduction, which is often employed in molecular films [11,12], is adequate as a main transport mechanism of MC film. Most of the hole carriers are trapped at the hopping sites of the potential well of the valence band or HOMO level. The carrier concentration estimated by the field-effect measurements in the previous section represents the number of hopping carriers. 3.2. Phthalocyanine evaporated films Phthalocyanine (Pc) films have been expected as gas sensors [6,7] and the coordinated metals in Pc are important factors for their electrical properties. It is necessary to investigate the relationship between their chemical structure and intrinsic electrical properties, especially without the influence of atmospheric gasses and impurities. As Pc derivatives, copper-phthalocyanine (CuPc), lead-phthalocyanine (PbPc), metal-free phthalocyanine (H2 Pc), and fluoro-phthalocyanine (F16 CuPc) were examined here, and n-type behavior of F16 CuPc was reported [13,14]. Fig. 7 shows the molecular structures of the Pc derivatives and these materials were purified by the sublimation method. Typical FET characteristics (IDS vs VDS as a function of VG ) of a CuPc sample after the deposition at Tsub of 373 K are shown in Fig. 8(a). IDS increases with negative
Electrical characterization of organic semiconductor films
163
Fig. 6: Temperature dependence of electrical parameters obtained by in situ field-effect measurement.
VG and the CuPc FET operates in an enhancement mode. This result indicates that negative gate voltages enlarge the conduction channel due to the formation of a hole accumulation layer, thus, CuPc films show p-type semiconducting properties. We have also investigated the effect of oxygen gas and annealing on the electrical properties of the films. Fig. 8(b) and (c) shows FET characteristics of the CuPc sample after the oxygen gas exposure and after the thermal treatment in vacuum, respectively. Although the IDS increases after the oxygen gas exposure, the IDS decreases by the thermal treatment at 373 K in vacuum for 1 h. It was also confirmed that H2 Pc and PbPc showed p-type semiconducting properties in the absence of atmospheric gasses. However, the
164
K. Kudo
Fig. 7: Molecular structures of Pc derivatives.
effect of oxygen gas on the FET characteristics depends on the molecular species. Comparing FET characteristics of these p-type materials, the effect of oxygen on PbPc film is significant but that on H2 Pc films is small. On the other hand, typical FET characteristics of F16 CuPc films are shown in Fig. 9. Electrical properties of F16 CuPc films are rather stable against the oxygen gas exposure. The experimental results in Table 2 and Fig. 9 indicate that the influence of oxygen gas exposure is small and the fluorine atoms in F16 CuPc molecules seem to prevent oxygen adsorption. The carrier mobility, µ, conductivity, σ , and carrier concentration, N obtained by the in situ field-effect measurements are shown in Table 2. Fig. 10 shows the variations of N and µ of as-grown sample, after oxygen gas exposure and after annealing in vacuum. It is noteworthy that the most significant change in N occurs upon the exposure to oxygen gas. Particularly in PbPc films, N increased when the sample was exposed to oxygen gas and decreased when the sample was annealed in vacuum. The effect of oxygen on N is marked in PbPc films compared with those in H2 Pc and F16 CuPc films. The effect of oxygen on CuPc film is between those of the PbPc and H2 Pc films. Thus, oxygen gas acts as an acceptor impurity and increases the net charge carriers. It is considered that electrons transfer from phthalocyanine molecules to oxygen molecules
Electrical characterization of organic semiconductor films
165
Fig. 8: Typical FET characteristics of a CuPc sample, (a) as-grown at Tsub of 373 K, (b) oxygen exposure (5 h), (c) annealing at 373 K in vacuum (1 h).
166
K. Kudo
Fig. 9: Typical FET characteristics of F16 CuPc films, (a) as-grown, (b) oxygen exposure (5 h), (c) annealing at 373 K in vacuum (1 h).
Electrical characterization of organic semiconductor films
167
Table 2 Electrical parameters of phthalocyanine derivatives obtained by in situ field-effect measurements Material
As-grown
O2 gas exposure
Annealing in vacuum
Hole mobility (cm2 /V s) PbPc 373 CuPc 373 H2 Pc 373 F16 CuPc 373
Tsub (K)
3.8 × 10−6 2.4 × 10−6 2.0 × 10−6 1.6 × 10−3
5.7 × 10−5 1.2 × 10−5 2.1 × 10−6 1.0 × 10−3
3.2 × 10−5 1.1 × 10−5 3.6 × 10−6 1.8 × 10−3
Conductivity (S/cm) PbPc 373 CuPc 373 H2 Pc 373 F16 CuPc 373
1.3 × 10−9 2.1 × 10−10 1.1 × 10−10 5.6 × 10−5
7.7 × 10−7 4.6 × 10−10 1.4 × 10−10 2.1 × 10−5
6.6 × 10−8 1.3 × 10−9 7.5 × 10−11 4.4 × 10−5
Carrier concentration (cm−3 ) PbPc 373 CuPc 373 H2 Pc 373 F16 CuPc 373
2.1 × 1015 5.4 × 1014 3.4 × 1014 2.2 × 1017
8.4 × 1016 2.4 × 1015 4.1 × 1014 1.2 × 1017
1.3 × 1016 8.2 × 1014 1.3 × 1014 1.5 × 1017
and oxygen molecules are directly related to the composition of the central metal of phthalocyanine molecules. This result is closely related to reports that the interaction between H2 Pc and oxygen is weak [15] or that the adsorption site of H2 Pc is different from that of metal phthalocyanines [16]. F16 CuPc has the largest mobility and H2 Pc the smallest, indicating that the carrier mobility is dependent on the coordinated metal and fluorine atoms. These seem to be intrinsic properties of the materials measured, since the effect of oxygen was excluded during measurements. It should be noted that the deposition at high temperature gave high µ for all Pc films. Evaluation of the thermal activation energy (E a ) is important to discussion of the carrier transport mechanism in organic semiconducting films. In particular, the temperature dependence of σ , µ, and N is necessary, because E a estimated by σ contains both µ and N , as expressed in Eq. 3. Fig. 11 shows E a (σ ), E a (µ), and E a (N ) estimated by Arrhenius plots between 295 and 380 K for CuPc film. Plots of log σ , log µ, log N against 1/T follow a straight line. The carrier mobility of CuPc film has an activation energy of about 0.26 eV. E a (µ) obtained by the field-effect measurement can be interpreted in terms of a barrier which prevents the carrier transport at the interface between the micrograins or between the organic film and the electrodes. The effect of oxygen on E a (σ ) and E a (N ) depends on the chemical structure of the Pc molecules. Although E a of F16 CuPc and H2 Pc remain almost the same values, those of PbPc and CuPc change to lower values after the oxygen exposure. In particular, E a (σ ) of PbPc film varied from 0.53 to 0.23 eV by introducing oxygen gas. After thermal treatment in vacuum, E a (σ ) recovers to almost the same value of 0.54 eV. These results indicate that oxygen molecules interact with the PbPc molecules and a shallow acceptor level is formed near the valence band edge or HOMO (highest occupied molecular orbital) level
168
K. Kudo
Fig. 10: The variations of N and µ of as-grown sample, after oxygen gas exposure and after annealing in vacuum.
of the PbPc. Thus, oxygen gas acts as an acceptor impurity and increases the net charge carriers. 3.3. Hole transporting materials for EL devices Organic electroluminescent (EL) devices have already been put to practical use for display panels. However, basic electrical parameters of organic materials used in EL devices have been hardly reported. Thin-film transistors (TFTs) using hole transporting materials of organic EL devices were fabricated and the electrical parameters were estimated by in situ field-effect measurements. Fig. 12 shows molecular structures of TPD (N ,N -diphenyl-N ,N -di(3-methylphenyl)-1,1 -biphenyl-4,4 -
Electrical characterization of organic semiconductor films
169
Fig. 11: Arrhenius plots of σ , µ, and N between 295 and 380 K for CuPc film.
diamine), a-NPD (N ,N -diphenyl-N ,N -di(1-napthyl)-1,1 -biphenyl-4,4 -diamine), and m-MTDATA (4,4 ,4 -tris-(3-methylphenyl-phenylamino)triphenylamine). The growth temperatures (Tsub ) of these materials were selected below the glass transition temperature (Tg ). The glass transition temperatures of TPD, a-NPD, and m-MTDATA are 63, 96, and 75°C, respectively.
170
K. Kudo
Fig. 12: Molecular structures of TPD, a-NPD, and m-MTDATA.
Typical FET characteristics of TPD, a-NPD and m-MTDATA just after deposition are shown in Figs. 13(a), (b), and (c), respectively. For all the materials, IDS increases with increasing negative VG . These results indicate that a hole accumulation layer is formed by negative gate voltages and evaporation films of hole transport materials have p-type semiconducting properties. From these results, the estimated field-effect carrier mobilities of TPD, a-NPD, and m-MTDATA were 1.6 × 10−6 , 9.3 × 10−8 , and 6.3 × 10−7 cm2 /V s, respectively (Table 3). These values of µ are smaller than those obtained by the time-of-flight method [17]. The difference in the µ values using the FET
Electrical characterization of organic semiconductor films
Fig. 13: Typical FET characteristics of TPD, a-NPD and m-MTDATA just after deposition.
171
172
K. Kudo
Table 3 Electrical parameters of hole transporting materials obtained by in situ field-effect measurements Material
As-grown
O2 gas exposure
Annealing in vacuum
Hole mobility (cm2 /V s) TPD 333 α-NPD 353 m-MTDATA 343
Tsub (K)
1.6 × 10−6 9.3 × 10−8 6.3 × 10−7
1.3 × 10−6 2.3 × 10−7 6.0 × 10−7
2.3 × 10−6 3.2 × 10−7 6.2 × 10−7
Conductivity (S/cm) TPD 333 α-NPD 353 m-MTDATA 343
1.3 × 10−11 1.1 × 10−12 1.7 × 10−10
8.1 × 10−12 9.4 × 10−13 2.4 × 10−10
7.6 × 10−12 6.5 × 10−12 1.3 × 10−10
Carrier concentration (cm−3 ) TPD 333 α-NPD 353 m-MTDATA 343
2.2 × 1013 7.6 × 1013 1.7 × 1015
2.6 × 1013 3.6 × 1013 2.5 × 1015
1.1 × 1013 1.3 × 1014 1.3 × 1015
and time-of-flight method is probably due to the difference in the electric field, current direction though the organic films, etc. Fig. 14 shows the variations of N and µ of as-grown sample, after oxygen gas exposure and after annealing in vacuum. As shown in Fig. 14(a) and (b), these hole transporting materials show stable against the oxygen exposure. Fig. 15 shows the electrical properties of TPD, a-NPD, and m-MTDATA films as a function of substrate temperature (Tsub ) during the deposition. The conductivity of m-MTDATA is higher than that of the other materials. The conductivities of a-NPD and m-MTDATA have weak dependence on Tsub below Tg . However, the conductivity of TPD has a peak at around 55°C. The conductivities of all the materials show a drastic change at Tsub over Tg . The carrier mobilities of these materials show clearer Tsub dependence. The carrier mobility of a-NPD and m-MTDATA changes to a low value under Tg , the carrier mobility of m-MTDATA is one order larger than that of a-NPD. On the other hand, the carrier mobility of TPD has a sharp peak at Tsub around 55°C. Similar phenomena were observed by the thermal treatment in air after the film deposition. The electrical properties of all the materials show a drastic change after the thermal treatment over Tg . Fig. 16 shows AFM (atomic force microscope) images of TPD films after thermal treatment below and above Tg . Though the surface of the TPD film after the thermal treatment at 50°C (< Tg ) was completely flat (Fig. 16(a)), the morphology of the film after the thermal treatment at 70°C (> Tg ) changed to that of an island-like film (Fig. 16(b)). The steepness of the change in electrical properties at a higher temperature is mainly due to the discontinuity of the conductive channel caused by the formation of island-like grains after the thermal treatment (Fig. 17). These results indicate that the degradation of organic EL devices at higher temperatures is closely related to the formation of island-like grains in hole transport layers at higher temperature over Tg .
Electrical characterization of organic semiconductor films
173
Fig. 14: The variations of N and µ of as-grown sample, after oxygen gas exposure and after annealing in vacuum.
3.4. Perylene evaporated films Three kinds of perylene derivatives (PTCDI, PTCDA, BPPC) were used and n-type characteristics were reported [18–20]. The chemical structures are shown in Fig.18. These materials purified by sublimation under an argon flow. Fig. 19 shows typical FET characteristics of PTCDI films. The channel conduction of all perylene derivatives increases with positive VG and the FETs operate in an enhancement mode. From the results of field-effect measurements, as deposited films of
174
K. Kudo
Fig. 15: Electrical properties of TPD, a-NPD, and m-MTDATA films as a function of substrate temperature (Tsub ) during the deposition.
Electrical characterization of organic semiconductor films
175
Fig. 16: AFM (atomic force microscope) images of TPD films after thermal treatment below and above Tg .
PTCDI, PTCDA, and BPPC showed n-type semiconducting properties in the absence of atmospheric gasses. The introducing of oxygen gas to the n-type films works to decrease the conductivity and the degree of change depends on the molecular species. FET characteristics of PTCDI film obtained just after the deposition were completely
176
K. Kudo
Fig. 17: Change of conduction channel due to island-like grain formation.
Fig. 18: Molecular structures of perylene derivatives (PTCDI, PTCDA, BPPC).
ruined by the oxygen gas exposure (Fig. 19(b)) and did not recover to the same values as those of as-grown samples. The hole mobility, µ, conductivity, σ , and carrier concentration, N , obtained by the in situ field-effect measurements are shown in Table 4. The effect of oxygen gas exposure on perylene films is in a marked contrast to that of p-type materials. The variations of N and µ in perylene films are shown in Fig. 20, for as-grown sample, after oxygen gas exposure and after annealing in vacuum. In some cases, oxygen gas acts as an acceptor impurity and produces good results for p-type materials, but oxygen
Electrical characterization of organic semiconductor films
Fig. 19: Typical FET characteristics of PTCDI films.
177
178
K. Kudo
Fig. 20: The variations of N and µ in perylene films.
gas exposure causes serious problems for n-type materials. These phenomena can be explained as follows: oxygen gas acts as an acceptor impurity and carrier compensation occurs in n-type films. The variation of the electrical properties, however, depended on the molecular structure and the growth condition of the films.
4. Conclusions The basic electric parameters of several kinds of organic thin films were evaluated by in situ field-effect measurement. The effects of thermal treatment and introducing oxygen gas on the electrical properties were also investigated. A marked change in the electrical parameters corresponding to the adsorption and desorption of oxygen molecules was observed and the variation of the electrical properties strongly depended
Electrical characterization of organic semiconductor films
179
Table 4 Electrical parameters of perylene derivatives obtained by in situ field-effect measurements Material
Tsub (K)
As-grown
O2 gas exposure
Annealing in vacuum
Electron mobility (cm2 /V s) PTCDI 373 PTCDA 373 BPPC 373
2.9 × 10−5 4.4 × 10−6 2.4 × 10−7
< 10−7 2.4 × 10−7 < 10−7
3.3 × 10−6 1.4 × 10−6 1.1 × 10−8
Conductivity (S/cm) PTCDI 373 PTCDA 373 BPPC 373
2.1 × 10−7 6.1 × 10−7 3.5 × 10−11
< 10−12 8.2 × 10−10 < 10−12
6.6 × 10−10 7.4 × 10−9 1.4 × 10−12
Carrier concentration (cm−3 ) PTCDI 373 PTCDA 373 BPPC 373
4.5 × 1016 8.7 × 1017 9.0 × 1014
< 1014 2.1 × 1016 < 1014
1.2 × 1015 3.4 × 1016 8.2 × 1014
on the molecular structure and the growth conditions of the films. These results demonstrate that the influence of atmospheric gases is significant for organic device applications and the in situ field-effect measurement is a powerful method to investigate the fundamental properties of organic materials.
References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20.
C.D. Dimitrakopoulos and P.R.L. Malenfant, Adv. Mater. 14, 99 (2002). H. Klauk, D.J. Grundlach, J.A. Nichols, C.D. Sheraw, M. Bonse, and T.N. Jackson, Solid State Tech. 43, 63 (2000). C.R. Kagan, D.B. Mitzi, and C.D. Dimitrakopoulos, Science 286, 945 (1999). K. Kudo, M. Yamashina, and T. Moriizumi, Jpn. J. Appl. Phys. 23, 130 (1984). K. Kudo, T. Sumimoto, K. Hiraga, S. Kuniyoshi, and K. Tanaka, Jpn. J. Appl. Phys. 36, 6994 (1997). C. Hamann, A. Mrwa, M. Muller, W. Gopel, and M. Rager, Sens. Actuat. B 4, 73 (1991). A. Wilson, J.D. Wright, and A.V. Chadwick, Sens. Actuat. B 4, 499 (1991). A.K. Ghosh, D.L. Morel, T. Feng, R.F. Shaw, and C.A. Rowe Jr., J. Appl. Phys. 45, 230 (1974). K. Kudo, T. Shinohara, T. Moriizumi, K. Iriyama, and M. Sugi, Jpn. J. Appl. Phys., Suppl. 20-2, 135 (1981). S.M. Sze, Physics of Semiconductor Devices (Wiley, New York, 1969) p. 425. K. Kaneto, K. Yamanaka, K. Rikitake, T. Akiyama, and W. Takashima, Jpn. J. Appl. Phys. 35, 1802 (1996). H. Bassler, G. Schonherr, M. Abkowitz, and D.M. Pai, Phys. Rev. B 26, 3105 (1982). Z. Bao, A.J. Lovinger, and A.Dodabalapur, Adv. Mater. 9, 42 (1997). Z. Bao, A.J. Lovinger, and J. Brown, J. Am. Chem. Soc. 120, 207 (1998). J. Simon and J.J. Andre, Molecular Semiconductors (Springer-Verlag, Berlin, 1985) p. 116. J.P. Contour, P. Lenfant, and A.K. Vijh, J. Catal. 29, 8 (1973). S. Naka, H. Okada, H. Onnagawa, Y. Yamaguchi, and T. Tsutsui, Synth. Met. 111–112, 331 (2000). G. Horowitz, F. Kouki, P. Spearman, D. Fichou, C. Nogues, X. Pan, and F. Garnier, Adv. Mater. 8, 242 (1996). A. Brown, D.M. de Leeuw, E.J. Lous, and E.E. Havinga, Synth.Met. 66, 257 (1994). T. Suga, M. Iizuka, S. Kuniyoshi, K. Kudo, and K. Tanaka, Synth. Met. 102, 1050 (1999).
Part C
Smart Soft Materials
Nanotechnology and Nano-Interface Controlled Electronic Devices Editors: M. Iwamoto, K. Kaneto and S. Mashiko © 2003 Published by Elsevier Science B.V.
CHAPTER 11
Introducing ruber into the Langmuir–Blodgett technique H. Xu a , R. Heger b , F. Mallwitz a , M. Blankenhagel b , C. Peyratout b , and Werner A. Goedel a a Macromolecular
and Organic Chemistry, OC3, University of Ulm, Germany für Kolloid- & Grenzflächenforschung, Berlin, Germany
b Max-Planck-Institut
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
183 184 188 189 189
1. Introduction The Langmuir–Blodgett (LB) technique offers the opportunity to generate suspended membranes by assembling a monolayer at the air–water interface and transferring it to cover a hole in a solid substrate. However, the preparation of suspended membranes via LB-transfer is generally more difficult than LB-transfer of thin organic coatings onto continuous smooth surfaces: Because it is not supported by an underlying substrate, the suspended membrane itself must be tough enough to withstand mechanical stress during fabrication and final use. Monolayers that are made from low molecular weight compounds or from liquid polymers easily rupture during transfer across a hole. Suspended membranes have been fabricated using glassy polymers, often stabilised by cross-linking [1,2]. These membranes are usually rigid. For certain applications, like membranes in micro mechanical valves and pumps, it might be advantageous to have elastomeric thin membranes available and to take advantage of the comparatively large reversible deformation of these materials. Here, we show that tough and mechanically stable freely suspended membranes – spanning millimetre sized holes in solid substrates – can be obtained from cross-linked monolayers of low Tg polymers with ionic head groups.
184
H. Xu et al.
2. Theory When applied to a water surface, liquid polymers like perfluoropolyethers [3], polyisoprene [4], polybutadiene [5] or polydimethylsiloxane [6] with ionic head groups easily form smooth and continuous monolayers. By variation of the polymer chain length and surface concentration, the thickness of these monolayers can easily be tuned in the range of 10 to 100 nm thickness [7]. If these monolayers are transferred to substrates with openings (e.g. an electron microscopy grid) as schematically depicted in Fig. 1 they initially cover the openings as approximately 50 nm thin bilayers. However, these membranes rupture within minutes after transfer. As an example two images of a membrane made via LB-transfer of polyisobutene with a single head group and a chain length of 300 repeat units, obtained shortly after transfer and 20 min later are shown in Fig. 2. Within 30 min, all membranes covering holes in the grid rupture. This rupture of the membrane can be expected; the membrane closely resembles a soapy membrane made out of a water core that is coated from both sides with a liquid layer of amphiphiles. Such membranes usually are metastable and rupture, especially if the water of the core region evaporates. The rupture can be suppressed, however, and one can obtain stable membranes if the monolayer is solidified before or shortly after transfer. This solidification has been achieved using three different principles: vitrification, photochemical cross-linking and physical cross-linking. Freely suspended membranes stabilised by vitrification have been prepared from monolayers of poly-4-n-butylstyrene with trimethylammoniumbromide head groups. Polybutylstyrene has a glass transition temperature of 25°C. Hence, at elevated temperatures polybutylstyrenes with ionic head groups, applied to a water surface, behave essentially like polyisoprenes at room temperature and form smooth and continuous
monolayer
solid support
transfer to grid Water Fig. 1: Scheme of the formation of a freely suspended membrane via Langmuir–Blodgett transfer of an anchored polymer monolayer to substrates with openings.
Introducing ruber into the Langmuir–Blodgett technique
185
Fig. 2: A freely suspended membrane generated by Langmuir–Blodgett transfer of a monolayer of polyisobutene with a single ionic head group ruptures within 30 min. (Light microscopy image with top and bottom illumination.)
Fig. 3: Light microscopy image of an approximately 20 nm thick freely suspended membrane of polybutylstyrene-N+ prepared by spreading at 40°C, cooling to 10°C and transfer to an electron microscopy grid.
monolayers. Upon cooling, the polybutylstyrene monolayers vitrify to room temperature and thus can be transferred to yield solid freely suspended membranes [8] (Fig. 3). Freely suspended membranes stabilised by chemical cross-linking have been obtained by irradiation of monolayers of polyisoprenes with ionic head groups and anthracene side chains. Upon irradiation with soft UV-light, the anthracene side chains dimerise (see Fig. 4). This dimerisation of side chains gives rise to permanent cross-linking points. Since the polyisoprene chains have a low glass transition temperature, this cross-linking transforms the initially liquid monolayer into a thin layer of an elastomer. This layer can easily be transferred across openings in solid substrates [9]. The resulting freely suspended membranes are long-term stable (at least several months).
186
H. Xu et al.
Fig. 4: Scheme of the preparation of elastomeric membranes via cross-linking of the side chains of anthracene tagged polyisoprene with ionic head groups.
Fig. 5: Top and side view of an elastomeric cross-linked polyisoprene membrane reversibly deformed by a small overpressure from below.
The elastomeric properties of these freely suspended membranes can be shown by applying a small pressure from one side. When the pressure is applied, the membrane bulges. When the pressure is released, the membrane flattens itself reversibly (see Fig. 5). In this procedure the monolayer is cross-linked and converted into a solid layer on the water surface; after cross-linking it can sustain neither shear flow nor extensional flow. When a monolayer is transferred to a substrate, which is smaller than the Langmuir trough, it has to undergo two-dimensional flow. Otherwise it will develop stress or will wrinkle, especially if several substrates are coated consecutively. Thus,
Introducing ruber into the Langmuir–Blodgett technique
187
Fig. 6: Scheme of physical cross-linking of a suspended membrane of polymer chains with multiple head groups. Upon drying of the water core, which is initially present in the transferred membrane, the head groups aggregate and form physical cross linking sites.
Fig. 7: Top view of an elastomeric polyisobutene membrane cross-linked via aggregation of multiple ionic head groups. (a) Chemical formula of the polyisobutene. (b) Membrane spanning a 300 µm hole in a brass plate. (c)–(f) Membrane bulges upward upon applying a small pressure from below.
in the experiments depicted above only small substrates were coated and most of the monolayer had to be discarded. The two-dimensional-flow problem can be avoided by transferring liquid monolayers and cross-linking them shortly after the transfer. This can be accomplished quite easily by using polymers with more than one ionic group per chain. On the water surface, these polymers behave similar to the polymers with single ionic head groups. Like the linear polymers depicted in Fig. 2, the monolayers of the three-arm-star polymers can be transferred to cover holes in solid substrates. In the case of the star polymers with several ionic head groups, however, the ionic groups form inverted micelles when the membrane dries. These inverted micelles efficiently cross-link the polymer and thus give rise to the formation of elastomeric membranes without irradiation being necessary [10] (see Fig. 6). Like in the case of photochemically cross-linked membranes, these physically crosslinked membranes are elastomeric. In Fig. 7(b) to (d) a continuously increasing pressure
No Fig. 10. Inserted s. 8 and 9. Please eck.
188
H. Xu et al.
Fig. 8: Scheme of generating porous membranes via incorporation of colloids into polymeric monolayers, followed by cross-linking, transfer and removal of the colloids.
Fig. 9: (a) Hybrid monolayer composed of anchored polymers and silica colloids. (b) Porous monolayer obtained after removal of the colloids.
of approximately 10–100 Pa is applied to a freely suspended membrane. The higher the applied pressure, the more does the membrane bulge upward. Upon release of the pressure this deformation is completely reversible and can be repeated multiple times. Cospreading polymers with ionic anchor groups and hydrophobised silica colloids on a water surface, followed by transfer to solid substrates of electron microscopy grids gives rise to mixed monolayers. In these monolayers, domains of silica particles are embedded in a continuous matrix of polymeric monolayers (see Figs. 8 and 9). Exposure of these membranes to hydrofluoric acid vapour removes the silica particles. This gives rise to porous monolayers and porous membranes of controlled porosity with a uniform pore size distribution [11]. These membranes are promising for applications like ultrafiltration, bio-encapsulation and as masks and moulds for generating new nanoscopic and mesoscopic structures and surface patterns.
3. Experimental Linear polyisoprene with a sulfonate head group and anthracene side groups has been synthesised via living anionic polymerisation followed by platinum catalysed hydrosilylation of the sulfonate terminated parent polyisoprene as published in [4] and [9]; polybutylstyrene with ammonium head group has been synthesised via living
Introducing ruber into the Langmuir–Blodgett technique
189
anionic polymerisation as described in [8]; polyisobutenes with sulfonate head groups have been synthesised via living cationic polymerisation as described in [12–14]. Monolayers on a water surface were prepared using a 20 cm × 46 cm rectangular Langmuir trough made of polytetrafluoroethylene, equipped with one compression barrier and a floating barrier for the detection of the surface pressure via the Langmuir method (Lauda FW2, Germany). The polymers were usually spread from chloroform solutions which contained 0.05 wt% of polymer and 10 wt% of ethanol (polyisoprene, polybutylstyrene) or from 4 × 10−4 wt% solutions in ethanol/pentane mixtures (1/50 by weight) (polyisobutenes). UV-illumination was made through the thermostatted, transparent lid of the trough using an array of four 30 cm long fluorescence lamps mounted parallel in an (40 × 40 cm) aluminium housing (Philips TL 36D 25/09N). The emission of the lamps was between 305 to 420 nm, maximum emission was at λmax = 355 nm. 30 min before and during illumination, the air space above the air–water interface was flushed with nitrogen (5 l/min). Silica colloids coated with polyisobutene amphiphiles (mean radius = 70 nm, polydispersity = 11%, suspended in cyclohexane) were obtained from Utrecht Colloid Synthesis Facility, Van ’t Hoff Laboratory for Physical and Colloid Chemistry, Utrecht University, The Netherlands.
Acknowledgements This work was partially conducted in the Max-Planck-Institute of Colloids and Interfaces and in the University of Ulm. The support by H. Möhwald, M. Antonietti, M. Möller, by the Max-Planck Gesellschaft and the Deutsche Forschungsgemeinschaft (Go 693/1, Go 693/6, SFB 569) is gratefully acknowledged. We thank S. Förster (MPI-KGF) for introducing us into living anionic polymerisation and M. Grasmüller and O. Nuyken (TU-Munich) for introducing us to living cationic polymerisation. We thank Dr. Carlos van Kats, and Dr. Judith Wijnhoven, A. van Blaadern, A. Phillipse (Utrecht Colloid Synthesis Facility) for providing the silica colloids and for helpful discussions.
References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12.
M. Seufert, C. Fakirov, and G. Wegner, Adv. Mater. 7, 52 (1995). M. Kunitake, T. Nishi, H. Yamamoto, K. Nasu, O. Manabr, and N. Nakashima, Langmuir 10, 3207 (1994). W.A. Goedel, C. Xu, and C.W. Frank, Langmuir 9, 1184 (1993). R. Heger and W.A. Goedel, Macromolecules 29, 8912 (1996). P. Christie, M.C. Petty, and G.G. Roberts, Thin Solid Films 134, 75 (1985). T.J. Lenk, D.H.T. Lee, and J.T. Koberstein, Langmuir 10, 1857 (1994). H. Baltes, M. Schwendler, C.A. Helm, R. Heger, and W.A. Goedel, Macromolecules 30, 6633 (1997). W.A. Goedel, C. Peyratout, L. Ouali, and V. Schädler, Adv. Mater. 11, 213 (1999). W.A. Goedel and R. Heger, Langmuir 14, 3470 (1998). F. Mallwitz and W.A. Goedel, Angew. Chemie Int. Ed. 40, 2557 (2001); Angew. Chemie 113, 2716 (2001). Hui Xu and W.A. Goedel, Langmuir 18, 2363 (2002). R. Santos, J.P. Kennedy, and M. Walters, Polymer Bull. 11, 261 (1984).
190 13. 14.
H. Xu et al. J.P. Kennedy, L.R. Ross, J.E. Lackey, and O. Nuyken, Polymer Bull. 4, 67 (1981); J.P. Kennedy, L.R. Ross, and O. Nuyken, Polymer Bull. 5, 5 (1981). R.F. Storey and Y. Lee, J. Polym. Sci. Polym. Chem. 29, 317 (1991).
Nanotechnology and Nano-Interface Controlled Electronic Devices Editors: M. Iwamoto, K. Kaneto and S. Mashiko © 2003 Elsevier Science B.V. All rights reserved
CHAPTER 12
Design of functional interface between living systems and semiconductor nano-structures Motomu Tanaka Lehrstuhl für Biophysik, Technische Universität München, James-Franck-Str. 1, D-85748 Garching, Germany E-mail:
[email protected]
1. 2.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Engineering of semiconductor surfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Grafting of alkylsiloxane monolayers on ITO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. Electrochemical passivation of bulk GaAs electrodes with thiol monolayers . 2.3. Passivation of semiconductor nano-structures close to surface . . . . . . . . . . . . . . 2.4. Outlooks – engineering of GaAs with functional 4,4 -mercaptobiphenyls . . . 3. “Soft cushions” at interface – biocompatible polymer films as physical models of extracellular matrices (ECMs) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Hydration and wetting of polysaccharide films . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Chemical control of wetting interaction – grafting of synthetic polymer brushes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3. Outlooks – chemical “switching” of cell/surface interactions . . . . . . . . . . . . . . . 4. Physical models of cell surface glycocalix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1. Thermodynamics and hydration of glycolipid monolayers . . . . . . . . . . . . . . . . . . 4.2. Morphology of glycolipids – structural basis of carbohydrate complexes . . . . 4.3. Viscoelasticity of oligosaccharides on cell membranes – rheology at the interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5. Native and model cell membranes on semiconductors – potential candidates for sensorics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1. Uptake of antibiotic peptides into model membranes on semiconductor electrodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2. Novel charge sensor based on bio-membrane/semiconductor hybrids . . . . . . . 5.3. Orientation-selective immobilization of native cell membranes on ultra-thin polymer films . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
192 193 193 194 197 199 201 202 205 207 209 209 211 214 215 218 221 223 227 228
192
M. Tanaka
1. Introduction Design of novel hybrid devices by the functional combination of living systems (e.g. proteins, cells) and nano-structured semiconductors is attractive from scientific aspects as well as from technological point of view [1] (Fig. 1). Scientifically, glycolipids, membrane receptors, or proteins of the extracellular matrix immobilized onto semiconductor devices can provide physical models of cell- and tissue surfaces, which allows the investigation of the basic principles of their complex functions in nature. With the aid of various surface sensitive techniques, (a) quantitative detection of protein–protein recognition processes at membranes and (b) fine-tuning of the adhesion forces between cells and surfaces by the interplay of specific (lock-andkey) forces and universal interfacial forces are possible. Technological applications should profit from the possibilities to immobilize various membrane proteins under non-denaturing conditions to obtain fast screening of the uptake of drugs/toxins and highly sensitive detection of “local” binding and adsorption to the surfaces. Especially, model cell membranes reconstituted on bulk and nanostructured semiconductors (e.g. ITO, GaAs-, and Si-based) have a large potential towards intelligent sensors based on electro-optical transducers. For example, isolated single quantum dots could be utilized as point-like luminescence sources, while the transport of the charged species across the bio-membranes can be detected by in-plane gate transistors with narrow (width ∼ 100 nm) channels. To date, such strategies have been impeded upon (1) the electrochemical decomposition of semiconductors (e.g. GaAs) under physiological conditions, (2) the strong, nonspecific Van der Waals attractions (hydrophobic–hydrophobic interactions), and (3) the toxicity of semiconductor surfaces, which often leads to the denaturing of proteins and cell death (apoptosis). Thus, to achieve this goal, fabrication of biocompatible and functional inter-layers to bridge “wet and soft” biological matters and “dry and hard” solid surfaces is required.
hν
Ψs
S
G D
G
Fig. 1: Schematic representation of a bio-membrane/semiconductor hybrid, based on stratified molecular constructs. Micro- or nano-structured semiconductors can be used as “local” sensors to detect the membranemediated processes, e.g. optical sensing of protein functions with near-surface quantum dots, detection of changes in surface potentials due to ligand–receptor coupling, monitoring selective material transport across the membrane using in-plane FET and 2 dimensional electron gas (2DEG), etc.
Living systems and semiconductor nano-structures
193
This review deals with the step-wise construction of hierarchical molecular assemblies, starting from surfaces of semiconductor nano-structures up to cell membranes. Required functionalization steps include: (A) engineering of the semiconductor surfaces with organic molecules; (B) deposition of biocompatible polymer films and control their interfacial characteristics; (C) physical modeling of cell surface glycocalix; (D) reconstitution and characterization of native- and model cell membranes. To optimize the interfacial interaction between soft and stratified layers, systematic combination of different surface-sensitive methods is necessary. Several examples will be introduced in the following sections.
2. Engineering of semiconductor surfaces Self-assembled monolayers (SAMs) are suitable both to render the surface hydrophobic for the later modification and to form very thin insulating layers that prevent leak currents, unspecific adsorption, and the surface decomposition in aqueous electrolytes [2]. To date, the most intensively studied systems along this strategy are SAMs of alkanethiols on gold electrodes, which is mainly due to the chemical stability of the interfaces. The investigations of blocking efficiencies of the SAMs against heterogeneous electron transfer and ion penetration enable to determine the surface coverage quantitatively [3–7]. Another class of SAMs widely used is based on alkylsiloxane monolayers on different oxide surfaces (e.g. SiO2 , Al2 O3 , SnO2 , TiO2 , etc.) [8,9]. A variety of characterization techniques have been applied to characterize the monolayer structures, such as contact angle measurements [10,11], X-ray reflectivity or ellipsometry [12]. In addition, these molecules can provide self-assembled monolayers with different functional groups [13]. To date, there have been only a few reports concerning the electrical properties of SAMs on semiconductors. One example is the electrical characterization of alkyltrichlorosilane monolayers on Si/SiO2 , discussed in terms of the suppression of charge carrier tunneling [14,15]. Nevertheless, systematic studies under physiological conditions (in aqueous electrolytes, near neutral pH) are still missing. 2.1. Grafting of alkylsiloxane monolayers on ITO Indium-tin-oxide (ITO) is stable under physiological conditions because of its polarizable properties [16,17], maintaining high sensitivity without insulating oxide layers. Furthermore, ITO is transparent to visible light, which enables multiple parameter measurements by using optical and electrical techniques. Previously, we reported the deposition of octadecyl trichlorosilane (OTS) via silylation with surface oxides [18]. The local defect area ratio, i.e. surface coverage, was quantitatively evaluated using impedance spectroscopy by measuring changes in the charge transfer resistance in the presence of redox couples (Fe2+/3+ ), yielding a defect area ratio of about 0.9%. However, the electrical properties of OTS monolayers were not observed in buffered electrolytes. Indeed, OTS is highly reactive and often makes poly-
194
M. Tanaka 15 10
Current [µA]
5 0 -5 -10
-0.4
-0.2
0.0
0.2
0.4
Bias potential [V]
Fig. 2: Cyclic voltammogram of ITO electrodes at scan rate 25 mV/s. Solid line: Bare ITO in 10 mM Hepes buffer (pH = 7.5); broken line: bare ITO in 10 mM Hepes buffer containing 1 mM K3/4 Fe(CN)6 (pH = 7.5); dotted line: ITO coated with octyltrimethoxysilane (OTMS) monolayer in redox buffer (pH = 7.5). The voltage was applied versus a Ag/AgCl reference electrode.
meric siloxane structures, and decomposition of ITO by a sub-product of the coupling reaction (HCl) led to a poor reproducibility of the preparation. In a more recent study [19], SAMs of alkylsilanes with trimethoxy coupling groups (octyltrimethoxysilane, OTMS) have been deposited onto ITO surfaces. Hydrophobicity and homogeneity of the surfaces are discussed by measuring contact angles between the ITO electrodes and water droplets before and after the monolayer deposition. Electrochemical properties of the SAMs (e.g. resistance, capacitance, defect area ratio, ion penetration) were measured quantitatively by using cyclic voltammetry and impedance spectroscopy (Figs. 2 and 3). The voltammograms proved the polarizable properties of ITO electrodes in buffered electrolyte over a large potential range. In addition, cyclic voltammetry in the presence of redox couples demonstrated the significant passivation effect of the SAMs against surface electrochemistry, in spite of the intrinsic roughness of the ITO electrodes (r.m.s. ∼ 2.5 nm). The impedance spectrum of the SAMs exhibited an obvious difference compared to that of the pure electrode in the frequency region between 1 kHz and 1 Hz. The measured spectra were analyzed in terms of equivalent circuit models (Fig. 4) that the monolayers behave as a diffusion barrier for ions in the electrolyte. The effects of alkyl chain length on the electrochemical properties were also studied using a silane with a longer chain (octadecyltrimethoxysilane, ODTMS). The impedance measurements in the presence of redox couples yielded a defect area ratio of as low as 0.2% for the ODTMS monolayer (Fig. 5). 2.2. Electrochemical passivation of bulk GaAs electrodes with thiol monolayers Gallium arsenide (GaAs) has been claimed to realize high sensitivity due to its high electron mobility in nano-structures such as a two-dimensional electron gas (2DEG)
Living systems and semiconductor nano-structures
10
4
0. 0
2
-0.2
6
-0.4
8 6 4
-0.6
2
-0.8
8
-1.0
5
phase [rad ]
absolute impedance [Ohm]
10
195
6
10
4
-1.2
2
-1.4
4
-1.6 10
-1
10
0
1
2
10 10 frequency [Hz]
10
3
10
4
Fig. 3: Absolute impedance and phase shift of the ITO electrodes before () and after () the deposition of the OTMS monolayer in 10 mM Hepes buffer (pH = 7.5). The symbols represent the measured data, while the lines are corresponding to the fits using the equivalent circuit models from Fig. 4. The bare ITO surface could be analyzed by the equivalent circuit I (solid lines). Fitting of the spectrum after the silanization with the ideal equivalent circuit II shows clear deviations from the measured data (solid lines). However, by introducing a Randles circuit for the monolayer (equivalent circuit III in Fig. 4) the fit could be improved (dotted lines).
Equivalent circuits Hepes buffers Ro
Redox buffers
CIF
I.
CIF Ro RS
Ro
CIF
IV.
W Rct
II.
(α ) CS / CPE
CIF CIF
III.
W RPT
RS
Ro
CS Ro
ZW
V.
W Rct
ZW
CS
ZW
Fig. 4: Summary of the equivalent circuits used in this study. Equivalent elements: R0 electrolyte resistance; CIF semiconductor/electrolyte interface capacitance; RS SAM resistance; CS SAM capacitance; CPE constant phase element; RPT phase-transfer resistance between SAM and electrolyte; Z W Warburg impedance; Rct charge transfer resistance.
196
M. Tanaka 0. 0
6
10
8 6
-0.2
2
-0.4
5
10
8
-0.6
6 4
-0.8
phase [rad ]
absolute impedance [Ohm]
4
2 4
-1.0
10
8 6
-1.2
4
10
-1
10
0
10
1
2
10 frequency [Hz]
10
3
10
4
Fig. 5: Absolute impedance and phase shift of the ITO electrodes before () and after () the deposition of the ODTMS monolayer with 1 mM K3/4 Fe(CN)6 in 10 mM Hepes buffer (pH = 7.5). The measured impedance data were analyzed by the equivalent circuits IV and V. The deposition the ODTMS monolayer led to an increase in the charge transfer resistance from Rct0 = 110 cm2 to Rct = 50 k cm2 , yielding the local defect area ratio of 0.2%.
[20] and a quantum well (QW) [21]. However, the application of GaAs to living systems is still difficult because of the complex electrochemical processes at the GaAs/electrolyte interface. Indeed, most of the electrochemical studies of GaAs have been performed in acidic or basic solutions [22–24]. There have been several reports on the functionalization of GaAs surfaces with various types of sulfides and mercaptos in contact with air or with metals, and the passivation effects were mostly discussed in terms of photoluminescence (PL) from the bulk GaAs [25,26] or electrical properties such as the Schottky barrier height [27–29]. Nevertheless, systematic studies on the functionalization of GaAs surfaces under physiological conditions are still missing. We reported the coating of n-type GaAs electrode by deposition of octadecylthiol (ODT) monolayers, which showed high stability both in air and in aqueous electrolytes [30,31]. Prior to the surface coating, four different wet chemical etching procedures were attempted to optimize the surface pre-treatment. The chemical composition of the surface was evaluated by X-ray photoelectron spectroscopy (XPS), demonstrating that the photochemical etching procedure (named “etch P” in this study) can generate a surface enriched with arsenides, which can serve as the binding sites for sulfides (Fig. 6). At the next step, the surface prepared by etch P was coated with an ODT monolayer. The monolayer showed high stability in air, as monitored by the constant ellipsometric thickness (18∼20 Å). Cyclic voltammetry showed that both oxidation and reduction current at the interface were significantly suppressed by monolayer deposition
Living systems and semiconductor nano-structures
197
As 3d
Ga 3d
Etch P
Intensity [a.u.]
Etch C
Etch B
Etch A As-oxide Untreated
45
40
35
25
20
15
Binding Energy [eV] Fig. 6: Ga 3d and As 3d core-level spectra by XPS (survey scan) for the untreated surface (+), and for the surfaces prepared by four different etching procedures; etch A (), etch B ( ), etch C (◦), and etch P (). The symbols represent the measured data points, while the solid lines correspond to fitting. The arsenide-oxide peak (E B ∼ 45 eV, indicated by an arrow) had disappeared after all the etching procedures.
(Fig. 7). The electrochemical passivation of the monolayer-coated surface was verified by impedance spectroscopy under current minimum potential (U j =0 = −360 mV, determined by cyclic voltammetry) for more than 20 h (Figs. 8 and 9). Thickness of the monolayer can be estimated by the change in interface capacitance (18∼22 Å), which is in good agreement with the ellipsometric thickness measured in air. The stability of the interface was further monitored under different bias potentials. Electrochemical passivation of the GaAs surface has been achieved for the first time under physiological conditions, which allows the potential application of GaAs electrodes to biological systems. 2.3. Passivation of semiconductor nano-structures close to surface Towards novel hybridization of (bio-)organic molecular assemblies and semiconductor nano-structures, the surface of low-dimensional semiconductor was also chemically modified. As the first step, photoluminescence properties of the surface-near indium
198
M. Tanaka
Current [µA]
100
without ODT
50
0 with ODT -50 -1.0
0.0 -0.5 Potential [V]
0.5
Fig. 7: Cyclic voltammograms of n-GaAs before (dotted line) and after (solid line) the deposition of ODT monolayer in a PBS buffer (pH = 7.5). For both cases, the surface was prepared by Etch P. The current observed before the deposition was significantly suppressed.
Absolute Impedance [Οηµ]
arsenide quantum dots (InAs QDs) and other nano-structures grown on GaAs [100] substrates were investigated as a function of the distance to the surface [32] (Fig. 10a, b). The photoluminescence signal could be detected even in the very vicinity of the surface, when the distance to the surface was as close as d = 10 nm (Fig. 11). By transferring almost the same functionalization protocols, ODT monolayers were deposited on the surface of surface-near InAs QDs [33]. The monolayer deposition resulted in a significant enhancement in the photoluminescence from the QDs, which can be attributed to the effective suppression of the surface state densities by arsenide–sulfide coupling (Fig. 12). It is notable that the shown luminescence signals from chemically modified quantum dots were reproducible for more than 30 days in ambient atmosphere (Fig. 13), and exhibited only a slight decrease (∼ 10%) even after rapid thermal
C
107
Rs Rp
105
103 10-3
10-1
101
103
105
Frequency [Hz] Fig. 8: Impedance spectra of n-GaAs directly after Etch P (), the same sample after 22 hours (), and n-GaAs functionalized with ODT (•) are shown. The symbols represent the measured data, while the solid lines correspond to fits according to the equivalent circuit model composed of a serial resistance, Rs , a capacitance, C, and a parallel resistance, Rp . As the surface of freshly etched GaAs was not stable, the data of this surface were measured in a smaller frequency range from 50 mHz to 10 kHz. After the functionalization, the spectrum was totally stable for more than 24 hours in the frequency range from 1 mHz to 100 kHz.
Capacitance [µF cm-2]
Parallel Resistance [Ohm cm2]
Living systems and semiconductor nano-structures
199
5
106 5
(a)
105 5
0
4
8
12 16 20 Time [hours]
24
2.0
(b) 1.5
1.0 0.8 0
4
8
12 16 20 Time [hours]
24
Fig. 9: Changes in (a) parallel resistance, Rp , and (b) interface capacitance, C, versus time before (◦) and after (•) the monolayer deposition. Before the functionalization, Rp0 = 63 k cm2 continuously increased to Rp1 = 0.56 M cm2 after 24 hours, while C decreased from C0 = 2.0 µF cm−2 to C1 = 1.2 µF cm−2 . After the monolayer deposition, both Rp and C kept constant; Rp2 = 4.0 M cm2 and C2 = 0.81 µF cm−2 , respectively. The dotted lines are given to guide the eye.
annealing (10−3 mbar, 573 K). Furthermore, the enhancement in the luminescence signal was more significant when the QDs approached closer to the surface. In fact, the enhancement factor P (luminescence intensity normalized by that of as-grown samples) takes its maximum for the dots that are closest (10 nm) to the surface, P ∼ 1.9. Since the hydrophobic surface of the monolayer can be functionalized with polymer films and model cell membranes, this strategy is promising for the design of local detectors in the very proximity of the surface. 2.4. Outlooks – engineering of GaAs with functional 4,4 -mercaptobiphenyls One of the promising candidates for the further surface engineering of GaAs is a derivative of rigid 4 -substituted 4-mercaptobiphenyls, which provides various surface functions by flexible substitution. Furthermore, Gölzhäuser et al. demonstrated that this aromatic thiol monolayer could be cross-linked with low-energy electrons, which can be applied as negative resists for nanolithography. Recently, we demonstrated the highly
200
M. Tanaka
(a)
GaAs
(b)
5 nm
AlGaAs
5 - 25 nm
AlAs/GaAs
hvex
hv
Surface
30 nm
GaAs
CB Growth Direction
InAs
VB GaAs GaAs
AlGaAs
GaAs
E
z
Fig. 10: Schematic illustrations of (a) the layered nano-structure and (b) the band diagram of the InAs quantum dots (QDs) in the vicinity of the surface. The carriers are created in a GaAs layer between QDs and AlAs/GaAs superlattice structures. The QDs were grown at T = 803 K in Stranski–Krastanov growth mode, resulting in ∼ 2 × 1010 dots per cm2 . Wavelength [µm]
Wavelength [µm] 1.3
1.2
1.1
1.3
1.0
T = 4.2 K
(a)
1.1
QD/I30
100
QD/I20 50
QD/I10
T = 300 K
QD/I30
(b) Intensity [a.u.]
Intensity [a.u.]
1.2
75
150
QD/I20
50
25 QD/I10
0
0 1.0
1.1 Energy [eV]
1.2
0.9
1.0
1.1
1.2
Energy [eV]
Fig. 11: Photoluminescence spectra of QD/I10, 20, and 30 at T = 4.2 K (a) and 300 K (b), excited by a 50 W cm−2 Ar+ laser (513 nm). The continuous decrease in the intensity due to the non-radiative recombination via surface states and the red shift of the peak due to the local strain near the dots were observed. The photoluminescence signal could be detected when the distance to the surface was as close as d = 10 nm.
stable surface coating of GaAs [100] surface with a mercaptobiphenyl monolayer [34]. Homogeneity and hydrophobicity of the surfaces were characterized by atomic force microscopy (AFM) and contact angles to a water droplet. Electrochemical properties of the monolayer such as resistance and capacitance were quantitatively measured by cyclic voltammetry and impedance spectroscopy. The current–voltage scans showed that the monolayer deposition led to a remarkable suppression of electrochemistry at GaAs/electrolyte interface. The electrochemical stability of the monolayer-coated GaAs was carefully verified by impedance spectroscopy in the wide frequency region between
Living systems and semiconductor nano-structures
201
Wavelength [µm] 1.4
1.3
1.2
Intensity [a.u.]
monolayer-coated as-grown
0.9
1.0 Energy [eV]
Fig. 12: Photoluminescence spectra of the InAs QDs as-grown (broken lines) and after monolayer deposition (solid lines). Here, QDs are 10 nm to the surface. Noteworthy, the luminescence peak showed a spectral red shift of 15 meV after the monolayer coating, which can be attributed to the reduction in distance to the surface by wet chemical etching of native oxide.
100 kHz and 10 mHz. Such a strategy will allow systematic control of surface dipoles, surface charges, and surface free energy (i.e. wetting properties).
3. “Soft cushions” at interface – biocompatible polymer films as physical models of extracellular matrices (ECMs) Cell surface glycocalix and extracellular matrices (ECMs) in nature maintain high local disjoining pressures and generate hydrated “cushions” between cells and tissues by the combination of weak, generic forces at interfaces [35,36]. One of the possible strategies to mimic such interlayers is the deposition (e.g. grafting, casting, spin-coating and layerby-layer transfer) of thin (d < 100 nm) hydrated polymer films on solid surfaces. This strategy can be applied for proliferation and stress-free immobilization of cells, enzymes and receptors under non-denatured conditions, as well as for creation of model ECMs to control cell–cell and cell–tissue interactions by the chemical nature and morphology of polymers. Nature utilizes such soft interlayer at cell–cell and cell–tissue interfaces, and controls their metabolism and material transport. For example, at repulsive disjoining pressures, a cell can keep a certain distance from other cells and proteins via hydrated “cushions”. However, strong non-specific adsorption or “black-hole” formation can take place
202
M. Tanaka
Wavelength [µm] 1.3
1.2
Intensity [a.u.]
right after coating after 30 days
0.9
1.0
1.1
Energy [eV] Fig. 13: Photoluminescence spectrum of QD/I10 right after the coating (broken line) and that of the same sample stored under ambient conditions (in air, at room temperature) for one month (solid line). No changes in intensity or in the peak position were observed.
when the interfacial interaction becomes too attractive [37]. Indeed, recent studies have demonstrated that the phase separation induced by strong adhesion can even be interpreted as the first-order wetting/de-wetting transition [38]. Thus, the precise control of “wetting affinity”, i.e. interfacial forces at soft interface under water, is very important to fabricate thermodynamically and mechanically stable biological molecular composites [1]. To date, a sufficient biocompatibility that prevents from nonspecific adsorption of proteins and cells could be achieved by grafted films of dextran [39] and poly(ethylene glycol) (PEG) derivatives [40], which have been used in numerous fields. When the films of functionalized PEG derivatives are covalently end-grafted, the protein/cell resistance of the films can be interpreted in terms of phenomenological steric repulsion forces [41–44], and therefore, strongly depends on the chain length [45], grafting density [46,47], surface coupling groups [48–50], and morphology of chains. In this section, some of our recent studies on the “wetting” of biocompatible polymers with surface elastic fluids (e.g. lipid membranes) will be introduced. 3.1. Hydration and wetting of polysaccharide films Recently, we studied the static and dynamic swelling behaviors of the Langmuir– Blodgett (LB) films of regenerated cellulose [51], which is a major component of cell walls of plant cells. In the 1990s, Wegner and his colleagues have established chemical
Living systems and semiconductor nano-structures
203
OH O O
SiO OSi
n
Fig. 14: Chemical structure of trimethylsilylcellulose (TMSC), which has the degree of substitution D.S. = 2.0. Hydrophobic trimethylsilyl side chains can be cleaved by treatment with HCl vapor, resulting in regenerated cellulose. Thicknesses of monolayers are about 0.9 nm (before regeneration) and 0.5 nm (after), respectively.
modification of cellulose derivatives and studied the supra-molecular architectures of LB films [52–54]. We also demonstrated the biocompatibility of the regenerated cellulose by deposition of native and model cell membranes [18] (ref. Section 3). Several types of cellulose derivatives with rigid ’rod-like’ backbones and ’hairy’ hydrophobic side chains have been applied. Since the thickness of the monolayer is 4∼9 Å, the total film thickness can be controlled in nm accuracy simply by changing the number of deposited monolayers. Furthermore, intra- and inter-layer structures can be stabilized by cross-linking of the side chains or by hydrogen bonding. Here, appropriate numbers of trimethylsilylcellulose (TMSC) layers (10 and 20 layers, with thicknesses of about 10 and 20 nm, respectively) were transferred successively onto silicon wafers, that were hydrophobized with ODTMS monolayers (ref. Section 1.1). After the deposition, hydrophobic trimethylsilyl side groups can be regenerated, resulting in the original cellulose with thickness of around 5 (10 layers) and 10 nm (20 layers), respectively (Fig. 14). To investigate static and dynamic swelling behaviors of such ultra-thin polysaccharide films under controlled thermodynamic conditions, the film thickness was measured quantitatively by ellipsometry coupled to a climate chamber (Fig. 15a, b). Ellipsometry is a non-invasive, powerful technique with a high thickness resolution (±0.1 Å) to study soft, hydrated polymer films. The static swelling behavior was studied by measuring the equilibrium film thickness as a function of the relative humidity to obtain force–distance curves, i.e. disjoining pressure plotted as a function of film thickness. The dynamic swelling behavior was studied by monitoring the change of the film thickness as a function of time after an ’osmotic shock’ (i.e. abrupt change of the relative humidity) to obtain the kinetics of the hydration. Interestingly, both the maximum swelling ratio ρmax = dmax /d0 (∼ 1.7) and the normalized decay length λ∗ = λh /d0 (∼ 0.12) were independent from the number of layers, i.e. initial thickness of the films (Fig. 16a, b). This could be attributed to the layer-by-layer structure of LB films that is maintained even after the regeneration. The dynamic swelling experiments suggested that the characteristic time constant of the water uptake of the cellulose film could be much smaller than that of the atmosphere exchange (τ ∼ 30 s). Further improvement to trace faster kinetics of hygroscopic polymers is required. As physical models of heterogeneous (i.e. multi-functional) cellular surfaces pos-
204
M. Tanaka
film lift
(a)
CCD camera
(b) detector
humidity chamber
air in tube lens analyzer
air out air in analyzer
air out
microscope objective
sample sample
polarizer
polarizer He-Ne-laser (HBO lamp) He-Ne-laser
Fig. 15: Schematic views of (a) a point-like ellipsometer and (b) an imaging ellipsometer coupled to a climate chamber. The static swelling behavior can be obtained by measuring the equilibrium film thickness as a function of the relative humidity, while dynamic swelling behavior was monitored as changes in the film thickness as a function of time after an ’osmotic shock’ (i.e. abrupt change of the relative humidity). When coupled to a cooled CCD camera instead of a rotating analyzer, two ellipsometric parameters ∆ and Ψ can be obtained for each pixel on the image.
(a)
(b) disjoining pressure [Pa]
layer thickness [nm ]
6
18
10 layers cellulose 20 layers cellulose
16 14 12 10 8 6
π~ρ
2
-n
10 layers cellulose 20 layers cellulose
8
10
8 6 4
2
10
7 8 6 4
π = π0 exp (-ρ/λ0)
2
10
6
}
Repulsion regime
}
Hydration regime
8 6
0
20
40
60
80
relative humiditiy X [%]
100
1.0
1.2
1.4
ρ
1.6
1.8
swelling ratio
Fig. 16: (a) Swelling curve (i.e. equilibrium thickness plotted as a function of relative humidity) and (b) force–distance relationship (disjoining pressure versus swelling ratio) of a cellulose LB film (10 and 20 monolayers).
sessing self-assembled patterns of functional domains (e.g. rafts, clustered ligands), it is interesting to generate micro-patterns of polysaccharide films with different sizes and distributions by UV photolithography (Fig. 17). When coupled to a cooled CCD camera, two ellipsometric parameters ∆ and Ψ can be obtained locally by applying the rotating analyzer principle for each pixel on the image. Different from scanning probe microscopy such as atomic force microscopy (AFM), this technique allows the non-invasive study on structure and kinetics of swelling behavior of structured, soft interfaces. Mathe et al. reported “local” swelling of physisorbed films of dextran and
Living systems and semiconductor nano-structures
205
Fig. 17: Micro-patterned polysaccharide films as multi-functional cell and tissue surfaces. Native and model cell membranes (ref. Section 4) can selectively immobilized due to the wetting affinity at the interface.
(a) Delta Image
(b) Psi Image
20 µm
Fig. 18: Ellipsometric images of a micro-structured cellulose LB film (20 monolayers, d = 10 nm. In the presented (a) ∆ and (b) Ψ maps, one can gain ellipsometric parameters within a lateral resolution of 1 µm and a vertical resolution of 1 nm.
hyaluronic acid by imaging ellipsometry, where not only the “swelling” in the vertical direction but also the “melting” of the edges of the micro-patterns could be observed [55]. As shown in Fig. 18, we could successfully resolve the ellipsometric images (Ψ and ∆ maps) of structured cellulose films, whose initial dry thickness was about 10 nm (i.e. 20 layers of cellulose). 3.2. Chemical control of wetting interaction – grafting of synthetic polymer brushes In contrast to the commonly used PEG brushes, poly(2-alkyl-2-oxazoline)s, synthesized by living cationic polymerization, provide the possibility to tailor macromolecules by the choice of different side substitutions and terminal groups [56,57]. Since the length of the alkyl chain in the 2-position determines the hydrophilicity of each monomer
206
M. Tanaka
H3 CO R1
OCH3 Si
N
N nH
R2
OCH3
R1- = CH3R2- = CH3-, C2H5-
O
n = 15, 30 Fig. 19: Structures of silane-functionalized poly(2-alkyl-2-oxazoline)s. PMOX15, poly(2-methyloxazoline) with n = 15; PEOX15, poly(ethyloxazoline) with n = 15; and PEOX30, poly(2-ethyloxazoline) with n = 30.
unit [58], chemical control of the hydration is possible. In fact, biocompatibility of poly(2-oxazoline) derivatives was investigated, where lipid vesicles with poly(2oxazoline) lipopolymers showed remarkably larger blood circulation times than ordinary phospholipid vesicles [59], that are even comparable to conventional PEG lipopolymers. In our recent study [60], we grafted 2-methyl- and 2-ethyl-2-oxazoline brushes with trimethoxy coupling groups (n = 15 and 30, PDI = 1.09–1.24) onto silicon wafers, and studied their static and kinetic hydration by ellipsometry (Fig. 19). The maximum swelling ratios ρmax of the shorter polymer chains are comparable (ρmax = 2.7–3.5) in spite of their differences in the side chains, while ρmax of the PEOX30 film, ρmax(PEOX30) = 1.6–1.8, was obviously smaller than the corresponding values obtained for the films of PMOX15 and PEOX15 (Fig. 20). At high disjoining
polymer thickness [nm]
5.5
1.52
5.0 1.50 4.5 1.48 4.0
thickness [nm] refractive index
3.5
1.46 1.44
3.0 1.42 2.5 1.40
2.0 0
20
40
60
80
100
relative humiditiy [%] Fig 20: Equilibrium thickness (left ordinate) and refractive indices derived from the Garnett equation (right ordinate) of a grafted PMOX15 film plotted versus relative humidity of the atmosphere. Significant changes in the film thickness were observed under relative humidity conditions > 90%.
Living systems and semiconductor nano-structures
207
Table 1 Physical parameters of the grafted poly(2-oxazoline) films measured by ellipsometry. PMOX15, poly(2methyloxazoline) with n = 15; PEOX15, poly(ethyloxazoline) with n = 15; and PEOX30, poly(2-ethyloxazoline) with n = 30. Initial dry thickness d0 was determined at a relative humidity of ∼ 4%, by assuming the refractive index n 0 = 1.52. ρmax is the maximum swelling ratio, n 1 and n 2 stand for exponents for the power-law potential P(ρ) ∼ ρ −n calculated for two high disjoining pressure regimes: (i) 5 × 108 ∼2 × 108 Pa and (ii) 2 × 108 ∼1 × 107 Pa, respectively. Normalized decay constant λ∗ was calculated in the lower disjoining pressure regime, P < 1 × 107 Pa. Polymer
d0 [Å]
ρmax
n1
n2
λ∗
PMOX15 PEOX15 PMOX30
18.2–21.7 12.4–20.8 24.4–39.7
2.7–3.4 2.8–3.5 1.6–1.8
18–24 12–34 32–33
6.6–8.5 6.5–8.4 12
0.49–0.56 0.48–0.63 0.16–0.18
pressure conditions (P > 1 × 107 Pa), we observed two regimes where strong repulsive forces obey a power-law, P(ρ) ∼ ρ −n : the first regime with large exponents (n 1 = 34– 12) can be almost explained by a theta function, while the exponents in the second regime were apparently smaller. On the other hand, the third regime, P < 1 × 107 Pa, is dominated by hydration forces, which can be described by Ph = P0 exp(−ρ/λ∗ ). Hydration of polymers with different initial thickness was compared by the normalized swelling ratio ρ. The decay constant λ∗ also exhibited a clear dependence on the chain length: λ∗ = 0.5–0.6 for PMOX15 and PEOX15, and λ∗ = 0.2 for PEOX30, respectively. These observations suggested that the static swelling behaviors of the poly(2-oxazoline) films are not dependent on the side chains but on the chain length (Table 1). Semi-quantitative analysis of the dynamic swelling behavior of the poly(2-oxazoline) brushes suggested that the hydration kinetics also seemed to be dependent on the chain length, but is not affected by the side chains. The characteristic time constant for PEOX30 films (τPEOX30 = 200 s) is larger than those for the shorter PEOX15 and PMOX15 (τ = 140–150 s) (Fig. 21). Since poly(2-oxazoline) used in this study can be terminally functionalized with alkyl chains, this system is also a promising candidate for the fabrication of “tethered” lipid bilayers with polymer spacers. The chemical and morphological modification (e.g. control of hydrophilic/hydrophobic balance, grafting density, and chain length) allows a precise control of interfacial properties of the resulting brushes. Recently, we could accomplish a homogeneous and fluid lipid bilayer with short (n = 10) PEOX spacers (Purrucker et al., submitted) (Fig. 22). 3.3. Outlooks – chemical “switching” of cell/surface interactions Recently, we proposed a unique “chemical switching” of the protein–surface interaction based on weak cationic polymer brushes. The monolayer of diblock copolymers (poly(2-(dimethylamino)ethyl methacrylate-block-methyl methacrylate), synthesized by the group of S.P. Armes, Univ. Sussex), whose hydrophilic part can be charged/decharged by pH titration under physiological conditions (pH = 5∼9) [61], is deposited from the air/water interface onto the hydrophobized silicon wafers by the Langmuir–
208
M. Tanaka
dry
thickness [arb. units]
humid
0
400
800
1200
1600
time[s] Fig. 21: Dynamic swelling curves of a PEOX15 film under several cycles of ‘osmotic shocks’. The film thickness (in arbitrary units) is plotted as a function of time. A clear increase in the “dry” film thickness (d0 = 2–3 Å) was observed when the humidity condition was switched back to 4%, which could be recovered by drying the sample at 70°C for 3 h.
Fig 22: Tethered lipid bilayers with poly(oxazoline) spacers. Thickness of the water reservoir and the fluidity of the membranes can be controlled by grafting density, side chain functionalities, and mixing behavior with matrix lipids.
Schaefer method. Preliminarily, we measured the pressure–area isotherms of the diblock monolayers at the air/water interface to confirm the stability of the films and their thermodynamic properties. Despite the “glassy” nature of MMA backbones, the monolayer showed no compression/relaxation hysteresis or material loss until it collapsed at 50 mN/m, when the barrier was driven at 20 µm s−1 . The application of reflectivity techniques, such as neutron reflection in D2 O, will enable us to observe the conformational changes of weak polyelectrolyte brushes and interaction with water soluble proteins (lectins, fibronectin, etc.) under pH titration.
Living systems and semiconductor nano-structures
209
4. Physical models of cell surface glycocalix Cell surface glycocalix is made of oligo- and poly-saccharide chains of glycolipids, glycoproteins, and of trans-membrane proteoglycans [35]. For example, glycolipids in animal cells are localized in the outer leaflet of the lipid bilayer and stabilize the structure of plasma membranes by a combination of various physical forces (e.g. electrostatic force, long-range van der Waals interaction, hydrogen bonding). Such intra- and intermolecular binding enables the carbohydrate to take a distinct conformation and to interact with counter-part lectins and cell adhesion receptors [35,62]. Examples are blood group- and tumor-associated antigens such as sialyl-LewisX and sialyl-Lewis-A that interact specifically with selectins [63–65]. Interestingly, besides carbohydrate–protein interactions, it has been demonstrated that cell surface carbohydrates can selectively bind to complimentary carbohydrates of another cell [66]. Technical advances during the past several decades (e.g. oligosaccharide synthesis, purification, and analysis) have led to an interdisciplinary approach in the field of glycoscience. However, the physical basis of glycocalix functions has not yet been understood experimentally. As physical models of glycocalix, we have investigated the interfacial properties of synthetic glycolipids, such as phase behavior, polymorphism, and rheological properties. 4.1. Thermodynamics and hydration of glycolipid monolayers Thermodynamic properties and swelling behavior of lipids covalently attached to lactose oligomers (named Lac N , the number of lactose units N = 1, 2, 3) were studied [67] (Fig. 23). Since each lactose unit takes a linear, cylindrical conformation, they are expected to be rather simple glycocalix models. In this study, Langmuir pressure–area isotherms were first measured at several temperatures to estimate thermodynamic and structural parameters of the monolayers at the air/water interface. The molar transition entropy and the molar latent heat were calculated by applying the Clausius–Clapeyron equation. It has been demonstrated that the phase behavior of the glycolipid monolayers is comparable to that of ordinary phospholipids [68], in spite of the lower degree of cooperativity between the larger head groups (Fig. 24a–c). These results suggest entropic effects of the head groups on the interaction between the neighboring molecules. At the
OH
OH
OH O
HO
O O
OH
OC16H33
O
HO OH
N
OC16H33
N=1-3 Fig. 23: Structures of the glycolipids studied, Lac N (N = 1, 2, and 3).
210
M. Tanaka 35
a
π [mN/m]
30
303 K 298 K 293 K 288 K
25 20 15 10 5 0 40
60
80
100
120
2
Area [ Å ] 35
35
π [mN/m]
303 K 298 K 293 K 288 K
25 20 15
c
30
π [mN/m]
b
30
303 K 298 K 293 K 288 K
25 20 15
10
10
5
5 0
0 40
60
80
100 2
Area [ Å ]
120
40
60
80
100
120
2
Area [Å ]
Fig. 24: Langmuir isotherms of the monolayers of: (a) Lac 1, (b) Lac 2, and (c) Lac 3 at various subphase temperatures. The phase coexistence line was fitted by a polynomial of 4th order (broken line). At high temperature and pressure conditions, the coexistence line can be approximated as parabola, whose vertex coincides with the critical point. The points of intersection of the linearly extrapolated lines in (a) were taken to define ALE and ALC . To estimate the deviation in A, several points were taken.
next step, these glycolipid monolayers were transferred onto hydrophilic silicon wafers. Using the same manner as presented in Section 2, the equilibrium thickness of the saccharide layer was measured as a function of relative humidity by ellipsometry. Here, since the transfer ratio was 1.0 for all the samples studied, the surface grafting density (i.e. area per molecule) was precisely controlled by the surface pressure at which the monolayer was transferred. The obtained swelling curve of Lac 1 was similar to those of other phospholipids, reflecting its “rod-like” lactose head groups. On the contrary, the force–distance relationships of Lac 2 and Lac 3 fitted well to conventional polymer “brush” theories [69–72], even though the statistical limit N 1 is hardly fulfilled. Such unique properties of these glycolipids, like (i) “phospholipid-like” phase behaviors and (ii) “polymer-like” swelling behaviors, might explain the distinct conformation of oligosaccharides on cellular surfaces, which can be recognized by proteins and carbohydrates selectively.
Cp [cal/mol ˚C]
Living systems and semiconductor nano-structures
12000
211
(a)
8000 4000 0 20
40
60
80
T [ ˚C]
(b)
SAXS
WAXS
Log (I) [arb. units]
4.57 Å
80 ˚C
80 ˚C
4.45 Å 60 ˚C
3.76 Å
20 ˚C 40 ˚C
7.50 Å
20 ˚C 0.1
0.2
0.3
0.4
0.5
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
q [Å-1] Fig. 25: (a) Differential heat capacity scan of the Lac 1 dispersion (1 mg/mL) recorded at a heating rate of 20°C/h, exhibiting a sharp transition at Tt = 74°C and a phase transition enthalpy of H = 30 kcal/mol. (b) Powder-averaged small-angle X-ray scattering (SAXS) data of the lamellar dispersion of Lac 1 at T = 20, 40, 60, and 80°C (left). The lamellar spacing showed a transition between 60°C (dSAXS = 68 Å) and 80°C (dSAXS = 60 Å). Wide-angle X-ray scattering (WAXS) data suggested a transition between the crystalline LC phase and the fluid Lα phase (right).
4.2. Morphology of glycolipids – structural basis of carbohydrate complexes To obtain structural information of generic carbohydrates complexes, polymorphism of Lac N lipids (N = 1, 2, 3) was studied in lamella phase [73]. By a combination of differential scanning calorimetry (DSC) and small- and wide-angle X-ray scattering experiments (SAXS and WAXS), the effect of hydrophilic/hydrophobic balance on their thermotropic phase behaviors was systematically studied. The dispersion of Lac 1 exhibited the crystalline–fluid phase transition (Fig. 25a, b). Both the thermotropic phase
M. Tanaka
Cp [cal/mol ˚C]
212
6000
(a)
4000 2000 0 20
40
60
80
T [˚C]
(b)
SAXS
4.45 Å
Log (I) [arb. units]
80 ˚C
WAXS
80 ˚C 60 ˚C
40 ˚C
20 ˚C
4.17 Å
20 ˚C 0.1
0.2
0.3
0.4
0.5
0.8
0.6
q
1.0
1.2
1.4
1.6
1.8
2.0
[Å-1]
Fig. 26: (a) Heat capacity trace of the Lac 2 dispersion (1 mg/mL), where the phase transition enthalpy was also clearly reduced to H = 9.2 kcal/mol. (b) SAXS diffraction patterns of the lamellar dispersion of Lac 2 at T = 20, 40, 60, and 80°C (left). The lamellar spacing showed a transition between 40°C (dSAXS = 87 Å) and 60°C (dSAXS = 78 Å). WAXS peaks suggested a transition between the gel phase and the fluid phase (right).
transition temperature (Tt = 74°C) and the transition enthalpy (H = 30 kcal/mol) of the Lac 1 dispersion are higher in comparison to those of other synthetic lipids with dihexadecyl chains, Tm = 40∼50°C and H = 8∼9 kcal/mol. Here, the alkyl chains are strongly correlated by van der Waals interaction, which even enabled them to form crystalline-like tight packing with almost no tilting. The additional enthalpic contribution can be due to the hydrogen bonding between the Lac 1 head groups that are free from dipoles, in contrast to phospholipids with P–N dipoles. In the case of Lac 2, the hydrophilic/hydrophobic balance between the head group and the alkyl chains is shifted to the hydrophilic side. This shift in the balance reduces the cooperativity between the alkyl chains, resulting in a decrease in the transition temperature and the phase transition enthalpy (Fig. 26a, b). The strongly crystallized alkyl chain packing was
Cp [cal/mol ˚C]
Living systems and semiconductor nano-structures
4000
213
(a)
2000 0 -2000 20
40
60
80
T [˚C]
(b)
4.46 Å 4.19 Å 7.61 Å
Log (I) [arb. units]
SAXS
WAXS
80 ˚C 80 ˚C 60 ˚C
20 ˚C
40 ˚C 20 ˚C 0
0.1
0.2
0.3
0.4
0.5
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
q [Å-1] Fig. 27: (a) DSC trace of the Lac 3 dispersion (1 mg/mL), showing no evidential endothermic peaks. (b) SAXS diffraction patterns of the Lac 3 lamellar dispersion at T = 20, 40, 60, and 80°C (left). The lamellar spacing showed no transition at all measurement conditions, dSAXS = 108 Å. The WAXS peaks suggested that the Lac 3 lamellar takes crystalline-like phase and no chain melting takes place (right).
modulated to the gel phase, which allows the hydration of the head groups. Different from the first two systems, the DSC trace of Lac 3 showed much less remarkable peaks. The WAXS data did not reveal any transition in the chain ordering, exhibiting two sharp scattering peaks due to the slightly tilted alkyl chains and one sharp correlation peak due to the head groups. The small-angle scattering also demonstrated the highly ordered 3D lamellar structure with more than ten sharp diffraction peaks with a constant distance, 108 Å (Fig. 27a, b). In this case, the very strong correlation between the hexasaccharide head groups forced the alkyl chain to take the distorted, crystalline-like packing, which is obviously different from the ideal hexagonal lattice. Since the attractive interaction between the head groups is so strong, the hydration does not take place any longer. To gain more insight in the structural characteristics of this phase, further conformational analysis by spectroscopic techniques such as FTIR and the systematic link to the morphology in two dimensions, i.e. morphology of monolayers, is further progressed.
214
M. Tanaka
4.3. Viscoelasticity of oligosaccharides on cell membranes – rheology at the interface One of the experimental approaches to study the complex interplay of various forces (e.g. electrostatic interaction, van der Waals force, and hydrogen bonding) operating on cellular surfaces is the measurement of the rheological (viscoelastic) properties of glycocalix. Several studies have been conducted on the morphology and elasticity of bacteria surfaces using atomic force microscopy (AFM) [74–76]. However, extremely toxic fixatives (e.g. glutaraldehyde, osmium tetroxide) or strong mechanical stresses often used in the traditional fixation procedures can damage the cell irreversibly. Although there are some studies on the elasticity of the isolated sheath [77] or sacculi [78], there has been no quantitative study on rheology of artificial glycocalix models, such as the monolayers of synthetic glycolipids at the air/water interface. The rheological study of insoluble monolayers (Langmuir monolayers) at the air/water interface can provide a quantitative measure of hydrogen bonding between neighboring molecules, physical entanglement, and cross-linking under dynamic conditions. In comparison to other conventional devices [79], a new class of rheometers, referred to as interfacial stress rheometer (ISR) [80], developed by G. Fuller’s group, allows for highly sensitive and real-time measurements of viscoelastic parameters at different frequencies under controlled thermodynamic conditions (surface pressure, temperature) (Fig. 28). Using this device, the rheology of Lac N monolayers at the air/water interface was investigated [81]. The viscous and elastic surface moduli of the monolayers were measured as a function of the length of the linear oligosaccharide head groups quantitatively. The Lac 1 monolayer was highly viscoelastic, which can be attributed to strong chain–chain correlations. The introduction of another lactose unit further reduced the chain–chain correlation, and the head groups are more hydrated (Fig. 29a). This results in a fluid nature of the Lac 2 monolayer. In contrast, a clear rheological transition of Lac 3 monolayers from a viscous to an elastic film could be observed for the first time at a surface pressure of 6∼8 mN/m, suggesting the formation of a cross-linked physical gel (Fig. 29b). On the other hand, if one considers the short (the stretched length of Lac 3 head group is still less than 4 nm) and cylindrical head group of Lac 3, this transition is obviously not caused by a physical entanglement of the oligosaccharide head groups. Interestingly, the surface pressure at which the rheological transition of the Lac 3 monolayer takes place (π = 6∼8 mN/m) corresponds to the end point of the coexistence of the liquid expanded and the liquid condensed phase. Above this transition, the hydrating water is excluded and hydrogen bonding “bridges” the Lac 3 head groups during lateral compression to higher pressures (Fig. 30). This is in contrast to the monolayers of Lac 1 and Lac 2, where the phase transition to the liquid condensed phase results in a significant increase in the film viscosity. Thus, it can be concluded that the rheological transition of the Lac 3 monolayer is not caused by the correlation between the condensed alkyl chains, but by the strong coupling between the linear hexasaccharide (Lac 3) head groups, which might mimic a function of glycocalix to introduce elasticity to the plasma membrane. Indeed, the obtained results are in good agreement with our recent X-ray scattering experiments on Lac N lipid dispersions (ref. Section 3.2). Along these lines, we are studying the effects of subphase conditions (e.g. pH conditions and ionic strength) on the film viscoelasticity. Moreover, the introduction
Living systems and semiconductor nano-structures Helmholtz Coils
B
215 Light Source
Pressure Sensor
Glass Wall Barrier
Magnetic Rod
Detector (Photodiode)
Glass Wall
L ~ 30 mm W ~ 20 mm
Fig. 28: Schematic overview (a) and close-up (b) of the interfacial stress rheometer (ISR).
of functional head groups (such as Lewis-X and sialyl-Lewis-X) [82] will enable us to investigate the effects of carbohydrate–carbohydrate and carbohydrate–protein interaction on the membrane rheology, proposing a realistic physical model of cell surface glycocalix (Fig. 31).
5. Native and model cell membranes on semiconductors – potential candidates for sensorics In nature, the plasma membrane defines the periphery of each cell and separates its cytoplasmic contents and extracellular spaces. The fundamental components of plasma membranes are lipid bilayers, which support the hierarchical structures of various membrane proteins. Artificial lipid monolayers and bilayers on solid supports are treated as plasma membrane models [83,84], which can be deposited onto solid surfaces by one of the following methods: (i) monolayer transfer, (ii) vesicle fusion, (iii) single bilayer spreading, or (iv) solvent exchange method (Fig. 32). The first method includes transfer of the lipid monolayers from air/water interface by the Langmuir–Blodgett
C
G', G'' [mN/m]
1.2
G' G''
1.0 0.8
30
C
1.4
0 60
80
100
120
Area per Molecule [Å2]
C B
0.4
G' G''
1.2
A
10
40
0.6
(a)
B
20
G', G'' [mN/m]
1.4
Lateral Pressure [mN/m]
M. Tanaka Lateral Pressure [mN/m]
216
1.0 C 0.8
(b)
20
B
10
A
0 40
0.6
60
80
100
Area per Molecule [Å2]
120
C 0.4
0.2
C
30
gelation
B
0.2 A
B 0.0
A
0.0 38
39
40
41
42
38
43
40
Area per Molecule [Å2]
42
44
46
48
50
52
54
56
58
Area per Molecule [Å2]
Fig. 29: (a) Dynamic moduli of a “fluid” Lac 2 monolayer, measured at T = 20°C, ω = 1 rad/s. The monolayer remained viscous fluid (G < G ) through the measurements. (b) Rheological transition of a Lac 3 monolayer (at T = 20°C, ω = 1 rad/s). At a surface area of about 0.5 nm2 (corresponding to a surface pressure of 6 to 8 mN/m), the monolayer became elastic (G > G ).
π = 20 mN/m
0.16
Lac2
G'' [mN/m]
0.14
Lac3 0.12 0.10 0.08 0.06 0.04 0.2
0.3
0.4
0.5
0.6
0.7
0.8
ω [rad/s]
Fig. 30: The loss modulus G of Lac 2 ( ) and that of Lac 3 () at π = 25 mN/m, plotted as a function of oscillation frequency.
and Langmuir–Schaefer methods. This is a laborious procedure, but enables one to deposit asymmetric bilayers [85]. In the second method, lipid vesicles are spread onto the substrate from vesicle suspensions [86]. By adjusting the adhesion forces to the substrate (e.g. electrostatic attraction, wetting conditions), vesicles rupture on the surface and form adherent bilayer patches, which fuse with each other to form continuous bilayers. The third method, single bilayer spreading, can be achieved by pasting a lipid reservoir onto substrates, followed by the swelling under water. A single bilayer is spontaneously pulled over the surface by adhesion forces [87]. The fourth method results in the formation of lipid membranes simply by exchange of solvent from alcohols (e.g. ethanol, isopropanol) to aqueous buffers [88]. According to the latter three methods, membranes are formed according to the self-assembling process to heal their local defects and pores.
Living systems and semiconductor nano-structures
(a)
OH HO
OH O
Me OH
OH
HO
O O
O OH
OH
(b)
O
AcNH
NHAc
O
O
O OH
CO2Ð
OH
OH O
HO
O
OH
O(CH2)15CH3
O
O OH
OHOH
O(CH2)15CH3
OH HO
OH O
HO
217
Me OH
OH
O O
HO
NHAc
O
O
O OH
OH
OH O
HO
O O
O
OH
OH
OHOH
O(CH2)15CH3 O(CH2)15CH3
Fig. 31: Other functional glycolipids with (a) sialyl Lewis-X and (b) Lewis-X head groups with lactose spacers. Synthetic modification of head groups and hydrophilic/hydrophobic balance even enables to form functional domains (artificial rafts) at the interface.
(i) Step 1
(ii) Step 1
Step 2
Langmuir-Blodgett (vertical lifting)
Langmuir-Schääaefer (holizontal dipping)
Step 2
adsorption, rupture
spreading, fusion
(iv)
(iii) Step 1
Step 2 dry lipid reservoir
Step 1
alcohol
Step 2
aqueous buffer
swellng, spreading
Fig. 32: Deposition of model cell membranes (lipid mono- and bilayers) on solid supports: (i) monolayer transfer, (ii) vesicle fusion, (iii) single bilayer spreading, or (iv) solvent exchange method.
To introduce biological functions to the supported lipid membranes, it is necessary to incorporate membrane proteins in non-denatured state, maintaining their native orientation and function (Fig. 33). One of the less-laborious methods is the direct insertion of proteins (peptides) from the solution into the pre-formed membranes. This is suitable especially for the incorporation of small antibiotic peptides [89], but it requires
218
M. Tanaka
(a)
(b)
Fig. 33: Introduction of bio-functionalities to model membranes: (a) incorporation of transmembrane proteins/peptides, and (b) docking of proteins onto a membrane surface.
optimal solvent and concentration not to disturb their native structure and functions [90–92]. Another way to reconstitute transmembrane proteins is to incorporate them into artificial lipid vesicles (proteoliposomes) and to deposit them onto substrates [93– 98]. To “dock” proteins without hydrophobic cores onto a membrane surface, several natural and synthetic anchors can be used: biotin tethers for avidin/streptavidin [99], glycan–phosphatidyl inositol (GPI) linkages [100], and nitrotriacetic (NTA) anchors against recombinant proteins with histidine tags [101,102]. Although it is challenging and practically difficult, the direct deposition of native cell membranes on solid surfaces [103] includes numerous advantages to maintain the original orientations and populations of the transmembrane proteins. In the following sub-sections, several experimental approaches towards the biomembrane–semiconductor hybrids are presented. 5.1. Uptake of antibiotic peptides into model membranes on semiconductor electrodes Previously, we deposited artificial lipid bilayers onto ITO electrodes [18] which were functionally modified with alkylsiloxane monolayers (ref. Section 1.1) and LB films of regenerated cellulose (d = 5 nm, ref. Section 2.1). ITO is transparent and allows for in situ investigations both with optical microscopy and electrical techniques. Since the regenerated cellulose films are hydrated and behave like an electrolyte, this system can be characterized as an electrolyte/membrane/electrolyte/semiconductor structure (EMES) (Fig. 34). Lipid membranes have been deposited by vesicle fusion, and the homogeneity and fluidity of the resulting membranes were confirmed by fluorescence microscopy and fluorescence recovery after photobleaching (FRAP). Reflection interference contrast microscopy (RICM) can be used to monitor the spreading of a lipid bilayer on a polymer cushion, exhibiting the self-healing of bilayers due to lateral diffusion (Fig. 35). In fact, deposition of the lipid bilayer resulted in a significant change in interface resistance to Rm = 0.44 M cm2 , which is slightly smaller than that of solvent-free black lipid membranes. Noteworthy, the electrical properties achieved here remained stable for more than a week, suggesting thermodynamic and mechanical stability of the membranes. Compared to the work of Gritsch et al. [17], where the same lipid bilayer was directly deposited onto ITO electrodes, the membrane resistance is
Living systems and semiconductor nano-structures
219
Electrolyte Membrane Electrolyte Semiconductor
EMES Structure Fig. 34: Polymer-supported lipid bilayer deposited onto an ITO electrode. As the hydrated cellulose films behave like an electrolyte, this system can be characterized as an electrolyte/membrane/electolyte/ semiconductor (EMES) structure.
Fig. 35: The self-healing of a large hole in a single bilayer by sliding over a cellulose LB film, observed by reflection interference contrast microscopy (RICM). The time difference between the four images was 2 min.
increased by nearly a factor of 5. The high resistance and stability achieved here can be attributed to the reduction of surface roughness by cellulose “cushions” as well as to the wetting affinity between cellulose and lipid bilayers. As a preliminary check of the membrane quality for the protein incorporation, an antibiotic peptide, gramicidin D, was incorporated into membranes by incubating small vesicles with gramicidin D for 6∼10 hours. Although function and ion selectivity of the incorporated channels were reasonable from their native functions, the amount of the incorporated channels was
220
M. Tanaka reference
counter
flow cell
Fig. 36: Schematic drawing of the system studied. A lipid bilayer was deposited on Si/SiO2 electrodes coupled to a flow cell by vesicle fusion, and a trifluoroethanol solution of antibiotic peptide gramicidin D was injected. The formation of a continuous membrane, insertion kinetics of peptides, and their functionalities can be monitored by impedance spectroscopy.
certainly small. By assuming a single ion channel conductance of 0.7 pS, the surface concentration of active gramicidin dimers was estimated as 3.5 × 107 cm−2 . We recently reported the deposition of artificial lipid bilayers directly onto Si/SiO2 electrodes [104]. In spite of numerous advantages due to their mechanical and chemical stability, flexibility for the manufacturing of a variety of device structures [105], there have been not so many applications of Si/SiO2 in biological systems, except for living cells on transistors [106,107]. It is still difficult to assign the electrical properties of membranes separately from the background signals [108,109], especially when the capacitance of the oxide layer is competitive or even less than that of the lipid membranes, 0.7 µF cm−2 [110]. Fromherz et al. had overcome this problem by touching a giant vesicle to the open gate of a field-effect transistor [111]. The lipid bilayer kept its continuous shape without rupturing, and adhered onto the cationic poly-lysine film due to electrostatic attraction. We took a different strategy to deposit lipid membranes onto highly doped ptype Si/SiO2 electrodes. First, important physical parameters of the substrates such as the doping ratio and thickness of the oxide layer were carefully optimized by a combined study using ellipsometry and impedance spectroscopy. A lipid bilayer was then deposited by fusion of small unilamellar vesicles on the electrode (Fig. 36). Self-assembling of the bilayer patches and the continuous growth of a homogeneous membrane could be monitored by measuring electrical impedance over a wide frequency range (from 100 kHz to 1 mHz). Despite of a relatively larger active electrode area (0.50 cm2 ), the resistance of the membrane can be compared to that of the solvent-free, black lipid membrane, 1 M cm2 . To increase the incorporation amount and to accelerate the uptake, gramicidin D was incorporated into this membrane from a trifluoroethanol solution. By measurement of membrane resistance as a function of time, we could monitor the kinetics of peptide uptake (Fig. 37). Furthermore, the membrane with gramicidin showed certainly high conductivity and reasonable selectivity to monovalent cations (Na+ , K+ ) against anions (Cl− ) (Fig. 38). Such strategy can be applied in fast screening of the uptake of toxins and other antibiotic peptides as well as in the in situ investigation of pore formation in bio-membranes.
Living systems and semiconductor nano-structures
221
1000
Membrane Resistance [ kΩcm 2 ]
900
injection
800 700 600 500 400 300 200 100 0 0
20
40
60
80
Time after Injection [min ] Fig. 37: Insertion of gramicidin D, monitored as a change in membrane resistance as a function of time. The insertion resulted in a rapid decrease of the resistance from 0.98 M cm2 , reaching its equilibrium (80 k cm2 ) in 45 min.
5.2. Novel charge sensor based on bio-membrane/semiconductor hybrids Recently, we reported the design of a novel membrane charge sensor by deposition of highly insulating polymer/lipid composite films on indium-tin-oxide (ITO) semiconductor electrodes [112] (Fig. 39). The lipid monolayers were deposited on LB multilayers of cellulose derivatives (isopentylcellulosecinnamate, IPCC) [113] by continuous exchange of solvent. Prior to membrane deposition, the intra- and inter-membrane structures were stabilized by cross-linking cinnamoyl side chains under UV illumination. The resulting polymer/lipid composite system showed an electric resistance of 2.5 M cm2 and a lateral diffusion constant for the lipids of 0.1 µm2 /s, which were obtained by impedance spectroscopy and FRAP, respectively. Such highly insulating and fluid composite films on semiconductors can be utilized as a membrane charge sensor to detect changes in surface charge by treating this electrolyte/(organic) insulator/semiconductor (EIS) system as an analogue of the metal/oxide/semiconductor (MOS) system. For this purpose, we incorporated 10 mol% of lipids with chelator headgroups in order to switch the membrane charge reversibly. In the presence of Ni2+ -loaded and EDTA-loaded (EDTA: ethylenediaminetetraacetic acid) buffers, significant changes in the impedance spectra could be observed (Fig. 40). Noticeably, the “switching” of electrical properties between “uncharged (Ni2+ -loaded)” and “charged (unloaded)” states was reversible and reproducible for more than 2 weeks, showing no degradation. Semi-quantitative Mott–Schottky analysis (due to a high electron density (1021 cm−3 ) of our ITO, it cannot be quantitatively treated as a “standard” semiconductor, but a “degenerated” semiconductor) demonstrated that the difference in surface charge density of Q ∼ 10−6 C/cm2 changed the flat band potential of the EIS system by nearly 50%. From our measurement accuracy to
222
M. Tanaka
7
Absolute Impedance |Z| [Ω]
10
bare electrode membrane with gramicidin, Cl with gramicidin, Na + with gramicidin, K +
6
10
5
10
4
10
3
10
-2
10
10
-1
0
10
1
10
2
10
3
4
10
10
Frequency [Hz] Fig. 38: Impedance spectra of the supported lipid bilayer before (◦) and after incorporation of gramicidin D at Ubias = 0 mV. The ion selectivities of the incorporated channels were reversible and reproducible for more than a week in the presence of Na+ , K+ , and Cl− ions.
-
Ni-loaded
Unloaded O
O O
Ni O
O
O
-
O
O
N
R
∆q = 1e
R
-
H N+
O
O
E
-
-
- - - - - -
-
+
+
O O-
- - +
Ni2+
I
EDTA
S Uncharged
Charged
Fig. 39: Principle of membrane charge sensor based on biological electrolyte/(organic) insulator/semiconductor (EIS) system, as an analogue of the metal/oxide/semiconductor (MOS) system. In this study, changes in surface charge density due to loading/de-loading of NTA lipids can be sensitively detected as changes in capacitance of space charge region or flat-band potential of semiconductors. Here, reversible complexation of nickel ions changes the molecular net charge of the NTA-lipid by 1 e− .
Absolute impedance |Z| in [Ohm ]
Living systems and semiconductor nano-structures
10
223
Cs
7
Uncharged (Ni-NTA) 10
10
10
10
6
5
Charged (with EDTA)
4
3
10
-2
10
-1
10
0
10
1
10
2
10
3
10
4
Frequency [Hz] Fig. 40: Dramatic changes in global shapes of impedance spectra. The lipid monolayer contains 10 mol% NTA-DOGS and was deposited on 6 layers of isopentylcellulose cinnamate (IPCC). In the presence of nickel ions (Ni-loaded), the lipid monolayer is baring 1 e− /NTA. After treatment with EDTA, the NTA head groups are unloaded and the monolayer is charged with 2 e− /NTA. A semi-quantitative Mott–Schottky plot under potential sweeps revealed that changes in surface charge density of Q ∼ 10−6 C/cm2 changed the flat band potential of the EIS system from UFB (loaded) = −500 mV to UFB (unloaded) = −730 mV. Furthermore, such “switching” of surface charge states can be reproduced for more than two weeks.
Ufb ∼ 10%, we can roughly estimate our sensitivity limit to be around 0.03 charges per nm2 . The obtained result suggests that the sensitivity limit of this strategy is promising to detect selective coupling of proteins on a membrane surface. In fact, the chelator complex of NTA headgroups can serve as docking sites for the recombinant proteins with his-tags (Fig. 41). Thus, EIS systems based on polymer supported monolayers with NTA lipids can be used as a recyclable platform for the selective detection of charged protein binding towards non-invasive applications in pharmaceutical screening tests. 5.3. Orientation-selective immobilization of native cell membranes on ultra-thin polymer films In comparison with various methods to reconstitute transmembrane proteins in artificial model membranes, the direct deposition of “native” cell membranes includes several
224
M. Tanaka
(III) Protein Bound
EDTA
- - - -- - - -- - - -(II) Uncharged - --
- - - -- --
charged proteins his-tag
Ni2+ (I) Charged EDTA - +-
Fig. 41: Potential application of biological EIS systems for sensing single protein binding. As the chelator complex of NTA can serve as docking sites for the recombinant proteins with his-tags, such systems can be used as a recyclable platform for selectively detecting binding of charged proteins.
inside label
monoclocal AB polyclocal AB with FITC
outside label neuraminidase
lectin
glycocalix sialic acid residue lectin with FITC Fig. 42: Immune fluorescence identification of asymmetric orientation of human erythrocyte membranes. The cytoplasmic domain of transmembrane Band 3 protein can be recognized with a specific coupling of a first monoclonal mouse antibody and a second fluorescence labeled polyclonal goat anti-mouse IgG antibody (inside label, top). On the other hand, the extracellular glycocalix of erythrocyte membranes can be detected by fluorescence labelled peanut agglutinin after cleaving sialic acid residues by neuraminidase (outside label, bottom).
Living systems and semiconductor nano-structures
225
20 µm Fig. 43: Immobilized RSO ghosts incubated with the poly-lysine films on planar glass slide. After the incubation, the cytoplasmic domains of Band 3 were identified with inside label. However, the membrane patches did not fuse with each other even after prolonged incubation time. The patches that might correspond to the ruptured ghosts (∼ 145 µm2 ) are highlighted by white circles. Noteworthy, outside labelling resulted in no fluorescence signal, indicating that all the adherent erythrocytes rupture and invert the orientation.
strong advantages. The lateral ordering and surface density of the immobilized proteins are the same as those in native cells, as well as the asymmetric orientation of transmembrane proteins to the membrane surface is strictly controlled by nature. In our recent study [114], right-side-out (RSO) human erythrocyte ghost membranes were deposited on three types of planar solid supports: (1) plain glass slides, (2) physisorbed films of poly-lysine, and (3) Langmuir–Blodgett (LB) films of cellulose derivatives. The resulting membrane orientation was identified with selective fluorescence markers (Fig. 42). The extracellular glycocalix of erythrocyte membranes can be detected by fluorescence labeled peanut agglutinin (outside label), while the cytoplasmic domain of transmembrane Band 3 protein can be recognized with a specific coupling of a first monoclonal mouse antibody and a second fluorescence labeled polyclonal goat anti-mouse IgG antibody (inside label). When RSO erythrocyte ghosts were incubated with planar glass cover slides, no adsorption or rupture of erythrocytes could be observed. To increase the interfacial attraction between cells and the surface, two types of hydrated polymer films were deposited on the glass cover slides. On poly-lysine films, patches of the ruptured membranes could be observed, exposing their cytoplasmic side to the bulk electrolyte (Fig. 43). However, the surface coverage still remained poor. Indeed, the membrane patches did not fuse to heal the defects, even after prolonged incubation time. This might be attributed to the strong attraction between negatively charged glycocalix and positively charged poly-lysine surfaces, which “pin” the adhered membrane patches [103,115]. Actually, such a stable, strong coupling (de-wetting) had been often reported for the strongly adsorbed polyelectrolytes on oppositely charged surfaces [116,117]. It is also well worth noting that the outside label did not yield any fluorescence signals, suggesting all of the immobilized membrane patches inverted
226
M. Tanaka
RSO Ghost
adhesion polymer support
rupture
spreading
Fig. 44: Schematic illustration of the orientation selective immobilization of erythrocyte membranes on the planar surface: (1) approach of vesicles, (2) adhesion to the surface, (3) rupture of vesicles, and (4) lateral spreading and membrane fusion.
their orientation (Fig. 44). On the other hand, RSO ghosts were likely to coat the surface of cellulose films more continuously (Fig. 45). The immune-fluorescence labeling demonstrated that immobilized erythrocyte membrane selectively inverted their native orientation, where the cytoplasmic side of the membrane is exposed to the bulk electrolyte. Tentatively, we interpreted this larger surface coverage on the cellulose film in terms of the “weak attraction” between the cell surface glycocalix and the polysaccharides [38] (ref. Section 2). Such ultrathin (thickness 5–10 nm), biological polysaccharide films have a large potential to immobilize native cell membranes without denaturing their structure, membrane orientation, and functions. Further challenges along this line will be the electrical detection of membrane proteins on semiconductor devices as
Living systems and semiconductor nano-structures
227
20 µm
Fig. 45: Fluorescence image of RSO ghosts deposited on the glass cover slide coated with the Langmuir– Blodgett (LB) film of regenerated cellulose. The cytoplasmic domains of Band 3 were labeled with lectin. Compared to the immobilization on poly-lysine films, the surface coverage was larger and more homogeneous.
Fig. 46: Dynamic accumulation of diffusive membrane proteins under external fields (e.g. electric field, chemical potential gradient) will provide micro-arrays of proteins on semiconductor devices.
well as the separation of membrane proteins by lateral external fields (e.g. electric field, chemical potential gradient) to fabricate “protein arrays” on semiconductor device surfaces (Fig. 46).
Acknowledgements I am sincerely indebted to E. Sackmann for his warm encouragement and fruitful comments while carrying out these studies. I am also grateful to my collaborating
228
M. Tanaka
students (H. Hillebrandt, K. Adlkofer, M.F. Schneider, F. Rehfeldt, O. Purrucker, M. Hochrein) for their enthusiastic challenges. Regarding Section 1, my gratitude is due to G. Abstreiter’s group (Walter–Schottky Institute, Tech. Univ. Munich, fabrication and characterization of semiconductor nano-structures) and T. Bolom and S. Veprek (Tech. Univ. Munich, XPS). The works presented in Section 2 were carried out under the collaborations with G. Wegner (Max-Planck Institute for Polymer Research, cellulose chemistry) and R. Jordan’s group (Tech. Univ. Munich, synthesis of poly(oxazoline) derivatives). All the glycolipids studied in Section 3 have been synthesized by C. Gege and R.R. Schmidt (Univ. Konstanz), and X-ray diffraction studies were carried out with the aid of R. Zantl. I am also thankful to K. Lim and G.G. Fuller (Stanford Univ.) for ISR and G. Mathe for ellipsometry experiments. Contributions of S. Kaufmann and J. Nissen on immobilization of human erythrocytes are deeply acknowledged. These works are financially supported by Deutsche Forschungs Gemeinschaft, DFG (SFB 563, Ta 259/1, Ta 259/2, Sa 246/30), Fonds der Chemischen Industrie, and National Science Foundation (NSF) through the Center on Polymer Interfaces and Macromolecular Assemblies (CPIMA). Last but not least, I am grateful to Japan Society for Promotion of Science (JSPS) and Alexander von Humboldt Foundation for the postdoctoral fellowship and DFG for the Habilitation fellowship (Emmy Noether Program).
References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20.
E. Sackmann and M. Tanaka, Trends Biotechnol. 18, 58 (2000). A. Ulman, An Introduction to Ultrathin Organic Films: From Langmuir–Blodgett to Self-Assembly (Academic Press, 1991). M.D. Porter, T.B. Bright, D.L. Allara, and C.E.D. Chidsey, J. Am. Chem. Soc. 109, 3559 (1987). C.E.D. Chidsey and D.N. Loiacono, Langmuir 6, 682 (1990). C. Miller and M. Grätzel, J. Phys. Chem. 95, 5225 (1991). E. Sabatani, J. Cohen-Boulakia, M. Bruening, and I. Rubinstein, Langmuir 9, 2974 (1993). H.O. Finklea, D.A. Snider, J. Fedyk, E. Sabatani, Y. Gafni, and I. Rubinstein, Langmuir 9, 3660 (1993). R. Maoz and J. Sagiv, J. Coll. Inter. Sci. 100, 465 (1984). E.P. Pluedemann, Silane Coupling Agents (Plenum Press, 1990). S.R. Wasserman, Y.T. Tao, and G.M. Whitesides, Langmuir 5, 1074 (1989). A.N. Parikh and D.L. Allara, J. Phys. Chem., 98, 7577 (1994). S.R. Wasserman, G.M. Whitesides, I.M. Tidswell, B.M. Ocko, P.S. Pershan, and J.D. Axe, J. Am. Chem. Soc. 111, 5852 (1989). N. Balachander and C.N. Sukenik, Langmuir 6, 1621 (1990). P. Fountain, D. Goguenheim, D. Deresmes, D. Vuillaume, M. Garet, and F. Rondelez, Appl. Phys. Lett. 62, 2256 (1993). C. Boulas, J. V. Davidovits, F. Rondelez, and D. Vuillaume, Phys. Rev. Letters 76, 4797 (1996). O.M.-R. Chyan and K. Rajeshwar, J. Electrochem. Soc. 132, 2109 (1985). S. Gritsch, P. Nollert, F. Jähnig, and E. Sackmann, Langmuir 14, 3118 (1998). H. Hillebrandt, G. Wiegand, M. Tanaka, and E. Sackmann, Langmuir 15, 8451 (1999). H. Hillebrandt and M. Tanaka, J. Phys. Chem. B 105, 4270 (2001). P. Baumgartner, C. Engel, G. Abstreiter, G. Böhm, and G. Weimann, Appl. Phys. Lett. 66, 751 (1995).
Living systems and semiconductor nano-structures 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65.
229
P. Baumgartner, C. Engel, G. Abstreiter, G. Böhm, and G. Weimann, Appl. Phys. Lett. 64, 592 (1994). S. Menezes and B. Miller, J. Electrochem. Soc. 130, 517 (1983). I. Uhlendorf, R. Reinecke-Koch, and R. Memming, J. Phys. Chem. 100, 4930 (1995). Z. Hens and W.P. Gomes, J. Phys. Chem. B 103, 130 (1999). S.R. Lunt, G.N. Ryba, P.G. Santangelo, and N.S. Lewis, J. Appl. Phys. 70, 7449 (1991). K. Asai, T. Miyashita, K. Ishigure, and S. Fukatsu, Surf. Sci. 306, 37 (1993). O.S. Nakagawa, S. Ashok, C.W. Sheen, J. Märtensson, and D.L. Allara, Jpn. J. Appl. Phys. 30, 3759 (1991). C.W. Sheen, J.X. Shi, J. Martenson, and D.L. Allara, J. Am. Chem. Soc. 114, 1514 (1992). E.A. Miller and G.L. Richmond, J. Phys. Chem. B 101, 2669 (1997). K. Adlkofer, M. Tanaka, H. Hillebrandt, G. Wiegand, E. Sackmann, T. Bolom, R. Deutschmann, and G. Abstreiter, Appl. Phys. Lett. 76, 3313 (2000). K. Adlkofer and M. Tanaka, Langmuir 17, 4267 (2001). E.F. Duijs, F. Findeis, R.A. Deustchmann, M. Bichler, A. Zrenner, G. Abstreiter, K. Adlkofer, M. Tanaka, and E. Sackmann, Phys. Stat. Sol. B 224, 871 (2001). K. Adlkofer, E.F. Duijs, F. Findeis, M. Bichler, A. Zrenner, E. Sackmann, G. Abstreiter, and M. Tanaka, Phys. Chem. Chem. Phys. 4, 785 (2002). K. Adlkofer, W. Eck, M. Grunze, and M. Tananka, J. Phys. Chem. B, in print (2003). H.J. Gabius and S. Gabius, Glycoscience (Chapmann & Hall, 1997). W.D. Comper, Extracellular Matrix (Harwood Academic Publishers, 1996). G. Elender and E. Sackmann, J. Phys. II France 4, 455 (1994). R. Bruinsma, A. Behrisch, and E. Sackmann, Phys. Rev. E. 61, 4253 (2000). S. Löfas and B. Johnson, J. Chem. Soc., Chem. Commun. 21, 1526 (1990). J.M. Harris, Poly(ethyleneglycol) Chemistry (Plenum Press, 1992). H.J. Taunto, C. Toprakcioglu, L.J. Fetters, and J. Klein, Nature 332, 712 (1988). S.I. Jeon and J.D. Andrade, J. Colloid Interface Sci. 142, 159 (1991). S.I. Jeon, J.H. Lee, J.D. Andrade, and P.D. de Gennes, J. Colloid Interface Sci. 142, 149 (1991). A. Halperin, Langmuir 15, 2525 (1999). Z. Yang, J.A. Galloway, and H. Yu, Langmuir 15, 8405 (1999). C.G.P.H. Schröen, M.A. Cohen Stuart, K. van der Voort Maarchalk, A. van der Padt, and K. van’t Riet, Langmuir 11, 3068 (1995). S.J. Sofia, V. Premnath, and E.W. Merrill, Macromolecules 31, 5059 (1998). B.J. Herren, S.G. Shafer, J. van Alstine, J.M. Harris, and R.T. Snyder, J. Colloid Interface Sci. 115, 46 (1987). E. Kiss and C.G. Gölander, Coll. Surf. 49, 335 (1990). N.L. Burns, J.M. van Alstine, and J.M. Harris, Langmuir 11, 2768 (1995). F. Rehfeldt and M. Tanaka, Langmuir, in print (2003). W. Wegner, Thin Solid Films 216, 105 (1992). M. Schaub, G. Wenz, G. Wegner, A. Stein, and D. Klemm, Adv. Mater. 5, 919 (1993). V. Buchholz, G. Wegner, S. Stemme, and L. Ödberg, Adv. Mater. 8, 399 (1996). G. Mathe, A. Albersdörfer, K. R. Neumaier, and E. Sackmann, Langmuir 15, 8726 (1999). S. Kobayashi, S. Iijima, T. Igarashi, and T. Saegusa, Macromolecules 20, 1729 (1987). T. Lehmann and J. Rühe, Macromol. Symp. 142, 1 (1999). T.G. Bassiri, A. Levy, and M. Litt, J. Polym. Sci. B 7, 463 (1969). M.C. Woodle, C.M. Engbers, and S. Zalipsky, Bioconjugate Chem. 5, 493 (1994). F. Rehfeldt, L. Pagnoni, R. Jordan, and M. Tanaka, Langmuir (2002), in print. A.S. Lee, A.P. Gast, V. Butun, and S.P. Armes, Macromolecules 32, 4302 (1999). W. Curatolo, Biochim. Biophys. Acta 906, 137 (1987). P.R. Crocker and T. Feizi, Curr. Opin. Struct. Biol. 6, 679 (1996). A. Varki, Proc. Natl. Acad. Sci. USA 91, 7390 (1994). J. Vogel, G. Bendas, U. Bakowsky, G. Hummel, R.R. Schmidt, U. Kettmann, and U. Rothe, Biochim. Biophys. Acta 1372, 205 (1998).
230
M. Tanaka
66. 67.
S. Hakomori, J. Biol. Chem. 265, 18713 (1990). M.F. Schneider, G. Mathe, M. Tanaka, C. Gege, and R.R. Schmidt, J. Phys. Chem. B 105, 5178 (2001). H. Möhwald, in Structure and Dynamics of Membranes, edited by R. Lipowsky and E. Sackmann (Elsevier, 1995). S. Alexander, J. Phys. II France 38, 983 (1977). P.G. De Gennes, J. Phys. 37, 1445 (1976). S.T. Milner, T.A. Witten, and M.E. Cates, Macromolecules 21, 2610 (1988). S.T. Milner, T.A. Witten, and M.E. Cates, Europhys. Lett. 5, 413 (1988). M.S. Schneider, R. Zantl, C. Gege, R.R. Schmidt, and M. Tanaka, Biophys. J. (2003), in print. A. Razatos, Y.-L. Ong, M.M. Sharma, and G. Georgiou, Proc. Natl. Acad. Sci. USA 95, 11059 (1998). Y.-L. Ong, A. Razatos, G. Georgiou, and M.M. Sharma, Langmuir 15, 2719 (1999). M. Arnordi, M. Fritz, E. Bäuerlein, M. Radmacher, E. Sackmann, and A. Boulbitch, Phys. Rev. E 62, 1034 (2000). W. Xu, P.J. Mulhern, B.L. Blackford, M.H. Jericho, M. Firtel, and T.J. Beveridge, J. Bacteriol. 178, 3106 (1996). X. Yao, M. Jericho, D. Pink, and T. Beveridge, J. Bacteriol. 181, 6865 (1999). B. Warburton, Curr. Opinion Colloid Int. Sci. 1, 481 (1996). C.F. Brooks, G.G. Fuller, C.W. Franck, and C.R. Robertson, Langmuir 15, 2450 (1999). M.F. Schneider, K. Lim, G.G. Fuller, and M. Tanaka, Phys. Chem. Chem. Phys. (2002), in print. C. Gege, G. Bendas, J. Vogel, U. Rothe, and R.R. Schmidt, Chem. Eur. J. 6, 111 (2000). L.K. Tamm and H.M. McConnell, Biophys. J. 47, 105 (1985). E. Sackmann, Science 271, 43 (1996). R. Merkel, E. Sackmann, and E. Evans, J. Phys. France II 50, (1989). E. Kalb, S. Frey, and L.K. Tamm, Biochim. Biophys. Acta 1103, 307 (1992). J. Nissen, S. Gritsch, G. Wiegand, and J. Rädler, Eur. Phys. J. B 10, 335 (1999). C. Miller, P. Cuendet, and M. Grätzel, J. Electroanal. Chem. 278, 175 (1990). D.W. Urry, D.F. Mayers, and J. Haider, Biochemistry 11, 487 (1972). H. Tournois, J.A. Killian, D.W. Urry, O.R. Bokking, J. de Gier, and B. de Kruijff, Biochim. Biophys. Acta 905, 222 (1987). J.A. Killian, K.U. Prasad, D. Hains, and D.W. Urry, Biochemistry 27, 4848 (1988). M. Bouchard and M. Auger, Biophys. J. 65, 2484 (1993). A.A. Brian and H.M. McConnell, Proc. Natl. Acad. Sci. USA 81, 6159 (1984). P. Chan, M.B. Lawrence, M.L. Dustin, L.M. Ferguson, D.E. Golan, and T.A. Springer, J. Cell Biol. 10, 245 (1991). P.B. Contino, C.A. Hasselbacher, J.B.A. Ross, and Y. Nemerson, Biophys. J. 67, 1113 (1994). R. Naumann, A. Jonczyk, R. Kopp, J. van Esch, H. Ringsdorf, W. Knoll, and P. Gräber, Angew, Chem. Int. Ed. Engl. 34, 2056 (1995). J. Salafsky, J.T. Groves, and S.G. Boxer, Biochemistry 35, 14773 (1996). G. Puu, E. Artursson, I. Gustafson, M. Lundström, and J. Jass, Biosens. Bioelectron. 15, 31 (2000). D. Savage, G. Mattson, S. Desai, G. Nielander, S. Morgensen, and E. Conklin, Avidin–Biotin Chemistry: A Handbook (Pierce Chemical Company, 1992). G.A.M. Cross, Annu. Rev. Cell Biol. 6, 1 (1990). A.L. Beauchamp, J. Israeli, and H. Saulnier, Can. J. Chem. 47, 1269 (1969). L. Schmitt, C. Dietrich, and R. Tampé, J. Am. Chem. Soc. 116, 8485 (1994). B.S. Jacobson and D. Branton, Science 195, 302 (1976). O. Purrucker, H. Hillebrandt, K. Adlkofer, and M. Tanaka, Electrochim. Acta 47, 791 (2001). S.M. Sze, Semiconductor Devices: Physics and Technology, (John Wiley & Sons, 1985). P. Fromherz, A. Offenhäuser, T. Vetter, and J. Weiss, Science 252, 1290 (1991). H.M. McConnell, J.C. Owicki, J.W. Parce, D.L. Miller, G.T. Baxter, H.G. Wada, and S. Pitchford, Science 257, 1907 (1992). M. Stenberg, H. Arwin, and A. Nilson, J. Colloid Interface Sci. 72, 255 (1979).
68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. 98. 99. 100. 101. 102. 103. 104. 105. 106. 107. 108.
Living systems and semiconductor nano-structures 109. 110. 111. 112. 113. 114. 115. 116. 117.
231
A. Sinner and A. Offenhaeusser, Thin Solid Films 327, 758 (1998). J.P. Dilger, S.G. McLaughlin, T.J. McIntosh, and S.A. Simon, Science 206, 1196 (1979). P. Fromherz, V. Kiessling, K. Kottig, and G. Zeck, Appl. Phys. A 69, 571 (1999). H. Hillebrandt, M. Tanaka, and E. Sackmann, J. Phys. Chem. B 106, 477 (2002). M. Seufert, F. Christo, and G. Wegner, Adv. Mater. 7, 52 (1995). M. Tanaka, S. Kaufmann, J. Nissen, and M. Hochrein, Phys. Chem. Chem. Phys. 3, 4091 (2001). N.E. Thomas, W.T. Coakley, and C. Winters, Coll. Surf. B 6, 139 (1996). T. Afshar-Rad, A.I. Baily, P.F. Luckham, W. MacNaughtan, and D. Chapman, Biochim. Biophys. Acta 915, 101 (1987). G. Decher, H.D. Hong, and J. Schmitt, Thin Solid Films 210–211, 831 (1992).
Nanotechnology and Nano-Interface Controlled Electronic Devices Editors: M. Iwamoto, K. Kaneto and S. Mashiko © 2003 Elsevier Science B.V. All rights reserved
CHAPTER 13
Structural color forming system composed of polypeptide-based LB films Takatoshi Kinoshita a , Shujiro Hayashi b , Yoshiyuki Yokogawa b , and Shintaro Washizu c a Department
of Materials Science and Engineering, Nagoya Institute of Technology, Gokiso-cho, Showa-ku Nagoya 466-8555, Japan E-mail:
[email protected] b National Institute of Advanced Industrial Science and Technology, Hirate-cho 1-1, Kita-ku, Nagoya 462-8510, Japan c Fujinomiya Research Laboratories, Fuji Photo Film co., LTD., Fujinomiya Shizuoka 418-8666, Japan
1. 2. 3.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Self-organized two-dimensional patterning by α-helical block-copolypeptides . . . . . . Two-dimensional structural color formation and regulation using polypeptide LB films . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
233 236 242 252 252
1. Introduction Fig. 1 shows an example of an intelligent function of a living system. Living cells sense their environment as a stimulus and respond to environmental changes for the maintenance of life. For example, a physical or chemical stimulus is accepted by the signaling cell and the induced signal is transmitted to the nerve system with a nerve impulse, and finally, to the muscle cell with an accompanying contraction as response. One of the important points for us was that these sensory systems are installed at biological interfaces such as the biological membranes. 30 years ago, Singer and Nicolson [1] predicted that biological membranes are an assembly of molecular machines, each responsible for an elementary function. This idea has been reasonably accepted now. And it is recognized that the biological membranes are a self-organized system composed of amphiphilic molecules such as lipids and membrane proteins, and the specific location and/or molecular orientation of these elements are essential to lead the intelligent functions at the biological interfaces.
234
T. Kinoshita et al.
Fig. 1: Transmembrane signaling and sensation.
As a part of the membrane mimetic chemistry or supramolecular science, monolayer, bilayers and vesicles have been extensively studied using lipid-like amphiphiles, and more recently, the formation of molecular membrane systems composed of α-helical polypeptides is also investigated. For example, Kimura et al. [2] reported on a α-helix regularly standing system as a self-assembled monolayer (SAM) on Au substrate (Fig. 2). Higashi et al. [3] showed the preparation of a α-helix rich PLGA monolayer and its enantioselectivety. For the last few years, we [4] have tried to construct novel artificial membrane or interfacial systems using polypeptides and their derivatives as a very simple model polymer of membrane proteins. And at the same time we proposed two approaches for mimicking the intelligent functions of biological membrane systems (Fig. 3). One is functional modeling such as stimulus–response coupling, information transfer, and so
Structural color forming system composed of polypeptide-based LB films
235
Fig. 2: Schematic illustration of a SAM of a helical peptide.
Fig. 3: A fundamental concept for mimicking biological membranes and interfaces.
on. The other is structural modeling, i.e., mimicking the design of biological membrane components such as membrane proteins and the surface and/or interface structure of living systems. Based on this concept, we will show here two types of polypeptide membrane systems (Fig. 4). One is the self-organized two-dimensional patterning in an amphiphilic helix monolayer. And the other is the polypeptide Langmuir–Blodgett (LB)
236
T. Kinoshita et al.
Fig. 4: Molecular membrane system composed of α-helical polypeptide.
film. For the former topic, we will show a possibility of nanometer-scale control of two-dimensional structure. And for the latter, structural color formation and regulation will be shown. 2. Self-organized two-dimensional patterning by α-helical block-copolypeptides A monolayer was firstly prepared by a diblock-copolypeptide composed of hydrophilic and hydrophobic α-helix segments [5]. The hydrophobic segment is poly(L-lysine) derivative and the hydrophilic segment is L-glutamic acid copolymer. The stable α-helix structure was confirmed by circular dichroism measurement of the TFE solution of the polymer. So this polymer, as schematically shown in Fig. 5, is composed of a hydrophobic larger-diameter helix and a smaller-diameter and longer hydrophilic helix rod. The monolayer was prepared by spreading a DMF/benzene mixed solution of the polymer on a water surface at pH 5 (Fig. 6). And the surface pressure was monitored by compressing the monolayer by the Wilhelmy method. Fig. 7 shows the surface pressure–area isotherm obtained at pH = 5. The limiting area was estimated to be ca. 6 nm2 /molecule. The calculated limiting area when the helix is normal to the air/water interface is shown in Table 1. We expected the normal orientation of the helix after compressing the membrane, however, the big difference between the calculated and observed limiting area indicated that it is impossible to obtain a standing helix monolayer using this system.
Structural color forming system composed of polypeptide-based LB films
237
Fig. 5: Chemical structure and schematic illustration of PLLZ25 –P(MLG42 /LGA18 ) (poly(ε-benzyloxycarbonyl L-lysine)25 –poly[(γ -methyl L-glutamate)42 /( L-glutamic acid)18 ]).
Fig. 6: Surface pressure measurements.
238
T. Kinoshita et al.
Fig. 7: π – A isotherm for a monolayer of PLLZ25 –P(MLG42 /LGA18 ) with 0.1 mol/l KCl in aqueous solution at pH 5. Table 1 Limiting area estimated from π – A isotherm.
PLLZ25 –P(MLG42 /LGA18 )
Observed (nm2 /molecule) ApH=5
Calculated (nm2 /molecule) A⊥
6.26
2.39
So the monolayer was transferred onto a mica surface at 25 mN/m to get an LB film with a single layer. And the morphology of the LB film was observed by atomic force microscopy (AFM). Fig. 8 shows the AFM image of the LB film on µm scale. From the depth of the cavity that was made by scratching with a cantilever, the thickness of the membrane was estimated to be ca. 1.4 nm. This value is almost consistent with the diameter of the helix, indicating that the α-helix rods lie down on the mica surface. Fig. 9 is a nanometer-scale image of the LB film. A stripe pattern composed of thick and thin layers was observed. And the difference of the thickness was ca. 0.3 nm. This value is almost consistent with the difference between the radii of the hydrophilic and hydrophobic helix rods. This may mean, therefore, that the thick layer is a molecular
Structural color forming system composed of polypeptide-based LB films
239
Fig. 8: AFM image (1.25 µm × 1.25 µm) of PLLZ25 –P(MLG42 /LGA18 ) LB film on mica (pH 5).
array of the hydrophobic larger-diameter helix and the thin layer that of the hydrophilic smaller helix. The interval of the stripes was estimated to be ca. 24 nm. This value is almost the twice the length of this polymer. These results suggest that this diblockcopolypeptide aggregates by head to head and tail to tail, resulting in the formation of a nanophase-separated structure as is shown in Fig. 10.
240
T. Kinoshita et al.
Fig. 9: AFM image (65.1 nm × 65.1 nm) of PLLZ25 –P(MLG42 /LGA18 ) LB film on mica (pH 5).
Fig. 10: Schematic illustration of the nanophase-separated structure of PLLZ25 –P(MLG42 /LGA18 ) LB film on mica substrate.
However, Fig. 11(a) is the AFM image within 140 nm2 , a larger area than that of Fig. 9. A disordered structure was observed. That is, except for the regular stripe domain a branching pattern was also made. Fig. 11(b) shows a schematic illustration of the speculated monolayer structure. We think that the hydrophobic interaction and hydrogen bonding between helix rods promote the phase-separated regular pattern, however, the large difference in diameter between the two helix units is not convenient, in this case, for regular packing of the helix rods over a wide range. So as a next step, the monolayer was prepared by a triblock-copolypeptide composed
Structural color forming system composed of polypeptide-based LB films
241
Fig. 11: AFM image (140 nm × 140 nm) of PLLZ25 –P(MLG42 /LGA18 ) LB film on mica substrate (a) and its schematic illustration (b).
of hydrophobic, hydrophilic and hydrophobic helix segments (Fig. 12). We expected that this dumbbell shape will block the development of the branching pattern. The hydrophobic segment is poly(L-leucine) and the hydrophilic segment is poly(L-glutamic acid). The stable α-helix structure was confirmed by FT-IR measurements of the LB film on Au surface. In this case, the monolayer on aqueous solution at pH = 4.0 was transferred on an Au surface for the FT-IR measurements. So this polymer in the acidic condition, as schematically shown in Fig. 12, is composed of a hydrophobic larger-diameter helix and a smaller-diameter hydrophilic helix. And then, the monolayer on aqueous solution at pH = 4.0 was transferred onto a mica surface to get the LB film with the single layer for AFM measurements. Fig. 13(a)
242
T. Kinoshita et al.
Fig. 12: Chemical structure and schematic illustration of PLL54 –PLGA80 –PLL54 .
shows the AFM image of the monolayer within 140 nm2 , indicating a stripe pattern composed of thick and thin layers. The interval of the stripes, in this case ca. 30 nm, is consistent with the length of the triblock-copolypeptide. So, the nanophase-separated structure of the monolayer is schematically shown in Fig. 13(b). As is expected, the triblock-copolypeptide made a more sophisticated stripe pattern with very few branching structures. We think that this is a good example showing a relation between the molecular shape and character of the polypeptides and their self-organized two-dimensional structure. It is expected, therefore, that the size of this pattern can be regulated by the degree of polymerization of each helix and the functions might be controllable by the side chain structure of the α-helix units. Thus, we showed the possibility of nanometer-size control of two-dimensional structure for the formation of novel functional interfaces.
3. Two-dimensional structural color formation and regulation using polypeptide LB films Recently, new types of display systems, such as colored liquid crystal and organic EL (electroluminescence), are rapidly developed and penetrating into our lives as displays of portable telephones, computer games, and displays of sound systems in automobiles. Furthermore, it is expected that the novel display systems should answer the problems of ecology, energy, and human health, especially for the eyes in this case. It is well-known that nature has unique display systems such as the structural color of butterflies, beetles, and tropical fish and shells. The structural color or interference color, in this case, is based on a nanometer-scale layered structure of the body surface, without any pigments.
Structural color forming system composed of polypeptide-based LB films
243
Fig. 13: AFM image (140 nm × 140 nm) of PLL54 –PLGA80 –PLL54 LB film on mica substrate (a) and its schematic illustration (b).
When light with a certain incident angle (α) is reflected at the surface layer, it can be emphasized or weakened depending on the wavelength (λ) of the light (Fig. 14). The λ-value of emphasized or weakened light is given by these two equations [6]. 2h 2 λ= n − sin2 α (emphasized) (1) m 4h 2 λ= n − sin2 α (enfeebled) (2) 2m − 1 where h is thickness of the corresponding layer, n is refractive index of the layer, and m is natural number (m = 1, 2, . . .). For α-helix LB films, in our experimental range of h, m ≤ 2, these two equations can be written as these equations, where d is the diameter
244
T. Kinoshita et al.
Fig. 14: Principle of structural color formation.
of the helix rod and l is the number of monolayers in the LB film. Thus we can see the particular color from the surface layer according this optical principle. Structural color materials are also investigated and applied as a color coating technique for ceramic surfaces, such as for example the structural color formation of oxidized titanium [7]. The temperature, in this case, can control the thickness of the oxidized layer and it systematically controls various structural colors. Furthermore, organic materials are also investigated, for example, solution [8] and gel systems [9] for cosmetic application, and structural color fibers are also introduced recently. However, structural color LB systems have not been fully investigated. So recently, we have tried to construct structural color LB films of stimulus–response type as a novel display system [10]. We are using, in this case also, α-helical polypeptide, because the polypeptide can produce a well-defined layered structure on the substrate by the LB method, as we showed above. The preparation and characterization of polypeptide LB films were investigated by Wegner’s group [11–14] using polypeptides with long alkyl side chains (Fig. 15). They call it “hairy-rod”. And they showed the supra-molecular structure and optical properties of these LB films, however, did not show the structural color formation. The long alkyl side chain of polypeptides is easier to crystallize in a film, even at room temperature. This will have a negative influence on the optical properties, so we selected poly(n-hexyl L-glutamate) (PHeLG, Fig. 16), which has not so long alkyl chains, as the LB film component. In this case, the stable α-helix conformation of this polymer was also confirmed by CD measurements. The monolayer was prepared by spreading a DMF/benzene mixed solution of this polymer on a water surface. And the surface pressure was monitored by compressing the monolayer by the Wilhelmy method at 20°C. Fig. 17 shows the surface pressure–area isotherm of this monolayer. The limiting area was estimated to be 26.2 nm2 /molecule. This value is exactly consistent with the calculated limiting area when the helix rod is parallel to the air/water interface. So the monolayer at 15 mN/m was transferred onto a silicon substrate. The substrate is hydrophobized before use by coating a silane-coupling agent with octadecyl moiety and heat-fixed at 110°C. Fig. 18 shows the relation between
Structural color forming system composed of polypeptide-based LB films
245
Fig. 15: Schematic illustration of polymer of the “hairy-rod” type, and its orientation on an air/water surface.
Fig. 16: Chemical structure of poly(n-hexyl L-glutamate) (PHeLG).
the deposition ratio and the number of layers. The first step was down mode of the substrate, the second was up mode, and then this repeated alternately. Each deposition ratio is ca. 1, indicating the monolayer was transferred onto the substrate keeping the structure on the water surface. The structure of PHeLG on the substrate was characterized by FT-IR measurements. Fig. 19 shows a transmission FT-IR spectrum of 120 layers of the LB film. There are four major peaks: a side chain C=O peak, two amide I bands associated with the α-helix structure and the β structure, respectively, and amide II of the α-helix structure. The α-helix amide I band is much larger than that of the β-structure, indicating PHeLG kept the α-helix structure in the LB film after the deposition. The thickness of LB film on the substrate was measured by AFM. Fig. 20(a) shows an AFM image of the 20 layers LB film with the boundary between the top of the 20 layers and the surface of the substrate itself. The difference in height between them is
246
T. Kinoshita et al.
Fig. 17: Surface pressure–area isotherm of PHeLG monolayer at air water interface.
Fig. 18: Deposition ratio of PHeLG monolayer onto silicon substrate as a function of number of layers.
Structural color forming system composed of polypeptide-based LB films
247
Fig. 19: The transmission FT-IR spectrum of 120 layers of PHeLG LB film.
Fig. 20: (a) AFM image of PHeLG LB film; a boundary between 20 layers of PHeLG and the substrate surface. (b) Section of the line in (a).
ca. 32 nm (Fig. 20(b)). This value is corresponding to 20 layers. Therefore, the thickness of one layer is estimated to be ca. 1.6 nm. This value is almost consistent with the diameter (1.5 nm) of PHeLG obtained from the limiting area of the π–A isotherm of this monolayer. Thus, we can confirm a well-defined layered structure in the LB film. Furthermore, we could find the color formation shown in Fig. 21. The LB films showed various colors depending on the number of layers; brown at 40 layers, dark blue at 80 layers, yellow at 120 layers, and red at 160 layers. As a next step, the color formation behavior could be quantitatively analyzed by the reflective VIS spectra of these LB films. Fig. 22 shows reflective VIS spectra of the LB films with the incident light angle at 10°. The 80 layers film shows a peak at 418 nm
248
T. Kinoshita et al.
Fig. 21: Structural color of PHeLG LB films.
Fig. 22: Reflective VIS spectra of 40, 80, and 120 layers of PHeLG LB films. Incident light angle was 10°.
corresponding to dark blue. Oppositely, the 40 layers film shows a minimum peak at 456 nm corresponding to brown or dark orange which is the complementary color to blue. The 120 layers film shows a peak at 619 nm corresponds to yellow. As a result,
Structural color forming system composed of polypeptide-based LB films
249
Fig. 23: Measured and calculated wavelength of interference colors as a function of number of layers.
we can compare these peak positions with the calculated values from Eqs. 1 and 2. Fig. 23 shows the measured and calculated wavelengths of colors as functions of number of layers. For example, in the case of 40 layers there is one emphasized peak and one enfeebled peak. And for 80 layers, there are two emphasized peaks and two minimum peaks. The solid lines were obtained by Eq. 1, i.e., λ of the emphasized light, and the dashed lines are λ of the weakened light obtained from Eq. 2, using the parameters (t = 1.6 nm and n = 1.6). The calculated lines are almost consistent with the measured peaks. Furthermore, Fig. 24 shows observed and calculated peak positions as a function of incident light angle. The open circles and squares are maximum peaks and the filled circles and squares are minimum peaks of these layers, respectively. And the lines A and B were obtained using Eqs. 1 and 2 under the condition A and under the condition B. The condition B yields the same calculated line B. The observed peak positions are almost consistent with the calculated values for all the incident light angles. These results suggest that we can get a structural color forming system by polypeptide-based LB films. Thus, we obtained a structural color forming system based on the structural modeling approach in Fig. 3. So, we tried to implement stimulus–response coupling functions in our prepared structural color forming system. First trial is photo-regulation of the structural color. For that, azobenzene containing polypeptides were prepared. It is expected that the photoisomerization of the azobenzene moiety can induce changes in the structure of the LB films (Fig. 25). This side-chain azobenzene system has been examined. As a result, poly(γ -methyl L-glutamate) containing 28% azobenzene side-chains showed photo-induced changes in the reflective spectrum of the LB film, ca. 10 nm peak shift to lower wavelength. It was also confirmed that the original spectrum can be recovered in the dark. Fig. 26 shows a shift of the λmax value induced by dioxane vapor sorption, i.e., the solvent-induced swelling of the LB film shifted the λmax to higher value. This is an example of a chemo-responsive structural color forming system. According to the
250
T. Kinoshita et al.
Fig. 24: Measured and calculated wavelengths of interference colors as a function of incident light angle.
evaporation of the solvent during this measurement, this λmax was apparently lowered. So we roughly calculated a shift of λmax using Eq. 1. In this case, the swelling of the layer was estimated by the experimental value of the amount of sorption of dioxane to this LB film, ca. 5% when the relative vapor pressure of dioxane is 0.2 [15]. And the refractive index of the solvent is almost the same as that of the LB matrix. As a result, It was shown that the color should change from the yellow to the red region. We directly observed the color change of the LB film in the sorption chamber. The vapor pressure was controlled by the temperature. As a result, we could observe the expected color change, yellow to pink, thus confirming the solvent-induced color change based on the swelling of the LB film. On the other hand, methanol shifted λmax to lower wavelength. Opposite to the shift from dioxane. In this case, methanol is a poor solvent, so there is no significant swelling of the layer, so we think that the lower refractive index of the methanol may induce the shift of λmax to lower values.
Structural color forming system composed of polypeptide-based LB films
251
Fig. 25: Control of interference color by azobenzene containing polypeptide LB Films.
Fig. 26: Reflective VIS spectra of 160 layers of PHeLG LB films on silicon substrate and with 1,4-dioxane. Incident light angle was 10°.
252
T. Kinoshita et al.
Fig. 27: Control of interference color by stimulus–response coupling.
These results indicate that these LB films sense the chemicals by their color change like litmus paper. We are now trying to construct such stimulus–response LB films using light, temperature and chemicals as a stimulus (Fig. 27). This type of novel display and novel sensing system will be more sophisticated in 1–2 years, based on the fundamental results today.
Acknowledgements This study has been supported by the New Energy and Industrial Technology Development Organization (NEDO), Japan, for the project on Technology for Material Structure Control.
References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15.
S.J. Singer and G.L. Nicolson, Science, 175, 720 (1972). Y. Miura, S. Kimura, Y. Imanishi, and J. Umemura, Langmuir 14, 6935 (1998). N. Higashi, T. Koga, Y. Fujii, and M. Niwa, Langmuir 17, 4061 (2001). T. Kinoshita, Prog. Polym. Sci. 20, 527 (1995); J. Photochem. Photobiol. B 42, 12 (1998). H. Yokoi, T. Kinoshita, Y. Tsujita, and H. Yoshimizu, Chem. Lett., 1210 (2000). E. Hecht and A. Zajac, Optics (Addison-Wesley, Reading, MA, 1974). S. Sakka, K. Kamiya, and T. Yoko, ACS Symposium Series 360 (American Chemical Society, Washington, DC, 1988) pp. 345–353. K. Naitoh, Y. Ishii, and K. Tsujii, J. Phys. Chem. 95, 7908 (1991). M. Hayakawa, T. Onda, T. Tanaka, and K. Tsujii, Langmuir 13, 3595 (1991). T. Kinoshita, S. Hayashi, and Y. Yokogawa, J. Photochem. Photobiol. A 145, 101 (2001). G. Duda, A.J. Schouten, T. Arndt, G. Lieser, G.F. Schmidt, C. Bubeck, and G. Wegner, Thin Solid Films 159, 221 (1988). W. Hickel, G. Duda, M. Jurich, T. Kröhl, K. Rochford, G.I. Stegeman, J. D. Swalen, G. Wegner, and W. Knoll, Langmuir 6, 1403 (1990). A. Mathy, K. Mathauer, G. Wegner, and C. Bubeck, Thin Solid Films 215, 98 (1992). M. Büchel, Z. Sekkat, S. Paul, B. Weichart, H. Menzel, and W. Knoll, Langmuir 11, 4460 (1995). T. Kinoshita, N. Okazaki, A. Takizawa, and Y. Tsujita, Polymer 20, 791 (1979).
Nanotechnology and Nano-Interface Controlled Electronic Devices Editors: M. Iwamoto, K. Kaneto and S. Mashiko © 2003 Elsevier Science B.V. All rights reserved
CHAPTER 14
Generation of a strong dipole layer and its function by using helical peptide molecular assemblies Shunsaku Kimura, Tomoyuki Morita, and Kazuya Kitagawa Department of Material Chemistry, Graduate School of Engineering, Kyoto University, Yoshida Honmachi, Sakyo-ku, Kyoto 606-8501, Japan
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. Primary amphiphilicity and orientation on water . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. Molecular orientation of helical peptide SAM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. In situ FTIR-RAS measurement on subphase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5. Helical peptide monolayer spread at limited area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6. Helical peptide monolayer spread on a mixture of water and methanol . . . . . . . . . . . . . . 7. Multilayer formation on gold . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8. Surface potential of helical peptide SAM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9. Photocurrent generation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10. Future aspects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
253 254 256 258 263 264 264 265 265 266 267
1. Introduction α-Helix is one of the major secondary structures found in proteins. The molecular shape of the helix is cylindrical, and the two ends, the N and C terminals, have an opposite kind of partial charge to each other. The N terminal is charged positively because the amide protons point to the N terminal. On the other hand, the C terminal is charged negatively due to the direction of the carbonyl oxygens of amides toward the C terminal [1]. As a whole, the helix possesses a large dipole moment along the helix axis. Hol has pointed out that the large dipole moment of the helix should contribute to the protein functions such as binding a substrate to the active center and electron transfer in the photosynthetic center, etc. [2]. Molecular assemblies of helical peptides have been investigated since the 1950s (references in [3]). For example, poly(γ -methyl L-glutamate) was spread on water, and the π–A isotherm was measured. It was found that helix axes in the monolayer lay
254
S. Kimura et al.
flat on the water surface. Upon compression, change from monolayer to bilayer was observed [4]. These phenomena are commonly observed with other helical peptides, and it is a general understanding that it is a hard task to obtain helical peptide layers with a vertical orientation. There are several factors which influence the orientation of helical peptides at an interface, which are introduced in the following sections.
2. Primary amphiphilicity and orientation on water Molecular designing, which considers the balance of hydrophilicity and hydrophobicity in the molecule, is presumably the most reasonable strategy for obtaining a monolayer with a vertical orientation. With respect to this point, naturally occurring biologically active peptides tell us many things. There are many naturally occurring peptides which are composed both of hydrophilic and hydrophobic amino acid residues in regular spatial arrangements. The arrangement is classified into two types: primary and secondary amphiphilicity. The former type has a block property where hydrophilic and hydrophobic residues form each cluster and are arranged in series. On the other hand, the latter type shows hydrophilic and hydrophobic surfaces in the molecule when it takes a secondary structure such as α-helix. It has been pointed out that amphiphilicity of peptides is crucial in the biological activity, for example, by taking an active conformation for binding to a receptor. An interesting and useful idea, which correlates the amphiphilicity of peptides with the biological activity, has been proposed [5]. The idea is called the membrane compartment concept, and describes in detail the way of induction of active conformation on the basis of the amphiphilicity and other properties of the peptide especially when the peptide binds to a cell membrane [6]. According to the concept, an interface provides a gradient environment to induce specific orientation of the peptide. For example, a helical peptide with primary amphiphilicity will be incorporated into a membrane surface as if the peptide feels a torque force due to the tendency of the hydrophilic part to be exposed to aqueous compartment and the hydrophobic part to be buried in the hydrophobic core of the cell membrane [7]. Under certain conditions, the helical peptide therefore takes a vertical orientation against the surface. An air/water interface is also regarded as gradient environment in terms of amphiphilicity where the hydrophilic part of a primary amphiphilic peptide tends to be distributed to the water subphase and the hydrophobic part to air. On the basis of this idea, several amphiphilic helical peptides were designed and synthesized (Fig. 1) [8–10]. The hydrophilic part of the peptides was either ammonium or a tripeptide of sarcosine (hydrophilic residue) with ammonium. The hydrophobic part was composed of hydrophobic amino acid residues and either a methyl, benzyl, pentafluorobenzyl, or adamantyl group. Leu as well as Ala was used as a component amino acid of the peptides because of favorable intermolecular interaction for obtaining regular molecular assemblies. All the peptides took α-helical conformation in solution as found by CD measurement. These peptides were spread on water, and the orientation of the helical peptides on water was determined by measurement of the π–A curve. A typical π–A curve (Fig. 2) shows a little mound due to phase transition from a liquid state to a solid state [10]. Observation of the peptide monolayer by fluorescence
Generation of a strong dipole layer and its function
255
Fig. 1: Molecular structure of α-helical peptides having primary amphiphilicity for the preparation of a peptide monolayer on water.
256
S. Kimura et al.
Fig. 2: A π – A curve of Boc-( L-Leu-Aib)12 -OBzl spread on water showing the phase transition from a liquid condensed state to a crystalline state. Reprinted with permission from Ref. [10].
microscopy revealed that phase separation took place at the mound in the π–A curve, indicating formation of a two-dimensional crystal of helical peptides. The orientation of helical peptides in the monolayer was determined to be parallel to the water surface, because the molecular area around the phase transition coincides with the molecular sectional area along the helix axis. Some peptides, AdmA16OH and HA16FB, showed a slow slope with a relatively low collapse pressure in the π–A curve, suggesting that the monolayer stayed at a liquid state and no phase transition to a solid state occurred. Another molecular design was made to strengthen the primary amphiphilicity of the peptide by connecting crown ether at the C terminal (Fig. 3). 18-Crown-6-ether is well known to form complex selectively with K+ ions. It was expected that the crown ether part would form a complex on water subphase containing KCl and would become hydrophilic to promote vertical orientation of the hydrophobic peptide part. However, the π–A curves of the peptide–crown ether conjugates indicated parallel orientation to the surface even after complexation with K+ . These results strongly suggest that there should be other factors which influence the orientation of helical peptides on water subphase. After many attempts, the authors have succeeded in obtaining a helical peptide monolayer with a vertical orientation by taking other factors into consideration. Before stating that, however, self-assembled monolayers (SAMs) of helical peptides on gold are explained in the following section to clarify the self-assembling property of helical peptides.
3. Molecular orientation of helical peptide SAM Alkanethiols are well known to form SAMs on gold by anchoring alkane via formation of Au–S bonds. The assembling property of the alkane part is very important for obtain-
Generation of a strong dipole layer and its function
257
Fig. 3: Molecular structure of the helical peptide–crown ether conjugates for the preparation of peptide monolayer on water.
ing a well-packed monolayer free from defects. Helical peptides having lipoic acid at the N terminal, Lipo-(Ala-Aib)8 -OBzl (Lipo and OBzl represent lipoic acid and benzyl ester, respectively; LipoA16B) and Lipo-(Lys(Z)-Aib)8 -OMe (Z and OMe represent benzyloxycarbonyl and methyl ester, respectively; LipoKZ16M), were synthesized and SAMs were prepared on gold (Fig. 4) [11]. The orientation of the helical peptides on gold was determined by FTIR-reflection absorption spectroscopy (RAS). The tilt angles of the helix axes from the surface normal in the LipoA16B and LipoKZ16M SAMs were
Fig. 4: Molecular structure of the helical peptides with a disulfide group for the preparation of peptide SAM on gold.
258
S. Kimura et al.
36° and 55°, respectively. The latter value means that the distribution of helix axes is random. The difference in helix orientation between the two peptides should be ascribed to the difference in the size of the side chain between LipoA16B and LipoKZ16M. The large side chain of LipoKZ16M should be less effective in promoting self-assembling of helices with a vertical orientation. Indeed, the monolayer of Boc-(Lys(Z)-Aib)8 -OMe on water stayed in a liquid condensed state upon compression, while the monolayer of Boc-(Ala-Aib)8 -OMe formed a two-dimensional crystal. The highly self-assembling property of the peptide itself is therefore an important factor for obtaining a SAM with a vertical orientation. The self-assembling property of the peptide is affected by the solvent system in the preparation of the SAM. For example, LipoA16B SAM, which was prepared from an ethanol solution of the peptide, showed a tilt angle of 36°, but that increased to 63° when the SAM was prepared from a N ,N -dimethylformamide solution [11]. N ,N -Dimethylformamide is a good solvent for the peptide to affect the self-assembling property of the peptide probably by favorable solvation to the peptide. Therefore, the good self-assembling property appearing in ethanol should contribute to the vertical orientation of the peptides more than the chemical reaction of a disulfide group with gold. The obvious effect of the chain length of the peptides on the helix orientation in the SAMs was also observed. FTIR-RAS spectra of the SAMs of Lipo-(Ala-Aib)n -OBzl (n = 6, 8, 12) revealed that the tilt angle of the helix axis from the surface normal decreases in the order of Lipo-(Ala-Aib)8 -OBzl > Lipo-(Ala-Aib)10 -OBzl > Lipo-(AlaAib)12 -OBzl. The smallest tilt angle was obtained with the 24mer peptide. The higher self-assembling property of the longer peptide should be due to the larger Van der Waals interaction to promote a tight packing. In the cases of helical peptide SAMs, the orientation of the helix on gold is therefore determined by several factors such as the type of component amino acid, solvent for preparation, and chain length. On the other hand, the helical peptides, even those that took a vertical orientation on gold, lay down on water to form the monolayer. Probably, the formation of an Au–S linkage would help the peptides stand up on gold by using the reaction energy to increase the peptide density as high as possible. What is then the factor which promotes horizontal orientation of hydrophobic helical peptides on water? The orientation of helical peptides on water is generally evaluated from the π–A isotherm comparing the molecular area with the sectional area of the molecule. This method is rather indirect to get information on molecular orientation. To overcome the ambiguity of the evaluation method, we have used in situ FTIR-RAS to determine the molecular orientation because that method is powerful to obtain directly the tilt angle from the surface normal. The principle has been already clarified and reported by several researchers. The application of this method to the helical peptide monolayer is useful and described in the following section.
4. In situ FTIR-RAS measurement on subphase The molecular orientation of helical peptides on subphase can be determined by in situ FTIR-RAS. The tilt angle of the helix axis from the surface normal is, therefore,
Generation of a strong dipole layer and its function
259
theoretically related with the intensity ratio of amide I and II bands in the RAS spectra which are obtained by using polarized incident light (s or p). The theoretical ratio of the band intensities is simulated for the peptide layer with a defined tilt angle, and is compared with the experimental value, giving the tilt angle of the helix axis on subphase. For in situ FTIR-RAS measurements, a trough with a small surface area with a monolayer/grazing angle accessory attached is used, and a wire-grid polarizer is set just before the subphase surface to obtain the polarized incident light. First, the background spectrum of the subphase itself is collected, and then the peptide monolayer is prepared on the subphase. After equilibrium, the sample spectrum is collected and transformed to the absorbance using the background spectrum. Spectral simulation is carried out by using computer programs principally according to the mathematical formalism developed by Ohta and Ishida [12]. This formalism is based on the Abelès matrix method [13] describing stratified layers of homogeneous films, and includes appropriate modifications to interpret absorbing and anisotropic properties of the layers [14]. An optical model of a three-phase (air/monolayer/water) system is considered. The optical property of the jth medium is described by the anisotropic complex refractive index, nˆ j c = n j c + ik j c
(1)
where i 2 = −1, n j and k j are the refractive index and the extinction coefficient of the jth medium, respectively, and c represents x, y, z coordinates. Although air ( j = 0) is a non-absorbing and isotropic medium and water ( j = 2) is also isotropic, the monolayer ( j = 1) has uniaxial symmetry with the surface normal. The respective refractive indices are, thus, expressed as Eq. 2: nˆ 0x = nˆ 0y = nˆ 0z = nˆ 0 = n 0 ,
nˆ 1x = nˆ 1y = nˆ 1z ,
nˆ 2x = nˆ 2y = nˆ 2z = nˆ 2
(2)
In the formalism, the relation between the amplitudes of the electric fields of the incident wave E 0+ , reflected wave E 0− , and transmitted wave after the monolayer E 2+ is expressed as Eq. 3: + E0 C0,1 · C1,2 E 2+ = (3) tˆ0,1 · tˆ1,2 E 0− E 2− with
C0,1 = C1,2 =
exp(−iδ0 ) rˆ0,1 exp(−iδ0 ) exp(iδ0 ) rˆ0,1 exp(iδ0 ) exp(−i δˆ1 ) rˆ1,2 exp(−i δˆ1 ) rˆ1,2 exp(i δˆ1 ) exp(i δˆ1 )
(4)
where rˆj −1, j and tˆj −1, j are the Fresnel reflection and transmission coefficients, respectively, between the ( j − 1)th and jth medium. These coefficients for s- and p-polarized
260
S. Kimura et al.
light are given by Eqs. 5 and 6, respectively: for s-polarized light: rˆj −1, j =
nˆ j −1 y cos θˆj −1 − nˆ j y cos θˆj nˆ j −1 y cos θˆj −1 + nˆ j y cos θˆj
tˆj −1, j =
2nˆ j −1 y cos θˆj −1 nˆ j −1 y cos θˆj −1 + nˆ j y cos θˆj
(5)
for p-polarized light: nˆ j −1 x cos θˆj − nˆ j x cos θˆj −1 2nˆ j −1 x cos θˆj −1 (6) tˆj −1, j = nˆ j −1 x cos θˆj + nˆ j x cos θˆj −1 nˆ j −1 x cos θˆj + nˆ j x cos θˆj −1 The complex refractive angles θˆj in different media are related to the incidence angle θ0 by Snell’s law: for s-polarized light: n 0 sin θ0 = nˆ j y sin θˆj rˆj −1, j =
for p-polarized light:
n 0 sin θ0 = nˆ j z sin θˆj
(7)
The term δˆj represents the phase thickness of the jth medium, and is expressed as Eq. 8: for s-polarized light: for p-polarized light:
δˆj = 2πνd j nˆ j y cos θˆj δˆj = 2πνd j nˆ j x cos θˆj
(8)
where ν and d j represent the wave number of the incident light in the vacuum and the thickness of the jth medium, respectively. δ0 is equal to zero because air is the initial semi-infinite medium. E 2− is equal to zero in Eq. 3 because there is no reflection in the water subphase. By taking the matrix product as Eq. 9: a b (9) C01 · C02 = c d the reflection coefficient (ˆr ) and reflectance (R) of the system are given by Eq. 10: c rˆ0,1 exp(−i δˆ1 ) + rˆ1,2 exp(i δˆ1 ) E− R = |ˆr|2 (10) rˆ = 0+ = = a E0 exp(−i δˆ1 ) + rˆ0,1rˆ1,2 exp(i δˆ1 ) On the other hand, the reflection coefficient (ˆr0 ) and reflectance (R0 ) of the water subphase are easily obtained from the Fresnel formula: n 0 cos θ0 − nˆ 2 cos θˆ2 R0 = |ˆr0 |2 (11) rˆ0 = n 0 cos θ0 + nˆ 2 cos θˆ2 The absorbance of the system is, therefore, given as Eq. 12: R (12) A = − log10 R0 The simulation was performed with taking the refractive index of air being unity. The refractive index and extinction coefficient of water were taken from the literature values [15]. The anisotropic extinction coefficient of the peptide monolayer along the c axis k1c (ν) (c = x, y, z) was calculated by using a uniaxial symmetrical model, in which helical peptides are inclined from the surface normal direction (z) with a tilt angle (φ) and free rotation around the z axis. The transition dipole moments of the amide I and II bands are expressed by two components, parallel (w) and perpendicular (u) to the
Generation of a strong dipole layer and its function
261
helix axis [16]. The wave numbers, maximum extinction coefficients, and peak widths at half-height of the transition dipole moments of the amide I and II bands were taken from literature values which have been reported with those of poly(γ -benzyl-L-glutamate) in a film [17,18]. The maximum values of k1c (ν) of helical peptide with a tilt angle φ from the surface normal are expressed by using the transition dipole moment components (TDM = Iw , Iu , IIw and IIu ) as Eqs. 13 and 14: k1xmaxTDM = k1ymaxTDM = kmaxTDM 13 (1 − f ) + 12 f sin2 α (13) 1 k1zmaxTDM = kmaxTDM 3 (1 − f ) + f cos2 α (14) with the orientational distribution order parameter: f = 12 (3 cos2 φ − 1)
(15)
where α represents the tilt angle of each transition dipole moment component from the helix axis, and this value is 0° for the transition parallel to the helix axis and 90° for the transition perpendicular to the helix axis. Each extinction coefficient over the whole wave number is expressed as an antisymmetrical linear combination of two Lorentzian functions [19]: k1cTDM (ν) =
k1cmaxTDM (fwhh/2)2 k1cmaxTDM (fwhh/2)2 − (ν − ν0 )2 + (fwhh/2)2 (ν + ν0 )2 + (fwhh/2)2
(16)
where fwhh is the peak width at half-height, ν0 is the center wave number of the absorption band. The refractive indices are obtained from the Kramers–Kronig transformation of Eq. 16: n 1cTDM (ν) = n ∞ −
k1cmaxTDM (ν − ν0 )(fwhh/2) k1cmaxTDM (ν + ν0 )(fwhh/2) + (ν − ν0 )2 + (fwhh/2)2 (ν + ν0 )2 + (fwhh/2)2
(17)
where n ∞ is the constant refractive index in the near-infrared region and was set to be 1.50 for the simulations. The sum of the extinction coefficients and refractive indices for the each transition dipole moment gives the complex refractive index of the peptide monolayer along the c axis: k1c (ν) = k1cIw (ν) + k1cIu (ν) + k1cIIw (ν) + k1cIIu (ν) n 1c (ν) = n 1cIw (ν) + n 1cIu (ν) + n 1cIIw (ν) + n 1cIIu (ν) nˆ 1c (ν) = n 1c (ν) + ik1c (ν)
(18)
The simulated RAS spectra of a helical peptide monolayer for s- and p-polarized light with varying the tilt angle of the helix axis from the surface normal are shown in Figs. 5 and 6, respectively. These simulations were performed with taking the incidence angle being 65° because a larger angle makes the reflected light from the water surface stronger and 65° is the maximum value possible for our experimental apparatus. Under this condition, the reflectance of s-polarized light is about 10 times larger than that of p-polarized light. In order to increase the sensitivity, measurements using s-polarized light were, therefore, chosen for the determination of the molecular orientation. The relationship between the intensity ratios of the amide I and II bands and the tilt angles
262
S. Kimura et al.
Fig. 5: The simulated in situ FTIR-RAS spectra in the region of the amide I and II bands of a helical hexadecapeptide monolayer with a thickness of 3 nm in the case of s-polarized light being used for the measurement. The numbers in the figure represent the tilt angles of the helix axis from the surface normal. Reprinted with permission from Ref. [20].
Fig. 6: The simulated in situ FTIR-RAS spectra under the same conditions described in Fig. 5 except p-polarized light is used for the measurement.
Generation of a strong dipole layer and its function
263
Fig. 7: The theoretically calculated relationship between the tilt angle of helical peptides from the surface normal and the intensity ratio of the amide I and II bands in in situ FTIR-RAS measurements on water subphase using s-polarized light. Reprinted with permission from Ref. [20].
for s-polarized light is shown in Fig. 7. The thickness of the peptide monolayer in the simulations is set to be 3.0 nm, which does not influence the ratio of the band intensities.
5. Helical peptide monolayer spread at limited area When Boc-(Leu-Aib)n -OBzl (n = 8, 12, 16) were spread on water initially in the gas phase, they took a parallel orientation even after compression. The conventional method for preparation of monolayers was not successful for obtaining helical peptide monolayers with a vertical orientation even though hydrophobic peptides with a cylindrical shape were used. However, the peptides took a vertical orientation when the peptide was deposited in the liquid phase instead of the gas phase in the preparation of the monolayer from the beginning. For example, the intensity ratio of the amide I and II bands in an in situ FTIR-RAS spectrum was 1.47 under the condition that Boc-(Leu-Aib)8 -OBzl was spread at the molecular area of 1.5 nm2 . According to the relationship between the intensity ratio and the tilt angle, Boc-(Leu-Aib)8 -OBzl was found to stand up on water with a tilt angle of 28°. The helical peptides may be forced to take a vertical orientation and to be packed as densely as possible during the evaporation process of the spreading solvent [20].
264
S. Kimura et al.
Among the three peptides, Boc-(Leu-Aib)8 -OBzl stood up more vertically on the subphase than the other longer peptides. For example, the tilt angle of Boc-(LeuAib)16 -OBzl monolayer was 39°, which is larger than the 28° of Boc-(Leu-Aib)8 -OBzl monolayer. In the preparation process, a large number of rod-shaped molecules were spread over a limited area. Under this condition, strong intermolecular interaction of the longer peptides may work locally among the peptides, which may hinder an ordered alignment of the peptides in a wide area of the monolayer.
6. Helical peptide monolayer spread on a mixture of water and methanol Helical peptides favorably took a parallel orientation to the interface, indicating that there must be an important factor which contributes to the parallel orientation. Molecules with dipole moment deposited on gold or water should experience an electrostatic attractive force as if there were an image dipole at the same distance on the other side of the interface. The attractive force should act to bring the parallel orientation, because the interaction energy with the image dipole becomes larger with decreasing the distance of the dipole from the interface. However, this interaction energy will become smaller with lowering the dielectric constant of the substrate or subphase, because the image dipole is proportional to (εs − εm )/(εs + εm ) (εs and εm represent dielectric constants of substrate or subphase and monolayer, respectively). Therefore, using subphase having a lower dielectric constant may promote the formation of vertically oriented monolayer. On the basis of this idea, helical peptides were spread on a mixture of water and methanol (1/1 v/v). The tilt angles of helical peptide monolayers spread on a mixture of water and methanol were 23° for Boc-(Leu-Aib)8 -OBzl and 33° for Boc-(Leu-Aib)16 -OBzl, which indeed became smaller than 28° and 39° for the respective peptide spread on pure water. Even though other factors such as surface tension may change upon the replacement of the subphase and influence the orientation, the reduction of the dipole–image-dipole interaction is considered to contribute significantly to the vertical orientation of the helical peptides on the subphase. This interpretation is also applied successfully to the case of multilayer formation on gold.
7. Multilayer formation on gold The helical peptide monolayers on subphase were transferred on gold by evaporation of the subphase which was directly put on gold [20]. With repeating the evaporation of the subphase deposited with a peptide monolayer, multilayers of helical peptides were obtained on gold as shown by the increase of the amide I band intensity by FTIR-RAS measurement. The intensity ratio of the amide I and II bands measured on gold was used to determine the tilt angle of the helix from the surface normal [21]. The effect of the subphase solution on the tilt angle was significant. The average tilt angle of 10-layer Boc-(Leu-Aib)8 -OBzl was 17° by using a mixture of water and methanol as subphase, whereas 46° by using water as subphase. The interaction between the dipole and the
Generation of a strong dipole layer and its function
265
image dipole should be diminished by using a mixture of water and methanol to induce the vertical orientation. The tilt angle was dependent on the number of helical peptide layers on gold. The tilt angles for the monolayer and the 5-layer Boc-(Leu-Aib)8 -OBzl are 45° and 26°, respectively. When the helical peptide was transferred directly on gold, the vertical orientation of the peptide monolayer on the subphase was affected probably because of the close location of the peptide to gold. However, in the case of a thick membrane, the intervening peptide layer should weaken the interaction of dipole of the transferring peptide layer with the image dipole in gold to promote the vertical orientation. The driving force to align the peptides horizontally therefore becomes insignificant for the thick multilayer.
8. Surface potential of helical peptide SAM The helical peptide SAMs with a vertical orientation are theoretically indicated to generate a surface potential due to the large dipole moment. The surface potentials of LipoA16B and BA16Lipo SAMs in vacuum were measured by the Kelvin probe method to be about −120 mV and 400 mV, respectively [22]. LipoA16B was connected to gold via the N terminal and exposed the C terminal to the outside to generate the negative surface potential. On the other hand, the positive surface potential was generated by exposing the positively charged N terminal to the surface. Therefore, the helix peptide layer with a parallel and vertical orientation indeed generates a surface potential due to the dipole moment. However, the observed values were lower than the calculated values based purely on the dipole moment. The difference is successfully explained by two ways depending on the direction of the dipole moment in the SAM. In the case of the peptide SAMs immobilized via the N terminal, the effects of the mutual depolarization and the ionic property of Au+ –S− linkage should work on the surface potential to reduce the absolute values. Whereas, the peptide SAMs immobilized via the C terminal may promote electron transfer from gold to the peptide layer due to the large positive surface potential. The generation of the surface potential based on the dipole moment should be compromised by the electron transfer. The effect of the electron transfer should become larger with increase of the dipole moment. Indeed, similar positive surface potentials were observed independently of the peptide layer thickness as long as the dipoles are aligned to point to the vacuum side.
9. Photocurrent generation Helical peptides may mediate electrons through hydrogen bonds with combination of the through space mechanism. Such electron transfer should be accelerated in the presence of the dipole moment. Indeed, Fox et al. reported that the electron transfer from N ,N dimethylaniline to pyrene was faster along the dipole moment of a helical peptide than that against the dipole moment [23]. A helical peptide carrying an N -ethylcarbazole group at the C terminal and a disulfide group at the N terminal was synthesized, and a
266
S. Kimura et al.
Fig. 8: Schematic illustration of photocurrent generation by the helical peptide SAM. Anodic current in the presence of an election donor in the aqueous phase was observed with the peptide SAM where the dipole moment is directed to the gold.
peptide SAM was prepared on gold. Interestingly, an anodic photocurrent was observed in the presence of an electron donor (ethylenediamine tetraacetate or triethanolamine) in the aqueous phase (Fig. 8), but a cathodic photocurrent was not clearly detected upon photooxidation of the N -ethylcarbazole by an electron acceptor (methylviologen) in the aqueous phase [24]. The situation was reversed when the helical peptide carrying an N ethylcarbazole group at the N terminal was immobilized on gold through the C terminal. A cathodic photocurrent was observed in the presence of methylviologen in the aqueous phase, but an anodic photocurrent was small in the presence of triethanolamine in the aqueous phase. The direction of the electron transfer coincides with that of the dipole moment, indicating that the dipole moment is crucial for determining the direction of the photocurrent.
10. Future aspects A surface potential of a few hundred mV was generated by helical peptide SAMs having a few nm thickness. The electric field therefore amounts to more than 106 V/cm. Due to the large electric field, the electron transfer through the helical peptide SAM is accelerated or hindered depending on the direction of the electron transfer for or against the electric field in the case of photocurrent generation using the helical peptide SAM in aqueous solution. On the basis of these results, further study is progressing on electric
Generation of a strong dipole layer and its function
267
properties of a single helical peptide immobilized on gold. The helical peptide SAMs are now subjected to scanning probe microscopies such as STM, STS, and KFM. The effect of the dipole moment of the helical peptide on the electron transfer through a single peptide molecule is now under investigation. The helical peptide may be a good candidate for a molecular device which shows a diode property.
References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24.
A. Wada, Adv. Biophys. 9, 1 (1976). W.G. Hol, Prog. Biophys. Mol. Biol. 45, 149 (1985). S. Kimura, In Handbook of Surfaces and Interfaces of Materials, edited by H.S. Nalwa, (Academic Press, San Diego, 2001) pp. 207–231. F. Takeda, M. Matsumoto, T. Takenaka, and Y. Fujiyoshi, J. Colloid Interface Sci. 84, 220 (1981). R. Schwyzer, Biopolymers (Peptide Science) 37, 5 (1995). D.F. Sargent and R. Schwyzer, Proc. Natl. Acad. Sci. USA 83, 5774 (1986). R. Schwyzer, Biochem. 25, 6335 (1986). K. Fujita, S. Kimura, Y. Imanishi, E. Rump, and H. Ringsdorf, Langmuir 10, 2731 (1994). K. Fujita, S. Kimura, Y. Imanishi, E. Okamura, and J. Umemura, Langmuir 11, 1675 (1995). K. Fujita, S. Kimura, Y. Imanishi, E. Rump, and H. Ringsdorf, Langmuir 11, 253 (1995). Y. Miura, S. Kimura, Y. Imanishi, and J. Umemura, Langmuir 14, 6935 (1998). K. Ohta and H. Ishida, Appl. Optics 29, 1952 (1990). F. Abeles, Ann. Phys. 3, 504 (1948). W.N. Hansen, J. Opt. Soc. Am. 58, 380 (1968). J.E. Bertie and M.K. Ahmed, J. Chem. Phys. 93, 2210 (1989). T. Miyazawa, J. Chem. Phys. 35, 693 (1961). T. Buffeteau, E. Le Calvez, B. Casteno, B. Desbat, D. Blaudez, and J. Dufourcq, J. Phys. Chem. B 104, 4537 (2000). T. Buffeteau, E. Le Calvez, B. Cesbat, I. Pelletier, and M. Pezolet, J. Phys. Chem. B 105, 1464 (2001). K. Ohta and H. Ishida, Appl. Spectrosc. 42, 952 (1988). K. Kitagawa, T. Morita, J. Umemura, and S. Kimura, Polymer 43, 3533 (2002). Y. Miura, S. Kimura, Y. Imanishi, and J. Umemura, Langmuir 15, 1155 (1999). Y. Miura, S. Kimura, S. Kobayashi, M. Iwamoto, Y. Imanishi, and J. Umemura, Chem. Phys. Lett. 315, 1 (1999). E. Galoppini and M.A. Fox, J. Am. Chem. Soc. 118, 2299 (1996). T. Morita, S. Kimura, S. Kobayashi, and Y. Imanishi, J. Am. Chem. Soc. 122, 2850 (2000).
Part D
Interfacial Dynamic Technology
Nanotechnology and Nano-Interface Controlled Electronic Devices Editors: M. Iwamoto, K. Kaneto and S. Mashiko © 2003 Published by Elsevier Science B.V.
CHAPTER 15
Guided mode studies of liquid crystal layers Fuzi Yang a and J.R. Sambles b a Liquid
Crystal Research Centre, Chemistry Department, Tsinghua University, Beijing 100084, China b Thin Film Photonics, School of Physics, University of Exeter, Exeter, EX4 4QL, UK
1. 2.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Optical guided waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Optical waveguide modes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. The field distributions of optical guided modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3. Coupling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. Liquid crystal waveguide geometries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Fully guided geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Fully leaky geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3. Half-leaky guided mode geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4. Improved fully leaky guided mode geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. Dynamic guided mode technique . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
271 274 274 277 282 286 288 291 293 301 306 308 310 310
1. Introduction Although the liquid crystalline state of matter has been recognised for over 100 years [1,2] the explosive growth in the application of such materials as primary components in flat panel displays having low power consumption and compact dimensions has occurred only during the last 30 years. The worldwide demand for flat panel displays is huge and continues to drive further scientific investigations in liquid crystal (LC) science and technology. This has resulted in developments in materials synthesis giving rise to novel materials and new discoveries in the fundamental science of liquid crystal phases. In addition there has been substantial new device structure development strongly pushed by requirements from the display market. Liquid crystals have the ability to flow while displaying anisotropic properties. They respond to (realign in) externally applied electric (or magnetic) fields. It is
272
F. Yang and J.R. Sambles
this response, coupled with the optical anisotropy, which has led to applications in flat panel display technology, including alpha-numeric displays, high resolution TV, data projection systems and monitors for desk and laptop computers, etc. They also change their behaviour quite markedly with temperature and they are readily aligned by molecular scale surface structures. Consequently LC materials are also being used as sensors for temperature, stress and flow, etc. For many of these applications, especially for displays, the optical behaviour of liquid crystals is vitally important and the optical response of a given LC geometry to changing conditions is a major issue. It may, for example, be important to know how some parameters of a liquid crystal, such as refractive index, dielectric permittivity, viscosity and elasticity coefficients, etc., change with temperature, or how the apparent optical response of a thin LC layer changes under application of a field. From both a fundamental and a device perspective we may wish to investigate how the response of LC layers vary with surface treatment, the specifics of the cell geometry, or the elastic and/or viscous properties of the material. Knowledge of the form of the LC director profile in cells and subsequently its change with applied field is essential for developing a full understanding of device function from which further devices and applications may arise. For example for fast devices details of the electroclinic coefficients may be required, or viscosities may be needed. This may demand determining both the static and dynamic director response. In order to characterise fully the optical properties of a given liquid crystal, in a given cell geometry, with defined thickness and boundary treatments it is obvious that some form of appropriate structural study needs to be undertaken. In many liquid crystal research laboratories the most common method used to explore the optical structure within a LC cell is that of polarised microscopy [3]. This provides a quick and simple procedure for exploring the approximate director profile allowing the study of cell uniformity, defect structures, phase transitions, the influence of aligning layers and also gives much of information on voltage controlled switching processes. However, by its very nature, polarised microscopy is an integral technique, which, through the transmitted intensity, provides an integrated optical response through the cell as a whole. Thus it cannot readily be used to resolve details of the director profile through the cell thickness or its variation with time during switching. A second procedure, often used for the study of liquid crystals structures, even with thin cells (< 10 µm), is X-ray scattering [4]. In bulk materials the X-ray scattering gives the symmetry of the phase and in thin layers it is primarily used to explore only the density wave layering found in SA , SC , S∗C and other more ordered phases, for example in the study of ferroelectric LC materials [4]. This method may then give very useful information on the elastic deformation of the density wave, the layers, but it says almost nothing directly as regards the optical properties of the LC cell, which are vitally important for display applications. At the same time as new display technologies using liquid crystalline materials were being developed in the early 1970s, new optical techniques were being introduced in the broad area of light guiding. This is particularly obvious in the area of optical fibre communications. Accompanying this development of optical fibre technology light guiding in thin films and layered structures has also received considerable attention both for integrated optics and other applications. During these developments a new optical
Guided mode studies of liquid crystal layers
273
probing technique, the guided mode technique [5,6], for investigating the optical index profiles in guiding structures has materialised. This technique utilises the fact that a series of discrete (quantised waveguide momentum) modes may be excited in thin films or fibres, their mode spectrum being dependent upon the refractive index distribution in, and the dimensions of, the structure. By exploring in depth the guided mode spectrum it is possible to characterise in some detail the optical properties of thin films or fibre guides. At the heart of LC displays are liquid crystals in the form of a thin layer, with a thickness of the order of a few microns with a director structure which varies little laterally, but may vary substantially through the cell. This layer is sandwiched between two glass plates (the two cell walls) which have some transparent conductive coatings (e.g. ITO) and liquid crystal alignment layers on them. Thus although most liquid crystal display structures are not designed as optical waveguides they often are just that, albeit rather ‘leaky’ ones. So in parallel with the development of new liquid crystal devices we have seen the evolution of a range of optical waveguide characterization techniques which allow the probing of the details of the director structure in these devices. It is these new techniques which are the primary focus of this chapter and which will be discussed later in detail. In the early 70s, both guided wave optics (often narrowly labelled integrated optics) and liquid crystal studies were in their infancy and, based upon the material syntheses and the requirements of display devices, much work centred simply on optical microscopy of the nematic mesophase of liquid crystals. Theoretical modelling of the nematic phase was developed quite early and the director profile in waveguides made of such nematics is quite simple, even in the case of finite surface tilt and twist in the cell. Because of this, even though the optical guided mode technique has great potential as a tool for studying both physical and chemical processes in thin films [7] there have been few such investigations of nematic liquid crystal cells. Thus in addition to polarised microscopy only the monitoring of the transmitted intensity for normal incidence light together with modelling the optical response of liquid crystal cells based upon 2 × 2 Jones’ matrix are used as standard procedures in most liquid crystal research laboratories. Often liquid crystal ‘waveguides’ are used as devices, as displays, in modulators [8,9], switches [10] or deflectors [11] but with limited studies of details of the optical tensor distribution in cells except for a few measurements of refractive indices and surface pre-tilt angle. However, several new areas of liquid crystal research have led to the guided mode technique becoming more and more useful and important. Firstly, in the early 80s the prediction [12] and experimental observation [13] of the surface stabilised ferroelectric, S∗C , liquid crystal (SSFLC) state introduced a challenge to unravel and understand the complex optical tensor profile in such LC cells. In view of the fact that thin cells containing S∗C liquid crystals may have many applications in fast optical switching, displays and TV and also because currently, unlike in the nematic case, there is still not a convenient theoretical model, then the experimental unravelling of the director structures in such cells is vital. Secondly, even for some simpler phases, such as nematic or SA phases, the director structure through the whole cell is not necessarily straightforward. For example there may exist coexisting nematic and SA phases in a twisted liquid crystal cell [14] in which the director profile through the LC cell
274
F. Yang and J.R. Sambles
is stepped in several sections, which can only be found in detail by the guided mode techniques. Any other method, such as polarised microscopy or monitoring the transmission intensity from normal incidence, cannot distinguish this stepped structure. Also from theoretical modelling and experimental demonstration the method based on monitoring the transmission intensity from normal incidence and 2 × 2 Jones’ matrix which treats the LC cell as a simple slab may sometimes introduce serious errors in the measured parameters [15]. By contrast the guided mode technique together with theoretical modelling based upon the 4 × 4 Berreman’s matrix method, treating the LC cell as a real multi-layer optical system will give the correct results. Finally, in dynamic studies of liquid crystal films the director profile and its change with time and field are very complex. The serious profile degeneracy problems associated with the usual optical procedures may only be solved by use of the recently developed dynamic LC guided mode technique [16,17]. In this chapter we first introduce briefly the background of the optical guided mode technique, including the guided mode spectrum, the optical field distribution for different order modes and various optical coupling methods to couple the radiation into the waveguides. Then we illustrate four different types of liquid crystal waveguide geometry, including the fully guided mode geometry, the fully leaky mode geometry, the half-leaky guided mode geometry and the improved fully leaky guided mode geometry. Finally we briefly discuss the dynamic LC guided mode technique. Various experimental results obtained recently using different types of LC optical guided mode techniques are presented to show the power of the techniques.
2. Optical guided waves [18] 2.1. Optical waveguide modes To introduce the background to optical waveguides let us first consider the simplest geometry comprising a planar slab of perfectly transparent isotropic dielectric surrounded by semi-infinite blocks of perfectly transparent isotropic dielectric. The geometry is invariant in the plane normal to the page, i.e. the y–z plane as shown in Fig. 1. The three materials in the waveguide geometry have been labelled as the cladding layer, of index n c , the guiding layer, of index n g and the substrate layer of index n s . For full-
Fig. 1: The geometry of a planar optical waveguide, with cladding area, guiding layer and substrate area.
Guided mode studies of liquid crystal layers
275
Fig. 2: Guided optical modes in a planar waveguide: (a) full guided mode, (b) substrate radiation mode, and (c) substrate-cladding radiation mode.
waveguiding the situation of n g > n s and n g > n c is required. For the ensuing discussion we further suppose n s > n c , unless explicitly stated otherwise, so n g > n s > n c . A ray optics picture is first used here to describe the optical modes [5] propagating in the planar waveguide. The situation to be considered is that of a plane wave of radiation, with an incident angle β (angle between the wavefront normal and the normal to the boundary of the geometry) to the interface, propagating inside the guiding layer as shown in Fig. 2. According to Snell’s Law (basically conservation of momentum in the y–z plane) the critical angles at the top and bottom interfaces of the geometry are given by nc (1) βc = sin−1 ng and −1
βs = sin
ns ng
(2)
where βs > βc . In Fig. 2 there are three different zigzag pictures for different ranges of the internal angle β. When βs < β < π/2 the light is primarily contained inside the guiding layer by total internal reflection at both the top and bottom boundaries. According to the ray optics picture the light propagates along the zigzag path shown in Fig. 2a. This case corresponds to a fully guided mode. Even if the ray is totally inside the guiding layer, the optical radiation field is not entirely constrained to the slab since there are evanescently decaying fields present in both the substrate and cladding layers. However, if there are no imaginary parts to the refractive indices of the two layers the optical energy flow of the modes will be strictly along the z-direction and no radiation propagates away in either of the semi-infinite blocks. Secondly when βc < β < βs , as shown in Fig. 2b, only one interface, the cladding interface, acts as a totally reflecting surface, part of the radiation energy of the modes escapes into the substrate area. This situation corresponds to a substrate radiation mode, and it is obvious that the radiation energy rapidly leaks out of the guiding layer. This type of waveguide may be labelled as a half-leaky waveguide – it only leaks radiation into one half space of the geometry.
276
F. Yang and J.R. Sambles
Finally for β < βc the radiation energy of the modes will leak into both the substrate and cladding half spaces across both interfaces as shown in Fig. 2c. Thus this is labelled a fully leaky waveguide and the confinement of radiation within the guided layer is quite weak. In all three cases described above, but primarily in the first case of fully guiding, since the light is propagating either entirely or partially in the guiding layer, then exploring modes excited in the geometry will give information on the optical properties of the guiding layer. Different order modes, which propagate with different momenta along the z-direction in the waveguide geometry, will have different numbers of optical field maxima across (in the x-direction) the guiding layer. Both the in-plane momentum and the optical field distribution of different order modes are dependent upon the thickness and refractive index profile of the guiding layer. Thus each mode will have a different sensitivity to particular portions of the guiding layer. This idea will be discussed later. For the moment the guiding condition, i.e. the reason for creating different order modes, will be explored in more detail. For the fully guiding situation, even though the saw-tooth ray picture clearly gives a description of a mode with oblique-up and oblique-down beams, from the wave-nature of the light a sustainable mode propagating in the guiding layer is only one that does not destructively interfere with itself. Thus there will only be a few angles of propagation for the zigzag which give correct constructive interference, i.e. there can only be a finite set of waveguide modes for a given guiding geometry. Let us consider only the component of momentum normal to the waveguide plane, that is in the x-direction. One might at first anticipate a very simple condition for constructive interference, that is mπ (3) kx = d where m is an integer and d is the thickness of the guiding layer. In Eq. 3, however, the phase shifts at the top and bottom interfaces are ignored. Incorporating correctly the phase shifts (−2Φgs ) and (−2Φgc ) at the substrate and cladding interfaces a more general equation can be given as 2kx d − 2Φgs − 2Φgc = 2mπ
(4)
Of course kx is simply given through the relationship kx = k0 n g cos β
(5)
where k0 is 2π/λ and λ is the light wavelength in free space. If we join Eqs. 4 and 5 together, then an even stricter and preferred form of the equation involving the waveguide momentum, kz , can be given as 1/2 mπ + Φgs (β) + Φgc (β) 2 2 2 (6) k z = k0 n g − d where 2Φgs and 2Φgc are also functions of k0 itself. This equation still gives a discrete set of guided modes but it is clearly not that trivial to solve. We will take this equation a stage further by replacing Φgs (β) and Φgc (β) by the appropriate functional forms
Guided mode studies of liquid crystal layers
277
found from Fresnel equations. According to the Fresnel equations it is clear that both Φgs (β) and Φgc (β) will be polarisation sensitive and it is generally the situation that there will be two families of discrete momentum modes, one set transverse electric (TE or s-polarised) and one set transverse magnetic (TM or p-polarised). For the situation of both half-leaky and fully leaky waveguides these standing wave mode solutions, excited at specific in-plane momenta, are no longer so well defined. In principle leaky modes exist for any momentum within the waveguide provided it is less than the cutoff momentum corresponding to the critical angle defined by βs . Nevertheless, depending on the refractive index profile of the guiding layer, and the geometry of the waveguide, there will still be some selected in-plane momenta at which there are stronger optical fields confined in the guiding layer. 2.2. The field distributions of optical guided modes [5,6] In order to show how different order guided modes are sensitive to different spatial regions of a waveguide wave optics should be used and optical field profiles in a slab waveguide need to be explored. From Maxwell’s electromagnetic theory the two important equations in isotropic (non-magnetic) lossless dielectrics are ∂ H(r, t) (7) ∂t ∂ E(r, t) (8) ∇ × H(r, t) = ε0 n 2 ∂t where ε0 and µ0 are the dielectric permittivity and magnetic permeability of free space respectively and n is the refractive index of the dielectric. When a plane wave propagates along the z-direction (Fig. 1) with the propagation constant γ (= kz ) then the electromagnetic field may be expressed as ∇ × E(r, t) = −µ0
E = E(x, y) exp[i(ωt − γ z]
(9)
H = H(x, y) exp[i(ωt − γ z] (10) which combined with Eqs. 7 and 8. Being aware that E z = Hz = 0, ∂/∂t ≡ iω; ∂/∂ z ≡ −iγ ; ∂/∂y = 0, gives us two independent solutions of the form ∂2Ey + [k02 n 2 − γ 2 ]E y = 0 (11) ∂x2 ∂ 2 Hy + [k02 n 2 − γ 2 ]H y = 0 (12) ∂x2 These two equations are the solutions for transverse electric and transverse magnetic optical fields, respectively. To obtain the optical field distribution in the waveguide appropriate boundary conditions have to be imposed to give E y and H y as functions of x. The boundary conditions are conservation of tangential E and H and also conservation of normal D and B. For the waveguide geometry mentioned above the fields are not zero in the surrounding media, they simply decay exponentially into these two regions. (This, in effect, is identical in form to the quantum mechanical boundary conditions for a
278
F. Yang and J.R. Sambles
non-infinite potential well.) For the TE case E y and ∂ E y /∂ x are continuous across the boundary while for the TM case H y and ∂ H y /∂ x are continuous across the boundary. The phase shifts at the interfaces are introduced by these continuity conditions leading to the waveguide equation Eq. 4, where, as before, m is an integer running from 0, 1, etc., and of course the phase factors are different for TE and TM modes. In the TE case −1 δs (13) Φgs = tan kx δc (14) Φgc = tan−1 kx and for the TM case n g 2 δs −1 (15) Φgs = tan n s kx 2 n δ g c Φgc = tan−1 (16) n c kx where
1/2 δs = γ 2 − k02 n 2s
(17)
and
1/2 δc = γ 2 − k02 n 2c
(18)
are the exponential decay coefficients for the evanescent fields in the substrate and cladding media, respectively. From Eqs. 17 and 18 it is clear that as γ continues to diminish (β continuously becoming smaller) δs and δc will become imaginary and one moves over to the ‘leaky’ situation. When γ < k0 n s the situation becomes a continuous spectrum in γ rather than the discrete values found for the trapped modes. The various field profiles of the electromagnetic modes for different ranges of the propagation constant (in-plane momentum) are shown in Fig. 3 for the TE case. As expected it is clear that in the guided mode range the optical fields are concentrated in the guiding
Fig. 3: The optical electric field profile, perpendicular to the mode propagation direction, for different TE modes propagating in a planar waveguide with different propagation constants. (a) substrate-cladding radiation mode, γ < k0 n c , (b) substrate radiation mode, k0 n c < γ < k0 n s , (c) guided mode with k0 n s < γ < k0 n g .
Guided mode studies of liquid crystal layers
279
Fig. 4: The field distributions for TE modes, produced by Eqs. 11 and 12, with mode orders of m = 0, m = 1 and m = 2. The corresponding ray optics models are also shown in each case.
layer with evanescently decaying fields in the cladding and substrate areas. In the radiation mode range the optical fields of course propagate out into the cladding and substrate media. For the guided mode situation the field profiles produced from Eqs. 11 and 12 for TE modes of order 0, 1 and 2 with the equivalent ray optics model are shown in Fig. 4. From Eq. 6 it is apparent that the fundamental mode, with m = 0, has the largest propagation constant γ (= kz ), close to the limit value of a plane wave in the guiding layer, n g k0 . This mode, in the ray picture, has correspondingly the largest internal angle, β, close to π/2 and the longest ‘wavelength’ in the x-direction. For the TM modes the situation is much the same. The next order mode, with m = 1, has a smaller propagation constant than that of the mode m = 0 and will generally have zero optical field near to the centre of the guiding layer as shown in Fig. 4. Thus it is clear that different order optical guided modes will have different field distributions through the guiding layer. As shown in Fig. 4 the zero order mode will be much more sensitive to the centre of the guiding layer than would be the first order mode. Therefore it would be not difficult to see that this guided mode technique applied to the study of a liquid crystal layer having a complex director distribution through a cell may allow discrimination of details of the optical tensor varying through the layer. This information would be unobtainable by integral (polarised microscopy, etc.) techniques. According to the mode equation Eq. 4 it is also abundantly clear that the thicker the guiding layer the more modes it will support. In addition more modes can also be excited for shorter probing wavelengths of the incident light unless there is very strong optical dispersion of the guide indices with wavelength. Further details of optical waveguide theory and technology may be found in a range of review articles and books [5,6]. Up to now our discussion concerning optical waveguides has been limited to isotropic, lossless and uniform materials, however, in an optical waveguide having
280
F. Yang and J.R. Sambles
Fig. 5: Geometry for an isotropic–uniaxial anisotropic–isotropic system.
liquid crystals as the guiding layer it is highly unlikely that these assumptions will hold. Of the three assumptions that of low loss is generally not too worrying. If the losses are not too large, as in liquid crystals in the optical range, they only change the mathematics somewhat and make otherwise infinitely sharp modes available for excitation and detection. However, the non-uniformity is of fundamental interest and it is the anisotropy which most significantly affects the discussion thus far. These two points are briefly discussed in the following. In general the eigenmodes of a waveguide geometry having an anisotropic guiding layer will not be pure TE or TM polarisations except for some special, high-symmetry structures. Thus the complex eigenmode solutions can not be readily be simply analysed using Maxwell equations, even though some methods based upon electromagnetic field theory have been developed [19] to try to solve this problem. Here we will explore a few situations which allow us to illustrate the behaviour of an anisotropic waveguide. Let us restrict the discussion to just uniaxiality in the guiding layer itself. Consider a uniformly aligned uniaxial layer of liquid crystal as shown in Fig. 5 surrounded by semi-infinite cladding and substrate media with isotropic refractive indices n c and n s , respectively. The liquid crystal is specified by indices parallel and perpendicular to the director, the optical axis N, n e and n o , respectively, and we assume n e > n o , i.e. the liquid crystal has positive anisotropy. For the general situation the director of the liquid crystal layer is tilted by θ from the x-axis and twisted by ϕ from the xoy plane as shown in Fig. 5. The line AO is the wave-front normal for the eigenmode in the liquid crystal. One of the two semi-axes (OF and OB) of the ellipse, formed by the intersection of the plane perpendicular to the wave front AO and the index ellipsoid of the liquid crystal, gives the refractive index for this eigenmode propagating in the uniaxial layer. Firstly, choosing the very special case of the optic axis along the z-axis then, from Fig. 5, it is simple to see that the TE guided modes depend only on n o , while the TM eigenmodes
Guided mode studies of liquid crystal layers
281
depend on both n e and n o , with the lower order modes (with high β) depending mostly on n o . Thus for this simple case the effective index for the TM eigenmodes changes with mode order. If now the optic axis is tilted so that it is still in the x–z-plane nothing fundamentally changes. The TE eigenmodes will still sense n o while the TM eigenmodes sense a different combination of n e and n o . Secondly, if the optical axis lies instead along the x-axis, then the lowest order TM mode will sense n e with the higher order modes becoming more sensitive to n o . Of course all order TE modes will still sense n o . In practice this means that when the optic axis is along the z-axis the TE and TM modes will have the same upper momentum limit, while for any other tilt of the optic axis in the x–z-plane the limit of the TM modes will, for n e > n o , move above the TE limit. Thirdly, if the optic axis lies along the y-axis, which is the simplest case for the optical axis lying out of the incidence plane xoz, now the TE modes are given by n e while the TM modes are given by n o , a very simple situation. For these three special cases mentioned above the eigenmodes propagating in an anisotropic waveguide geometry are pure TE or pure TM modes, even though their propagation constants may vary. However, as soon as the optic axis is rotated out of the y-axis to some arbitrary angle in the x–y plane, or to some arbitrary angle in the y–z plane, or both, i.e. the optical axis is at a general position in the frame as shown in Fig. 5, the eigenmodes are no longer pure TE and TM. Thus an experimental investigation of such a system using radiation of a given linear polarisation, either TE or TM, will lead to polarisation conversion, the output radiation having some of the orthogonal polarisation component present. The signals from the polarisation conversion are very useful for investigating the director structure of liquid crystal waveguides since they are so clearly sensitive to both tilt and twist of the optical axis (the director) out of the plane of incidence of the exciting radiation. Some analytical explanations can also be provided for the situation of polarisation conversion in the uniaxial guiding layer [20,21]. The geometry of Fig. 5 gives cos ψ = cos θ cos β − sin θ sin β sin ϕ
(19)
where ψ is the angle between the optical axis (the director) and the wave-front normal of the eigenmode in the uniaxial layer. The index of the extraordinary eigenmode n e (βe ) is defined by the semi-major axis OF of the ellipse formed by the intersection of the plane perpendicular to the wave front AO and the index ellipsoid of the uniaxial layer. From Fig. 5 this gives none . (20) n e = n 2o sin2 ψ + n 2e cos2 ψ Of course the extraordinary eigenmode index lies within the range n e ≥ n e ≥ n o and will correspond to a pure TE or a pure TM eigenmode for the three special situations mentioned above. In Fig. 5, s-polarised radiation has its E field along the y-axis, whereas p-polarised radiation has its E field in the x–z plane. For either form of incident radiation, two eigenmodes are excited in the uniaxial layer, one with the E field along the short semi-axis OB in the ellipse BOF, and normal to the plane AON, and a second with the E field along the major semi-axis OF in the ellipse BOF, and normal to OB.
282
F. Yang and J.R. Sambles
So the angle Ω between Oy and OB gives a measure of the s-to-p or p-to-s, conversion signal when either pure s or pure p radiation enters the uniaxial layer. From Fig. 5 we obtain sin β cos θ + sin θ cos β sin ϕ (21) cos Ω = 1 − (cos θ cos β − sin θ sin β sin ϕ)2 Obviously, only when Ω equals 0 or π/2 is there no polarisation conversion. This corresponds to one of three special cases; (1) ϕ = π/2, i.e. the optic axis lying in the incidence plane xoz, (2) θ = 0, i.e. the optic axis being along the x-axis, and (3) ϕ = 0, θ = π/2, i.e. the optic axis being along the y-axis. These are exactly the situations mentioned above. Of course the amplitudes and the phases of different order guided modes, propagating in the uniaxial guiding layer are more complex than described above because of the interference between the reflections at the two interfaces. However, whether or not there is creation of polarisation conversion is still essentially correctly described by Eq. 21. In addition to all the above considerations we may need to incorporate biaxiality, this may be found for the low-symmetry SC , S∗C and some special nematic phases. In addition in real liquid crystal cells significant variations of the director twist and tilt through the layer will generally exist and these very important situations have to be considered. Thus for most investigation of liquid crystal waveguides simple analytic expressions tend not to be utilised, instead full multilayer optics theory [22–25] is used to model reflectivities, transmissivities and optical field profiles. This then allows the incorporation of the full optical tensor with a spatially varying (through the cell) director profile, allowing the prediction of optical response functions which may be used to compare with data. Up to now the optical waveguide has only been considered as isolated from the outside environment by infinite dielectric slabs. We now need to describe the experimental procedures used to couple incident radiation into the guiding layer to allow a detailed probing of its optical tensor structure. 2.3. Coupling It is clear that a true guided wave cannot be directly excited by light from the cladding or substrate area unless the light is introduced either from the ends of the guide (endcoupling) or by some secondary mechanism such as fluorescence in the guiding layer. However, end-coupling or fluorescence excitation inside the guiding layer is highly impractical with many liquid crystal waveguides, thus some light-coupling mechanism, which will inevitably perturb the waveguide, has to be introduced to allow experimental exploration of the waveguide. For conventional procedures used in laboratories there are, broadly speaking, two mechanisms, prism or grating coupling, for coupling external radiation to the guiding layer. Both give the possibilities of enhancing the momentum of the incident radiation along the propagating direction of the guided waves, essential if the external radiation is to be coupled into a truly guided system. For the first, prism-coupling, the momentum of the incident radiation is enhanced by the refractive index of the prism. The momentum
Guided mode studies of liquid crystal layers
283
Fig. 6: Geometry for a prism-coupled waveguide system, with a low index tunnel barrier of thickness w.
of the incident radiation of the second, grating-coupling, is enhanced by multiples of the grating momentum. In the prism-coupling procedure a high-index prism is generally used to excite the guided waves by providing the necessary phase matching between the evanescent fields of the incident radiation and a guided wave [26]. A geometry for this type of coupling is shown in Fig. 6. A high-index prism, ideally with n p > n g , is put in close proximity to an air-clad waveguide and radiation is made incident at an incident angle β so that its momentum along the interface matches that of the guided mode, that is kz = n p k0 sin β = γ
(22)
The propagation constant, γ , of the waveguide mode excited via evanescent coupling across the air-gap thickness d is modified from its original value by the proximity of the coupling prism. Of course as the air-gap tends to infinity γ tends to γ while the coupling tends to zero. In most experiments the incident angle β is varied and the coupling to the waveguide is monitored in some way, then the observed features which correspond to resonant mode coupling can yield information on the mode structure. As shown in Fig. 6 the true external incidence angle, β , against the entrance face of the prism is related to the internal angle β by the prism angle σ and the refractive index n p of the prism through
σ −1 n p β = sin sin 90° − β − (23) n0 2 where σ is the apex angle of the symmetric prism and n 0 ≈ 1 is the index of air. According to this equation if incident radiation is normal to the entrance prism face, then for a given β corresponding to the critical angle between the prism and the guiding layer the angle σ satisfies
σ n g = (24) cos 2 np Thus for a typical guiding layer with an index of the order 1.55 and a prism of index 1.80 a symmetric prism with apex angle of the order 60° may be used to couple the radiation into the waveguide. This is quite easy to fabricate.
284
F. Yang and J.R. Sambles
By the same evanescent mechanism which allows the radiation to be coupled into the waveguide the prism can also couple the radiation out of the waveguide. Two independent prisms, one for coupling in and one for coupling out, can be used, particularly if the propagation distance of the guided modes is greater than a few microns and some absorption aspects of the waveguide are to be investigated. However, for convenience a single symmetric prism is often used for both coupling in and coupling out in standard experiments. Of course, for practical liquid crystal waveguides, apart from free-standing films, an air coupling gap is not practicable since a constraining wall is needed to contain the liquid crystal. Thus another substance is required to give the low index tunnel barrier, this might be a dielectric, like a silicon dioxide layer, or a thin metallic film, in either case deposited by evaporation on to the base of the prism. For more convenience a similar thin layer may be deposited on to a surface of a high index glass plate which is then index matched with a high index matching fluid to the coupling prism. In this fashion a planar liquid crystal cell can first be fabricated by a normal commercial-like procedure and then studied by optical coupling through a prism and matching fluid. We shall return to the specific geometries needed for liquid crystal studies in the next section. The second important technique for coupling the incident radiation into the waveguide is grating-coupling [6]. In this arrangement a grating, which may be an amplitude modulation (surface grating) or a phase modulation (index grating), is used to give extra momentum to the incident radiation enabling it to couple to the guided modes. As a simple example to give a basic illustration of the grating coupling effect a grating surface modulated waveguide is shown in Fig. 7, in which the grating profile is located at the interface between the waveguide layer and the cladding, although the waveguide–substrate interface will be the same as usual. If a TE or s-polarised plane wave is incident with an angle β upon such a grating geometry having pitch (or wavelength), Λ and height g as shown in Fig. 7, then the incident optical field on the interface of the grating may be written E y = A{i[(k0 n c sin β)z − (k0 n c cos β)x − ωt]} where A is an amplitude coefficient and, for a sinusoidal grating, we have g 2π z x= 2 Λ
(25)
(26)
Fig. 7: Geometry of a grating-coupled waveguide structure. The grating wavevector 2π/Λ adds momentum to the incident radiation allowing it to couple to the guided mode.
Guided mode studies of liquid crystal layers
285
It is clear that if the boundary between the cladding and guiding layer is a flat surface with x not a function of z, i.e. g = 0, the propagation constant in the z direction will simply be k0 n c sin θ which will not be big enough, according to the definition of the critical angle for true waveguiding, to couple to guided modes. However, if g is finite then at the grating boundary x is a function of z and for a sinusoidal shape of the grating then
g 2π z k0 n c cos θ sin − ωt (27) E y = A exp i (k0 n c sin θ)z − 2 Λ and it is obvious that the propagation constant can no longer be readily extracted since there is no simple term just multiplying z in the equation. However, if we expand the above equation in terms of Bessel functions as follows
∞
g 2π Jm − k0 n c cos θ exp i k0 n c sin θ + m z − ωt (28) Ey = A 2 Λ m=−∞ then the problem can be overcome and, as expected, an infinite series of propagation constants has been introduced in the grating coupling 2π kzm = k0 n c sin θ + m (29) Λ where m = 0, ±1, ±2, etc., gives the order of diffraction. For the situation of a surface modified by varying gradient the simple planar boundary condition found in Fresnel equations become more complicated. The consequence is that a single incident plane wave will perhaps produce several diffracted plane waves, as well as local evanescent diffracted fields. The coupling mechanism of the grating-coupling is that if we can find an integer m such that kzm is equal to the propagation constant of a waveguide mode then the mode matching condition will be satisfied and some incident radiation may be coupled into this guided mode. Of course the radiation propagating in the waveguide can also equally be coupled out of the guide, just as in the case of evanescent prism coupling. As mentioned above for a planar structure the strength of coupling is dictated by the coupling gap and easy modelled by Fresnel equations. However, for grating-coupling it is the amplitude of the grating which dictates the coupling strength and with a non-planar boundary involved the optics is much more difficult to model and to compare with experimentally recorded data. In practice these two coupling methods mentioned above can also be combined to give prism–grating coupling systems, or, in addition, holographic couplers [27]. For completeness two other simple coupling methods should also be briefly introduced. One is the end-coupling method [28], in which radiation with a field profile similar in form to the field profile of a guided mode is fed into a waveguide through its end-face. The mode propagation direction in the waveguide is normal to the end-face as shown in Fig. 8. By using a focusing lens the incident radiation coupling is localised at the end of the waveguide. A high quality, defect free end-face, normally produced by polishing or cleaving, of the waveguide is required in this technique. Thus it will be extremely difficult to arrange a geometry to end-couple incident radiation into a liquid crystal waveguide, even though such a procedure may easily be applied to an optical fibre or a semiconductor laser. Another simple coupling technique is tapered coupling [29]. The principle of this coupling method is illustrated in Fig. 9. As shown in the
286
F. Yang and J.R. Sambles
Fig. 8: Geometry of an end-coupled system for a waveguide in which a lens is used to focus the incident beam of radiation onto the end of the guiding layer.
Fig. 9: The tapered-coupling geometry for waveguides where the light is converted by reflection from being radiative to being totally internally reflected in the guiding layer.
picture the incident radiation is coupled by total internal reflection with a zigzag picture into a thin waveguide terminated by a taper. Due to the slowly changing propagation constant in the wedge section of the waveguide end the wavefront is not exactly a plane wave. Because fabricating a cell to give appropriately graded faces to the liquid crystal layer is not particularly easy, this method is also mainly used to couple the radiation into a optical fibre or some solid waveguide in integrated optics. As mentioned above it is very clear that for the optical probing of liquid crystal waveguides prism or grating coupling are by far the most appropriate techniques. Of course prism-coupling is a simple and convenient method for investigating a fabricated liquid crystal cell, however, grating coupling in some senses is also quite convenient since the grating may be fabricated within the cell during construction and thus no extra optical elements are needed in the study of the cell. In addition aligning liquid crystals in the cell may be achieved by grating and there can be found several studies of the using of grating/liquid crystal geometries [30–33]. However, the cost and the complex procedure of grating fabrication, the extra complexity of the overall optical response of the cell when used as a device may be drawbacks of the grating-coupling technique. In addition the geometry of a liquid crystal waveguide with a grating structure is much more difficult to model optically by comparison to the rather simple prism-coupled planar geometry. Since only a limited amount of quantitative work has yet to materialise using grating-coupling to detail the director profile in a liquid crystal cell we will largely confine our attention to the prism-coupling technique in the next section.
3. Liquid crystal waveguide geometries From the optical waveguide theory mentioned in the last section the guided mode spectra are sensitive to the parameters of the guiding layer including the profiles of
Guided mode studies of liquid crystal layers
287
the optical indices (tensor) through the layer and the thickness of the layer. This arises particularly from the fact that each different order mode will be sensitive to different parts of the guiding layer through different optical field distributions for each the guided mode. Of course, this sensitivity is only in one dimension, through the thickness of the guiding layer, but since a mono-domain of liquid crystal in a cell should be invariant in the plane of the guiding layer, then this is exactly the sensitivity required for studying liquid crystal thin films. In addition, comparing with other optical methods the guided mode is the only technique which is going to yield the required spatial selectivity through the thickness of the guiding layer. Over recent years four somewhat different guided mode geometries have been exploited by using prism-coupling to liquid crystal films [34]. For all of the four geometries which are described in more detail below, the essential experimental procedure comprises that of monitoring the angle dependent reflectivity and/or transmissivity of a plane parallel, monochromic linearly polarised optical beam, incident through a coupling-in prism at the glass/liquid crystal layer boundary. According to the optical waveguide models mentioned above, at certain angles of incidence the momentum of the incident radiation along the surface will match that of one of the guiding modes in the layered planar waveguide structure. Then if the geometry is appropriate there will be a reduction in the polarisation-conserving reflectivity at these angles. Of course some related change in the polarisation-conserving transmissivity at these angles will also occur, if another coupling-out prism is used at the bottom of the waveguide structure. Hence by simply monitoring the reflectivity and/or transmissivity as a function of angle of incidence we will find all the mode momentum, the momentum spectra, of the waveguide. These momentum spectra are generally enough to determine the optical properties of the guiding layer for isotropic and lossless materials as may be used in normal integrated optics. However, for a waveguide structure incorporating liquid crystals the situation is more complex. The momentum spectra are only a minimum set of information and although useful, will not readily give the full director profile of the liquid crystal through the cell. If the director is twisted out of the plane of incidence or tilted from the surface the modes in the cell are neither pure TE or TM and the reflected and transmitted radiation will generally have a polarisation converted component. This will not only create polarisation-conversion reflectivity and transmissivity signals but it will also lead to more complex polarisation-conserving reflectivity and transmissivity spectra. More generally, instead of just discussing the mode momentum, one monitors accurately the angle dependent reflectivity (and/or transmissivity) over a wide range of angles and then fits these data to predictions from a multilayer optical model of the geometry. In particular, using angle dependent polarisation conversion reflectivity and/or transmissivity, the sensitivity to the director twist/tilt is greatly enhanced and fitting of this sort of data may yield, in exquisite detail, the director profile through the cell. The resolution of the director profile detail depends on the particular geometry studied, of which, as mentioned above, there are essentially four.
288
F. Yang and J.R. Sambles
Fig. 10: The geometry for a metal-clad liquid crystal waveguide. Here a thin metal film on the high index pyramid acts as a tunnel barrier and the metal film on the substrate is optically thick.
3.1. Fully guided geometry A series of fully guided modes may be excited in the waveguide having a low refractive index cladding, a high index guiding layer and low index substrate. Although in principal any low index materials can be used as cladding and/or substrate, in general metal-clad waveguides have been the focus of attention in this geometry. There are two main reasons for using metal-clad liquid crystal waveguides in the fully guided geometry. One is that the metal layers may be used as the electrodes as well as providing a low refractive index (n real < 1) and the other is that surface plasmons may be excited at the interface between the metal and the alignment film/liquid crystals. With their exponentially decaying optical fields these surface plasmon excitations can be used to explore the director profile near the aligning surface of the liquid crystals. A typical sample geometry for a metal-clad liquid crystal waveguide is illustrated in Fig. 10. This is a nearly symmetric metal-clad dielectric waveguide. In this geometry the thin, about 30–50 nm, metal layer coated directly on to a high index pyramid, acts as both a mirror for trapping guided modes and as a tunnel barrier for radiation coupling. The thickness of this top metal coating is critical and depends on the kind of metal and the wavelength of the radiation. Too thick a metal layer will result in weak coupling to the guided modes, while too thin a layer will allow the guided modes to become more ‘leaky’ and thus be much broader in angle response, giving a less sensitive experiment. Generally, the metal layer on the substrate is thick enough, optically opaque, to act as a high quality mirror, if the transmission data are not needed. Of course, two aligning films, e.g. SiO X , are coated on to these metal surfaces respectively to align the liquid crystal in the cell. For suitable coupling of the incident radiation into the guided modes supported by the liquid crystal layer the refractive index of the pyramid should be chosen to have a higher value than any other layer in the geometry. Both high index glass and some high index anisotropic crystal materials, e.g. sapphire, may be used as coupling prisms, although care has to be taken in production to ensure the optic axis of any anisotropic crystal is orthogonal to the plane of incidence. In this metal-clad waveguide geometry, provided the thickness of liquid crystal layer is greater than the cutoff thickness [6], which is nearly always the case for
Guided mode studies of liquid crystal layers
289
visible radiation, a series of sharp fully guided modes should be excited. In addition, with silver (or aluminium, or gold) films acting as mirror surface in the visible, a broader surface plasmon–polariton (SPP) resonance may also be excited [35] at the metal/alignment layer/liquid crystal interfaces using TM polarised incident radiation. While each guided mode is sensitive to different parts of the director profile through the cell, the TM polarised surface plasmon resonance will be sensitive to the director at the aligning surface as well as the optical properties of the aligning film [36]. Further, by switching the input polarisation from TM to TE, the mode spectra become dictated by the transverse optical index of the system and so allow further details of the director profile to be extracted. Hence it is very clear that only using this relatively simple reflectivity-monitoring technique the director profile inside a cell may be unravelled in some detail. A significant amount of work has been done using this novel technique. For the simple nematic phase, the fully guided technique has been first used to explore the director reorientation at the surface [37] allowing an estimation of the surface anchoring energy [38]. The voltage response of a 90° twisted nematic (TN) cell has also been explored by the metal-clad waveguide technique [39], while, by the use of pulsed voltages very small field-induced changes in the optical permittivity of a nematic have also been investigated in some detail [40,41]. This has allowed the study of macroscopic director effects as well as other induced order-parameter effects. Other examples include the study of bulk director reorganisation in a nematic having finite surface tilt [42], and a very sensitive measurement of the electro-optic pretransitional effects in the isotropic phase using a differential variant [43] of the direct waveguide technique. Comparing with studying the rather simple nematic phase the guided mode technique is an even more powerful and important tool for experimentally unravelling the director profile of the more complex chiral smectic C (S∗C ) liquid crystal inside cells. By observing the guided mode spectra with the metal-clad fully guided geometry the director alignment in the ‘chevron’ structure of a surface stabilised ferroelectric liquid crystal (SSFLC) cell was first optically confirmed [44]. For a structure as shown in Fig. 10 with silver (metal) films, silicon oxide aligning layers and homogeneously aligned layer of ferroelectric liquid crystal (FLC) SCE3 the angle dependent reflectivities have been first experimentally recorded. At a radiation wavelength of 632.8 nm (He–Ne) with a temperature of 31.1°C in the S∗C phase the p-polarised reflectivity (Rpp ) data as a function of angle of incidence are shown fitted by multilayer optics modelling theory in Fig. 11 [44]. A series of sharp resonance dips in the reflectivity are due to the excitation of fully guided modes in the guiding ferroelectric liquid crystal layer, while the broad resonant dip at about 66° is due to the SPP on the silver/SiO X /FLC interface. In the case of the alignment direction at the surfaces being parallel to the plane of incidence, Fig. 11a, some apparent ‘mode-splitting’ occurs in the guided mode area of the reflectivity curve. This splitting, in the Rpp signal, indicates that there is some p-to-s (or s-to-p) polarisation conversion, which in turn implies that there is twist of the director out of the original nematic alignment direction as shown in Eq. 21. For the orthogonal plane of incidence, Fig. 11b, some sharp guided modes appear in the region of the broader SPP resonance dip. This indicates that the twist of the primary director of the FLC is not too far from the surface alignment direction since these s-like guided modes here (from the
290
F. Yang and J.R. Sambles
Fig. 11: The reflectivity for p-polarised light from a homogeneously aligned ferroelectric liquid crystal, SCE3, cell, in the S∗C phase at 31.1°C, with the surface alignment direction (a) parallel and (b) perpendicular to the plane of incidence. The solid lines indicate the fit of the data (crosses) to theory. The cell walls, first coated with silver, are overcoated with silicon oxide (evaporated obliquely at 60°) to form an aligning layer. The cell was 3.5 µm thick and measurements were made at λ = 632.8 nm. (From [44].)
p to s conversion) have quite large momenta which are close to that of the SPP which is largely dictated by the SiO X and the situation of the director near the surface. By using multi-layer optics, with the liquid crystal divided into a large number of sub-layers, to carefully fit such angle dependent data in detail, the director profile in the SSFLC cell is found to be largely a uniformly twisted slab, with a twist angle of 13° (from the original nematic alignment direction) with rather limited tilts of less than 2°. Two thin regions of order 100 nm are near both surfaces over which the director twists out from the original alignment direction. This distribution of the optical director across the cell is in good accord with the chevron model of the layers in such a cell found by X-ray scattering [4]. It should be pointed that the smectic layers cannot be directly determined by the optical studies, such as the guided mode technique here, the layer information can only deduced from the director profile determined if the information about the cone angle of the FLC material is also available. Many fundamental research studies about SSFLC cells have been performed with the metal-clad fully guided geometry, including studies of various SSFLC cells under
Guided mode studies of liquid crystal layers
291
applied AC and DC electric fields [45–47], half-splayed states [48], including a determination of the low level of optical biaxiality [49]. Further studies have also explored the homeotropically aligned state [50] finding for the smectic A phase that the smectic layers are not necessarily parallel to the cell walls. In view of the high definition of the optical reflectivity response, that is the resonant modes are relatively narrow in angle, the experiments mentioned above have been used to quantify in some detail the director distribution through liquid crystal cells. Thus the metal-clad liquid crystal waveguide technique is a powerful tool for studying the behaviour of liquid crystal inside cells, specially for the more complex ferroelectric phases. Unfortunately there are several limitations that inhibit the usefulness of the metal-clad fully guided technique for practical device investigation. Firstly, there are no metal layers in most real devices, and certainly no top, thin, metal film. Secondly, with the relatively soft silver layers it is quite difficult to use the strongly rubbed polymer alignment layers, which are found in most commercial cells on the transparent conducting electrodes (indium tin oxide – ITO) coatings. Even for those situations where quite strong gold films have been used with polymer alignment the alignment created may be different to that with ITO coatings. Thirdly, the thin metal tunnel layer gives very different optical response for the two orthogonal polarisation directions of the incident radiation, TE and TM. A thin metal layer, such as 40–50 nm of silver as typically used in the visible part of the radiation spectrum, tends to reflect strongly TE radiation. This results in rather weak coupling to TE-like modes in the guiding layer and more especially gives a weak polarisation conversion signals. Since it is just these conversion signals that are particularly sensitive to the director twist and/or tilt, then its weakness limits the detailed determination of the director profile through the cells by this guided technique. A way needs to be found to explore more realistic device-like structures. 3.2. Fully leaky geometry It is obvious that if the metallic layers are removed from the previous geometry than all three limitations will be avoided. However this will inevitably mean losing the surface plasmon resonance, and also all of the guided modes will now become leaky. Now the sample geometry becomes that of a liquid crystal layer sandwiched between a pyramid and a glass substrate plate, both with a higher index than the primary index of the liquid crystal and both coated with transparent ITO as a conducting film on top of which are thin transparent rubbed polymer aligning layers as shown in Fig. 12. The use of high index pyramid means that there are still critical edges available to help in quantifying the refractive index tensor of the liquid crystal. For this fully leaky geometry, when the incident angle β is smaller than the angle for total reflection between the pyramid and the liquid crystal layer, the incident light beam enters the liquid crystal layer and, at the substrate surface, it is partly reflected back into the liquid crystal again with some of the radiation being refracted into the substrate. It is clear that true guided modes cannot be excited and supported in this geometry. However when the electromagnetic wave reflected at the two liquid crystal/glass interfaces satisfies a constructive interference condition then a partial localisation of the radiation
292
F. Yang and J.R. Sambles
Fig. 12: Geometry for a fully leaky liquid crystal waveguide. The pyramid and substrate both have refractive indices greater than the liquid crystal layer.
inside the liquid crystal layer will occur. Of course, these are fully leaky guided modes, with propagating radiation leaking out both into the substrate (transmission) and cladding area (reflection). Thus, if the angle dependent reflectivity is once again recorded, then the sharp features obtained from the metal-clad waveguide are now completely absent and a series of much broader resonance will be obtained. However, in a similar fashion to the true guided modes, the electromagnetic waves interfering within the multi-layer system, giving these poorly defined resonances, will also have their own different field distribution across the liquid crystal layer. Hence a study of these fully leaky modes should also provide details of the director profile through the cell. This type of study is an extension of conoscopy, measuring much higher internal angles of incidence by using a matching fluid and coupling pyramid and also providing much more quantitative information concerning the director profile in a given cell. This fully leaky geometry has also been used to explore the director alignment in SSFLC cells [51,52]. A typical set of reflectivity data, compared with a multi-layer optics model is shown in Fig. 13. It is as mentioned above, the sharp reflectivity
Fig. 13: Typical reflectivity results for a ferroelectric liquid crystal in the S∗C phase at 38.3°C, using p-polarised radiation at λ = 632.8 nm. The solid line shows the fit of data (crosses) to theory. In this case rubbed polyimide was used to provide the alignment layers. (From [51].)
Guided mode studies of liquid crystal layers
293
features recorded for the metal-clad fully guided geometry are completely absent and the sensitivity of data fitting to the specifics of the director profile is much reduced. This limits the precision with which the technique may be used to determine the director twist/tilt structure through the cell. However unlike the metal-clad waveguides there is little constraint on the strength of the p to s conversion signal which may then be used to give some further information on the twist/tilt profile of the optical tensor in the cell [53]. Because the fully leaky geometry uses a simple prism and a matching fluid to couple in radiation it may be applied to commercial-like cells. In addition it also gives good p to s conversion signals for helping to determine the details of director profile through a cell. Thus this fully leaky geometry should be a favoured technique, if, that is, the serious profile degeneracy problem arising from fitting model data to the wide modes obtained experimentally can be overcame. However before moving on to an improved fully leaky guided mode geometry we will first introduce another very powerful technique – the half-leaky guided mode geometry in the next subsection. 3.3. Half-leaky guided mode geometry A third variant of the prism-coupled liquid crystal waveguide geometry is the half-leaky guided mode geometry which avoids drawbacks from both of the above geometries and approximates quite well to a real cell geometry, while at same time giving sharper resonant features and strong polarisation conversion signals. The chosen geometry for the half-leaky guided mode (HLGM) technique is that of a high index glass prism (which may for convenience of cell fabrication be replaced by a prism, matching fluid and a glass plate), ITO coating, rubbed polymer alignment layer, an aligned liquid crystal layer and a low index glass substrate with first ITO coating and then alignment layer on it. It is the asymmetry of the glass index which provides the essential new function of this geometry, since now there are a range of angles of incidence in the upper prism for which light is totally reflected at the liquid crystal/low index glass substrate interface. Thus this interface acts as a perfect mirror (dielectric/dielectric interface beyond critical angle) over a certain angle range providing a half-guided or half-leaky guide system. For this new geometry the high index glass should have an index, n c , which is greater than the highest index available in the liquid crystal while the low index glass should have an index, n s , lower than the lowest index available in the liquid crystal. The new half-leaky guided mode geometry is shown in Fig. 14. For the convenience of discussion we suppose that the liquid crystal is positive uniaxial with the extraordinary refractive index, n e , greater than the ordinary index, n o . This means that the ideal condition of the new geometry is n c > n e and n s < n o . According to the guided mode theory it is obvious that there can be no true guided waves in the liquid crystal guiding layer with these conditions satisfied. However, analytical treatment, as well as numerical modelling, of this situation shows [54] that there is a special wavevector range in which there are strong polarisation conversion signals in the reflectivity spectrum. This in-plane wavevector range is between k0 n s to k0 n where n is the maximum effective index of the liquid crystal probed by the radiation, n o < n < n e . When the angle of incidence (in-plane momentum) is in this window
294
F. Yang and J.R. Sambles
Fig. 14: Geometry for the half-leaky guided mode (HLGM) method with a high index prism and low index substrate.
the optical field will be evanescent in the substrate area since n c k0 sin β is greater than n s k0 (so it is beyond the prism/substrate critical angle) while it propagates in the liquid crystal layer because n c k0 sin β is less than k0 n (so it is below the prism/liquid crystal pseudo-critical angle). In this incident angle window the radiation reflected from the prism/liquid crystal boundary will interfere with that mirror-reflected from the liquid crystal/substrate boundary producing sharp interference features, ‘resonances’, in the angle dependent reflectivity response. From the illustration mentioned above it is clear that this special wavevector window may be quite narrow between βs = sin−1 (n s /n c ) and β = sin−1 (n /n c ), where βe ≥ β ≥ βo with βo = sin−1 (n o /n c ) and βe = sin−1 (n e /n c ). Then by scanning the incident radiation over this narrow angle window the sharp half-leaky guided mode reflectivity spectra will be experimentally recorded. Comparing with the fully guided and fully leaky geometries the major advantages of this half-leaky guided mode geometry are obvious. Firstly, since the high index glass plate has been replaced by a low index one from the fully leaky geometry then over the limited angle range the optical energy leaking into the substrate is zero. Thus the modes recorded in this range are quite sharp, and are hence more sensitive to the details of the director profile than the fully leaky geometry. Secondly, and possibly more importantly, because there are no metallic coatings needed the test cell can be constructed following the commercial cell fabrication process with ITO and rubbed polyimide alignment. Thirdly and also associated with the absence of metallic coatings, the polarisation conversion signal may now be quite strong, no longer limited by the reflectivity of the metal layer for TE radiation. This is very important for determining the director profile through cells in detail. If there is director twist and/or tilt from the plane of incidence, then in the half-leaky window strong resonant maxima will be recorded in the angle dependent TM to TE conversion reflectivity. Fitting these data provides even more detail on the twist/tilt director profile across the cell. The advantages of the HLGM technique have been analysed in detail [54]. From numerical modelling it has been found that the technique has very good sensitivity to all changes in the director twist and tilt, even less than 1°. It may be used to monitor optical biaxiality as low as 0.0002, and it also gives details of the director configuration quite near the bounding surfaces of the cell.
Guided mode studies of liquid crystal layers
295
From the experimental point of view a practical point worthy of note is that although the geometry has to have a high index prism there is no essential need to use very high index glass since for normal liquid crystals the refractive indices are not very high in the visible. We may use an index of 1.73 (632.8 nm, He–Ne) for which matching fluids are available, thus the prism used in the geometry such as shown in Fig. 14 may then be replaced by a high index upper glass plate, n = 1.730, matching fluid (e.g. CH2 I2 ) layer, and a high index, n = 1.730, prism. The advantage of this arrangement is not only for simpler cell fabrication but it also allows rotation of the cell under the prism thereby guaranteeing, provided there is not simple homeotropic liquid crystal alignment, that some strong TE to TM conversion signal may be obtained by a suitable choice of rotation of the cell. So by using two glass plates with the requisite high and low indices having similar thermal expansion coefficients, so that heating the cell does not cause mechanical stress problems, and following commercial procedures a HLGM test cell can be fabricated and the detailed characterisation of the director profile in such a structure undertaken. Since the HLGM technique was developed in 1993 [55] a significant amount of work has been undertaken using this powerful tool. The first study, using this technique, was to explore the detailed optical tensor configuration in a homogeneously aligned SSFLC (Merck-BDH SCE3) [55]. In the test cell the aligning surface layers were silicon oxide deposited by evaporation at 60° to create in-plane homogeneous alignment in the nematic phase. The experimental recorded data has shown as high as 60% p to s conversion reflectivity signals in the incident angle window as discussed above. Using the prediction from Fresnel multilayer optics theory to fit the angular dependent p to s conversion reflectivities yields, in unprecedented detail, the optical tensor configuration through the cell. From the fitting results the director profile of the SSFLC across the cell is that of a slightly bent ‘chevron’ director structure with a small tilt angle of order 1.5°, with near-surface region of order 0.3 µm in thickness. Additionally a permittivity biaxiality of the FLC material as small as 0.0035 can be found from the fits. In addition information on a small amount of optic tensor axes dispersion is also provided by using two wavelengths, 632.8 nm (He–Ne) and 514.5 nm (Ar-ion) in these experiments. For this SSFLC cell some typical experimental data for both wavelengths together with theoretical fits are shown in Figs. 15a and 16a, the corresponding director tilt and twist profiles being given in Figs. 15b and 16b, respectively. From Figs. 15b and 16b it is clear that there are finite tilt surface angles in the cell, which is contrary perhaps to earlier expectations, since Cognard [56] indicates that for 60° obliquely evaporated SiO X and a nematic liquid crystal, there will be little surface tilt. It looks like the evidence for tilt with the S∗C phase is overwhelming. From Figs. 15b and 16b the optic tensor axes dispersion with the wavelength of the radiation is also very clear, since there are different tilt angle distributions through the cell for two different wavelengths. The sensitivity of the half-leaky guided mode to the director profile in the cell can be confirmed by a particular example. Consider only the right hand side of Fig. 16a, which has been expanded and shown in Fig. 17a. In Fig. 17a a particular peak, indicated by an arrow, is very sensitive to director tilt. In this figure three model curves are compared, using the same parameters as used to generate Fig. 16a except for changes in tilt profile. For the short dashed line the tilt angle is everywhere modelled as zero, for the solid line
296
F. Yang and J.R. Sambles
Fig. 15: (a) The p to s conversion reflectivity from liquid crystal SCE3 using the HLGM technique with a wavelength of 632.8 nm. The solid line indicates the fit of theory to the experimental data (shown as crosses). (b) The twist and tilt profile in the cell determined from fitting the experimental data. (From [55].)
Fig. 16: (a) The experimental p to s conversion reflectivity obtained at 514.5 nm using the same cell as for Fig. 15. (b) Twist and tilt profiles determined by fitting experimental data to theory. (From [55].)
it is the chosen fit and for the long dashed line it is described in Fig. 17b by having a maximum tilt of 2°. It is very clear that the amplitude of this chosen half-leaky guided mode is very sensitive, with resolution much less than 0.5°, to the director tilt profile across the cell. This also strongly supports the director tensor axes dispersion shown in Figs. 15b and 16b. Comparing with a metal-clad waveguide, in which even though the modes may have been sharper, because of the weakness of the p to s conversion signal by the metal layer, the sensitivity to director twist/tilt profile would be much reduced, the relative value of the HLGM technique is very clear. The advantage of the sensitivity to the director twist means that the HLGM technique is well suited to the study of the electroclinic effect. This effect has been investigated for a homogeneously aligned SA phase near to the SA –S∗C phase transition point. Both materials with first order (material C7) [57] and second order (material C8) [58]
Guided mode studies of liquid crystal layers
297
Fig. 17: (a) Angular dependent reflectivity, for a wavelength of 514.5 nm, obtained by theoretical modelling using the twist profile in Fig. 16(b) and the tilt angle profiles shown in (b). (b) The tilt angle profiles used to generate the theoretical curves shown in (a). The reflectivities and corresponding profiles are indicated by the different dashed lines. (From [55].)
transitions have been studied. The extra director tilts induced by the electroclinic effect have been comparing with mean field theory and the electroclinic coefficients have also been quantified. From the experimental results strong surface anchoring constraints are also observed. Some studies of homeotropic S∗C alignments have also been explored by the HLGM technique. By using a DC in-plane field to unwind the helix of a S∗C homeotropically aligned layer, with or without lecithin surface layers [59–61] the director profile in such cell has been studied. Although the arrangement of the S∗C director should be particularly simple in this geometry the optical results show that, somewhat surprisingly, the smectic layers are not parallel to the cell walls but tilted by as much as 4°. When the cell is rotated so that the director lies in a plane perpendicular to the plane of incidence a finite p to s conversion signal appears to indicate this layer tilt and this phenomenon can be extended from the S∗C phase up into the SA phase. When we use an in-plane DC field to unwind the helical director structure of a homeotropically aligned FLC, then by fitting the reflectivity data recorded the cone angle of the FLC can be accurately obtained for a set of temperatures in the S∗C phase. The extended mean-field theory for a S∗C to SA transition has been confirmed by the above experimental results [60] and, in addition, detailed information on the optical tensor components of both the S∗C and SA phases has also been given [62]. In such experiments a p to s polarisation conversion signal can still be recorded even in the SA phase. This indicates that with respect to the cell wall the layer tilt of the S∗C phase will be retained into the SA phase [59,61] which would be expected to be aligned with its layers flat in the plane. For a homeotropically aligned FLC with no surface treatment the extra tilt induced in the SA phase by an in-plane field can be very accurately measured by the p to s conversion signal. Thus it is possible to quantify the electroclinic effect in this almost unconstrained environment, a simple linear relationship between the induced tilt and the DC field being found even under very weak fields [61]. The predictions from
298
F. Yang and J.R. Sambles
a second order Landau mean-field theory, which includes the coupling between the tilt angle and the DC field, has been confirmed in these experiments [61]. From the knowledge of the optical tensor and cone angles obtained for the S∗C material in the homeotropic geometry as mentioned above it will be relatively straightforward to fit with multiplayer optics modelling theory the more complex reflectivity data obtained as the unwinding DC field is removed. This then will give the pitch of the helical S∗C phase at zero voltage applied, and the ratio of the spontaneous polarisation PS to the twist elastic constant B3 will be given by fitting the distorted helix for finite fields. In additional PS can be found from dielectric measurement and then B3 is readily obtained [63]. The homeotropically aligned smectic-C cell has also been studied by the HLGM method [64]. Under the application of an AC voltage the director configuration changes over a time-scale of the order of seconds. Fitting model results to the recorded angle dependent reflectivity data indicates firstly a change of tilt angle of the primary director, suggesting perhaps a layer tilt, and secondly a reduction of the imaginary part of the optical permittivity, implying a suppression of fluctuations. However the expected Helfrich-like deformation is not recorded, with detailed analysis showing not an increase of layer tilt with field but a field-induced increase of cone angle. The examples given above are mainly for SC or S∗C phases, since we wished to show the power of the technique for exploring the director structures of complex phases. However, even for some simpler phases, such as the nematic or smectic-A phases, if the director structure is quite complex due to external constraints then the HLGM technique may provide a vital method for determining the profile. Two recent examples are given as follows. The first concerns the observation of coexisting nematic and smectic-A phases in a twisted liquid-crystal cell [65]. While the properties of twisted nematic cells, which are the important elements of the most widespread electro-optic liquid crystal display devices, are well understood, there is no information about the structure of such cells below the nematic–smectic transition temperature. Since the smectic layers cannot sustain twist the question is what does the system do on cooling? Does it become full of defects or can some other defect-free configuration exist? This phase transformation is also interesting from the fundamental point of view because it represents a nontrivial example of a transition in an inhomogeneous soft system in a confined geometry. The properties of such a transition are expected to be strongly dependent on the cell thickness. So a powerful technique for exploring this is required to accurately determine the director configuration through a thin cell in some detail. As mentioned above the HLGM geometry provides the required technique. The sample comprises a twisted homogeneously planar-aligned cell with an angle of 87° between the two rubbing directions of the top and bottom polyimide coated substrates. The liquid crystal material in this study is the ferroelectric liquid crystal SCE13. After filling the liquid crystal in the isotropic phase (∼110°C) the temperature is slowly reduced. Upon cooling into the N∗ phase a well-aligned monodomain forms. Then suitable angle dependent reflectivity, RSS , RPP and RSP (polarisation conversion) data are recorded at different temperatures. From close to the N∗ –SA phase transition data are taken at intervals of about 1° down to the S∗C phase. From fitting the angle
Guided mode studies of liquid crystal layers
299
Fig. 18: Director twist angle profiles φ(z) across the 1.6 µm twisted liquid-crystal cell obtained using the half-leaky guided mode method for three different temperatures. (1) T = 67.0°C; (2) T = 66.3°C; (3) T = 65.8°C. Note the untwisted region in the centre of the cell which is identified as a region of smectic-A material that grows with decreasing temperature. The insert shows a schematic of the geometry of the experiment. (From [65].)
dependent reflectivity data recorded several points have been obtained. Firstly, as expected, in the N∗ phase a uniformly twisted director profile is found in the cell. Secondly, from the data taken on slow cooling it is easy to identify the N∗ to SA phase transition point by the sudden appearance, between 67.0 and 67.8°C, of an extra optical mode feature in the HLGM reflectivity spectrum. Thirdly, when the temperature is lower than the N∗ to SA phase transition point there is still a good monodomain in the cell which is conformed by the very good guided mode features in the RSS , RPP and RSP reflectivities and very low background in the RSP data. Finally, in the temperature range of the SA phase the director profile has a very different form to the linearly twisted nematic. For a thinner cell with thickness of 1.60µm an untwisted homogeneous region forms in the centre of the cell. This has been identified as a region of SA material. This SA area is separated from the walls by thin regions of nematic with a higher twist gradient than before the SA nucleated. Upon further cooling the thickness of the SA region grows at the expense of the nematic as shown in Fig. 18. For increased cell thickness more separated SA areas appear in the cell. For the 2.0 µm cell there are two regions of uniform azimuthal angle (SA phase) separated by a twisted nematic with two twisted nematic boundary regions as shown in Fig. 19a, while for the 2.4 µm cell there are three SA regions separated by two twisted layers with two twisted boundary layers as shown in Fig. 19b. The uniqueness of the director profile through these test cells can only be achieved by the HLGM technique, since it is very sensitive to the variation of the director twist/tilt distributions across the cells. So the director profiles discovered in this work gives a picture of coexisting nematic and smectic-A phases in a twisted liquid crystal cell. The experimental results also show that comparing with the bulk phase sequence of the FLC material the SA phase exists over a very small temperature range in these twisted thin cells. This suppression in phase transition is caused by the high twist gradient in the liquid crystal and has to be expected since the SA phase cannot tolerate
300
F. Yang and J.R. Sambles
Fig. 19: Director twist angle profiles φ(z) in the 2.0 µm twisted cell with two separate smectic-A regions (a), and in the 2.4 µm cell with three smectic-A regions (b). (From [65].)
any twist. A theoretical model of these novel results is also presented in the work [65] to give a complete explanation. The second example concerns establishing the direction of the ‘easy’ axis at a twisted nematic liquid crystal wall determined by the half-leaky guided mode technique [66]. For the measurement of surface torsional or azimuthal anchoring energy, which is very important both from the fundamental science perspective as well as for the production of display devices, a geometrical technique using a thin twisted nematic (TN) cell has been widely used. Since no external field is needed to distort the director profile in the cell both the mathematics and the experimental procedures are quite simple. In this TN geometry the twist-off of the director at the surface of the cell is brought about by equilibrium between the two surface torsional anchorings mediated through the bulk twist elastic constant, K 22 . From continuum theory the key numbers required to quantify the torsional anchoring strength are the elastic constant K 22 , the gradient of the twist angle, dφ/dz, and the deviation of the twist angle from the easy axis, φ − φe , at the wall of the TN cell. K 22 is measured separately (given in the chemical suppliers data sheet) and the dφ/dz can be obtained from the total twist angle of the liquid crystal director, φt , through the thin TN cell and the LC layer thickness, dLC , by dφ/dz = φt /dLC . Then half of the difference between the two angles φt and φt0 , which is the angle between the two easy axes on the two interfaces of the cell, is taken as the deviation of the twist angle at both boundaries, if the alignments are assumed to be identical. Both φt and dLC can be accurately determined in the geometrical technique using for example the HLGM technique. However, when a thin TN cell has been assembled the twist angles of the director on the two boundaries always deviate from the easy axes, hence accurate determination of φt0 is not simple. It is in fact much easier to measure φt rather than φt0 . It is apparent that after rubbing the polyimide, assembling the cell and mounting it on a sample holder an accurate and direct determination of the easy axes directions of a TN cell is very important to allow characterisation of the surface torsional anchoring force by the geometrical technique. From numerical modelling using continuum theory for a thin TN cell it is apparent
Guided mode studies of liquid crystal layers
301
that under high voltage (a typical value for a practical cell about 4–5 V) the gradient dφ/dz near both surfaces is zero. This is because in the middle part of the cell the director, driven by the high electric field, is almost homeotropically aligned. This removes the influence of the two boundaries from each other through the twist elastic constant K 22 , i.e. now the direction of the director at the wall will coincide with the direction of the easy axis. Thus all that is required is some optical procedure for determining this director direction at the surface when a suitable voltage has been applied to the cell. There are three factors which may be used to determine the direction of the director at the surface: (1) For an interface between an isotropic and a uniaxial anisotropic medium if the optic axis of the uniaxial medium is in the incidence plane there will be no polarisation conversion reflectivity signal. By contrast a very small angle of optic axis twist out of the incidence plane creates a small but detectable polarisation conversion signal in the reflected beam. (2) If the director of the top surface is close to the incidence plane of the radiation beam then in the HLGM geometry there is a pseudo-critical angle dependent on the high index of the isotropic medium and the low ordinary index of the uniaxial medium. Also in an incident angle range beyond this pseudo-critical angle, for a s-polarised incident beam the optical field in the anisotropic medium exponentially decays away from the interface. (3) For a practical TN cell at high voltages the director lies in the direction of the easy axis for a distance of about 0.5 µm which is much greater then the decay distance mentioned above. Thus there will be no p to s (or s to p) conversion within this depth if the director lies in the incidence plane. According to the above three factors an experimental procedure is designed to accurately measure the easy axis at the cell surface [66]. Firstly matching fluid is placed between a high index prism and the top glass plate of the HLGM cell to allow the cell to be freely twisted against the incidence plane with the rubbing direction set near to the incidence plane. Then with the incident angle set greater than the pseudo-critical angle and under a high AC voltage (4.0 V, 1.0 kHz) the p to s polarisation conversion reflectivity signals are recorded for different cell twist angles. The minimum point of the polarisation conversion reflectivity signal accurately gives the director direction, the easy axis direction, at the top surface of the cell. The experimental results of the polarisation conversion reflectivity against the easy axis twist angle from the incidence plane are shown in Fig. 20 [66]. Using this procedure the surface torsional anchoring coefficient between a nematic liquid crystal (E7-BDH) and a rubbed polyimide layer has been determined [67]. All of the above applications of the HLGM technique show quite clearly that it is a very powerful procedure for exploring the optical tensor configuration of a liquid crystal in a thin cell. However, this powerful technique still has a limitation, i.e. the two glass plates from which the test cell is comprised are very different from each other and from commercial cell glass. So if the director profile of a real commercial device cell, which has low index, normally 1.52, glass plates, is to be explored by a guided mode technique, some further improvement is still required. 3.4. Improved fully leaky guided mode geometry If a standard commercial-like liquid crystal cell with low index glass plates is to be investigated by a guided mode technique to unravel the director profile through the
302
F. Yang and J.R. Sambles
Fig. 20: Experimentally recorded reflectivity data RPS from 62.5° to 65.0° of incident angle for a TN cell at 4.0 V. (a) Six RPS curves which correspond to the director in the incident plane and twisted away by 1.0°, 2.0°, 4.0°, 5.0° and 6.0° from bottom to top, respectively. (b) Six RPS curves which correspond to the director in the incident plane and twisted by the same angles in the opposite direction as (a). (c) The intensity of a selected p to s conversion mode against the twist-off angle of the director from the incident plane of the radiation. (From [66].)
cell, then only the low index fully leaky guided mode (FLGM) technique may be chosen. As mentioned before, because all the guided modes will now be leaky and will give correspondingly broad features in the reflectivity data, this may severely limit the precision with which the director twist/tilt structure through the liquid crystal cell may be determined. In addition, the use of low index glass means that no longer will there be any critical angle available to help determine the refractive indices of the liquid crystal. However, recently some improvements [21] have been introduced to the fully leaky geometry, in which two refinements to the original technique have been made to make it much more useful. Firstly, the full sets of both transmission, T , and reflection, R, data may be utilised, including all the polarisation-conversion signals RSP , RPS , TSP and TPS as well as the polarisation-conserving signals RPP , RSS , TPP and TSS . The polarisation-conversion signals are particularly sensitive to the director twist and tilt. Secondly, two matching prisms with matching fluid have been used to allow rotation of
Guided mode studies of liquid crystal layers
303
Fig. 21: The two prism-coupling cell geometry of the improved fully leaky guided mode technique.
the cell to a position that allows optimisation of sensitivity to director twist and tilt. In addition, this allows the acquisition of data for a set of different azimuthal angle settings that, by fitting of all the data sets, further removes ambiguity in determining the director profile. The new FLGM geometry is shown in Fig. 21. As mentioned above this is a symmetrical structure in which two identical low index prisms with matching fluid couple the radiation in and out from a symmetrical commercial-like liquid crystal cell comprised of two low index glass plates with suitable ITO coatings and alignment layers on their inner surfaces. Of course, comparing with the other three geometries an extra detector and an extra polarizer are needed in the experimental arrangement for detecting suitable transmission signals. Since the improved FLGM technique was introduced in 1999 [21] a substantial body of work has been done using this technique to investigate the commercial-like liquid crystal cells. To demonstrate the potential use of this new technique for the determination of the director profile in a liquid-crystal layer, a conventional surfacestabilised ferroelectric liquid crystal cell has been studied [21]. The glass of the cell is ordinary glass with an index close to 1.52, as are the two coupling prisms. The sample used in this study contains the ferroelectric liquid crystal SCE8∗ . The alignment is homogeneous with a slight pretilt achieved by the use of rubbed polyimide in a parallel arrangement. The glass plates, which are 1 mm thick and coated with ITO on the inner face, have an index of 1.517 at 632.8 nm. All measurements were conducted on a monodomain at room temperature, 23.7° C. Experimental data, including the polarisation-conserving signals (RPP , RSS , TPP and TSS ) and the polarisation-conversion signals (RSP , RPS , TSP and TPS ), are recorded with the cell configured so that the surfacealignment axis (rubbing direction) is close to the plane of incidence. An example of the polarisation-conversion reflectivity signals, RSP and RPS , are shown in Fig. 22. From Fig. 22 we note that these RSP and RPS signals are quite weak, being less than 1%. This indicates that the director of the ferroelectric liquid-crystal layer is only twisted a small
304
F. Yang and J.R. Sambles
Fig. 22: Experimentally recorded reflectivity conversion data, RPS and RSP , (crosses) and fitted theory curves (solid line) for a SCE8∗ cell. (From [21].)
amount from the alignment direction. Also note that there are differences between RSP and RPS , which implies a small tilt angle of the director through the cell. To theoretically model the optics of this cell structure and thereby predict the observed optical response, a scattering-matrix approach has been employed with the liquid-crystal being represented as 150 sub-layers. These sub-layers are part of an overall model structure that treats the liquid crystal as having several (up to 10) boundary matched regions in which the director twist and tilt vary linearly. This linear variation is represented by both start and finish tilt and twist angles and a layer thickness. This procedure reduces the number of variables for the liquid crystal layer to two real and imaginary permittivities and ten sets of linked layer parameters. The final director profiles, obtained by fitting the recorded data with the predictions from the multi-layer optics, are shown in Fig. 23a and b, in which there is a clear director-tilt chevron close to the centre of the cell with thin boundary layers with the liquid crystal being in the C2U state [68]. The smectic layer tilt, δ, may also be extracted by using a value for the cone angle of 19.50° at 23.7°C (BDH-Merck data) combined with the profiles of Fig. 23a and b [55]. The resulting layer tilt profile, the layer chevron, is shown in Fig. 23c. This shows primarily two tilts that are slightly different in the two parts of the cell. For the lower, slightly thicker, portion the layer tilt is 16.70° and for the thinner upper portion it is 17.30°. There is also a discernable, but small variation of this tilt near the upper bounding surface. Thus it is clear that the new FLGM technique, using both reflectivity and transmissivity signals, is capable of giving details of the director profile in a commercial-like cell to almost the same level of precision as had been previously obtained by use of either metal-clad guiding structure or cells with different index glass plates in the half-leaky geometry. Unlike the other two procedures, there are no very sharp features in the angledependent signals owing to the weak (leaky) nature of the guiding; furthermore, there are no true critical angles to yield the refractive indices of the liquid crystal because the
Guided mode studies of liquid crystal layers
305
Fig. 23: Fitted profile of the director in the SCE8∗ layer: (a) twist angle measured from the incident plane. (b) tilt angle measured from the plane parallel to the cell wall, and (c) layer tilt angle calculated with a cone angle of 19.50° together with the twist/tilt profiles. (From [21].)
glass of the cell and the coupling prisms have such a low index (1.517). Nevertheless, there is sufficient information in the eight data sets available at each azimuthal angle of study to yield all the requisite information. This means that more detail can still be extracted regarding the spatial distribution of the director profile through the cell by the new FLGM technique. After this first demonstration of the improved fully leaky technique some commercially like standard liquid-crystal cells have also been investigated. This includes the quantification of the azimuthal anchoring energy [69] and the surface- and bulk-order parameters [70] of a homogeneously aligned nematic liquid crystal under an in-plane electric field, and the determination of the polar anchoring energies of both homogeneously [71] and a homeotropically [72] aligned nematic liquid crystals. Some mixed alignment liquid crystal cells, which have a zero-order grating alignment on the superstrate and rubbed polyimide alignment on the substrate, have also been studied by
306
F. Yang and J.R. Sambles
this new technique to investigate the influence of the groove depth of the grating on the alignment [73] and the torsional anchoring energy [74]. Because the new technique allows the use of standard liquid crystal cells then combined X-ray scattering and fully leaky guided mode studies have been undertaken to explore the smectic layer and the optic tensor configuration in a ferroelectric liquid crystal cell [75]. All of this work, using the guided mode technique to explore conventional commercial-like cells, paves the way for detailed exploration of the behaviour of such cells under various conditions and also allows dynamic studies.
4. Dynamic guided mode technique As mentioned in the above sections the success of the guided mode techniques in determining the director distribution through a thin liquid crystal layer can be attributed to the fact that each optical mode has a different field intensity profile across the liquid crystal layer. Thus a clear ‘pictures’ of the director across the thin cell can be ‘observed’ by the results of fitting the recorded angle-dependent reflectivity and/or transmissivity data with the predictions from multi-layer optic theory. From both a fundamental scientific perspective and also for device development the switching process of a liquid crystal layer under an external voltage is very important. Thus establishing a clear picture of the transient director profile through a cell as it varies on time scales of order ms, the dynamic director profile, is very important. It is also obvious that the guided mode technique would be an attractive tool for this dynamic director profile determination. Some early studies successfully resolved the dynamic director profile across the liquid crystal cells in a metal-clad fully guiding geometry [76,77]. However, like the vast majority of liquid crystal waveguide techniques mentioned before, these studies were based upon the standard angle-scan collimated beam procedure. The experiment is a fairly simple optical arrangement in which a plane polarised collimated laser beam is incident through a coupling prism onto the liquid crystal cell. By rotating the liquid crystal cell and prism arrangement around an axis perpendicular to the incident wave vector, the angular dependent reflectivity and transmissivity features are recorded. Of course, this is a relatively slow data acquisition procedure, every switching process has to be repeat again for every rotating step, i.e. every incident angle, taking anywhere from several minutes to two hours to perform a single angle scan [77]. These ponderous studies of dynamic processes using a slow angle scan with time dependent data taken at each angle demand highly repeatable voltage cycling of the cells, complete thermal stability and total lateral invariance within a given cell. Thus an improved guide mode technique has been developed for fast dynamic studies. The new approach involves the use of a convergent beam and has several advantages over the standard collimated beam angle scan procedure [78,79]. This convergent beam guided mode (CBGM) technique uses a highly focused beam spot that allows simultaneously the excitation of many guided modes and produces reflectivity and transmissivity data over a wide incident angle range. There are several advantages to this procedure. Firstly the CBGM technique removes the need to physically rotate the liquid crystal cell geometry and consequently the focused beam remains completely stationary
Guided mode studies of liquid crystal layers
307
Fig. 24: The geometry for hemispherical coupling to a conventional FLGM cell.
on the liquid crystal layer. This allows the study of cells with lateral non-uniformity, opening up the potential for single pixel studies.To demonstrate the usefulness of the technique with either half-leaky [80] or fully leaky [81] geometries the static director profiles for nematic liquid crystal cells have been determined. Secondly, possibly even more importantly, data from the converged beam technique may be captured with a suitable charge coupled device (CCD) array in a time equal to the line transfer rate of the CCD array (< 0.1 ms for a typical array). Thus using a convergent beam technique, data may be acquired around five to six orders of magnitude faster than using the collimated beam procedure, allowing for real time studies of liquid crystal dynamics by the guided mode technique. Since the fully leaky geometry allows the study of standard commercial-like liquid crystal cells the first COGM study [16] of liquid crystal dynamics was undertaken with the improved fully leaky guided mode geometry. In this study the cell consisted of two ordinary glass plates (n = 1.52) coated with ITO upon which is a rubbed polyimide layer. The rubbing directions on the top and bottom plates are antiparallel, thus inducing a uniform, nearly planar alignment of the director through the cell. The cell is filled with the nematic liquid crystal ZLI-2293 (Merck). For directly coupling the convergent beam in and out from the sample geometry two low index hemispheres are optically matched onto the cell by use of a suitable low volatility silicon based oil as shown in Fig. 24. The use of the matching fluid not only provides a continuous optical medium for coupling light into and out of the waveguide modes but also enables the cell to be easily rotated to any azimuthal angle for optimising sensitivity to director twist and tilt. The experimental setup of the COGM technique is shown in Fig. 25. The He–Ne laser beam (632.8 nm, 75 mW) is expanded, polarised, and focused through the hemisphere onto the liquid crystal layer. The reflected and transmitted beams are then captured with a linear CCD array (DALSA SPARK). The diffuser arrangement situated in the beam expander serves two purposes. First, it breaks up the spatial and temporal coherence of the laser, and second, it provides an intensity profile across the expanded beam such that approximately the same intensity is available to excite each guided mode. The dynamic study in this work uses an integration time of 0.3 ms and therefore it is desirable for the rotating diffuser to complete at least one whole revolution in this time. A turbine
308
F. Yang and J.R. Sambles
Fig. 25: The experimental setup of the convergent beam guided mode technique.
dental drill (LARES APOLLO557) was used and once properly adapted gave a diffuser rotation rate of 120,000 rpm (period ∼0.5 ms). Initially, the total of eight data sets of the reflectivity and transmissivity were taken with no voltage and 1.5 V rms (10 kHz) applied to the cell to characterise the static optical parameters of the liquid crystal cell. Once the optical parameters of the LC cell and the static director profile at 0 and 1.5 V had been ascertained by fitting the data recorded with multi-layer optics theory, the dynamics of the director profile as it relaxes from the 1.5 V to the 0 V configuration was studied. The switching dynamics of the LC cell were recorded by synchronising the CCD array to capture data when the voltage across the LC cell was changed. After removal of the voltage both RPP and RSS signals were captured as a function of time and fitted together. The fitted experimental data for the RPP and RSS signals and the corresponding director profiles used to obtain each least squares fit are shown in Fig. 26. From the fitting results the tilt of the director in the middle of the cell against time can be extracted and then the exponentially decaying time constant of the relaxation process is revealed to be τ = 51.4 ± 0.1 ms. Then, from the relation between the time constant, τ , the thickness of the cell, d, the splay elastic constant, K 11 , and the effective rotational viscosity, γ ∗ , the γ ∗ value is deduced. Using a value for K 11 of 12.75 pN and a d value of 6.63 µm (determined from the static fits), γ ∗ is evaluated to be 0.147 ± 0.003 Pa s at 20.0°C. By using the same experimental arrangement as mentioned above the ‘back-flow’ phenomenon of the relaxation process in a twisted nematic liquid crystal cell, which was theoretically predicted by Berreman [82] in 1975 and indirectly confirmed by an optical ‘bounce’ in the light transmission [83,84], has now been directly and clearly ‘observed’ from the dynamic director profiles [17]. These experiments clearly show that the COGM arrangement with the FLGM geometry is a very powerful technique for investigating dynamic processes within liquid crystal cells in detail.
5. Conclusions The distribution of the liquid crystal optical permittivity tensor, the director profile, through an aligned cell and its static or dynamic response to changes in environment
Guided mode studies of liquid crystal layers
309
Fig. 26: Left column and middle column: fits to the transient RSS and RPP signals, respectively, for various times. Right column: the corresponding director profile used to obtain fits to experimental data. (From [16].)
(applied DC, AC or pulsed electric field, surface anchoring, temperature, cell geometry, etc.) is one of most important aspects of liquid crystal science. Much fundamental understanding, material parameter determination and device design procedures are dependent upon the knowledge of the director profile and its changes under various conditions. Because of the nature of optical guided modes the liquid crystal waveguide techniques provide an extremely powerful method for studying this tensor profile in detail allowing exploration of both static and dynamic processes. In this chapter first the principles of optical guided waves in general have been described with some discussion of optical field profiles, coupling procedures, etc., this has been followed by a fuller discussion of several types of waveguiding techniques used to investigate liquid crystals in aligned cells. These techniques include that of the metal-clad fully guiding geometry; the fully leaky waveguide with no metal layers but which gives poor resolution; the half-leaky waveguide technique which again has no metal layers but which yield sharper optical features giving much finer detail of the director profile; the improved fully leaky guided mode technique which can be directly used to investigate a commercial-like standard liquid crystal cell and still gives enough information for characterising the director profile through such cells, and finally the
310
F. Yang and J.R. Sambles
convergent beam guided mode technique with the improved fully leaky geometry to directly and rapidly study the dynamic process in a standard liquid crystal cell. A range of experimental results using these techniques have been presented to illustrate their usefulness with a range of cells structures and liquid crystal phases. All these guided mode techniques are very powerful tools for exploring the static or, very recently, the dynamic director profiles and determining the optical or physical properties of thin liquid crystal films. Which technique is chosen is dependent on the sample and the information required. However it seems likely that the advantages of being able to use off-the-shelf cells will see progressively more use of the fully leaky technique. It is very clear that by using a wide range of incident angles and multilayer optics theory to fit the obtained data in these guided mode techniques some of the uncertainties and erroneous conclusions associated with integrated techniques (e.g. simple crossed polariser microscopy), which often ignore the surface layers, are avoided. However, it should be added that, dependent on the sample geometry and the measurements taken, the guided mode technique may require elaborate data analysis based on the Berreman transmission/reflection matrix and a complex fitting procedure to obtain the required results. It is essential that sufficient consideration is given to the sample geometry suitable for obtaining the information required and as many of the unknown parameters are pre-determined by other procedures before any one particular optical guided mode technique is chosen. In this review chapter we have chosen not to discuss to any length the other very interesting and useful technique, grating-coupling of radiation into guided modes. This is primarily because this needs special cell fabrication and also the theoretical analysis of the optics is much more complicated. By contrast the prism-coupled procedures discussed use simple planar multilayer optics theory, leading to accurate detailed comparisons of predicted responses with those observed experimentally. This leads to substantial confidence in the director profiles thus deduced, which is fundamentally why these guided wave procedures provide an underpinning to liquid crystal science.
Acknowledgements The authors are extremely grateful to the project 10174044 supported by NSFC.
References 1. 2. 3. 4. 5. 6. 7. 8.
F. Reinitzer, Monatsh. Chem. 9, 421 (1888). O.Z. Lehmann, Phys. Chem. 4, 462 (1889). M.A. Handschy and N.A. Clark, Appl. Phys. Lett. 41, 39 (1982). T.P. Rieker, N.A. Clark, G.S. Smith, D.S. Parmar, E.B. Sirota, and C.R. Safinya, Phys. Rev. Lett. 59, 2657 (1987). P.K. Tien, Appl.Opt. 10, 2395 (1971). D. Marcuse (editor), Intergrated Optics, (IEEE Press, New York, 1973). H.A. Weakliem, D.J. Channin, and A. Bloom, Appl. Opt. 14, 560 (1975). D.J. Channin, Appl. Phys. Lett. 22, 365 (1973).
Guided mode studies of liquid crystal layers 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34.
35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56.
311
J.P. Sheridan, J.M. Schnur, and T.C. Giallorenzi, Appl. Phys. Lett. 22, 561 (1973). J.P. Scheridan, OSA Topical Meeting on Integrated Optics, New Orleans, (1974) TUA S-1. Chenming Hu, J.R. Winnery, and N.M. Amer, IEEE J. Quan. Elect. QE10, 218 (1974). R.B. Meyer, Mol. Cryst. Liq. Cryst. 40, 33 (1977). N.A. Clark and S.T. Lagerwall, Appl. Phys. Lett. 36, 899 (1980). L.Z. Ruan, M.A. Osipov, and J.R. Sambles, Phys. Rev. Lett. 86, 4548 (2001). Fuzi Yang and J.R. Sambles, Jpn. J. Appl. Phys., 37, 3998 (1998). N.J. Smith and J.R. Sambles, Appl. Phys. Lett. 77, 2632 (2000). N.J. Smith, M.J. Tillin, and J.R. Sambles, Phys. Rev. Lett., (2002), accepted. Fuzi Yang, J.R. Sambles, and G.W. Bradberry, The Optics of Thermotropic Liquid Crystals, edited by S.J. Elston and J.R. Sambles, (Taylor & Francis, London, 1998) Ch. 5. M.S. Kharusi, J. Opt. Soc. Am. 64, 27 (1974). Fuzi Yang and J.R. Sambles, J. Opt. Soc. Am. B 11, 605 (1994). Fuzi Yang and J.R. Sambles, J. Opt. Soc. Am. B 16, 488 (1999). R.M.A. Azzam and N.M. Bashara, Ellipsometry and Polarised Light, (North Holland, Amsterdam, 1979). M. Born and E. Wolf, Principle of Optics, (Pergamon, Oxford, 1964). J. Lekner and M.C. Dorf, J. Opt. Soc. Am. A 4, 2092 (1987). D.Y.K. Ko and J.R. Sambles, J. Opt. Soc. Am. A 5, 1863 (1988). J.H. Harris, R. Shubert, and J.N. Pelky, J. Opt. Soc. Am. 60, 1007 (1970). E.A. Ash, E. Seaford, O. Soares, and K.S. Pennington, Appl. Phys. Lett. 24, 207 (1974). R. Shubert and J.H. Harris, IEEE Trans. Microwave Theory Tech. MTT-16, 1048 (1968). N.H. Hartshorne and A. Stuart, Crystals and the Polarising Microscope, 2nd ed. (Arnold, London, 1950). A. Sugimura and T. Kawamura, Jpn. J. Appl. Phys. 23, 137 (1984). Y. Kawata, K. Takatoh, M. Hasegawa, and M. Sakamoto, Liq. Cryst. 16, 1027 (1994). G.P. Bryan-Brown, J.R. Sambles, and K.R. Wolford, J. Appl. Phys. 73, 3603 (1993). E.L. Wood and J.R. Sambles, J. Mod. Opt. 40, 493 (1993). Fuzi Yang and J.R. Sambles, Physical Properties of Liquid Crystals: Nematics, edited by D. Dunmur, A. Fukuda, and G. Luckhurst (INSPEC – The Institution of Electrical Engineers Publication, London, 2001) Ch. 7.3. E. Kretschmann, Z. Phys. 241, 313 (1973). G.J. Sprokel, R. Santo, and J.D. Swalen, Mol. Cryst. Liq. Cryst. 68, 29 (1981). K.R. Welford, J.R. Sambles, and M.G. Clark, Liq. Cryst. 2, 91 (1987). K.R. Welford and J.R. Sambles, Appl. Phys. Lett. 50, 871 (1987). Lizhen Ruan, S.J. Eolston, and J.R. Sambles, Liq. Cryst. 10, 369 (1991). Lizhen Ruan, G.W. Bradberry, and J.R. Sambles, Liq. Cryst. 11, 655 (1992). Lizhen Ruan, G.W. Bradberry, and J.R. Sambles, Liq. Cryst. 12, 799 (1992). Lizhen Ruan, T.W. Preist, Fuzi Yang, and J.R. Sambles, Liq. Cryst. 13, 541 (1993). E.L. Wood, J.R. Sambles, and P.S. Cann, Liq. Cryst. 16, 983 (1994). S.J. Elston, J.R. Sambles, and M.G. Clark, J. Mod. Opt. 36, 1019 (1989). S.J. Elston and J.R. Sambles, Appl. Phys. Lett. 55, 1621 (1989). S.J. Elston, J.R. Sambles, and M.G. Clark, J. Appl. Phys. 68, 1242 (1990). S. Ito, F. Kremer, E. Aust, and W. Knoll, J. Appl. Phys. 75, 1962 (1994). S.J. Elston and J.R. Sambles, Jpn. J. Appl. Phys. 29, 41 (1990). S.J. Elston and J.R. Sambles, Mol. Cryst. Liq. Cryst. 220, 99 (1992). M.N. Gong, J.R. Sambles, and Fuzi Yang, Liq. Cryst. 13, 637 (1993). C.R. Lavers and J.R. Sambles, Ferroelectrics 113, 339 (1991). C.R. Lavers and J.R. Sambles, Jpn. J. Appl. Phys. 30, 729 (1991). C.R. Lavers and J.R. Sambles, Liq. Cryst. 8, 577 (1990). Fuzi Yang and J.R. Sambles, J. Opt. Soc. Am. B 10, 858 (1993). Fuzi Yang and J.R. Sambles, Liq. Cryst. Opt. 13, 1 (1993). J. Cognard, Mol. Cryst. Liq. Cryst. 78, Supplement (1981).
312 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84.
F. Yang and J.R. Sambles Lizhen Ruan, J.R. Sambles, E.L. Wood, and J. Seaver, Liq. Cryst. 18, 401 (1995). Lizhen Ruan, J.R. Sambles, and J. Seaver, Liq. Cryst. 19, 133 (1995). Fuzi Yang and J.R. Sambles, Mol. Cryst. Liq. Cryst. 250, 143 (1994). Fuzi Yang, G.W. Bradberry, and J.R. Sambles, Phys. Rev. E 50, 2834 (1994). Fuzi Yang, J.R. Sambles, and G.W. Bradberry, Liq. Cryst. 18, 407 (1995). Fuzi Yang, J.R. Sambles, and G.W. Bradberry, J. Appl. Phys. 78, 2187 (1995). Fuzi Yang, G.W. Bradberry, and J.R. Sambles, Phys. Rev. 53, 674 (1996). L. Ruan, J.R. Sambles, and J.M. Towler, Liq. Cryst. 18, 81 (1995). L.Z. Ruan, M.A. Osipov, and J.R. Sambles, Phys. Rev. Lett. 86, 4548 (2001). F.Z. Yang, H.F. Cheng, H.J. Gao, and J.R. Sambles, J. Appl. Phys. 88, 4553 (2000). F.Z. Yang, H.F. Cheng, H.J. Gao, and J.R. Sambles, J. Opt. Soc. Am. B 18, 994 (2001). N. Itoh, M. Koden, S. Miyoshi, and T. Wada, Liq. Cryst. 15, 669 (1993). B.T. Hallam, Fuzi Yang, and J.R. Sambles, Liq. Cryst. 26, 657 (1999). B.T. Hallam, C.V. Brown, and J.R. Sambles, J. Appl. Phys. 86, 6682 (1999). Fuzi Yang, J.R. Sambles, Youmei Dong, and Hongjin Gao, J. Appl. Phys. 87, 2726 (2000). Fuzi Yang, Lizhen Ruan, and J.R. Sambles, J. Appl. Phys. 88, 6175 (2000). B.T. Hallam and J.R. Sambles, Phys. Rev. E 61, 6699 (2000). B.T. Hallam and J.R. Sambles, Liq. Cryst. 27, 1207 (2000). B. Hodder, J.R. Sambles, S. Jenkins, and R.M. Richardson, Phys. Rev. Lett., 85, 3181 (2000). M. Mitsuishi, S. Ito, and M. Yamamoto, J. Appl. Phys. 81, 1135 (1997). M. Mitsuishi, S. Ito, and M. Yamamoto, Appl. Phys. Lett. 69, 2199 (1996). E. Kretschmann, Opt. Commun. 26, 41 (1978). J.R. Sambles and N.J. Smith, Mol. Cryst. Liq. Cryst. 347, 37 (2000). N.J. Smith and J.R. Sambles, J. Appl. Phys. 85, 3984 (1999). N.J. Smith and J.R. Sambles, Mol. Cryst. Liq. Cryst. 347, 45 (2000). D.W. Berreman, J. Appl. Phys. 46, 3746 (1975). C.J. Gerritsma, C.Z. van Doorn, and P. van Zanten, Phys. Lett. A 48, 263 (1974). F. Nakano, H. Kawakami, H. Morishita, and M. Sato, Jpn. J. Appl. Phys. 19, 659 (1980).
Nanotechnology and Nano-Interface Controlled Electronic Devices Editors: M. Iwamoto, K. Kaneto and S. Mashiko © 2003 Elsevier Science B.V. All rights reserved
CHAPTER 16
Explanation of the static and dynamic director orientation in thin nematic liquid crystal films using deuterium NMR spectroscopy Akihiko Sugimura a and Geoffrey R. Luckhurst b a Department
of Information Systems Engineering, Osaka Sangyo University, 3-1-1 Nakagaito, Daito, Osaka 574-8530, Japan E-mail:
[email protected] b Department of Chemistry and Southampton Liquid Crystal Institute, University of Southampton, Highfield, Southampton, SO17 1BJ, UK E-mail:
[email protected]
1. 2. 3. 4.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Theoretical background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Static director distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1. Bistable director orientation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2. Continuous director orientation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5. Director dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1. Director dynamics for the orthogonal geometry of B and E . . . . . . . . . . . . . . . . 5.2. Director dynamics for the non-orthogonal geometry of B and E . . . . . . . . . . . 6. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
313 315 320 321 321 323 331 331 335 346 348 348
1. Introduction Nuclear magnetic resonance (NMR) is widely used in the study of liquid crystals. Deuterium NMR has proved to be especially important for the investigation of liquid crystals because the spectra of specifically or fully deuteriated materials are rather simple compared to the corresponding proton NMR spectra [1–5]. The quadrupolar splitting for deuterons observed in the liquid crystal phase is related to the second rank
314
A. Sugimura and G.R. Luckhurst
orientational order parameter for the C–D bond direction and so deuterium NMR has been widely used for the study of the orientational order of liquid crystals and their phase transitions [2]. The quadrupolar splitting is also determined by the angle between the director and the magnetic field. In consequence, deuterium NMR spectroscopy is providing to be a powerful method with which to investigate the director orientation and its distribution, as well as the director dynamics in nematic [6–17] and smectic liquid crystals [18–20]. In more recent years deuterium NMR spectroscopy, combined with continuum theory, has been applied successfully to investigate the static director distribution in thin nematic liquid crystal cells, with different film thicknesses and different surface anchoring strengths, subject to both magnetic and electric fields [11– 13]. Deuterium NMR spectroscopy has also been employed to investigate the dynamic director alignment process in a thin nematic film following the application or removal of an electric field [14–17]. This technique has the added advantage that the presence of the magnetic field of the spectrometer ensures that during the electric field-induced alignment the director rotates as a monodomain [13] which facilitates the analysis of the results. In this chapter, we describe some of our studies of the static director distribution in thin nematic liquid crystal cells with different film thicknesses and different surface anchoring strengths using a combination of deuterium NMR spectroscopy and continuum theory. The nematic liquid crystal, 4-pentyl-d2 -4 -cyanobiphenyl (5CB-d2 ) deuteriated in the α-position of the pentyl chain, was confined between two glass plates with both weak and strong anchoring conditions; the anchoring strengths were measured by using a saturation voltage method [21]. A series of deuterium NMR spectra was acquired as a function of the applied electric field, which can be used to explore the director deformation. We also describe the application of deuterium NMR spectroscopy to investigate the director dynamics in the same nematic liquid crystal (5CB-d2 ) confined between two glass plates and subject to magnetic, B, and AC electric, E, fields. The cell was set in the NMR probe with the electric field, whose direction is normal to the substrate surface, making an angle of about 45°, with the magnetic field. This experimental geometry allows a unique director motion in the alignment process. In the absence of the electric field the director for 5CB will align parallel to the magnetic field because the diamagnetic anisotropy, ∆χ, ˜ is positive. When an electric field, which is strong enough to overcome the magnetic torque, is applied then the director will make an angle with the electric field since the dielectric anisotropy, ∆˜ε , is also positive. After the electric field is switched off, the director will then move from being at an angle to the magnetic field to being parallel to it. The dynamics of the director relaxation can be followed by monitoring the NMR spectrum during this alignment process, as a function of time. That is, the time dependence of the director orientation during and after the application of an electric field is studied. This is possible because the NMR spectrum for a monodomain sample with one group of equivalent deuterons, having a negligible dipolar interaction, contains a simple quadrupolar doublet whose separation is determined by, among other things, the angle, θ, made by the director, n, with the magnetic field. The layout of this chapter is as follows. In the next section we give the theoretical background to both the NMR experiment and the continuum theory for the director distribution; this section is also concerned with the determination of the surface
Static and dynamic director orientation in thin nematic liquid crystal films
315
anchoring energy. The NMR experiments are described in Section 3. The results for the static director distribution are given in Section 4 where they are discussed in the context of the continuum theory analysis. The results for the director dynamics are described in Section 5 where they are discussed on the basis of the hydrodynamic theory analysis. This allows the ratio ∆˜ε /∆χ, ˜ the diamagnetic anisotropy ∆χ, ˜ the field-induced relaxation times and the rotational viscosity coefficient to be determined provided ∆χ˜ is known independently. Our conclusions are in Section 6. 2. Theoretical background Deuterium has a nuclear spin of one and so possesses a quadrupole moment, which interacts with the electric field gradient at the nucleus, to give a quadrupolar interaction tensor. This does not influence the number of lines in the deuterium NMR spectrum for a normal liquid because the rapid and random molecular motion averages the quadrupolar interaction to zero. The NMR spectrum of a single deuteron, therefore, contains a single line composed of a pair of degenerate transitions as indicated schematically in Fig. 1a. In a liquid crystal phase this degeneracy is removed because of the intrinsic long range orientational order combined with the quadrupolar interaction of the deuterium nuclei. For a monodomain sample, in which the director is uniformly aligned, the NMR spectrum consists of a single doublet (see Fig. 1b) this is also observed for sets of equivalent deuterons provided the dipolar interaction is negligible in comparison with the linewidth. The separation, ∆˜ν , between the quadrupolar split lines is related to the components of the Saupe ordering matrix and the quadrupolar tensor, q. However, for aliphatic deuterons the quadrupolar tensor is, to a good approximation, cylindrically symmetric
Fig. 1: A typical deuterium NMR spectrum of a single deuteron in an isotropic phase (a) and a partially ordered system (b).
316
A. Sugimura and G.R. Luckhurst
about the C–D bond direction, and then the quadrupolar splitting is given by [2] 3 (1) ∆˜ν0 = qCD SCD , 2 where SCD is the orientational order parameter for the C–D axis and qCD is the quadrupolar coupling constant. In this analysis the director is taken to be parallel to the magnetic field. However, when the angle the director makes with the magnetic field changes, the splitting will also change in a well-defined way. Thus if the director makes an angle, θ, with the magnetic field of the NMR spectrometer the quadrupolar splitting is given by [7] ∆˜ν (θ) = ∆˜ν0 P2 (cos θ),
(2)
where ∆˜ν0 is the splitting when the director is parallel to the magnetic field (see Eq. 1) and P2 (cos θ) is the second Legendre polynomial. As the director moves away from being parallel to the field the splitting is predicted and observed to decrease, pass through zero at the magic angle (θ = 54.74°) and then to increase to one half of the original splitting, ∆ν˜ 0 , when the director is orthogonal to the magnetic field. Strictly the quadrupolar splitting changes sign at the magic angle but the sign of the splitting is not directly available from the spectrum. The angular variation of the deuterium NMR spectrum predicted from Eq. 2 is illustrated in Fig. 2. One of the prime advantages in the use of NMR spectroscopy to determine the director orientation is that the form of the spectrum is also influenced by the distribution of the director with respect to the magnetic field. In other words we can see from the spectrum whether the sample is a monodomain and if not the form of the director distribution can be estimated given the aid of some theoretical prediction. This situation obtains because when the director is not uniformly aligned the observed spectrum is a weighted sum of the spectra from all director orientations provided the molecular diffusion between different director orientations is slow on the NMR timescale, which is usually the case. The form of the spectrum can be simulated provided the probability, P(θ), of finding the director at an angle θ to the magnetic field is known. Then the observed spectrum is given by (3) I (ν) = L(ν, ν˜± (θ), T2−1 )P(θ) sin θ dθ. In this expression L(ν, ν˜± (θ), T2−1 ) denotes the Lorentzian shape of a spectral line centered at either ν˜ + (θ) or ν˜ − (θ) and with a linewidth at half height of 2T2−1 . The form of the normalized Lorentzian lineshape is 2 T2−1 + (ν − ν˜ ± (θ))2 , (4) L(ν, ν˜± (θ), T2−1 ) = π −1 T2−1 where the angle dependent resonance frequency is ν˜ + (θ) = ν0 + (∆˜ν0 /2)P2 (cos θ),
(5)
with an analogous expression for the resonance frequency ν˜ − (θ) for the other component of the quadrupolar doublet. As an example of the spectrum observed when the director is not uniformly aligned we consider the limiting case when the director is randomly
Static and dynamic director orientation in thin nematic liquid crystal films
317
Fig. 2: The dependence of the deuterium NMR spectrum on θ , the angle between the director and the magnetic field. The value of θ is increased from 0 to 90° in steps of 2°. The simulated powder spectrum of a nematic sample in which the director is randomly distributed in three dimensions is also shown.
distributed in three dimensions, that is the normalized distribution, P(θ), is just 1/2. The simulated spectrum corresponding to this limit is also shown in Fig. 2. The spectrum contains two dominant features at its center which come from the director being perpendicular to the magnetic field and so have a splitting of approximately ∆ν˜ 0 /2. The other feature is seen at the extremes of the spectrum and has a separation of approximately ∆˜ν0 ; these weak shoulders originate from the director being parallel to the magnetic field. Deuterium NMR provides a very good method for investigating the director distribu-
318
A. Sugimura and G.R. Luckhurst
tion in a thin nematic cell subject to an electric field because of the sensitivity of the spectrum to the director orientation as indicated in Fig. 2. By varying the intensity of the electric field, the total balance of the magnetic field, the electric field, the surface force and the elastic torque can be controlled. In other words the application of an electric field makes it possible to investigate the variation of the director distribution in a thin nematic cell with the electric field even in the presence of the strong magnetic field of the NMR spectrometer. However, to help interpret the deuterium NMR spectra, especially when the director is not uniformly aligned it is valuable to have some theoretical guidance as to the form of the director distribution. We shall, therefore, now consider the continuum theory for the nematic director subject to both magnetic and electric fields as well as surface anchoring. We begin with the total free energy, G, of the system in order to understand the director distribution in a thin nematic cell subject to a variety of torques. Here G contains four terms associated with elastic distortions, G d , surface anchoring, G s , and with interactions of the nematic with the two fields, G m and G e , namely (6) G = (G d + G m + G e ) dz + G s , where the integral is over the z coordinate (corresponding to the direction normal to the electrodes). The coordinate system defined by our experiments and used in the calculations is shown in Fig. 2, in which the magnetic and electric fields are applied parallel and normal to the electrodes, respectively. Because of the large difference in the magnitudes of the electric and magnetic susceptibilities of organic materials, the field energy terms have different forms. For the magnetic field, the diamagnetic susceptibility and its anisotropy are small (∆χ˜ ≈ 10−6 ) so that the nematic does not significantly perturb the applied magnetic field. In contrast for the electric field, both the permittivity and the permittivity anisotropy are large so that the electric field and the displacement can be significantly altered by the presence of the nematic [22]. The contribution to the free energy associated with the magnetic field and the diamagnetic anisotropy can be written as ∆χ˜ (B · n)2 , (7) Gm = − 2µ0 where µ0 is the permeability of free space and ∆χ˜ is the anisotropic diamagnetic susceptibility of the nematic. The free energy contribution due to the electric field has the analogous form ε0 ∆˜ε (8) (E · n)2 , 2 where ε0 is the permittivity of free space. The application of a voltage across the nematic film results in an electric displacement D, Ge = −
Dα = ε0 ε˜ ⊥ E α + ε0 ∆˜ε n α n β E β ,
(9)
where ∆˜ε = ε˜ − ε˜ ⊥ , and ε˜ and ε˜ ⊥ are the principal values of the nematic dielectric susceptibility tensor, parallel and perpendicular to the director, respectively. Assuming that D and E vary only in the z direction, and neglecting the effects of space charge,
Static and dynamic director orientation in thin nematic liquid crystal films
319
∇ · D = 0 implies that Dz is a constant across the nematic film. Dz can be found from Eq. 9 as Dz = ε0 ε˜ ⊥ E α + ε0 ∆˜ε sin2 θ(z) E(z), (10) where the dependence of the director orientation on the position in the cell is made explicit. The surface anchoring energy, G S , may be expressed as [23] A G S = − (n · e)2 . (11) 2 Here A is the surface anchoring strength and e denotes the easy axis or the anchoring direction as given by de Gennes [24]. From the coordinate system defined in Fig. 2 the components of the director n, the electric field, E, and the magnetic field, B, can be written as n = (cos θ(z), 0, sin θ(z)),
(12)
E = (0, 0, E(z)), (13) B = (B, 0, 0). (14) The normal Euler–Lagrange approach to minimize the total free energy, including the unified surface anchoring energy, leads to the basic equations from which to calculate the director distribution θ(z) (see equations (18), (19), (22), and (23) in Ref. [25]). They can be expressed as
∆χ˜ 2 ε0 ∆˜ε 2 d2 θ(z) 1 d f (θ(z)) dθ(z) 2 f (θ(z)) + = B − (z) sin 2θ(z), (15) E dz 2 2 dθ dz 2µ0 2 f (θ(z))
A− dθ(z)
= sin 2(θ 0− − θ0− ), dz z=0 2
A+ dθ(z)
= − f (θ(z)) sin 2(θ 0+ − θ0+ ), dz z=
2 d ε0 ε˜ ⊥ + ε0 ∆˜ε sin2 θ(z) E(z) = 0, dz
V = E(z) dz,
(16)
(17)
(18)
0
where f (θ(z)) = K 1 cos2 θ(z) + K 3 sin2 θ(z),
(19)
and K 1 and K 3 are the splay and bend elastic constants, respectively, is the thickness of the nematic film, V is the voltage applied across the cell, and θ 0− and θ 0+ are the pretilt angles of the surface directors at z = 0 and z = , respectively; these angles can change depending on the torques acting on the surface directors. The angles made by the easy axis at z = 0 and z = are fixed; they are denoted by θ0− and θ0+ , respectively.
320
A. Sugimura and G.R. Luckhurst
A− and A+ are the surface anchoring strengths at z = 0 and z = , respectively. For our system we set θ0− = −θ0+ and A = A− = A+ . The torque balance equations at both surfaces (see Eqs. 16–17), give boundary conditions to solve the torque balance equation in the nematic film (see Eq. 15) and the perturbation of the electric field (see Eq. 18). Numerical solution of Eqs. 15–18 gives the director distribution across the nematic film as a function of the applied voltage. Many of the parameters occurring in the continuum theory analysis and needed to calculate the director distribution are already known. However, the determination of the surface anchoring energy, A, presents more of a challenge. There is a formal relationship between the anchoring strength and the saturation voltage, Vs , at which the director becomes completely homeotropic; that is, the director across the entire nematic film including the boundary layers is oriented along the field direction. This, for a nematic film with a uniform director orientation, gives a formal relationship between Vs and A [21], namely √ Vs ε0 ∆˜ε K 3 Vs ε0 ∆˜ε A= tanh . (20)
2 K3 The value of A can, therefore, be determined by measuring the saturation voltage. As expected from its definition, Vs can be determined precisely for the condition of zero optical retardation and then the unified surface anchoring energy can be estimated from Eq. 20. Because the optical retardation is inversely proportional to the applied voltage in the high voltage regime, the intersection of the extrapolated line for the values of the measured optical retardation and the horizontal inverse voltage axis gives the saturation voltage, Us .
3. Experimental The nematogen used for our studies of the static director distribution as well as for its dynamics was 5CB-d2 , which had been specifically deuteriated in the α-position of the pentyl chain. This was prepared using a procedure described elsewhere [26] but with the reduction of the ketyl group performed using lithium aluminium deuteride rather than the hydride. A thin nematic sandwich cell was prepared. The glass plates were coated with transparent In2 O3 to act as electrodes. The cell was held together by a special glue which is stable in the presence of the cyanobiphenyls and which can be cured using UV radiation for a few minutes. The saturation voltage method [21] was employed to measure the surface anchoring strength, A, at the interface of 5CB with the substrate surface. All of the measurements were made at different temperatures in the nematic phase. The arrangement of the nematic cell in the NMR spectrometer is shown schematically in Fig. 3. The spectra were recorded using a JEOL Lambda 300 spectrometer, which has a magnetic flux density, B, of 7.05 T. The spectra were obtained using a quadrupolar echo sequence, with a 90° pulse of 7.7 µs and interpulse delay of 40 µs. The post delay time was 30 ms. The number of free induction decays (FID) used to produce spectra with good signal-to-noise varied from 2,000 to 10,000 depending on
Static and dynamic director orientation in thin nematic liquid crystal films
321
Fig. 3: Experimental setup of the nematic cell in the superconducting solenoid of the NMR spectrometer.
the sharpness of the spectral lines observed during the alignment process. The nematic cell was held in the NMR probe head so that the electric field, whose direction is normal to the substrate surface, makes an angle with the magnetic field. An amplifier and a function generator were used to provide a 1 kHz sinusoidal ac electric field to the cell. This frequency is sufficient to overcome the alignment effects of ionic conduction. By changing the intensity of the electric field, the director orientation rotates in a plane defined by B and E, since the diamagnetic anisotropy and the dielectric anisotropy of 5CB are both positive.
4. Static director distribution 4.1. Bistable director orientation The nematogen studied in this experiments was 5CB-d2 . The glass surface of the nematic cell of 100 µm thick was not treated so that the effective anchoring strength was essentially zero. The nematic cell was arranged in the NMR probe head so that the glass plates and the rubbing direction were aligned parallel to the magnetic field. In our experiment the projection of the easy axis onto the glass surface is parallel to the magnetic field. The fine adjustment of the cell alignment was carried out by switching the electric field on while the sample was in the nematic phase and rotating the cell using the goniometer of the spectrometer until a doublet splitting was obtained which was 1/2 (to within ±0.2 kHz determined by the linewidth) of the splitting without the electric field (this uncertainty in the splitting means that the electric field can be arranged to be orthogonal to the magnetic field to within ±3°). The deuterium NMR spectra, measured as a function of the applied electric field, are given in Fig. 4. The spectra show that for small electric fields the director remains parallel to the magnetic field giving rise to the large quadrupolar splitting. Then when the potential is about 60 V, corresponding to a threshold value of the electric field, a
322
A. Sugimura and G.R. Luckhurst
Fig. 4: The deuterium NMR spectra measured as a function of the applied electric field at 298 K for the 100 µm nematic cell with untreated glass surfaces.
weak quadrupolar doublet appears in the spectrum with a splitting which is essentially half that of the original splitting. This shows that it comes from the director being orthogonal to the magnetic field and so parallel to the electric field. As the potential is increased so the intensity of the quadrupolar doublet associated with the director parallel to the electric field grows at the expense of that originating from the director parallel to the magnetic field. When the potential is 70 V the director would seem to be completely aligned parallel to the electric field. This voltage dependence of the quadrupolar splitting can be understood by considering Eq. 15, which applies when B and E are orthogonal to each other, since the surface anchoring strength (Eqs. 16–17) is zero in this thick cell and so director deformation due to surface effects can be ignored. Accordingly there is no director deformation in the nematic film and the director distribution should be uniform giving a monodomain in the film. That is, the left hand side of Eq. 15 can be set to be zero and this gives two kinds of solution. One is that the director is oriented parallel to the magnetic field. This gives a spectrum with the maximum splitting, which is shown by the lines with large dashes in Fig. 4. The other is with the director aligned parallel to the electric field and this gives a spectrum with half the value of the splitting in zero electric field; this is indicated by the lines with small dashes in Fig. 4. However, as we have noted there is a narrow voltage range over which
Static and dynamic director orientation in thin nematic liquid crystal films
323
Table 1 The nematic cells investigated
Group 1 weak anchoring
Group 2 strong anchoring
Cell no.
Thickness
(µm)
TNI (K)
Anchoring strength A (×10−4 J/m−2 )
(A) (B) (C) (D) (E) (F)
7.1 36.8 56.7 7.6 37.8 56.7
305.0 305.2 305.2 305.3 305.3 305.0
0.930 0.194 0.101 34.6 17.2 4.20
both splittings are found at the same time; these two doublets appear at V = 62 V. The simultaneous appearance of the two doublets is, we believe, due to the inhomogeneity in the thickness of the cell. The exact thickness of the nominal 100 µm cell is uncertain, as we do not have a method of accurately measuring the thickness and its variation over the entire cell; we are, therefore, using the nominal value of the spacer given by the supplier. If the thickness variation over the entire cell is say from 100 to 105 µm, the electric field strength over the nematic film has a gradient sufficiently large to give the two quadrupolar doublets shown in Fig. 4 in the voltage range 62 to 67 V. 4.2. Continuous director orientation The arrangement of the nematic cell in the NMR spectrometer is the same as that described in the previous subsection. Six thin nematic sandwich cells with different thicknesses and anchoring strengths were prepared. The In2 O3 coated glass surfaces in group 1, with (A) 7.1 µm, (B) 36.8 µm, and (C) 56.7 µm thick cells, listed in Table 1, were rubbed unidirectionally in a parallel manner to produce a uniform planar director alignment. The surface anchoring strength at the interface of 5CB-d2 and the substrate surface for each cell used in the present experiments was measured using the saturation voltage method described in Section 2. The surface anchoring strengths were found to be in the range of 1.01–9.30 × 10−5 J/m2 ; this corresponds to the so-called weak anchoring condition. The surfaces of the cells in group 2, with (D) 7.6 µm, (E) 37.8 µm, and (F) 56.7 µm thick cells, listed in Table 1, were coated with a thin polyimide film and also rubbed unidirectionally in a parallel manner to produce a uniform planar director alignment. The surface anchoring strength was found to be in the range of 4.20–34.6 × 10−4 J/m2 , which corresponds to the so-called strong anchoring condition. The results are shown as a function of film thickness in Fig. 5 where the solid lines are a guide to the eye. It is obvious from these results that the surface anchoring strength increases with decreasing nematic film thickness, as previously reported [26,27]. The nematic–isotropic transition temperature, TNI , for the nematic in each of the cells of the two groups are also given in Table 1 and are essentially the same as for the bulk sample. The deuterium NMR measurements were carried out at a constant temperature of 2 K below TNI for the cells in the two groups
324
A. Sugimura and G.R. Luckhurst
Fig. 5: The dependence of the surface anchoring strength for the cells listed in Table 1 on their thickness.
Fig. 6: The deuterium NMR spectra measured for three nematic cells (A) 7.1 µm thick, (B) 36.8 µm thick and (C) 56.7 µm thick with weak anchoring conditions, as a function of the electric field strength.
The deuterium NMR spectra measured for the three nematic cells with weak anchoring are shown in Fig. 6 (A) (7.1 µm thick), (B) (36.8 µm thick), and (C) (56.7 µm thick). The number of FIDs used to obtain spectra for each cell was 20,000. The voltage dependence of the spectra shows that with increasing electric field strength the quadrupolar splitting is reduced, passes through zero and then increases again to a value
Static and dynamic director orientation in thin nematic liquid crystal films
325
Fig. 7: The deuterium NMR spectra measured for three nematic cells (D) 7.6 µm thick, (E) 37.8 µm thick and (F) 56.7 µm thick with strong anchoring conditions, as a function of the electric field strength.
which is essentially half of that at zero electric field. It would seem that the director orientation changes more or less continuously from being parallel to the magnetic field to being orthogonal to it, as the electric field grows. This change of the quadrupolar splitting follows the pattern illustrated in Fig. 1 in accord with Eq. 2. It is of interest to note that the spectra are essentially, but not exactly, the same for the same electric field strengths irrespective of the cell thickness, as expected from theory. Also as the electric field is increased, the lines appear to broaden necessarily leading to a decrease in the signal-to-noise ratio. For higher electric fields, however, the lines sharpen again resulting in an improved signal-to-noise ratio. Thus the experimental results indicate a continuous change in the director orientation with increasing electric field and a slight broadening of the director distribution. This broadening is especially apparent in the center of the range when the director is at angle of 45° to the magnetic field, as expected theoretically [28]. The deuterium NMR spectra measured for the nematic cells with a strong anchoring are shown in Fig. 7 (D) (7.6 µm thick), (E) (37.8 µm thick), and (F) (56.7 µm thick). The number of FIDs used to record the spectra for each cell was again 20,000. Somewhat surprisingly the variation of the spectra with the applied electric field is similar to that found for the weak anchoring cells (cf. Fig. 6 (A), (B), and (C)). A major question is why do the quadrupolar splittings shown in the NMR spectra given in Figs. 6 and 7 change continuously with increasing electric field whereas for the thick cell with zero anchoring strength only two discrete bistable director orientations
326
A. Sugimura and G.R. Luckhurst
Fig. 8: The geometry used for the experiment and the analysis. The z-axis is taken as being parallel to the magnetic field. The z 1 -axis is taken as being normal to the glass plate. The magnetic field (B), electric field (E) and director (n) are in the xz-plane. The director and electric field make the angles θ and α, respectively, with the magnetic field. The director makes an angle, φ (= θ + 90° − α), with the substrate surface.
are observed. As we mentioned in the Introduction, the quadrupolar splitting is directly related to the director orientation. That is, the experimental results in Figs. 6 and 7 show that the director rotates continuously with increasing electric field from being parallel to the magnetic field to being orthogonal to it and hence parallel to the electric field. It is to be anticipated that the director deformation within the nematic film should be taken into consideration for the cells (A)–(F) with surface treatment because the surface anchoring affects the director distribution in the nematic film. In other words the elastic torque in Eq. 15 must be considered in order to understand the variation in the director orientation across the cell with strong surface anchoring. However, it is generally expected for the thicker cell and zero anchoring strength that the director deformation is limited to the vicinity of the substrate surface and that the director orientation is constant over almost the entire region of the nematic film. This behaviour is also consistent, as we shall see, with the following simulation results for the director distribution. Accordingly we think that one reason for the observed continuous rotation of the director is associated with the experimental geometry. As we have mentioned in the previous section, it is important for the observation of the bistable director orientation to set the cell precisely in the NMR probe so that B and E are orthogonal to each other. In a real experiment (as shown in Fig. 2) it is difficult to set the electric field exactly perpendicular to the magnetic field and so we consider here the more general experimental geometry given in Fig. 8 in which the electric field makes an arbitrary angle α with the magnetic field. Then the general torque-balance equation for the bulk nematics, given in Eq. 15, should be changed to
d2 θ(z) 1 d f (θ(z)) dθ(z) 2 + f (θ(z)) dz 2 2 dθ dz ∆χ˜ 2 ε0 ∆˜ε 2 =− B sin 2(θ(z) + α) − (21) E (z) sin 2θ(z). 2µ0 2
Static and dynamic director orientation in thin nematic liquid crystal films
327
In the limit that surface forces are negligible, as for the 100 µm sample, the director will be uniform and so the elastic terms on the left hand side will vanish. Then the solution to Eq. 21 will give a continuous variation in the director orientation with increasing electric field for all the angles α between the two fields except 90°. When the surface forces cannot be ignored it is necessary to solve the torque balance equation, numerically, as we shall see. In general the director orientation is not expected to be uniform over the entire nematic film because of surface anchoring. Then for this non-uniform state the observed NMR spectrum is a weighted sum of spectra from all director orientations as we saw in Section 2. The voltage dependence of the deuterium NMR spectra shown in Figs. 6 and 7 reflect the director distribution throughout the sample, which is determined by a combination of the unified surface anchoring, elastic, magnetic, and dielectric energies. However, the NMR spectra and their variation with the applied voltage suggest that the director is more or less uniformly aligned across the nematic film. Given the unexpected nature of this result we have sought to confirm it by simulating the spatial variation of the director. This is obtained theoretically via the numerical solution of Eqs. 16–18 and 21 as a function of the electric field. In the numerical calculation of the director distribution, Eqs. 16 and 17 give the boundary conditions to solve the torque balance equation 21. The pretilt angles, θ 0+ and θ 0− , of the surface directors in Eqs. 16 and 17 are variable because they can be changed by the external fields, both B and E. That is, Eqs. 16–18 and 21 have to be solved in a self-consistent manner. In our calculations the values B = 7.05 T, θ0+ = 6°, θ0− = −6°, ∆χ˜ = 1.5 × 10−6 [29,30], ∆˜ε = 6.75 (˜ε⊥ = 7.16), K 1 = 3.44 × 10−12 N, K 3 = 4.07 × 10−12 N [31], α = 88° (for the cells (D) and (E)), α = 85.5° (for the cell (F)) and A = 3.46 × 10−3 J/m2 were used. The values of α were determined from the experimental results. Figs. 9 (D), (E), and (F) show typical examples of the simulation results for the director distribution θ(z/ ) across the entire nematic film as a function of the electric field for the strong anchoring cells (D), (E), and (F), respectively. In these figures the profiles of the director distribution show a top-hat shape different from the bowler-hat shape, which is well-known as the director profile for the case with only an electric field. To emphasise the major effect of the magnetic field on the director orientation across the film we show in Fig. 9 (D ), (E ) and (F ) the analogous distributions calculated with B equal to zero. Now the director orientation undergoes a significant variation across the film which we describe as the bowler-hat distribution. In both the top-hat and the bowler-hat distributions there is clearly an asymmetry which results from the easy axis at the surfaces making angles which are opposite in sign. It may be helpful to discuss the origin of the major difference in the shape of the director profile before starting our main discussion on the director distribution. When a constant magnetic field is applied parallel to the substrate surface, the surface director tends to reduce its pretilt angle, θ 0± . In other words, the pretilt angle of the surface director is also a function of the magnetic field strength. This is a completely different situation to that without the magnetic field. This means that a strong deformation energy is localized near the substrate surface in the initial director distribution before the application of the electric field to the cell. This feeds back to the bulk torque balance equation 21 and the boundary conditions, that is, Eqs. 16 and 17 result in showing a top-hat shape for the director profile. Now we return to our
328
A. Sugimura and G.R. Luckhurst
Fig. 9: Typical continuum theory predictions for the director distribution θ (z/ ) across the entire nematic film as a function of the electric field. Figures (D), (E), and (F) correspond to the director distribution predicted for the strong anchoring cells (D), (E), and (F), respectively.
main discussion. The director is found to be aligned almost parallel to the magnetic field for low applied voltages. The pretilt angle of the director at the surface increases slightly with increasing electric field, because of the strong anchoring condition. As
Static and dynamic director orientation in thin nematic liquid crystal films
329
the electric field strength is increased so the director deviates to an increasing but essentially continuous extent from being parallel to the magnetic field. In the bulk of the nematic the director is more or less uniformly aligned at a constant angle with respect to the magnetic field. This angle increases with increasing field until it reaches its limiting value of α. Near the substrate surface the director orientation changes rapidly and to an extent which increases with the electric field. However, as the thickness of the nematic film increases so the relative distance at the interface where the director changes its orientation becomes smaller. Similar numerical calculations provide the director distribution θ(z/ ) over the entire nematic film for the weak anchoring cells as a function of the electric field. These simulations predict that the director distribution across the film is almost homogeneous except in the vicinity of the interface. The associated NMR spectrum should, therefore, consist of a single quadrupolar doublet. It seems clear both for weak and strong anchoring that the inhomogeneity of the director distribution near the interface causes the broadening of the spectral lines and the associated decrease of the signal-to-noise ratio. In order to test this interpretation quantitatively, it is necessary to simulate the deuterium NMR spectra directly by using the theoretical director distributions. The NMR spectrum can be simulated from a knowledge of the director distribution as we saw in Section 2. However, since we know the director orientation for discrete positions across the cell it is convenient to replace the integral representation of the lineshape in Eq. 3 by the sum I (ν) =
1
L ν, ν± (θ(z/ )), T2−1 .
(22)
z/ =0
Here θ(z/ ) is the angle made by the director with respect to the magnetic field for a particular slice in the nematic film at z/ . The deuterium NMR spectra have been simulated in this way from the theoretical results for the different director distributions θ(z/ ) and Eq. 22 as a function of the voltage for each cell. The number of slices used was 200 within the nematic film. The value of the linewidth parameter, T2−1 , was determined from the observed spectra to be 1 kHz and independent of the director orientation. The spectral range used in the simulation was taken to be −50 kHz ≤ ν ≤ 50 kHz and is analogous to that of the experimental spectra. The simulated spectra are shown in Figs. 10 (A), (B), and (C) for the weak anchoring cells, and Figs. 11 (D), (E), and (F) for the strong anchoring cells. In these figures the solid lines indicate the simulated spectra and the dashed lines show the experimental spectra which were taken from those shown in Figs. 4 and 5 in order to compare the simulated results with experiment. The simulated spectra at different voltages are found to be in surprisingly good agreement with the experimental spectra. The only significant exception to this is for the spectra with the director close to the magic angle where the lines in the experimental spectra are observed to be broader than those predicted. To illustrate the possible origins of such discrepancies we consider the results for cell A shown in Fig. 10 where the agreement between theory and experiment is at its worst. Cell A is 7.1 µm thick so that the amount of 5CB-d2 that it contains is small which makes it difficult to obtain spectra with good signal-to-noise ratios. This
330
A. Sugimura and G.R. Luckhurst
Fig. 10: Simulated deuterium NMR spectra for the three nematic cells (A) 7.1 µm thick, (B) 36.8 µm thick and (C) 56.7 µm thick with weak anchoring as a function of the electric field strength. The solid lines indicate the simulated spectra and dashed lines the experimental spectra (see Fig. 6).
Fig. 11: Simulated deuterium NMR spectra for three nematic cells (D) 7.6 µm thick, (E) 37.8 µm thick and (F) 56.7 µm thick with strong anchoring conditions as a function of the electric field strength. The solid lines indicate the simulated spectra and dashed lines the experimental spectra (see Fig. 7).
proves not to be a problem for the extreme voltages where the lines are relatively sharp and it is easy to process the FID leading to the spectrum. However, for electric fields of 0.79 and 0.90 V/µm the spectral baseline is not flat which results from the low spectral intensity. At 0.75 V/µm (see Fig. 10) the lines in the experimental spectrum
Static and dynamic director orientation in thin nematic liquid crystal films
331
have broadened and in consequence the baseline is severely distorted. The origin of the broad spectral lines is the deviation of the director from a uniformly aligned state. Any such broadening of the director distribution is especially apparent in the NMR spectrum when the director makes an average angle of 45° with the magnetic field. This occurs because the quadrupolar splitting is especially sensitive to small changes in the angle θ between the magnetic field and the director for this orientation (see Eq. (2)). It should be stressed, however, that the width of the spectral lines when the electric field is 0.75 V/µm although large would only require a broadening of several degrees to account for it. The origin of the increased breadth of the director distribution may result from the balance between the dielectric and magnetic torques at this point so that other less controllable factors could influence the director distribution. Similar comments can be made for the results obtained with cell E where the disagreement between theory and experiment is especially apparent at an electric field of 0.71 V/µm (see Fig. 11). Nonetheless it is important to recognize that overall there is really quite remarkable agreement between the experimental spectra and those simulated using the director distribution obtained from continuum theory for quite different film thickness, potentials and surface anchoring strengths. In other words the combination of deuterium NMR spectroscopy and continuum theory gives us a good understanding of the director distribution in a thin nematic film subject to competing constraints.
5. Director dynamics 5.1. Director dynamics for the orthogonal geometry of B and E A thin nematic sandwich cell 37.8 µm thick was prepared. The transparent electrodes were covered with polyimide its surface was rubbed unidirectionally in a parallel manner to produce a uniform director orientation. The surface anchoring strength at the interface of 5CB-d2 with the substrate surface is 1.2 × 10−4 J/m2 corresponding to strong anchoring. All of the measurements were made at 303 K. The nematic cell was held in the NMR probe head so that the glass plates and the rubbing direction were aligned parallel to the magnetic field (see Fig. 2). In this experiment the projection of the easy axis onto the glass surface was parallel to the magnetic field. An amplifier and a function generator were used to provide a 10 kHz sinusoidal ac electric field to the cell. This frequency is sufficient to provide a time resolution of 0.1 ms during the turn-on dynamics measurements. On applying or removing the electric field, the director orientation rotates in a plane defined by B and E. Figs. 12 (A) and (B) show schematically the pulse sequences for the observation of the spectra during (A) the turn-on and (B) turn-off processes. The triangular symbols in the 2 H observation pulse sequences indicate data acquisition during the free induction decay. ta is the time during which the external voltage is applied, tr is the time allowed for director relaxation, tD (∼ 10 µs) is the dead time of the receiver coil, and tac (∼ 55 ms) is the acquisition time for the FID. For the turn-on process the director relaxation was monitored at several values of ta between 0 and 3 ms following the application of the electric field (60.5 V). For the turn-off process, the director relaxation
332
A. Sugimura and G.R. Luckhurst
Fig. 12: The pulse sequences used for the observation of (a) the turn-on and (b) the turn-off processes. The triangular symbols in the 2 H observation pulse sequences denote data acquisition, ta is the time during which the external voltage is applied, tr is the time allowed for director relaxation, tD is the dead time of the receiver coil, and tac is the acquisition time for the free-induction decay.
was measured at several values of ton in the range 0 to 7 ms. An electric field of 100 V was applied for 3 ms to obtain the initial director alignment orthogonal to the magnetic field.
Static and dynamic director orientation in thin nematic liquid crystal films
333
Fig. 13: The deuterium NMR spectra for the turn-on process recorded at 303 K. The spectra were measured by changing the application time, ta , for the voltage.
The deuterium NMR spectra recorded during the turn-on and turn-off processes are shown in Figs. 13 and 14, respectively. In the turn-on process, the spectra show that the alignment process is complete within 3 ms when the quadrupolar splitting is reduced to half of its initial value (the director is now parallel to the electric field which is orthogonal to the magnetic field). Following the application of the electric field the director remains parallel to the magnetic field for up to 0.7 ms before it changes its orientation, i.e., there is an induction period of 0.7 ms for the turn-on process. Then after this induction period the lines begin to develop an asymmetric and powder-like shape as the director moves away from being parallel to the magnetic field. This major broadening of the spectral lines for times up to 1.1 ms or more shows the presence of a broad director distribution in which the director adopts a range of intermediate orientations between 0° and curiously the magic angle (∼ 54.74°). At 1.2 ms most of the director has achieved an orientation with θ ≈ 54.74°. After this the director appears to move now more or less as a monodomain until at t = 3 ms it is completely aligned along the electric field and orthogonal to the magnetic field. One other feature of the alignment process in this cell which is worth pointing out is the presence of a significant fraction of the sample with the director parallel to the magnetic field for a fairly long time (up to 1.1 ms).
334
A. Sugimura and G.R. Luckhurst
Fig. 14: The deuterium NMR spectra for the turn-off process recorded at 303 K. The spectra were measured at a time tr after the voltage was turned off.
In the turn-off process, on the other hand, see Fig. 14, changes in the director orientation are observed at 1 ms after the removal of the electric field. For times shorter than 1 ms, the director is observed to remain in its initial orientation parallel to the electric field. In other words, and as expected, the turn-off process also exhibits the expected induction period. At t = 1 ms a second quadrupolar splitting, with somewhat broader lines, is observed which corresponds to the director oriented parallel to the magnetic field. Furthermore, the spectrum at t = 1 ms shows that the original lines at θ = 90° become broader and there is an increase in the spectral intensity in the region between these two lines. This indicates a spread in the director distribution between the extremes of 90° and 0°. Indeed as time evolves there is an increase in the spectral intensity in the region between the parallel and the perpendicular features of the spectra showing that the director is increasingly adopting a range of orientations between 90° and 0° to the magnetic field. The form of this powder pattern changes further with time as the director distribution alters to give more of the director parallel to the magnetic field. The spectrum recorded after 2.5 ms is of interest as it shows a maximum spectral intensity corresponding to an intermediate director orientation between 0° and 90°. After a further 0.2 ms the intensity of the parallel features has increased at the expense of the
Static and dynamic director orientation in thin nematic liquid crystal films
335
perpendicular while at the same time the splitting originating from intermediate director orientations has increased showing that the angle between the director and the magnetic field has decreased. These processes continue until after 7 ms the director is almost uniformly aligned parallel to the magnetic field (although there are still two very weak peaks coming from the director at θ = 90°). In summary the director starts uniformly aligned perpendicular to the magnetic field then passes through a regime in which its orientational distribution is very broad before becoming aligned parallel to the magnetic field. Such changes are consistent with those observed for other systems where there is a degeneracy in the alignment pathway for the director [10]. The alignment process for the change in the director orientation following the turn-on and turn-off of the electric field is clearly very complex. The director does not realign as a monodomain and this is perhaps to be expected either because of the degeneracy in the alignment pathway or because of the surface alignment. In addition the alignment process appears to be qualitatively and quantitatively different for the turn-on and turn-off process presumably because of the difference in the magnitude of the field torques responsible for the alignment. Analysis of the spectral lineshapes which indicate a non-uniform distribution of the director can, in principle, yield not only information about the viscosity coefficients but also the elastic constants. 5.2. Director dynamics for the non-orthogonal geometry of B and E A thin nematic sandwich cell ( = 56.1 µm) was prepared. The transparent electrodes were not treated in any way. The surface anchoring strength was found to be 1.0 × 10−7 J/m2 corresponding to weak anchoring at the untreated electrodes. All of the measurements were made at different temperatures in the nematic phase of 5CB. The nematic cell was held in the NMR probe head so that the electric field, whose direction is normal to the substrate surface, makes an angle, α ≈ 45°, with the magnetic field (see Fig. 8). Fig. 15 shows the spectrum recorded without an electric field at 295 K; the quadrupolar splitting, ∆ν˜ 0 , is 54.2 kHz. The weak feature in the centre of the spectrum originates from a small amount of the isotropic phase of 5CB-d2 possibly dispersed in the glue holding the cell together. The final adjustment of the cell alignment to ensure that the electric field makes an angle of essentially 45° with the magnetic field was carried out by switching on a large electric field (100 V) and rotating the cell by a few degrees, clockwise or counter-clockwise, using the goniometer of the spectrometer until a doublet splitting was obtained which is 1/4 of the splitting without the electric field (this condition would mean that the substrate surface makes an angle about 45° with the magnetic field). However, the electric field will not be at exactly 45° with respect to the magnetic field because the value of 100 Vrms is insufficient to align the director parallel to the electric field. An amplifier and a function generator were used to provide a 10 kHz sinusoidal ac electric field (50 V) to the cell. On applying or removing the electric field, the director orientation is expected to occur in a plane defined by B and E. The quadrupolar echo pulse sequences used for the observation of the spectra during (a) the turn-on and (b) the turn-off processes are same as those described in the previous Subsection 5.1 (see Fig. 12a and b). For the turn-on process the director relaxation was monitored at several values of ton between 0 and 25 ms following the application of the
336
A. Sugimura and G.R. Luckhurst
Fig. 15: The deuterium NMR spectrum of 5CB-d2 recorded without an electric field at 295 K, the quadrupolar splitting frequency, ∆˜ν , is 54.2 kHz and the effective spin–spin relaxation time, T2 , is 1 ms.
electric field (50 V). For the turn-off process, the director relaxation was measured at several values of toff in the range 0 to 25 ms. An electric field of 50 V was applied for ta = 50 ms to obtain the initial director alignment for the turn-off experiments. Deuterium NMR spectra obtained during the turn-on and turn-off processes at 295 K are shown in Fig. 16a and b, respectively. In the turn-on process, the quadrupolar splitting decreases and then saturates with time after about 3 ms. In the turn-off process, the quadrupolar splitting increases because the director moves from being at approximately 30° to the magnetic field to being parallel to it. The spectra recorded in the fast time region for the turn-on and turn-off processes contain weak oscillatory spectral features associated with the director rotation during the acquisition for the free induction decay. The origin of the oscillatory spectral features is understood [32] but they are not of importance for this investigation. However, it is of interest to note that the oscillations appear on the inside of the quadrupolar doublet during the turn-on process but on the outside of this doublet for the turn-off process. They therefore give the sense of the director motion. The time resolved director orientation can be easily determined from our experimental results contained in the quadrupolar doublets because of the simple relation between the quadrupolar splitting and the director orientation, as we shall now discuss. If we neglect the director distribution that causes the slight broadening of the spectral lines [33], then an appropriate form of the time dependence of the director orientation can be determined relatively easily. This is a good assumption since there is no elastic deformation in the nematic slab used in our experiments. The observed quadrupolar splitting can then be used together with Eq. (2) and the value of the splitting when the director is parallel to the magnetic field to determine the director orientation. According to continuum theory [34], we should consider the one-dimensional
Static and dynamic director orientation in thin nematic liquid crystal films
337
Fig. 16: The deuterium NMR spectra for (a) turn-on and (b) turn-off processes recorded at 295 K for 5CB-d2 . In the turn-on process, the spectra were measured by changing the application time for the voltage (ta ). In the turn-off process, the spectra were measured at a time tr after the voltage was turned off.
distortion of the director across the cell (see Fig. 8). All of the deuterium NMR spectra in our study appear to be dominated by a single doublet which allows us to determine the director orientation associated with an essentially monodomain sample. In this analysis, we therefore treat the director as being uniformly aligned. As shown in Fig. 1, the electric field makes an angle α with the magnetic field. The rate of change of the director orientation is given, for the turn-on process, by the torque-balance equation [35] which for a monodomain nematic [14] is γ1
dθ(t) ε0 ∆˜ε 2 ∆χ˜ 2 B sin 2θ(t) + =− E sin 2(α − θ(t)). dt 2µ0 2
The solution of Eq. 23 is obtained analytically as [14] θ(t) = θ∞ + tan−1 tan(θ0 − θ∞ ) exp(−t/τ ) ,
(23)
(24)
where θ∞ is the limiting value of θ(t) when t tends to infinity, τ is the relaxation time, and θ0 is the initial angle. The limiting angle θ∞ is given by Um + Ue cos 2α cos 2θ∞ = 1/2 , Um2 + 2Um Ue cos 2α + Ue2
(25)
338
A. Sugimura and G.R. Luckhurst
where ∆χ˜ 2 ε0 ∆˜ε 2 B , Ue = E , 2µ0 2 are the strengths of the magnetic and electric interactions, respectively. The relaxation times for the turn-on (τon ) and turn-off (τoff ) processes are then −1/2 τon = τm−2 + 2τm−1 τe−1 cos 2α + τe−2 , (26) Um =
and τoff(≡ τm ) = γ1 /(2Um ),
(27)
(28) τe = γ1 /(2Ue ), respectively and τe is the electric field-induced relaxation time for the director. Eqs. 25–27 give the following simple relationships used in the analysis of our results, sin2 2α =
R 2 sin2 2θ∞ , R 2 − 2R cos 2θ∞ + 1
B2 Um = Ue µ0 ε0 E 2
∆χ˜ ∆˜ε
=
(29) 1
R 2 − 2R cos 2θ∞ + 1
,
(30)
where R = τoff /τon .
(31)
Eqs. 29 and 30 give the values of α and ∆χ˜ /∆˜ε , respectively, by using τon , τoff and θ∞ . It is also evident from Eqs. 27 and 30 that ∆χ˜ and γ1 can only be determined if ∆˜ε is known from an independent experiment. Fig. 17 shows the temporal variation in the ratio of the quadrupolar splittings, ∆ν(t)/∆˜ ˜ ν0 , determined from the time resolved deuterium NMR spectra shown in Fig. 16a and b for the turn-on and turn-off processes. We can see in Fig. 17 that for the turn-on process the director rotates from the initial angle θ0 = 0° for ∆ν˜ 0 = 54.2 kHz and then aligns at the limiting angle, θ∞ , of 29.8° for ∆ν˜ ∞ = 34.1 kHz (see Eq. 2), the limiting value of ∆ν(t) ˜ as t tends to infinity. The limiting value of θ∞ was obtained by using ∆˜ν∞ and Eq. 2. In the turn-off process, the time dependence of the director orientation was obtained in the same way and is shown in Fig. 17. The director rotates back parallel to the magnetic field and the time taken for the alignment process is slower in the turn-off process than in the turn-on process. This is clearly apparent from the results listed in Table 2 but is not so obvious from the different time dependences of the quadrupolar splittings shown in Fig. 17. The large difference in the relaxation times for the turn-on and turn-off processes follows from the theory. Thus the relaxation time, τoff, for the turn-off process is independent of the electric field (see Eq. 27) whereas τon for the turn-on process depends on both the electric field strength and its orientation, α, with respect to the magnetic field (see Eq. 26). In our experiments α is approximately 45° so that the cross term in the denominator is negligible, accordingly the quadratic term in τe simply adds to that in τm thus increasing the denominator and so decreasing the relaxation time τon in comparison with τoff , as we find experimentally.
Static and dynamic director orientation in thin nematic liquid crystal films
339
Fig. 17: The time dependence of the ratio, ∆˜ν/∆˜ν0 , for (a) the turn-on and (b) the turn-off processes determined from the deuterium NMR spectra of 5CB-d2 shown in Fig. 16a and b at 295 K, in which the experimental value, ∆ν˜ 0 = 54.2 kHz (see Fig. 2), has been used. For the turn-on process, the director aligns towards the limiting value (θ∞ = 29.8°) with time. For the turn-off process, θ relaxes back from θ0 = 28.8° to θ∞ = 0°. The solid lines in (a) and (b) are the best fits to ∆˜ν /∆˜ν0 = 3(cos2 θ (t) − 1)/2, where θ (t) = tan−1 [tan(−29.8°) e−(t (ms))/0.766 ] + 29.8° and θ (t) = tan−1 [tan(28.8°) e−(t (ms))/1.54 ], respectively. The time constants obtained from these fitts are τon = 0.766 ms for the turn-on and τoff = 1.54 ms for the turn-off.
From the experimental values of ∆ν˜ 0 and ∆ν˜ ∞ combined with Eqs. 2 and 24, the values of the two relaxation times, τon and τoff, were obtained by fitting the ratio of the quadrupolar splittings as a function of time for the turn-on and turn-off processes at each temperature. The solid lines in Fig. 17 show the best-fit lines giving the values of τon and τoff to be 0.766 ms and 1.54 ms, respectively. We have chosen to fit the time dependence of the quadrupolar splittings rather than that for the director orientation calculable from Table 2 The temperature dependence of the quadrupolar splitting, ∆ν˜ 0 , the ratio, ∆˜ν∞ /∆ν˜ 0 , of the quadrupolar splittings with and without the electric field, the turn-on, τon , and turn-off, τoff , relaxation times and the angle, α, between the electric and magnetic fields T (K)
∆ν˜ 0 (kHz)
∆ν˜ ∞ /∆˜ν0
τr (ms)
τd (ms)
α (°)
295 297 299 301 303 303.5 304 304.5 305 305.5
54.2 52.1 49.6 46.4 44.6 42.6 41.2 39.5 37.6 34.7
0.630 0.632 0.636 0.636 0.627 0.636 0.637 0.636 0.634 0.623
0.766 0.672 0.594 0.514 0.443 0.424 0.404 0.384 0.348 0.314
1.54 1.34 1.18 1.02 0.88 0.84 0.80 0.76 0.70 0.62
44.7 44.7 44.6 44.6 44.7 44.7 44.5 44.7 44.7 44.7
340
A. Sugimura and G.R. Luckhurst
Table 3 The temperature dependence of the dielectric anisotropy, ∆˜ε [29], the diamagnetic anisotropy, ∆χ˜ , the ˜ ε , and the ratio of the rotational viscosity coefficient, γ1 , the ratio of the anisotropic susceptibilities, ∆χ/∆˜ diamagnetic anisotropy to the quadrupolar splitting, ∆χ˜ /∆ν˜ 0 T (K)
∆˜ε
∆χ˜ (×106 )
γ1 (Pa s)
(∆χ˜ /∆˜ε ) (×107 )
(∆χ˜ /∆˜ν0 ) (×1011 kHz)
295 297 299 301 303 303.5 304 304.5 305 305.5
11.4 11.2 10.8 10.2 9.40 9.19 8.95 8.68 8.40 8.15
1.17 1.16 1.13 1.07 0.973 0.962 0.940 0.909 0.865 0.846
0.0710 0.0614 0.0526 0.0432 0.0339 0.0319 0.0297 0.0273 0.0239 0.0207
1.03 1.04 1.04 1.05 1.04 1.05 1.05 1.05 1.03 1.04
2.15 2.22 2.27 2.31 2.23 2.26 2.29 2.30 2.30 2.44
it because the splitting constitutes the primary experimental data and the absolute error associated with each point is more or less constant. The values for these relaxation times at other temperatures were obtained in the same manner and are listed in Table 2. An accurate value for the angle α between the magnetic and electric fields was calculated by substituting the values for θ∞ , τon and τoff into Eq. 29; they are listed in Table 2. Our results allow us to calculate the ratio, ∆χ/∆˜ ˜ ε , via Eq. 30, the results for this important quantity are shown in Table 3 and plotted against the reduced temperature, T /TNI , in Fig. 18, where the nematic–isotropic transition temperature for 5CB-d2 was taken as 306 K. It is immediately apparent that the ratio is independent of temperature within experimental error. Similar results have been reported in Refs. [5] and [36], however, Bunning et al. [37] have claimed a small but significant variation in this ratio
Fig. 18: The ratios ∆χ˜ /∆˜ε and ∆χ˜ /∆ν˜ 0 for 5CB-d2 as a function of the reduced temperature, T / TNI , where TNI denotes the nematic–isotropic transition temperature which is 306 K for 5CB-d2 .
Static and dynamic director orientation in thin nematic liquid crystal films
341
with temperature. This may have resulted from the fact that ∆˜ε and ∆χ˜ were measured in independent experiments unlike our results which are determined directly as the ratio from the NMR experiment. Although the results reported by Bunning et al. [37] are almost certainly in error, it is of interest to consider their explanation for the temperature dependence. They noted that mesogenic molecules are not cylindrically symmetric so that the contribution of the biaxial ordering of the molecule to the anisotropic susceptibilities must be included. In general the anisotropy ∆ A˜ (= A˜ − A˜ ⊥ ) of some ˜ is given by partially averaged second rank tensorial property, A, ∆ A˜ = (3/2)Szz Azz + (3/4)(Sx x − S yy )(Ax x − Ayy ).
(32)
Here A is the traceless tensor set in a molecular frame and S is the Saupe ordering matrix given in the same frame. For convenience the molecular frame x yz is usually taken to be the principal axis system for the ordering matrix; then Szz is the major order parameter and (Sx x − S yy ) is the biaxial order parameter and measures the difference in the ordering of the axes orthogonal to z, the axis which approximates closest to the molecular symmetry axis. The ratio, ∆χ˜ /∆˜ε , is then given by ) Szz χzz + (1/2)(Sx x − S yy )(χx x − χ yy ∆χ˜ = , + (1/2)(S − S )(ε − ε ) ∆˜ε Szz εzz xx yy xx yy
(33)
where the influence of internal electric fields have been ignored and ε is in essence an effective traceless dielectric tensor for a perfectly oriented system. Since the major and the biaxial order parameters have different temperature dependences it is to be expected that the ratio, ∆χ/∆˜ ˜ ε , will, in general, be temperature dependent even though ˜ ε χ and ε are usually temperature independent. The simplest way to ensure that ∆χ/∆˜ is temperature independent is if the biaxial terms are small in comparison with the major terms. Certainly for 5CB (Sx x − S yy ) is extremely small in comparison with Szz [38] and so provided the biaxialities in the associated molecular quantities are not abnormally large then the biaxial contributions to ∆χ/∆˜ ˜ ε (see Eq. 33) can be neglected leaving the major order parameter which will cancel and the ratio will be independent of temperature as we find. This independence would also be expected if the biaxial order parameter was linear in Szz which can occur for temperatures close to the nematic–isotropic transition which is the region in which our measurements are made. It seems likely that both factors contribute to the temperature independence which we have observed. The angle α between the electric and magnetic fields can also be determined from our results, without making any further assumptions, from Eq. 29. The results for α are listed in Table 2 and equal, on average, to 44.7° and as we can see they are also, as requires, independent of temperature. This temperature independence provides an internal check on our experiments and analysis for once the sample was positioned in the probe head of the NMR spectrometer this was not changed for the studies of the sample at different temperatures. Accordingly α should not change with temperature which is in accord with our results. In order to calculate the values of ∆χ˜ and γ1 from our results we need the value of ∆˜ε at the same temperature. For this we have used the dielectric anisotropies determined by Dunmur and Miller [29]. The resultant temperature dependences of ∆χ˜
342
A. Sugimura and G.R. Luckhurst
Fig. 19: The dependence of γ1 and ∆χ˜ for 5CB-d2 on the reduced temperature, T / TNI . The open circles and solid line indicate the temperature dependences of ∆χ˜ from the present work (see Table 3) and the literature [30,39], respectively. The open and closed triangles indicate the temperature dependence of γ1 from the present work (see Table 3) and the literature [31], respectively.
and γ1 are listed in Table 3 where ∆˜ε used in the calculation is given. Fig. 19 shows the dependence of ∆χ˜ and γ1 of the reduced temperature. The open circles and solid line indicate the temperature dependence of ∆χ˜ from the present work and the literature [30,39], respectively. The open and closed triangles indicate the temperature dependence of γ1 from the present work and the literature [31], respectively. In Fig. 19 the dashed lines for the present work are only a guide for the eye. We note that our values of ∆χ˜ and γ1 are in relatively good agreement with those obtained by other studies. The temperature dependence of ∆χ˜ will be discussed later. Of course, analogous information concerning the director dynamics could be obtained by other methods, such as capacitance and transmitted light measurements. However, it is difficult to determine both the diamagnetic anisotropy and the rotational viscosity at the same time. On the other hand, using deuterium NMR measurements, the diamagnetic anisotropy and the rotational viscosity are both determined using the simple relation between the relaxation times for the turn-on and turn-off processes. In addition, deuterium NMR spectroscopy presents the possibility of being able to explore the spatial distribution of the director in nematic cells as a function of time during the turn-on and turn-off processes. This is of special importance because it allows us to see if the director is uniformly distributed so that the hydrodynamic theory which we have used is strictly applicable. We now turn to a consideration of the temperature variation of the diamagnetic anisotropy. As we have noted ∆χ˜ should be linear in the second rank orientational order parameter for the effective symmetry axis. We can check this because the quadrupolar splitting, ∆ν˜ 0 , is also proportional to the same order parameter provided the molecule behaves as if it is cylindrically symmetric which as we have seen is likely to be the case. Accordingly the ratio, ∆χ/∆ ˜ ν˜ 0 , should be temperature independent. Its value for
Static and dynamic director orientation in thin nematic liquid crystal films
343
Fig. 20: Typical examples of the calculated relationship between α and θ∞ as a function of Um /Ue . The dashed line shows the linear relationship between α and θ∞ for the limit of an infinite electric field. The curves for Um /Ue = 0.6 and Um /Ue = 0.2 correspond to the present case and a higher electric field, respectively. For the simultaneous applications of both the magnetic and electric fields to the nematic film, the director is not oriented as being parallel to the electric field (θ∞ = α), but deviates from the direction of the electric field to that of the magnetic field (θ∞ < α). For Um ≥ Ue (e.g. Um /Ue = 1.2), θ∞ tends as being satisfied with the condition of θ∞ < α/2.
each temperature is listed in Table 3 and shown in Fig. 18 as a function of the reduced temperature. The data clearly show that ∆χ˜ /∆˜ν0 is indeed independent of temperature. We also discuss the physical meaning of Eq. 30 with Eq. 29. These equations give Um sin 2θ∞ = Ue sin 2(α − θ∞ ), which is equivalent to that derived from the time independent torque balance equation for the same coordinate system (see Fig. 8). For the critical case of α = 90°, it gives two solutions of θ = 0° and θ = 90° independent of the intensities of B and E. This prediction has been confirmed experimentally for a monodomain nematic using deuterium NMR spectroscopy [12]. For α = 90°, we find a relationship between α and θ∞ which depends on Um /Ue . Fig. 20 shows typical examples of the calculated α dependence of θ∞ for different values of Um /Ue , in which the dashed line shows the linear relationship between α and θ∞ in the limit of an infinite electric field, that is Um Ue . The curve for Um /Ue = 0.6 corresponds to the present experimental conditions. The value of θ∞ for α = 45° is given as 29.5°, which shows good agreement with the experimental result, namely θ∞ = 29.8°. Even if a higher electric field would be applied (Um /Ue = 0.2), θ∞ = 45°for α = 45°. In order to obtain θ∞ = 45° for α = 45°, the condition of Um Ue is required. The simultaneous applications of both the magnetic and electric fields to the nematic film, means that the director is not oriented parallel to the electric field (θ∞ = α), but deviates from this direction so that θ∞ < α. For Um ≥ Ue (for example Um /Ue = 1.2 in Fig. 20), θ∞ satisfies the inequality θ∞ < α/2. The NMR results obtained during the field-induced alignment of the nematic director
344
A. Sugimura and G.R. Luckhurst
have revealed several important features. First the widths of the spectral lines are symmetric and relatively sharp which indicates that the director distribution is narrow and does not change during the alignment process. In other words the nematic is aligned as a monodomain. This is important because it allows the simplified torque-balance equation in Eq. 23 to be used to describe the time dependence of the director orientation. The second feature of our results is that the time dependence of the director orientation predicted by theory is in good agreement with the experimental results. In an attempt to understand why the theory which neglects the elastic energy and surface anchoring is so effective in explaining our results we have calculated the time dependence of the director orientation across the cell without ignoring these two effects. According to the hydrodynamic theory, we should consider the time dependence of the onedimensional director distortion across the nematic film. Here we introduce an angle, φ(z 1 , t) (= θ(z 1 , t) + 90° − α), made by the director with the glass plate, and a new coordinate system, in which the z 1 -axis is taken as being normal to the plate (see Fig. 8). The rate of change of the director orientation for the transient process is given by the extended torque-balance equation [24]
∂φ(z 1 , t) ∂ 2 φ 1 ∂ f (φ) ∂φ 2 = f (φ) 2 + γ1 ∂t ∂ z1 ∂ z 1 2 ∂φ ∆χ˜ 2 ε0 ∆˜ε 2 + B sin 2(α + φ) + (34) E (z 1 ) sin 2φ, 2µ0 2 f (φ) = K 1 cos2 φ + K 3 sin2 φ.
(35)
An initial director distribution, φ(z 1 , 0), and the boundary conditions are needed to calculate the time-dependence of the director orientation numerically using Eq. 34. In our first attempt to calculate the time evolution of the director distribution we have used Eq. 34 without any further assumptions . The torque balance equations at both surfaces are given by Eqs. 16 and 17. These equations give the boundary conditions needed to solve the torque balance equation in the nematic film and the variation of the electric field across the film (see Eq. 18). Numerical solution of Eqs. 16–18 and 34–35 gives the one-dimensional director distortion across the nematic film as a function of time. In our calculations the values B = 7.05 T, = 56.1 µm, φ0 = 4° at z 1 = 0, φ0 = 4° at z 1 = , A = 1.0 × 10−7 J/m2 , V = 50 V, α = 44.7° (see Table 2), ∆χ˜ = 1.17 × 10−6 (see Table 2), γ1 = 0.0714 Pa s (see Table 2), K 1 = 8.42 × 10−12 N [31], K 3 = 9.96 × 10−12 N [31], and ∆˜ε = 11.4 [39] at 295 K were used. In the calculations the initial director distribution for the turn-on process (t = 0 ms in Fig. 21a) is obtained as that after t = 0.1 ms using Eqs. 16–17 and 34–35 for V = 0. The initial director distribution for the turn-off process (t = 0 ms in Fig. 21b) is obtained as that after the application of the electric field for 50 ms using Eqs. 16–18 and 34–35; this is the same condition as that used in our experiments. Fig. 21a for the turn-on process and 21b for turn-off process show typical examples of the simulation results for the director distribution θ(z 1 / ) across the entire nematic film as a function of time. These predicted distributions for the variation of the director orientation with time throughout the nematic film show an interesting form. In Fig. 21 (a) most of the director
Static and dynamic director orientation in thin nematic liquid crystal films
345
Fig. 21: Typical examples of the simulated director distribution θ (z 1 / ) across the entire nematic film as a function of time for (a) the turn-on and (b) the turn-off processes for 5CB-d2 at 295 K.
(> 99%) is uniformly aligned at θ = 6° after 0.2 ms while for the rest of the nematic film the director orientation in the vicinity of the glass surfaces deviates slightly from this angle because of the weak surface anchoring. The director orientation increases with time and saturates to θ∞ at t = 7.1 ms. It is apparent from this calculation that the director is almost completely oriented as a monodomain as found by the experiments. In the turn-off process the director is also oriented with time as a monodomain as shown
346
A. Sugimura and G.R. Luckhurst
in Fig. 21b. Most of the director (> 99%) is oriented at θ = 28° after 0.1 ms. The director orientation angle decreases with time and reaches θ∞ = 0 at t = 7.1 ms. We have also compared the time dependence of the director orientation obtained from the complete torque-balance equation with that predicted by the simplified version (see Eq. 23). We find that the results of the complete calculations are in excellent agreement with the predictions of the simplified hydrodynamic theory. These complete simulation results for the spatial and temporal variation of the director distribution support the analysis of the director dynamics for a thin nematic slab with weak boundaries using the torque-balance equation 23 for a monodomain. This behaviour is entirely analogous to that which we see in Section 4.2 for the static director distribution across a nematic cell for varying field strengths. The ability of the time resolved director distribution enables us to calculate the spectra by using Eq. 22. In the calculation the effective spin–spin relaxation time, T2 , is taken to be 1 kHz, whose value is obtained by the half width of the spectrum shown in Fig. 15. The simulated deuterium NMR spectra corresponding to these time resolved director distributions are shown in Figs. 22a for turn-on and 22b for turn-off. The time-resolved NMR spectra predicted by theory for the monodomain are clearly in a very good agreement with the experimental spectra shown in Fig. 16a and b. Thus although the recorded spectra contain weak oscillatory spectral features associated with the director rotation during the acquisition of the FID [32], the quadrupolar splittings of the calculated spectra at each time show a complete coincidence with those of the experimental spectra. It is apparent from this significant agreement that the physical parameters used in the simulations reflect the reliability of the values of the nematic for the physical parameters, especially for the magnetic anisotropy and rotational viscosity coefficient obtained from the experiment. This means that the hydrodynamic theory can account for the experimental results and so the reorientation of the director towards equilibrium follow exactly that predicted by Eq. 23.
6. Conclusion Deuterium NMR spectroscopy has been employed to investigate the spatial director distribution in a thin nematic film. The dependence of the quadrupolar splitting on the electric field strength has been measured for nematic cells, with no surface forces, as well as with weak and strong anchoring. When there are no surface forces we find for orthogonal fields that, at a certain critical value of the electric field strength, a sudden change of the director orientation from being parallel to the magnetic field to being parallel to the electric field. This behaviour can be explained by the absence of elastic deformations in the thin nematic film and the critical orthogonal relationship between the opposing fields. For cells with weak anchoring, the director orientation changes continuously from being parallel to the magnetic field to being orthogonal to it. One cell with strong anchoring showed not only a continuous change in the director orientation as for the weak anchoring cells and the other strong anchoring cells but also the appearance of a quadrupolar doublet due to the presence of a small, but significant, fraction of the
Static and dynamic director orientation in thin nematic liquid crystal films
347
nematic region which has the director still aligned more or less parallel to the surface of the glass plates. We have interpreted all of these results using continuum theory in terms of changes in the director distribution due to the presence of surface anchoring, elastic, electric and magnetic energies. The electric field dependence of the quadrupolar splitting can be understood from the continuum theory predictions of the director distribution. This reveals, in keeping with experiment, that the effect of the magnetic field on the director pretilt at the surface causes the director to be uniformly aligned across the nematic film which is an especially novel result. Our experimental results and their comparison with theory show that deuterium NMR provides a very valuable technique with which to investigate the director distribution in thin liquid crystal samples. Deuterium NMR spectroscopy has also been used to investigate the director dynamics in the nematic slab, 56.1 µm thick, of 5CB with weak anchoring and time-resolved NMR spectra have been obtained. The nematic cell was held in the NMR probe so that the electric field makes an angle of α = 44.7° with the magnetic field. This experimental geometry allowed us to avoid the degeneracy of the realignment pathway for the director found for larger angles. A series of deuterium NMR spectra, obtained using a quadrupolar echo sequence, was acquired as a function of time. When the electric field, whose intensity is controlled so that the director makes a non-zero angle with the magnetic field is applied to the nematic film, the director moves from being parallel to the magnetic field to being at an angle θ with respect to it. After the electric field is switched off, the director relaxes back to being parallel to the magnetic field. In this way a non-equilibrium director orientation with respect to the field can be created without causing flow to occur within the nematic sample. Deuterium NMR spectra were recorded during the turn-on and the turn-off realignment processes as a function of time. We have studied the time dependence of the director orientation for the turn-on and turn-off processes at different temperatures in the nematic phase. From the measured relaxation times, we have determined the rotational viscosity coefficient and the diamagnetic anisotropy of 5CB at different temperatures. Both of these quantities were found to be in reasonably good agreement with values reported in the literature. The ratio of the anisotropic susceptibilities, ∆χ/∆˜ ˜ ε , was found to be independent of temperature in agreement other but not all experimental studies; it can be understood in terms of the effective cylindrical symmetry of the 5CB molecules. The time and spatial variation of the complete director dynamics were predicted by a hydrodynamic theory and found to be in good agreement with our experimental results as well as the predictions of the simple theory in which elastic deformations and surface anchoring are ignored. The present results indicate that deuterium NMR spectroscopy provides a valuable technique with which to investigate the turn-on and the turn-off alignment processes as a function of time in thin nematic slabs. Deuterium NMR spectroscopy has also been used to investigate the dynamic director alignment process following the application or removal of an electric field in the nematic liquid crystal subject to orthogonal magnetic and pulsed electric fields [15]. The time dependence of the NMR spectra has been measured for a thin nematic film. The director was found to be rotated from being orthogonal to the magnetic field to being parallel to it when the electric field, whose direction was essentially perpendicular to that of the magnetic field, was switched off. The experimental spectra show that the director
348
A. Sugimura and G.R. Luckhurst
is uniformly aligned orthogonal to the magnetic field in the initial state and that the director uniformly oriented parallel to this field in the final state. The intermediate states are observed to correspond to a wide director distribution. In the realignment pathway, the directors can rotate equally probably clockwise or counterclockwise and this degeneracy induces a backflow effect. This can be included within the theory by using an effective value of the rotational viscosity which is less than the true value. The time dependence of the director distribution was simulated using the torque balance equation including a time dependent viscosity torque term with an effective rotational viscosity. For the experimental spectra found during the passage of the director from being orthogonal to being parallel to the magnetic field, however, the spectral lines are observed to be broader than those predicted. The domain walls resulting from the degeneracy in the director realignment could cause the failure of the theory to predict the broad director distribution. Although this particular investigation reveals the lack of a good theoretical analysis for the director distribution, deuterium NMR spectroscopy is clearly providing a very valuable technique with which to investigate the director distribution in nematic films both dynamic and static.
Acknowledgements This work was supported by a Scientific Grant-in Aid from the Japan Society for the Promotion of Science (JSPS) and was carried out as an Anglo-Japanese joint research project of the International Exchange program supported by The Royal Society and JSPS.
References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15.
J.C. Rowell, W.D. Phillips, L.R. Melby, and M. Panar, J. Chem. Phys. 43, 3442 (1965). G.R. Luckhurst, J. Chem. Soc. Faraday Trans. 2 84, 961 (1988). A.F. Martins, P. Esnault, and F. Volino, Phys. Rev. Lett. 57, 1745 (1986). M. Winkler, D. Geschke, and P. Holstein, Liq. Cryst. 17, 283 (1994). M. Bender, P. Holstein, and D. Geschke, J. Chem. Phys. 113, 2430 (2000). R. Stannarius, G.P. Crawford, L.C. Chien, and J.W. Doane, J. Appl. Phys. 70, 135 (1991). S.M. Fan, G.R. Luckhurst, and S.J. Picken, J. Chem. Phys. 101, 3255 (1994). J.R. Hughes, G. Kothe, G.R. Luckhurst, J. Malthete, M.E. Neubert, I. Shenouda, B.A. Timimi, and M. Tittelbach, J. Chem. Phys. 107, 9252 (1997). E. Ciampi, J.W. Emsley, G.R. Luckhurst, and B.A. Timimi, J. Chem. Phys. 107, 5907 (1997). P. Esnault, J.P. Casquilho, F. Volino, A.F. Martins, and A. Blumstein, Liq. Cryst. 7, 607 (1990). A. Sugimura, K. Nakamura, T. Miyamoto, P.J. Le Masurier, B.A. Timimi, T.H. Payne, and G.R. Luckhurst, Proc. of the 4th Int. Display Workshops, Vol. 65 (1997). G.R. Luckhurst A. Sugimura, and B.A.Timimi, Mol. Cryst. Liq. Cryst. 347, 297 (2000). G.R. Luckhurst, T. Miyamoto, A. Sugimura, T. Takashiro, and B.A. Timimi, J. Chem. Phys. 114, 10493 (2001). C.J. Dunn, G.R. Luckhurst, T. Miyamoto, A. Sugimura, and B.A. Timimi, Mol. Cryst. Liq. Cryst. 347, 167 (2000). G.R. Luckhurst, T. Miyamoto, A. Sugimura, and B.A. Timimi, Thin Solid Films 393, 399 (2001).
Static and dynamic director orientation in thin nematic liquid crystal films 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39.
349
G.R. Luckhurst, T. Miyamoto, A. Sugimura, and B.A. Timimi, Mol. Cryst. Liq. Cryst. 347, 167 (2000). G.R. Luckhurst, T. Miyamoto, A. Sugimura, and B.A. Timimi, J. Chem. Phys. 117, 5899 (2002). G.R. Luckhurst, Mol. Cryst. Liq. Cryst. 347, 121 (2000). G.R. Luckhurst, T. Miyamoto, A. Sugimura, and B.A. Timimi, Mol. Cryst. Liq. Cryst. 347, 147 (2000). G.R. Luckhurst, T. Miyamoto, A. Sugimura, and B.A. Timimi, Mol. Cryst. Liq. Cryst., in press. A. Sugimura, T. Miyamoto, M. Tsuji, and M. Kuze, Appl. Phys. Lett., 72, 329 (1998). H.J. Deuling, E.Guyon, and P.Pieranski, Solid State Comm. 15, 277 (1974). B. Jérôme, Rep. Prog. Phys. 54, 391 (1991). P.G. de Gennes and J. Prost, The Physics of Liquid Crystals, 2nd ed. (Oxford University Press, Oxford, 1993). A. Sugimura, G.R. Luckhurst, and Z. Ou-Yang, Phys. Rev. E 52, 681 (1995). G. Barbero and G. Durand, J. Appl. Phys. 67, 2678 (1990). A.A. Sonin, The Surface Physics of Liquid Crystals, (Gordon and Breach, 1995). S.A. Brooks, G.R. Luckhurst, and G.F. Pedulli, Chem. Phys. Lett. 11, 159 (1971). D.A. Dunmur and W.H. Miller, J. Physique 40, 361 (1979). B. Bahadur, Liquid Crystals – Applications and Uses, Vol.1, (World Scientific, 1990) p.156. K. Skarp, S.T. Lagerwall, and B. Stebler, Mol. Cryst. Liq. Cryst. 60, 215 (1980). A.M. Kantola, G.R. Luckhurst, A. Sugimura, and B.A. Timimi, Mol. Cryst. Liq. Cryst., in press. G.R. Luckhurst, T. Miyamoto, A. Sugimura, and B.A. Timimi, J. Chem. Phys. 116, 5099 (2002). F.C. Frank, Disc. Faraday Soc. 25, 19 (1958). G. Labrunie and J. Robert, J. Appl. Phys. 44, 4869 (1973); See for example, P.J. Collings and M. Hird, Introduction to Liquid Crystals, Chemistry and Physics, (Taylor & Francis, 1997) p. 197. Hp. Schad, G. Baur, and G. Meier, J. Chem. Phys 71, 3174 (1979). J.D. Bunning, D.A. Crellin, and T.E. Faber, Liq. Cryst. 1, 37 (1986). J.W. Emsley, G.R. Luckhurst, and C.P. Stockley, Mol. Phys. 44, 565 (1981). D.A. Dunmur, M.R. Manterfield, W.H. Miller, and J.K. Dunleavy, Mol. Cryst. Liq. Cryst. 45, 127 (1978).
Nanotechnology and Nano-Interface Controlled Electronic Devices Editors: M. Iwamoto, K. Kaneto and S. Mashiko © 2003 Elsevier Science B.V. All rights reserved
CHAPTER 17
MDC–SHG spectroscopy of organic monolayer film Atsushi Tojima a , Ryouhei Hiyoshi a , Takaaki Manaka a , Mitsumasa Iwamoto a,*, and Ou-Yang Zhongcan b a Department
of Physical Electronics, Tokyo Institute of Technology, O-okayama 2-12-1, Meguro-ku, Tokyo 152-8552, Japan * E-mail:
[email protected] b Institute of Theoretical Physics, Academia Sinica, P.O. Box 2735, Beijing, 100080, China
1. 2.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Analysis of polarization of monolayers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Monolayers with C∞ -symmetry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. Spontaneous polarization and orientational order parameter . . . . . . . . . . . . . . . . 2.3. Non-linear polarization and orientational order parameter . . . . . . . . . . . . . . . . . . 3. MDC and SHG measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Experimental system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. MDC–SHG spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1. Detection of phase transition in C∞v -symmetry and orientational order . . . . . 4.2. Detection of untilting to tilting phase transition in monolayer . . . . . . . . . . . . . . 5. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
351 353 353 355 357 359 359 362 362 362 365 373 374
1. Introduction Many organic materials that are interesting in terms of electronics have been synthesized and discovered during the past few decades [1,2]. One of the most remarkable achievements is the Nobel Prize in Chemistry 2000 awarded to Heeger [3], MacDiarmid [4] and Shirakawa [5], for their contribution to the discovery and development of conducting polymers. In the hope of observing novel and useful electrical and optical properties, many investigations have been carried out to fabricate organic devices. Plastic solar cells, flexible-type field effect transistors (FETs), electroluminescent (EL) devices and so on have been developed, along with the development of new organic materials [6]. Insightful ideas have also been proposed to open-up new methods in electronics, in
352
A. Tojima et al.
which the specific function of organic molecules is believed to play an important role. One example is the unimolecular rectification using a D + –π–A− system proposed by Aviram and Ratner [7], with which many experiments have been carried out towards the fruition of molecular diodes [8]. The idea of electron tunneling rectifier junctions based on electron tunneling via molecular energy states has also called much attention [9]. Furthermore, many investigations on the preparation of conducting molecular wires using carbon nanotubes, multiporphyrins and others have been carried out [10,11]. Similarly, many state-of-the-art techniques for successful preparation of tunneling spacers using insulating thin films such as polyimide have also been developed [12]. As mentioned above, many organic materials including conductors, semiconductors and insulators, have been synthesized and many efforts have been made to fabricate organic devices utilizing the specific properties of organic molecules. However, these are no longer sufficient. To find further new ways to molecular electronics, the relationship between the function and structure of molecules must be clarified on the molecular level. A very simple but fascinating way is to use monolayers for this study, because monolayers show various interesting behaviors that are not seen in bulk materials. It is known that monolayers at the air–water interface exhibit very interesting behaviors as two-dimensional (2D) systems, due to the symmetry breaking at the interface. Thus the physico-chemical properties of monolayers at the air–water interface have been a research subject for physicists, chemists and biologists since the discovery of the technique for the formation of floating monomolecular films by Langmuir [13]. For the past several decades, along with the development of a variety of experimental techniques including scattering, spectroscopic, electrical and optical techniques, the dynamical features of monolayer systems and their structures have been examined by many researchers [14–16]. However, the relationship between the structure of monolayers and the function of monolayers must be clarified from the viewpoint of molecular electronics. In other words, the dielectric property, optical property, electronic conduction, and others must be clarified on the nanometer scale for the breakthrough toward the 21st-century electronics. From the theoretical side, monolayer systems are viewed as quasi 2D systems and their states are characterized by the geometrical configuration and by the orientational distribution of the constituent molecules. For example, the state of floating monolayers composed of rod-like polar molecules is expressed using two kinds of order parameters [17]. One is the planar order parameter, i.e., the positional order parameter which expresses the molecular configuration describing the positional distribution of the heads of molecules on the water surface [18,19], and the other is the orientational order parameter which expresses the orientational distribution of rod-like polar molecules normal to the water surface. Among these parameters, the orientational order parameter is very important, because this parameter specifies the shape of rod-like molecules and it is naturally connected to the specific physico-chemical property of monolayers, e.g., the dielectric property of monolayers. In Section 2, the dielectric polarization of a monolayer with C∞ -symmetry is analyzed and it is expressed using orientational order parameters defined by Legendre polynomials. The specific properties of organic monolayers such as spontaneous polarization and non-linear polarization are interpreted using these parameters. Then the
MDC–SHG spectroscopy of organic monolayer film
353
importance of the development of an experimental method used for the determination of the orientational order parameters is stressed [16,20]. In Section 3, Maxwell displacement current (MDC)–optical second harmonic generation (SHG) measurement used for the determination of orientational order parameters and the detection of phase transitions of a floating monolayer is mainly discussed. In Section 4, some experimental results on 4 -n-octyl-4-cyanobiphenyl (8CB) and fatty acid are shown. In this chapter, discussion is confined to Langmuir-monolayers, but the knowledge obtained from this study will be useful for a better understanding of monoand multilayer films on solid surface.
2. Analysis of polarization of monolayers As mentioned in Section 1, until now, a variety of experimental methods including scattering, spectroscopic, electrical and optical techniques have been developed, and they have been employed to study the molecular structure and orientational phase transition of monolayers at the air–water interface [14,15]. Using X-ray diffraction, Brewster angle microscopy (BAM) and others, monolayer structure, texture of monolayers, and so on have been elucidated step by step. However, the physico-chemical properties of 2D systems have not been fully discussed yet from the viewpoint of dielectric physics. Obviously, the dielectric property of monolayers is dependent on the polarization of the monolayers. Thus experimental methods that allow dielectric polarization phenomena to be probed have called much attention. Among them are MDC [21–23] and SHG [24–28] measurements. MDC and SHG are related to dielectric spontaneous polarization and non-linear dielectric polarization phenomena, respectively, and they are generated only from polar non-centrosymmetric monolayers, although these are not generated from isotropic bulk materials. For a better understanding of the generation of MDC and SHG, it is convenient to start from the analysis of dielectric polarization of monolayers. For this analysis, a monolayer composed of rod-like polar molecules is considered and a monolayer with C∞ -symmetry is mainly discussed, but very important insight into the physics of monolayers can be obtained. Further, this analysis is easily extended to the case of monolayers with Cs -symmetry, whose director makes an angle θt from the direction normal to the water surface. 2.1. Monolayers with C∞ -symmetry Now let us consider that a monolayer is composed of molecules with permanent dipole moment µ along their molecular long axis, and an anisotropic electronic polarizability, i.e., an electronic polarizability α , parallel to the molecular long axis, and α⊥ , perpendicular to the molecular long axis (see Fig. 1). Further, the molecule (2) . The water surface and the floating has a molecular second-order susceptibility, αM monolayer film are considered as an infinite plane. The coordinate system is chosen in such a way that the monolayer plane is parallel to the x–y plane and the monolayer normal falls along the positive z-axis, as shown in Fig. 1a. The molecule at the origin is facing the water surface, and it tilts with an angle θ. The molecular area
354
A. Tojima et al.
z (zm)
z'
(a) α//
y'
n ,m θ
d
α⊥
y
Molecule
φ
x
x'
d = (sinθ cosφ, sinθ sinφ, cosθ) m = (0, 0, 1), n = (0, 0, 1)
(b)
z
m
θA
l x
untilting monolayer Fig. 1: Molecule of untilting monolayer with C∞v -symmetry. (a) The molecules in monolayers are assumed to be uniaxial, depicted by an ellipsoid of rotation for the electronic polarizability. (b)The molecules in an untilting monolayer.
A decreases and increases by monolayer compression and expansion, respectively. This molecule also orientates with an azimuthal angle π/2 − φ from the y-axis. The coordinate system (x , y , z ) is attached to the dipolar molecule at the origin with the → z -axis pointing toward the molecular long axis, d . Thus the direction of the molecule’s → d is expressed as (sinθ cos φ, sin θ sin φ, cos θ) in the coordinate system (x, y, z). As the molecule considered here is rod-shaped, the interaction with the water surface
MDC–SHG spectroscopy of organic monolayer film
355
should be independent of the azimuthal angle φ. The molecular average direction called m, is expected to be normal to the water surface. That is, the director the director, → n = (0, 0, 1) in the coordinate system (x, y, z), indicating that the points in the direction → monolayer has C∞ -symmetry (Fig. 1a). To specify the average orientation of molecules on the water surface, the macroscopic alignment parameters known as the orientational order parameters are used [17]. These parameters describe the alignment origin of the collective properties of molecules in monolayers [29,30]. The universal definition Sn ≡ Pn (cos θ) can be adopted as orientational order parameters to express monolayer systems with C∞ -symmetry. Here Pn (cos θ) is the nth Legendre polynomial, and denotes thermodynamic average over all molecular directions. 2 It is apparent that the orientational order parameter S2 ≡ P2 (cos θ) = 3 cos2 θ−1 corresponds to the Maier–Saupe order parameter introduced by Tsvetkov [31,32] into liquid crystals (LCs). For organic monolayers with C∞ -symmetry, the orientational order parameter S1 ≡ P1 (cos θ) = cos θ makes more sense to specify the characteristic property of monolayer films due to the symmetry breaking at the interface. Similarly, 3 cos θ specifies the characthe orientational order parameter S3 ≡ P3 (cos θ) = 5 cos θ−3 2 teristic property of monolayer films. As such, these two order parameters can be used to describe the specific dielectric property of monolayer films with C∞ -symmetry [16,20]. Suppose a monolayer is subjected to an external electric field E, the dielectric polarization of each molecule is expressed as [24,33] ↔
m = µ + α (1) · E + α (2) : E E + · · · , ↔(1)
(1)
where α is the linear electronic polarizability and α is the second-order susceptibility. Discarding the non-linear terms higher than second order, the polarization of monolayers, P, is described as (2)
↔
P = Ns m = Ns µ + Ns α (1) · E + Ns α (2) : E E,
(2)
where Ns is the surface density of the molecules. ↔ Calculating µ, α (1) and α (2) , the equation of dielectric polarization using orientational order parameter such as Sn (n = 1, 2 and 3) is derived. The main contributors to the MDC and SHG are µ and α (2) , respectively. In the following, we focus on µ and α (2) . 2.2. Spontaneous polarization and orientational order parameter Suppose the orientational distribution of constituent rod-like polar molecules in the monolayer film obeys the Boltzmann distribution rule, then the average of the dipole moment components is given by [17,34] µi f (φ, θ) sin θ dφdθ, (3)
µi = where f (φ, θ) is the distribution function and it is proportional to exp(−W/kB T ). Here W is the interaction energy that includes the interaction energy between molecules Win , the interaction energy between dipoles and field We , and others. kB is the Boltzmann constant, and T is the absolute temperature. µi (i = x, y, z) is the i-direction component of the permanent dipole µ.
356
A. Tojima et al.
As the direction of the permanent dipole moment µ coincides with the molecular → long axis, µ = → µ is expressed as µ = µ d . When an external electric field E = → E = (E x , E y , E z ) is applied to the monolayer film, the dipolar molecule at the origin experiences this electric field. The electric interaction energy We between the dipole and this external field is given by → → (4) We = − E · dm ≈ −µ d · E. It is reasonable to neglect the interaction energy part between the external field and ↔ the induced dipoles due to the electronic polarization α (1) and α (2) given by Eq. 1, because they are proportional to E i2 or E i3 (i = x, y, z) and as a whole only affect the third-order or fourth-order non-linearity, and they are usually very small. Due to the electric interaction We expressed by Eq. 4, the dipolar molecules prefer to reorient along the direction of the applied field. Using the Onsager approximation [35], we obtain the → dielectric polarization due to the orientational polarization Pd = Pd = (Px , Py , Pz ) as Pd = Ns m = Ns µS1 → n + Ns
µ2 ↔ → S ori · E, 3kT
where
(5)
S1 = P1 (cos θ) = cos θ =
cos θ f (φ, θ) sin θ dφdθ, →
(6)
↔
is the orientational order parameter when E = E = 0, and S ori is a tensor defined as 0 0 1 − S2 ↔ 0 1 − S2 (7) S ori = 0 2 0 0 1 − 3S1 + 2S2 with
3 cos2 θ − 1 3 cos2 θ − 1 S2 = P2 (cos θ) = = f (φ, θ) sin θ dφdθ. 2 2
↔
(8)
S ori describes the influence of the orientational distribution of the polar molecules in ↔ monolayer films. It is clear that S ori depends on the monolayer structure characterized by the two orientational parameters S1 and S2 . In Eq. 5, the first term is the spontaneous polarization and the second term is the orientational dipolar polarization due to the external electric field. It is also found from Eq. 7 that the contribution of the ↔ ↔ orientational dipolar polarization becomes µ2 /3kT if S1 = S2 = 0, i.e., S ori = I . This is nothing but the Debye–Langevin equation, which expresses the orientational polarization in isotropic bulk states [33,35]. In other words, a monolayer shows specific dielectric property originating from the non-zero orientational order parameters, S1 and S2 [16,20]. In the MDC measurement, an MDC generated across monolayers is monitored in → closed-circuit conditions, i.e., E = 0, whilst the monolayers are compressed, as will be described in Section 3. Thus only the first term of Eq. 5 plays an important role in this measurement, and the orientational order parameter S1 is determined from the MDC
MDC–SHG spectroscopy of organic monolayer film
357
over the entire range of molecular areas generated by monolayer compression. For a better understanding of orientational order parameter S1 of monolayers, it is instructive to discuss a special case where√the molecular orientational motion is restricted within the angle 0 < θ < θ A = arcsin A/A0 (A0 = πl 2 , l: the length of long axis of polar molecules) with θt = 0, due to the hardcore repulsive interaction among molecules as shown in Fig.1b. Using S1 = 0, Pz is given by Pz = Ns µS1 .
(9)
This situation appears when the molecular area A is smaller than A0 . On the other hand, when the molecular area A is greater than A0 , the electrostatic Coulomb attractive force working between molecules and the water surface makes a significant contribution, and the molecules lie on the water surface. The interaction energy between dipolar molecules and water is only dependent on θ and it is given by Wm (θ) = −(µ2 /16π0 ml 3 cos θ)[(w − m )/(w + m )], where 0 , m and w are the permittivities of free space, monolayer and water, respectively. With the apparent Boltzmann distribution function f (θ) ≈ exp[−Wm (θ)/kB T ], Wm (θ) becomes −∞ in the limit θ → π/2. Thus the molecular motion is restricted to only on the water surface, i.e., Pz becomes 0 [17] and eventually S1 = 0. It is postulated that the phase transition from planar to orientational alignment phase happens at the molecular area A0 during monolayer compression [17], and S1 and Pz are expected to change smoothly at this transition point. As MDC is generated due to the change of Pz , MDC is effective for the detection of the phase transition. 2.3. Non-linear polarization and orientational order parameter Whenever the dielectric property of monolayers is discussed, the non-linear polarization is inevitably involved, as the centrosymmetry is broken at the interface [24,26]. This ↔ higher order polarization arises from the non-linear electronic polarizability α (2) , the non-linear additional electronic polarizability produced by the local-field molecular interaction, and the non-linear orientation-induced dipole moment as a result of the interaction between the dipole moment and the external field. Usually, the contribution of these non-linear polarizations is rather small, but this situation changes under some special conditions. The non-linear electronic polarization originating from the quantum interaction between electrons and the external electric field is quite large, and can be detected as SHG signals. In SHG signals, the main contributor is the second-order ↔ non-linear electronic polarizability α (2) . The non-linear polarization is approximately written as ↔
P (2) = χ (2) : E E = Ns α (2) : E E.
(10)
The second-order non-linear susceptibility (SOS) χ ≡ [χi j k ] is related to the second-order non-linear electronic polarizability of molecules, expressed by the tensor i j k i j k (2) α ≡ [αi , j ,k ], by χi j k = Ns Ti j k αi j k . Ti j k describes the coordinate transformation between the molecular coordinates (x , y , z ) and the laboratory coordinates (x, y, z) in the case of a monolayer with C∞ -symmetry. There are many ways used to transform between laboratory and molecular frames, such as the irreducible tensor approach with (2)
358
A. Tojima et al.
the Wigner rotation matrix [36]. For the unification of the notation for the transformation performed in this chapter, the usual Euler rotation matrix R(φ, θ, ψ) = [Rii ] [37] is i jk i jk j adopted. The transformation Ti j k is expressed by Ti j k = Rii R j Rkk with cos φ cos ψ − sin φ cos θ sin ψ,
− cos φ sin ψ − sin φ cos θ cos ψ,
sin φ sin θ
sin θ sin ψ,
sin θ cos ψ,
cos θ
Rii = sin φ cos ψ + cos φ cos θ sin ψ, − sin φ sin ψ + cos φ cos θ cos ψ, − cos φ sin θ . (11)
In the expression of Eq. 10, is not taken into account the local field correction factors such as the Lorentz factor which arises due to the surface, but also, significantly, due to dipole–dipole interaction with neighboring molecules, where the latter will be obviously concentration dependent. However, this simplification does not lose underlying physics here, because one way of making the local field correction is to introduce a factor in the right-hand side of Eq. 10 [24,26]. From the SHG measurement, one can deduce χ (2) , but to obtain α (2) , one has to have i j k knowledge of Ti j k . After a lengthy calculation of the 93 components for obtaining χi j k [38], the complete expression for the macroscopic SOS tensor χ (2) of a monolayer with C∞ -symmetry on a material surface is derived as function of the components of the second-order non-linear electronic polarizability α (2) and the orientational order → parameters Sn ≡ Pn (cos θ) [17,29,30]. The non-linear polarization P (2) = P N is found →N to be given by the sum of the polarization P ch , associated with the chirality of the →N associated with the non-chirality of the monolayer. monolayer, and the polarization P ach These two polarization are expressed in a vectorial form as [27,38] →N P ch
→
→
→
→
→
→
= 12 s14 [( E · → n )( F × → n ) + (F · → n )( E × → n )] + 12 a14 ( E × F) →
→
+ 12 (a36 − a14 )→ n · ( E × F)→ n,
→N P ach
→ → → → →
(12) → → →
= (s33 − s15 − s31 )( n · E)( n · F) n + s31 ( E · F) n → →
→
→ →
→
→
+ 12 s15 [(→ n · F) E + (→ n · E) F] + 12 a15 ( E × F) × → n. →
(13)
Here E and F are external electric fields, and 7 independent non-zero elements are given by Ns s14 = S2 (σ14 − σ25 ) = χ123 + χ132 , 2 Ns Ns s15 = (S1 − S3 )(2σ33 − σ32 − σ31 ) + (3S1 + 2S3 )(σ24 + σ15 ) = χ131 + χ113 , 5 10 Ns Ns s31 = (S1 − S3 )(2σ33 − σ24 − σ15 ) + (4S1 + S3 )(σ32 + σ31 ) = χ311, 10 10 Ns Ns (S1 − S3 )(σ32 + σ31 + σ24 + σ15 ) + (3S1 + 2S3 )σ33 = χ333 , s33 = 5 5 Ns Ns (λ14 + λ25 + λ36 ) + S2 (λ14 + λ25 − 2λ36 ) = χ123 − χ132 , a14 = 3 6 Ns S1 (λ15 − λ24 ) = χ131 − χ113 , a15 = 2 Ns Ns (14) (λ14 + λ25 + λ36 ) + S2 (2λ36 − λ14 − λ25 ) = χ312 − χ321 , a36 = 3 3
MDC–SHG spectroscopy of organic monolayer film
359
where the two 3 × 6 matrices, (σi j ) and (λi j ), are defined from the molecular SOS tensor α (2) in the same way as the conventional contracted notation of (si j ) and (ai j ) defined from χ (2) , i.e., σ11 = α111 , σ14 = α123 + α132 , λ14 = α123 − α132 , etc. It is interesting here to note that s14 , a14 and a36 specify the chiral property of monolayers, whereas s15 , s31 , s33 and a15 specify the non-chiral property of monolayers. Eventually, from the view point of order parameters, S2 characterizes the orientational order of monolayers related to the molecular chirality, whereas S1 and S3 the orientational order of monolayers related to the molecular non-chirality. In the case of monolayers with C∞v -symmetry, where the constituent molecules are achiral (non→ → chiral), i.e., s14 = a14 = a36 = 0, the non-linear electric polarization in the SHG ( E = F) is reduced in a compact form as [27,38,39] →N
→
→ →
→ →
P = (s33 − s31 − s15 )(→ n · E)2 → n + s15 (→ n · E) E + s31 ( E · E)→ n.
(15)
It is found that the non-linear dielectric polarization is given as a function of the order parameters S1 and S3 .
3. MDC and SHG measurements 3.1. Method Fig. 2 shows a schematic diagram of MDC measurement coupled with SHG measurement, where electrode 1 is suspended in air and is placed parallel to the water surface, and electrode 2 is immersed in the water. These two electrodes 1 and 2 are connected to each other through an electrometer. The induced charge on electrode 1 changes in accordance with the orientational motion of molecules on the water surface and the change of surface density of the molecules. In the MDC measurement, monolayers are compressed with aid of two moving barriers. MDC flows through the closed circuit (see Fig. 3). The charge induced on electrode 1 suspended in air due to the spontaneous Iω
Ep
Es
SHG
M DC
I2ω Electrode 1
Q1
L
+
A -
water
Electrode 2
Fig. 2: Experimental setup for MDC and SHG measurements.
360
A. Tojima et al.
electrode 1 Incident light
pi n
δ
Ei
(ω)
n
z
Reflected SH p 2ω
γR
Eo (2ω)
s in
s2ω θin
barrier
+
θo
A
water
θo
p 2ω
x
E o(2ω)
γT
y s2ω
electrode 2
Transmitted SH
Fig. 3: Molecules and optical arrangement. The monolayer is compressed in the x-direction. The reflected and transmitted light can be detected.
polarization Pz , given by Eq. 9, is expressed as [16,42] Q 1 = −Pz B/L − Cφs ,
(16)
where B is the working area of electrode 1, C is the capacitance between electrode 1 and the water surface, L is the distance between electrode 1 and the water surface, and φs is the surface potential of water. Assuming that the molecules lie on the water surface in the range of low surface pressure due to the electrostatic Coulomb interaction working between the polar molecules and water, the order parameter of the monolayer is determined over the entire range of molecular area. For example, for monolayers with C∞v -symmetry, the orientational order parameter S1 defined in Eq. 9 is determined, assuming S1 = 0 in the range of low surface pressure before monolayer compression. Furthermore, it is clear from Eq. 16 that the MDC flows due to the change in the polarization of monolayers, and thus phase transitions such as the transition from planar isotropic phase to polar orientational phase of monolayers are easily detected. On the other hand, optical second-harmonic (SH) light is generated from non-centrosymmetric monolayers by laser irradiation. This SHG is due to the quantum interaction → between electrons in molecules and the external electric field, E [24]. The generation of SH signal depends on the states of monolayers. In more detail, if the local field correction factors such as Lorentz factor are not taken into account, the SH intensity in → e out depends on the term → e out · P N , where → e out is the unit vector. the direction → e out · → n and From Eq. 15, it is shown that the SHG signal is obviously dependent on →
MDC–SHG spectroscopy of organic monolayer film
361
→ →
E · n . Thus the SHG measurement is helpful for the detection of phase transitions such as the tilting to untilting (t–U) phase transition. For example, if the monolayer is in the planar isotropic phase, an SH signal should → → e out · P N = 0, whereas p-polarized radiation is not be generated because of P N = 0, i.e., → expected from monolayers with C∞v -symmetry by p- (and s-)polarized light irradiation [24,25]. Furthermore, the orientational order parameters such as S1 and S3 can be determined from the SHG measurements [16,27,28], as will be described below. Fig. 3 shows the experimental arrangement. θin and θo represent the incident and output angles, respectively. The angles δ and γ represent the polarized angles of incident light and output SH light, respectively. The input light is the sum of s- and p-polarized waves and it is expressed as → E in
= Sω → s in + Pω → p in ,
(17)
s in and → p in are unit vectors for s- and p-polarized waves, and Sω and Pω where → represent the amplitude of s- and p-polarized waves, respectively. Assuming that the incidence plane is the y–z plane (see Fig. 3), the elements of sin and pin are given by (1, 0, 0) and (0, cos θin , sin θin ), respectively. Further, the angle of the polarizer is chosen as δ for the input light (see Fig. 3), Sω and Pω are given by Sω = E in sin δ and Pω = E in cos δ, respectively. Similarly, the output light is the sum of the s- and p-polarized waves. Taking into → e out (2ω) · P N . account these, the output light intensity I2ω is found to be proportional to → Thus by choosing appropriate angles δ and γ , the orientational order parameters can be determined [22,27,28]. For example, in the special case that the second-order susceptibility α (2) is dominated by a single component, i.e. αz z z along the molecular long axis, the SH intensity generated from a monolayer with C∞v -symmetry is given by I2ω ∝ |(AS1 + B S3 )2 | · Iω2 ,
(18)
where A and B are obviously functions of the angles θin , θo , δ and γ . A and B are given by A = −2 sin θin cos θin cos θo cos2 δ cos γ + 2 sin θin sin δ cos δ sin γ + cos2 θin sin θo cos2 δ cos γ + sin θo sin2 δ cos γ + 3 sin2 θin sin θo cos2 δ cos γ ,
(19)
and B = 2 sin θin cos θin cos θo cos2 δ cos γ − 2 sin θin sin δ cos δ sin γ − cos2 θin sin θo cos2 δ cos γ − sin θo sin2 δ cos γ + 2 sin2 θin sin θo cos2 δ cos γ .
(20)
Thus by choosing the angles of optical arrangement to satisfy A = 0 (or B = 0), the orientational order parameters S3 (or S1 ) can be determined, respectively. It should be noted here that a similar expression as given by Eq. 18 can be obtained even when the local field correction factors such as Lorentz factor are taken into consideration in the analysis. That is, we can have the similar prediction that the orientational order parameters S1 and S3 can be obtained [40].
362
A. Tojima et al.
3.2. Experimental system The experimental setup consists of Langmuir-trough equipped with a two-electrode arrangement for the MDC measurement and optical measurement arrangement with a Q-switched Nd : YAG laser (e.g., wavelength 0.532 µm, pulse duration <7 ns, repetition pulse rate ≤15 Hz) for the SHG measurement (see Figs. 2 and 3). The rectangularshape Langmuir-trough (600 mm × 150 mm in length and width, 10 mm in depth) is composed of polytetrafluoroethylene (PTFE) and it is filled with pure water (electrical resistivity >17 M cm). One transparent silica glass plate is attached to the bottom of the Langmuir trough for the SHG measurement. In this experimental system, reflected and transmitted SH signals from floating monolayers can be detected at the same time. Thus order parameters S1 and S3 of monolayers with C∞v -symmetry can be determined [40]. For the MDC measurement, a transparent glass slide coated with indium tin oxide (ITO) is used as electrode 1. It is placed in air parallel to the water surface at a distance of 1 mm. Electrode 2 is a gold-wire (1 mm ∅ and 500 mm) and it is immersed in the water. Electrodes 1 and 2 are connected to each other through an electrometer (Keithley 617) whose internal electrical resistance is negligibly small. For the SHG measurement, the Q-switched laser irradiates onto the monolayer at an intensity of about 6 mJ with a pulse rate of 2 Hz. The laser spot size is about 56 mm2 . The surface pressure of the monolayer is measured during monolayer compression by a Wilhelmy plate.
4. MDC–SHG spectroscopy 4.1. Detection of phase transition in C∞v -symmetry and orientational order Fig. 4 shows the chemical structure of 4 -n-octyl-4-cyanobiphenyl (8CB) and the fatty acid molecules used here. The permanent dipole moments of methyl and cyano-groups in 8CB molecules make the contribution to the generation of an MDC, whereas the
C8H17
CN
8CB
COOH
fatty acid (octadecanoic acid) Fig. 4: 4 -n-octyl-4-cyanobiphenyl (8CB) and fatty acid molecules.
MDC–SHG spectroscopy of organic monolayer film
Surface Displacement pressure current [mN/m] [fA]
Dipole moment [mD]
Reflected Transmitted SH intensity SH intensity [a.u.] [a.u.]
4 3
363
2
1
15 p-p s -p
1
10
0.5
5
0 40
0 1.5
50
60
1
70
80
p-s s -s
0.5 0 600 400 200 0 100 50 0 4 2 0 40
60
80
100
120
2
Area per Molecule [Å ] Fig. 5: A typical example of SHG and MDC of 8CB monolayers during the course of monolayer compression. p-polarized SHG and MDC are generated at the same molecular area. Compression speed 4.2 Å2 min−1 molecule−1 .
electronic polarization of the biphenyl group with alkyl and cyano-groups makes the contribution to the SHG. As monolayers are placed on the water surface, the contribution of the hydrophilic part is diminished owing to the screening effect of the water with a high dielectric constant (≈ 78). As a result, although the dipole moment of methyl-group is very small in comparison with that for the cyano group, the long alkyl chain parts also make a contribution to the generation of MDC, in a manner to that seen in fatty acids monolayers [16,42]. Fig. 5 shows a typical example of the MDC–SHG measurement for 8CB monolayers. The experiment was carried out by monolayer compression at a speed of 0.042 nm2 min−1 molecule−1 . The incident and output light are polarized waves, and they are as indicated in the figure, such as p– p, s– p, etc. From bottom to top, the surface pressure–area, MDC–area, the vertical component of dipole moment–area, reflected SH intensity–area, and transmitted SH intensity–area
364
A. Tojima et al.
are plotted. As we can see in the figure, the MDC and SHG signals are generated in accordance with the change of molecular area. The isotherms can be divided into four regions. In region 1, where the surface pressure is immeasurably low, MDC is very small and almost zero. Similarly the SH intensity is very small. These results suggest that the 8CB molecules lie on the water surface and they are randomly distributed. In other words, the monolayer in region 1 is in the planar and isotropic state, i.e., Pz given → by Eq. 9 is 0 and the SOS components in P N given by Eq. 15 are zero. In region 2, the surface pressure is still immeasurably small, but the MDC begins to flow by monolayer compression, possibly due to the standing up of 8CB molecules on the water surface. As the MDC is generated due to the change of spontaneous polarization Pz , the MDC in region 2 reflects the change of the orientational order parameter S1 in Eq. 9. Integrating the MDC with respect to the molecular area, the vertical component of the dipole moment of 8CB molecules given by µS1 is calculated. The dipole moment– area isotherm in Fig. 5 represents the relationship between µS1 and the molecular area A (≡ 1/Ns ). It is found that S1 gradually increases with increasing monolayer compression, and finally saturates at the end of region 2. The SHG is observed for the p-polarized output light, whereas it is not observed for the s-polarized output light. These results can be explained using Eq. 15. Briefly, for the s-polarized wave → → s in , with irradiation of monolayers, the external electric field E is given by E in = Sω → → → → s in = (−1, 0, 0) in the laboratory frame (see Eq. 17). Substituting E in into Eq. 15, P N is →N calculated as P = s31 Sω2 → n because of → s in · → n = 0. Thus it is concluded that the s- and → → p-polarized radiation are proportional to P N · → s out = s31 Sω2 → n ·→ s out = 0 and P N · → p out = 2→ → s31 Sω n · p out = 0, respectively. Similarly, for the p-polarized wave irradiation, the → → external electric field E is given by E in = Pω → p with → p in = (0, cos θin , sin θin ) in the → in → laboratory frame (see Eq. 17). Substituting E in into Eq. 15, P N is calculated as →N P = (s33 − s15 − s31 )Pω2 (→ n ·→ p in )2 → n + s31 Pω2 (→ p in · → p in )→ n + s15 Pω2 (→ n ·→ p in )→ p in . Thus it is concluded that the s- and p-polarized radiation are proportional to →N → → P · s out = 0 and P N · → p out = 0, respectively, because of → n ·→ s out = 0, → n ·→ p out = → → → → → → n · p in = sin θin = 0, and p in · p out = cos(θin − θo ) = 0, p in · p in = 1. This theoretical prediction agrees well with the results observed in region 2, where p-polarized wave output is detected but s-polarized wave output is not observed. Based on these results, it is confirmed that 8CB molecules stand up on the water surface by monolayer compression, and non-linear second-order polarization expressed by Eq. 15 is induced by laser irradiation. That is, the monolayer of 8CB has C∞v -symmetry in region 2. Of course, it is necessary to check another possibility. This possibility is that domains with Cs -symmetry randomly distribute on the water surface and it looks like a C∞v -symmetry owing to the randomly distributed domains. If the domains make a significant contribution to the generation of MDC, MDC flows even in region 1 due to the condensation of domains possessing the spontaneous polarization [42], but such MDC is not seen in region 1 (see Fig. 5). Further it is instructive here to add the following discussion. As mentioned earlier, 8CB is a rod-like molecule, and the permanent dipole moment of alkyl in 8CB makes the contribution of the generation of MDC on the water surface, whereas the secondorder non-linear electronic polarization of the biphenyl group with alkyl and cyano
MDC–SHG spectroscopy of organic monolayer film
365
groups makes the contribution to the SHG. From the experiment shown in Fig. 5, it is clear that the generation of MDC and SH light is initiated at the same molecular area, corresponding to the molecular area of the onset of region 2, by monolayer compression. Thus it is concluded that the rod-like 8CB molecules, in which the core and the long alkyl are located in line, stand up by monolayer compression. In region 3, the surface pressure increases smoothly but MDC flow is steady at further compression. SHG signals are also generated steady, where the transmitted and reflected light are generated steady, and the intensities of these transmitted and reflected signals somewhat increase by compression, due to the condensation of molecules. These results indicate that the orientational order is nearly saturated in this region and only the density of molecules increases by monolayer compression. As the MDC and SHGs are proportional to the spontaneous polarization expressed by Eq. 9 and non-linear polarization given by Eq. 15, respectively, they are only proportional to the density of molecules when the orientational order does not change by monolayer compression. As a result, only very small increases of MDC and SHG are seen in region 3 by monolayer compression. This increase is obviously not so drastic in comparison with the increase of surface pressure. At the end of region 3, MDC decreases abruptly by further compression, whereas the SH signal is generated in a way similar to that in region 3. These experimental results suggest that the transition from one layer to an interdigitated three-layer is induced in region 4 during monolayer compression [43]. Briefly, the spontaneous polarization Pz given by Eq. 9 does not change due to the formation of the three-layer structure. Thus the MDC is not generated by further compression, since MDC should flow due to the change of the polarization Pz . Further the non-linear polarization given by Eq. 15 does not change, indicating that the density of molecules does not change due to this transition. As a result, the SH intensity does not change by further compression. The order parameters were estimated from the SHG and MDC measurements. Fig. 6 shows an example of the order parameters S1 and S3 for 8CB monolayers with C∞v symmetry. These order parameters were determined from the MDC and SHG, using the relationships given by Eqs. 9 and 18. In the SHG measurement, the optical arrangement was set to satisfy the relationship A = 0 or B = 0 for transmitted and reflected SH light. In region 1, S1 and S3 of SHG are very small although fluctuations are observed owing to the formation of domains on the water surface. In region 2, S1 and S3 increase gradually and reach their maximum. In region 3, they are nearly constant about 0.5. In contrast, S1 of MDC is nearly zero in region 1, and it increases monotonically in region 2. S1 is about 0.5 in region 3. From these results, it is postulated that molecules lying on a water surface in the range of immeasurably low surface pressure gradually stand up by monolayer compression, up to an average tilting angle of about 60° in region 2. 4.2. Detection of untilting to tilting phase transition in monolayer For the MDC–SHG spectroscopy of monolayers, it is informative to discuss the U → t phase transition observed in monolayers. This transition is observed for monolayers with C∞v -symmetry in U-phase. As mentioned in Section 2, using a variety of refined experimental methods coupled
366
A. Tojima et al.
4 3
2
1
S3,SHG
1 0.5
S1,SHG
0 1 0.5
S1,MDC
0 1
Surface Displacement pressure current [mN/m] [fA]
0.5 0
100 50 0 4 2 0 40
60
80
100
120
2
Area [Å /molecule] Fig. 6: Order parameters of 8CB monolayer determined from the MDC–SHG measurement. Compression speed 4.2 Å2 min−1 molecule−1 .
with π–A measurement [13,14], both in-plane 2D order of the hydrophilic polar heads of the amphiphiles and out-of-plane orientation order of their hydrophobic long axes have been recognized. Among various kinds of Langmuir-monolayers, the most fundamental one is fatty acids, and their monolayers have been extensively explored. The ordering structures of all phases of fatty acid monolayers measured by many researchers have been summarized (see, e.g., fig. 32 and table I in Ref. [15]). Among the determined phase transitions given in the phase diagrams, those located in the condensed region, such as CS (closely-packed solid), S (solid), and LS (super liquid) to both L2 and L2 (liquid condensed) phase transitions, are extremely interesting. In CS, S, and LS phases, the amphiphile long axes are standing vertically to the water surface [untilting phase (U)] and their polar heads are packed in a hexagonal lattice, while in both the L2 and L2 phase the long molecular axes are tilting [tilting phase (t)] uniformly to a nearest neighbor (NN) and a next nearest neighbor (NNN), respectively, and the polar heads of in-plane structures appear in a distorted hexagonal lattice, a stretching along the
MDC–SHG spectroscopy of organic monolayer film
367
d0
d0
a/cos θt
a m,n
m
n
β (A0)
θt A0/cosθt
A0 U-phase
T-phase
(a)
(b)
Fig. 7: (a) Model structures: U-phase, untilted phase, and (b) t-phase, tilt toward a nearest neighbor (NN).
tilt direction (Fig. 7) [15]. However these interesting phase transitions have not been discussed in association with the generation of MDC and SHG. As has been pointed out by Sirota [15,44], in the condensed phases where the long axes are like cylinders and they are closely packed, a tilt directly causes a distortion of the projection of the hexagon on the surface (Fig. 7). As such, the in-plane geometry relation of closely packed monolayers on the water surface is represented only by a tilting order parameter, the tilt angle θt from the water surface normal. The generation of MDC is intimately related to the microscopic orientational motion of polar molecules, MDC due to the U → t phase transition, and it can be discussed in conjunction with the tilt angle θt . Fig. 7a and b show model structures of U-phase (untilted phase), and t-phase (tilting phase, tilting toward a nearest neighbor (NN)), respectively. At the condensed U-phase, uniaxial rod-like molecules with a permanent dipole moment µ along the molecular long axis are orthogonally and hexagonally packed. The average
direction of molecular m is parallel to the unit vector → n . β(As ) = arcsin As /πl 2 is the maximum director → → Here l is the√ length of the rod-like angle between the molecular long axis and director m. molecules and As is the molecule area with the relation As = 2d02 / 3 (Fig. 7a). → is tilted uniformly to On the other hand, at the condensed t-phase, the director m a direction of the azimuthal angle φt from the NN direction and with the tilt angle θt from → n (Fig. 7b). The basic assumption here is that the collective tilt happens in a
368
A. Tojima et al.
way that the tilting causes the stretching of the projected lattice cell in the tilt direction with a ratio of 1/ cos θt . Thus, the U → t transition is considered microscopically as a geometric distortion of the orthogonally hexagonal orientation of the molecules by competition between the 2D packing entropy and the interaction between molecules in bulk and interface, where the tilting causes the stretching of the projected lattice cell on m, as shown in Fig. 6. Here the interaction water surface but not the section normal to → is described using the Lennard-Jones (L-J) potential and others. Interestingly, the tilt causes increase in the entropy of the monolayer, due to the stretching of the lattice. In the following, the L-J potential is adopted as the energy describing the interaction [45–49]. In the context of the mean field approach [45] the free energy of an amphiphilic U U U U = molecule in √ the U state is assumed √ 12 as F = W − T S . The internal energy W 1/6 6 3(−C/(2d0 / 3) + D/(2d0 / 3) ) describes the L-J potential with rmin = (2D/C) and Wmin = −3C 2 /4D where rmin is the inter-molecular distance when W U reaches its minimum Wmin defined by a characteristic temperature TLJ as Wmin = −kTLJ . Here k is the Boltzmann constant. The free energy of entropy −T S U = −kT ln A0 − kT ln 2π(1 − cos β(A0 )) contains both contributions from in-plane entropy (first term) and chainorientation entropy (second term) of the molecule. From Fig. 7, in the t-phase the enand, with tropy energy changes to −T S t = −kT ln(A0 / cos θt ) − kT ln 2π(1 √ − cos β(A0 ))6 3 t = −C( 3 cos θ /2d ) a lengthy derivation [45], the L-J potential becomes W t 0 i=1 (1 − √ 3 2 2 2 2 −3 12 −6 sin (φt + iπ/3) sin θt ) + D( 3 cos θt /2d0 ) (1 − sin (φ + iπ/3) sin θ ) where t t i=1 φt is the angle of molecular tilt from the NN direction. Thus, under these circumstances, this U → t transition can be discussed using only one parameter η = sin θt . Discarding the higher terms, the free energy change ∆F = F t − F U from untilted phase to tilting phase in amphiphile monolayers is approximately expanded as [45,50] ∆F ≈ Gη2 + Bη4 − Eη6
(21)
where 1 G = − kT + 3(1 − ξ )ξ kTLJ , 2 1 3 B = (13ξ − 4)ξ kTLJ − kT , 8 4
1 ξ (26 − 89ξ ) 1 E = kT − kTLJ − kTLJ ξ cos6 θt 14ξ cos6 θt − 5 (10 − cos 6φt ), (22) 6 4 8 √ 6 with ξ = (rmin /(2d0 / 3)) . This theory predicts the phase diagram in the relevant external parameters space, i.e.2 T2 and6 d0 . ∆F yields the most important result: When ξ − 13 − 13 , the minimum of ∆F is achieved at η = 0 (U B > 0, i.e. T /TLJ < 39 2
2 phase) for G > 0 (i.e. T /TLJ < −6 ξ − 12 + 32 ), as G changes sign, the minimum shifts continuously to η2 = sin2 θt = −G/2B, i.e., the equation of state of the system as
2 2 1 − η2 3 13η2 − 4 1 − η2
,
(23) T /TLJ = −6 ξ +2 2 1 + η2 13η2 − 4 13η2 − 4 1 + η2 and U → t phase transition occurs.
MDC–SHG spectroscopy of organic monolayer film
369
It is interesting here to note that this model predicts that the tilting phase (the low symmetry phase) is stable at high temperatures when compared to the untilting phase. This uncommon situation is due to the competition between the L-J potential and the chain-orientation entropy: At high temperature the tilt causes an increase of L-J potential but decreases the energy of orientation entropy. As mentioned earlier, the tilting causes a distortion of the projection of the hexagon on the surface, i.e., the increase in molecular area, and thus leads to the increase in entropy of monolayer systems. As a result, the free energy change ∆F gives the above mentioned result. Given these, it is interesting to show that the MDC experimental technique is helpful to detect such t → U transition predicted by the relationship between free energy ∆F and η. First, it is necessary to calculate the orientational order parameter S1,z = cos θ for monolayers in tilting phase [17], for analyzing the MDC across monolayers due to the U → t phase transition, where the induced charge on the suspended electrode is → proportional to S1,z . Fig. 8a shows a schematic diagram of the molecular orientation d and molecules in monolayers in tilting phase (see Fig. 8b). In the coordinate system → (x, y, z) in the laboratory frame, d is expressed as (sin θ cos φ, sin θ sin φ, cos θ). That is, the molecule tilts with an angle of θ from the normal direction (z-direction) to the water surface (x–y plane), and orients with an azimuthal angle π/2 − φ from the y-axis. → In another way, d is written as (sin Θ cos Φ, sin Θ sin Φ, cos Θ) in the coordinate m, it tilts with system (x m , y m , z m ), where the z m -axis is attached to the axis of director → an angle θt from the z-axis in the z–x plane. Thus the following relation is satisfied between the two coordinate systems, (x, y, z) and (x m , y m , z m ): cos θ = − sin Θ sin Φ sin θt + cos Θ cos θt . (24) Therefore the orientational parameter S1,z is written as S1,z = cos θ = − sin Θ sin Φ sin θt + cos Θ cos θt . (25) Here, sin Θ sin Φ and cos Θ are two orientational order parameters. Therefore the polarization Pz in the monolayer normal direction, corresponding to the 1st term of Eq. 5, is given by Pz = Ns µS1,z = Ns µ cos θ = −Ns µ sin θt sin Θ cos Φ + Ns µ cos θt cos Θ, (26) If the interaction between the molecular dipole and water surface is discarded, the distribution f (Θ) is isotropic in the U and t phases and it is not modified by the tilting process. Thus among the two orientational order parameters, the latter one just corresponds to the orientational order parameter S1 of monolayers in single monolayers in untilted phase (θt = 0, see Eq. 5). Looking at the experimental arrangement for MDC measurement, a charge Q is induced on electrode 1 and it is given by [23,17] N µS1,z Q=− , (27) L
370
A. Tojima et al.
zm z θt Θ m
(a)
Molecule
θ
d φ x
y (ym)
Φ
xm
d=(sinΘcosΦ, sinΘsinΦ, cosΘ) =(sinθcosφ, sinθsinφ, cosθ)
(b) z
θt m n
molecule
x
water
tilting monolayer Fig. 8: Molecules in tilting monolayer with a tilting angle of θt . The molecules in monolayers are assumed to be uniaxial.
using Eq. 16 under the assumption φs = 0. Here N is the number of molecules under electrode 1 and it is given as N = B/A. B is the working area of the electrode, L is the distance between suspended electrode and water surface, and A is the molecular area given by A = A0 / cos θt in the tilting phase (see Fig. 7). Therefore the MDC is calculated as Bµ d S1,z I= . (28) L dt A
MDC–SHG spectroscopy of organic monolayer film
Dipole Moment [mD]
200
4 3 2
371
Region 1
100
Surface Pressure [mN/m]
Displacement Current [fA]
0 60 40 20 0 -20 60 Compression speed 20 mm/min (11Å 2/min/molecule)
40 20 0 20
30 40 2 Area per Molecule [Å ]
50
Fig. 9: MDC due to tilting to untilting (t → U) phase transition observed for fatty acid monolayer.
Substituting the relation A = A0 / cos θt into Eq. 28, we obtain Bµ d
sin Θ sin Φ 2 cos Θ I = γ1 − cos θt sin θt cos θt , L dA0 A0 A0 dA0 . (29) with γ1 = dt It is found from Eq. 29 that the MDC changes at the molecular area A = A0 , where the tilt angle θt changes due to the U → t transition, especially due to the contribution of the first term. Focusing on the change observed in the condensed phase by monolayer compression, MDC measurement was carried out for monolayers of octadecanoic acid (C18) at the air–water interface [23]. The monolayers of C18 were compressed by moving a barrier at a constant speed, and the current flowing through the closed circuit is recorded with the surface-pressure– area isotherm. The MDC was generated in a manner similar to that in our previous study. Fig. 9 shows the experimental result, where surface-pressure–area (π–A) and MDC–A, and the vertical component of dipole moment–area (µS1z –A), proportional to S1z –A isotherms, were plotted from bottom to top, respectively. At a molecular area of 20 Å2 (region 2 → 3), MDC changes, i.e., first decreasing and then increasing during monolayer compression. The decrease is due to the condensation of molecules, whereas the increase is due to the concerned U → t phase transition. For simplicity, in the case where molecules can randomly orient over all orientational
372
A. Tojima et al.
directions θ < θ A , due to the hard-core repulsive interaction working among rod-like molecules, cos Θ, sin Θ sin Φ and A0 are given by (cos θ A + 1) /2, 0 and πl02 sin θ A , respectively [17]. Thus the MDC is approximately given as 1 + 1 − (A0 /πl02 )2 Bµ d dA0 cos2 θt , with γ1 = . (30) I = γ1 L dA0 A0 dt Therefore the generation of MDC is expected in the positive direction due to the t → U transition, i.e., θt → 0, as A0 decreases by monolayer compression, if µ > 0. The increase in the MDC observed in Fig. 9 supports this prediction well. In practical observation, the MDC changes abruptly at the transition. This seems contrary to Eq. 30, in which everything varies smoothly at the transition. It is speculated that the compression procedure is dynamically anisotropic and must create a small but finite θt tilt even in the U phase. Furthermore, it is experimentally and theoretically shown that the angle of the molecular long-axis from the normal direction to the water surface changes θc < 40° due to the U → t transition in the NN-direction [15,45]. Therefore, it is expected that the molecular area Ac at the transition point changes with a ratio of cos θc > 0.76. In other words, the molecular area changes about ∆ A (= A0 (1/ cos θt − 1)) < 6.0 Å2 from θt = 0 to θt = θc at the molecular area of around 20 Å2 where the MDC changes due to this transition. This prediction is true as shown in region 3
(see Fig. 9). Furthermore, for θt = 0 one finds from Eq. 30 I = −(γ1 Bµ/L A20 )(1 + 1/ 1 − (A0 /πl02 )2 ), so that for a negative pulse γ1 = dA0 /dt occurring in the t → U transition by compression the MDC appears as a positive pulse as shown in Fig. 9. As discussed above, the t → U phase transition is detectable by MDC measurement, and the generation of MDC due to this transition can be explained by a model using one parameter, η. The fatty acid monolayer is not active at laser irradiation, whereas 8CB monolayer is SH active. In this sense it is interesting to discuss the non-linear polarization of monolayers in tilting phase. As mentioned earlier, the spontaneous polarization is characterized using two orientational order parameters y 10 = cos Θ and y 11 = sin Θ cos Φ when the director is tilted from the monolayers with C∞v -symmetry. If we assume that the uniaxial symmetry → (Fig. 7b) is held in both tilting and untilting of the molecular distribution around m phases due to the repulsive interaction between the molecules of the monolayer and the attractive interaction between dipolar molecules and the interfaces [17], the orientation distribution function f (Φ, Θ) of molecules in the monolayer becomes independent of Φ, i.e., y 11 = 0. Thus when the director tilts with an angle θt from the normal (z-axis) in the x–z plane → (see Fig. 2, φt = 0), the non-linear polarization P N given by Eq. 15 changes to →N
→
→ →
→ →
→ P = (s33 − s15 − s31 )(→ m · E)2 m + s31 ( E · E)→ m + s15 (→ m · E) E. →
(31)
Here m is the unit director vector that corresponds to the tilting direction (see Fig. 7). By expanding Eq. 31 into the elements of Px , Py and Pz as a function of E x , E y , and E z (see equations (1) and (2) in Ref. [14]), we can easily check that the non-linear polarization given by Eq. 31 represents the non-linear polarization of monolayers with
MDC–SHG spectroscopy of organic monolayer film
373
Cs -symmetry in the laboratory frame (x, y, z) (see table I in Ref. [41]). In other words, ) of monolayers with Cs -symmetry are calculated the ten non-zero SOS components (s pq as s11 = (s31 + s15 ) sin θt cos2 θt + s33 sin3 θt s12 = s31 sin θt s13 = s31 sin3 θt + (s33 − s15 ) sin θt cos2 θt s15 = (2s33 − 2s31 − s15 ) sin2 θt cos θt + s15 cos3 θt s24 = s15 cos θt s26 = s15 sin θt s31 = s31 cos3 θt + (s33 − s15 ) cos θt sin2 θt s32 = s31 cos θt s33 = s33 cos3 θt + (s31 + s15 ) cos θt sin2 θt s35 = (2s33 − 2s31 − s15 ) cos2 θt sin θt + s15 sin3 θt
(32)
If the interaction between the water surface and molecules makes a significant contribution, probably, this is true in an actual monolayer system, the distribution function f (Φ, Θ) should be dependent on both Θ and Φ, and thus the monolayer changes symmetry from C∞v to Cs -symmetry. From Eq. 24, the angle θ is dependent on Φ and Θ if θt = 0. In this case, the spontaneous polarization of monolayers should be described using two order parameters, y 10 = cos Θ and y 11 = sin Θ cos Φ, and the non-linear polarization of monolayers should be described using six parameters, y 10 =
cos Θ, y 11 = sin Θ cos Φ, y 30 = (5 cos3 Θ − 3 cos Θ)/2, y 31 = (3 sin Θ(5 cos2 Θ − 1) cos Φ)/2, y 32 = 15(1 − cos2 Θ) cos Θ cos 2Φ, and y 33 = 15 sin3 Θ cos Φ [45]. However, the merit of the MDC measurement coupled with SHG measurement for the detection of the phase transition of monolayers [27,28,22] has already been shown by simply using the polarization approximately given by Eq. 31, in which only two order parameters S1 and S2 are used [27,51]. By using SHG measurement, the phase transition from C∞v → Cs has been confirmed in 4-heptyloxy-4 -cyanobiphenyl (7OCB) monolayers [52]. Further we may conclude that the interaction among molecules can be discussed using aforementioned six order parameters determined by MDC–SHG measurement.
5. Summary The dielectric polarization of organic monolayers at the air–water interface has been analyzed, assuming that the monolayers have a C∞ -symmetry. The formula of polarization of organic monolayers is derived and it is expressed using orientational order parameters. It was revealed that Maxwell displacement current (MDC) measurement coupled with optical second harmonic generation (SHG) measurement is helpful for the determination of these orientational order parameters as well as for the detection of phase transitions. Monolayers of 4 -n-octyl-4-cyanobiphenyl (8CB) and fatty acids at
374
A. Tojima et al.
the air–water interface were examined during monolayer compression and the molecular motion of monolayer was discussed on the basis of the orientational order parameters determined in the experiment.
References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39.
Special issue on Functional Organic Materials for Devices, J. Mat. Chem. 9, 1853 (1999). Organic Thin Films, Materials Chemistry Discussion No. 2, J. Mat. Chem. 10, 1 (2000). A.J. Heeger, Rev. Mod. Phys. 73, 681 (2001). A.G. MacDiamid, Rev. Mod. Phys. 73, 701 (2001). H. Shirakawa, Rev. Mod. Phys. 73 713 (2001). C.J. Brabec, N.S. Sariciftci, and J.C. Hummelen, Adv. Funct. Mater. 11, 15 (2001). A. Aviram and M.A. Ratner, Chem. Phys. Lett. 29, 277 (1974). C. Joachim, J.K. Gimzewski, and A. Aviram, Nature 408, 541 (2000) C.M. Fischer, M. Burghard, S. Roth, and K. v. Klitzing, Appl. Phys. Lett. 26, 3331 (1995). Y. Zhang and S. Iijima, Phys. Rev. Lett. 82, 3472 (1999). H.S. Nalwa, Supramolecular Photosensitive and Electractive Materials, (Academic Press, San Diego, 2001). M. Iwamoto and M. Kakimoto, in Polyimides: Fundamentals and Applications, edited by M.K. Ghosh and K.L. Mittal (Marcel Dekker, New York, 1996) pp. 815–884. G.L. Gaines, Insoluble Monolayers at the Liquid–Gas Interfaces (Interscience, New York, 1966). A. Ulman, Characterization of Organic Thin Films, (Butterworth-Heimann, Boston, 1995). V.M. Kaganer, H. Möhwald, and P. Dutta, Rev. Mod. Phys. 71, 779 (1999). M. Iwamoto and C.X. Wu, The Physical Properties of Organic Monolayers (World Scientific, Singapore, 2001). A. Sugimura, M. Iwamoto, and Z.C. Ou-Yang, Phys. Rev. E 50, 614 (1994). D.M. Taylor and G.F. Bayes, Phys. Rev. E 49, 1439 (1994). R.E. Collin, Field Theory of Guided Waves (McGraw-Hill, New York, 1960) Ch. 12. M.Iwamoto, IEICE Trans. on Electronics E83-C, 1062 (2000). M. Iwamoto et al., Nature (London) 353, 645 (1991). M. Iwamoto, Y. Majima, H. Naruse, T. Noguchi, and H. Fuwa, J. Chem. Phys. 95, 8561 (1991). M. Iwamoto and Y.Majima, J. Chem. Phys. 94, 5135 (1991). Y.R. Shen, The Principles of Nonlinear Optics (Wiley, New York, 1984). N. Blöembergen and P.S. Pershan, Phys. Rev. 128, 606 (1962). T.F. Heinz, in Nonlinear Surface Electromagnetic Phenomena, edited by H.E. Ponath and G.I. Stegemen (Elsevier Science, New York, 1991), pp. 397–398. A. Tojima, T. Manaka, and M. Iwamoto, J.Chem.Phys. 115, 9010 (2001). A. Tojima, Y. Matsuo, R. Hiyoshi, T. Manaka, Y. Majima, and M. Iwamoto, Thin Solid Films 393, 86 (2001). S. Chandrasekhar, Liquid Crystals (Cambridge Univ. press, London, 1977). P.G. de Gennes, The Physics of Liquid Crystals (Clarendon, Oxford, 1991). A. Saupe, Z. Naturforsch. 19a, 161 (1964). V. Tsvetkov, Acta Physicochim. (USSR) 16, 132 (1942). C. Kittel, Introduction to Solid State Physics (Wiley, New York, 1974). C.X. Wu, W. Zhao, M. Iwamoto, and Z.C. Ou-Yang, J. Chem. Phys. 112, 10548 (2000). H. Fröhlich, Theory of Dielectrics (Oxford Univ. Press, New York, 1958). G.R. Luckhurst and G.W. Gray, The Molecular Physics of Liquid Crystals (Academic Press, London, 1979). M.E. Rose, Elementary Theory of Angular Momentum (Wiley, New York, 1957). M. Iwamoto, C.X. Wu, and Z.C. Ou-Yang, Chem. Phys. Lett. 325, 545 (2000). C.C. Wang, Phys. Rev. 178, 1457 (1969).
MDC–SHG spectroscopy of organic monolayer film 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52.
375
A. Tojima, T. Manaka, and M. Iwamoto, Rev. Sci. Instrum., to be published. J.A. GiordMaine, Phys. Rev. 138, A1599 (1965). M. Iwamoto, T. Kubota, and M.R. Muhamad, J. Chem. Phys. 102, 19368 (1995). J. Xue, C.S. Jung, and M.W. Kim, Phys. Rev. Lett. 69, 474 (1992). E.B. Sirota, Langmuir 13, 3849 (1997). M. Iwamoto and Z.C. Ou-Yang, J. Chem. Phys. 117, 7705 (2002). T. Kihara, Rev. Mod. Phys. 25, 831 (1953). K.M. Aoki, Y. Tabe, and H. Yokoyama, Mol. Cryst. Liq. Cryst. 367, 2979 (2001). J.L. Barrat and L. Bocquet, Phys. Rev. Lett. 82, 4671 (1999). See, for example, N. Israelachivichi, Intermolecular and Surface Forces (Academic Press, London, 1992). M. Iwamoto, T. Manaka, A. Tojima, and Z.C. Ou-Yang, Chem. Phys. Lett, 359, 169 (2002). M. Iwamoto, Y. Mizutani, and A. Sugimura, Phys. Rev. B 54, 8186 (1996). M. Iwamoto et. al., submitted.
Nanotechnology and Nano-Interface Controlled Electronic Devices Editors: M. Iwamoto, K. Kaneto and S. Mashiko © 2003 Elsevier Science B.V. All rights reserved
CHAPTER 18
Light-driven dynamic controls in nano-hybrid materials Takahiro Seki * Chemical Resources Laboratory, Tokyo Institute of Technology, 4259 Nagatsuta-cho, Midori-ku, Yokohama, Japan
1. 2.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Photoalignment of polysilane chain by Az monolayer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Procedures and preparative conditions of polysilane film . . . . . . . . . . . . . . . . . . . 2.1.1. System and principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1.2. Photoalignment procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1.3. Effects of film thickness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. Optimum conditions of the azobenzene monolayer. . . . . . . . . . . . . . . . . . . . . . . . . 2.2.1. Lateral packing density . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.2. Tail length . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.3. Micro-patterning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. Facilitated photoinduced mass migration by nano-hybridization . . . . . . . . . . . . . . . . . . . . 3.1. Binary component hybrid materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1.1. Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1.2. SRG Formation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1.3. Thermal Properties and SRG Stability . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Soft Crosslinkable Liquid Crystalline Polymer. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.1. Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.2. SRG formation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.3. Crosslinking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
377 378 378 378 379 381 382 383 383 384 385 385 385 387 390 391 391 391 392 393 394 394
1. Introduction In biological systems, nano-scale elements such as proteins, DNA, and lipid membranes are assembled in very sophisticated ways so that they exert various functions for life. * Present address: Department of Applied Physics, Graduate School of Engineering, Nagoya University, Chikusa, Nagoya 464-8603, Japan.
378
T. Seki
This could be the ultimate example for the utility of hybridization. In materials science and technology, hybridization of materials at nanometer scales is of key essence in many aspects for technological innovations, for example, creation of nano-composite for high modulus materials, smart and intelligent materials that find themselves a suitable response to an environmental stimulus, high performance optical and electrical devices. These materials and devices strongly rely on the hybridization and interfacial controls between different materials. This chapter introduces our on-going two topics on light-responsive, smart, soft material systems in which the interface and hybridization play the essential roles in the generation of functions. In both subjects, the dynamic photofunctions are produced by the trans/cis isomerization of azobenzene (Az). In the former part, photoresponsive azobenzene monolayers that are capable of controlling the orientation of polymer chains are mentioned. It is revealed that the in-plane orientation of polysilane (silicon catenated polymer) backbone is controlled through transfer from a pre-oriented Az monolayer by linearly polarized light (LPL). In the latter part, the photoinduced mass migration in soft liquid crystalline Az polymers will be introduced. By nano-hybridization with low molecular mass liquid crystal (LC) molecules, the efficiency of mass migration is greatly improved, by more than three orders of magnitude.
2. Photoalignment of polysilane chain by Az monolayer 2.1. Procedures and preparative conditions of polysilane film 2.1.1. System and principle The surface-mediated photoalignment has recently become an important technology in liquid crystalline materials [1,2]. The orientational control of polymer chains by such photochemical procedure is an alluring and challenging target because it may provide new technologies for micro-processing of polymer films. Here, the photoalignment behavior of (polydi-n-hexylsilane) (PDHS) by an azobenzene monolayer (6Az10-PVA) is introduced [3–5]. Illustrative representations of the system and chemical structure of materials are shown in Fig. 1. Use of a polymeric material for the Az monolayer is of particular significance for the following reasons. First, the reversible trans/cis photoisomerization reaction readily proceeds due to the amorphous nature of the monolayer. Second, the polymeric monolayer is mechanically and environmentally so robust that the monolayer is not damaged by the successive solvent spin-cast procedure. Irradiation of LPL to the monolayer induces the orientation of Az moiety preferentially in the perpendicular direction to the polarization plane of actinic light with respect to the in-plane component. This effect is called the photoinduced optical anisotropy or Weigert effect (Fig. 2) [6,7]. The transition moment of Az is nearly parallel to the direction of the long axis of the rod-like molecule. When the randomly oriented Az units are irradiated LPL, the Az units whose long axis are in the parallel direction are preferentially excited. This situation causes the reorientation of the Az side chains to a non-excitable direction, namely orthogonal to the light polarization direction. In the present case, the light illumination at both wavelengths (365 and 436 nm) is performed
Light-driven dynamic controls in nano-hybrid materials
379
Fig. 1: Schematic illustration of the surface-mediated photoalignment of poly(di-n-hexylsilane).
with an ultrahigh-pressure Hg lamp. Fig. 3 shows the polarized absorption spectra of the 6Az10-PVA monolayer after irradiation with LPL at 436 nm. In this typical example the dichroic ratio (DR = [Abs(⊥) − Abs()]/[Abs(⊥) + Abs()]) is 0.43. This large in-plane anisotropy cannot be obtained with simple irradiation with 436 nm LPL from the initial trans state. The higher in-plane orientation is only obtained after non-polarized 365 nm light is pre-irradiated to the Az monolayer. This may be ascribed to the fact that the cis-Az film is more fluid [8], having larger motional freedom. This can lead to efficient molecular reorientation. 2.1.2. Photoalignment procedure A spin-cast film of PDHS (Mw = 2.5 × 104 , film thickness = 45 nm) from a hexane solution is subsequently prepared onto this photooriented Az monolayer. Use of hexane
380
T. Seki
Fig. 2: Schematic illustration of photoinduced optical anisotropy by linearly polarized light in photochromic molecular systems. The molecules reorient perpendicular to the polarization plane of the irradiating light.
Fig. 3: Photoinduced in-plane anisotropy in the 6Az10-PVA monolayer. A⊥ and A indicate the absorption spectra taken with perpendicular and parallel vector of the probing light.
as the cast solvent does not damage or destroy the photoinduced orientation of the Az monolayer. Storage of the PDHS film in the dark at room temperature for 2 days allows sufficient crystallization of this polymer giving the absorption peak around 360 nm. As shown in Fig. 4a, the film shows no preferential in-plane orientation just after the solvent casting. After crystallization the PDHS film exhibited a strong in-plane anisotropic nature (Fig. 4b). Since the transition moment of the Si backbone is along the backbone direction, the polarized absorption spectra indicates that the Si main chain is aligned perpendicular to the polarization plane of the actinic light. The aligned direction of PDHS main chain agrees with that of the Az orientation on the substrate. This fact indicates that the crystallization of PDHS chain occurred on the photooriented
Light-driven dynamic controls in nano-hybrid materials
381
Fig. 4: Photoalignment behavior of PDHS film on the LPL-irradiated 6Az10-PVA monolayer. (a) Immediately after solvent cast, (b) after first crystallization of PDHS, (c) after annealing, and (d) after successive second crystallization of PDHS.
Az monolayer in an epitaxial manner. The orientational order of PDHS film was further enhanced upon annealing and successive cooling. Upon heating, the backbone conformation adopts the helical gauche state (320 nm absorption) with enhanced the in-plane orientational order (Fig. 4c). Successive crystallization leads to a more highly oriented PDHS film (Fig. 4d). 2.1.3. Effects of film thickness Since the in-plane-component orientation of the Az monolayer is transferred to the PDHS chain in the contacting region, the orientational order of the entire film should be dependent on the film thickness [4]. PDHS films with varied thickness from 10 to 100 nm are prepared on the pre-irradiated Az monolayer. Fig. 5 shows the order parameter (S) as a function of film thickness for the PDHS films of the lower molecular weight (Mw = 2.5 × 104 ) (a) and the higher one (Mw = 1.4 × 106 ) (b). In both cases, the orientational order of the PDHS backbone becomes larger for the thinner films. The order parameter is enhanced after the second crystallization for the lower molecular weight sample (triangles in Fig. 5a). The orientational order is enhanced upon annealing and the S for thicker films became comparable with those of the thinner ones. It is reasonably assumed that the influence of the photooriented Az layer does not reach
382
T. Seki
Fig. 5: Orientational order parameter (S) of photoaligned PDHS film on the 6Az10-PVA monolayer as a function of film thickness. The circle and triangle represent data after the first and second crystallization, respectively. Mw s of PDHS are 2.5 × 104 (a) and 1.4 × 106 .
the overall thickness only after the first crystallization. However, the lateral packing correlation of the disordered part is improved by heating, leading to an enhancement of orientational order at least within the thickness less than 100 nm. This interpretation is rationalized by AFM topographic observations. In the case of the high molecular weight material, the photoalignment of PDHS is considerably suppressed (Fig. 5b). When the PDHS film becomes thicker than 30 nm, essentially no alignment in the PDHS chain is induced. 2.2. Optimum conditions of the azobenzene monolayer Much knowledge has been accumulated on the LPL induced molecular reorientation in photochromic LB films (mostly Az systems) [9–12]. However, little attention has been paid as to what extents the lateral packing density and structural modification of the molecular structure affect the photoorientation behavior. Changes in the molecular structure at the alkylene spacer and alkyl tail attached to the Az part have been found to influence the molecular cooperativity with contacting liquid crystal molecules in the command surface systems [13,14].
Light-driven dynamic controls in nano-hybrid materials
383
Fig. 6: In-plane dichroic ratio (DR) of Az monolayers (open) and A of PDHS films (closed) as a function of occupying area of Az unit. The occupying area is controlled by pre-mixing of the spreading solutions containing trans- and cis-6Az10-PVA.
2.2.1. Lateral packing density The lateral packing density of an LB monolayer can be readily changed by the position of the moving barrier on the water surface. The molecular occupying area (Aoc ) can be varied by changing the surface pressure for transfer [15]. An alternative way to control Aoc is to vary the mixing ratios of the trans/cis isomers of Az in the spreading solution. Since the cis-Az monolayer occupies much larger areas than the trans one at low pressures, deposition of mixed monolayers at a fixed low surface pressure (5 mN m−1 ) allows large variations of Aoc . The data of S for the PDHS film on the photooriented Az monolayer are shown in Fig. 6 together with the DR data of the Az monolayer [5]. Here, the area is changed by mixing the two isomerized states at different ratios. A good correlation was obtained between the profiles of DR for the Az monolayer and S for the crystallized PDHS film. These parameters commonly show a maximum around 0.4 nm2 per Az unit. 2.2.2. Tail length The length of tail part is anticipated to influence the order of photoinduced anisotropy because the length of the alkyl chain greatly affects the molecular packing. In the present system, the Az monolayer is anchored to the hydrophilic substrate surface via the polar PVA backbone and the tail part is positioned to the outermost surface. PDHS should interact directly with the tail part of the Az side chain. In this context, an exploration on the tail effect is of great interest. DR of the Az layer after photoorientation (436 nm LPL, 3.0 J cm−2 ) and the resulting S of PDHS (thickness = 25 nm ) as a function of the carbon number of the tail part are shown in Fig. 7. In all cases, the PDHS backbone is commonly aligned parallel to the direction of the Az monolayers, namely perpendicular to the polarization direction of the actinic light. As obviously shown, the highest DR of Az monolayers and S of PDHS are obtained for the Az monolayer having the C8 tail [5]. DR and S are 0.56
384
T. Seki
Fig. 7: DR of Az monolayers (open) and S of PDHS films (closed) as a function of tail length (m) in the mAz10-PVA monolayer.
and 0.68, respectively, for the C8 tail. The magnitude of the DR has a good relationship with the degree of spectral shift. The monolayer showing the larger hypsochromic shift of λmax provides the larger DR. Thus, the higher orientational order is obtained for the monolayer containing more amounts of H-aggregated Az in the common lateral density of 0.4 nm2 per Az unit. 2.2.3. Micro-patterning A great benefit to apply the photoprocess is the feasibility of micro-patterning. Fig. 8 shows polarized optical microscopic images of a locally photoaligned PDHS film as a bent line. In this experiment, the optimized conditions of photoalignment are employed: The Az layer with the C8 tail deposited at Aoc = 0.4 nm2 per Az unit is used. The irradiation to the Az monolayer is first performed in the above conditions through a photomask placed in contact with the surface. Onto this Az layer, the PDHS film (thickness = 25 nm) is prepared by solvent spin-casting. The rotation of the crossed polarizers at 45° indicates the clear switching of emergence of disappearance of the line. This fact indicates that the bright line area is birefringent and the PDHS chains are uniaxially aligned. The dark parts do not change its low transmission by rotation, indicative of the random orientation of the polymer at micrometer levels. Thus, the micro-patterning of polymer alignment at 6 µm resolution can be successfully achieved [5]. In the photoprocess, basically any desired pattern is applicable. It would be impossible or laborious to attain such locally addressed orientation via mechanical or flow processes.
Light-driven dynamic controls in nano-hybrid materials
385
Fig. 8: Polarized optical microscopic images of a microscopically photoaligned DHS film (thickness = 25 nm) via LPL irradiation through a photomask. The width of bright line is ca. 6 µm.
3. Facilitated photoinduced mass migration by nano-hybridization 3.1. Binary component hybrid materials This section gives another good example that exhibits a strong molecular cooperativity in dynamic processes observed in solvent cast films. The essentials have much in common to monolayer systems [16,17]. 3.1.1. Background Az polymers are potentially useful as materials for reversible holographic information storage and photonic devices [18–22]. Surface relief grating (SRG, regular topological surface modification) formed via the irradiation with an interference pattern of coherent
386
T. Seki
Fig. 9: Schematic illustration of photoinduced surface relief gratings (SRG) and an typical example of the SRG structure in 6Az10-PVA/5CB nano-hybrid film observed by AFM.
light has been demonstrated only recently [23–25] and is perhaps the most attracting target in the current research of Az polymers (Fig. 9). A great deal of data has been accumulated quite rapidly due to its basic phenomenological interest [26–30] and also to attractive technological applications [30]. This process has particular technological advantages since (i) it offers a facile, all-optical and single step fabrication process that does not require a wet development procedure, and (ii) the surface topology is erasable by application of circularly polarized light or heating above the glass transition temperature (Tg ), which realizes the repeatable utilization. It is of no doubt that the SRG is formed via large-scale polymer chain migration, however, the precise mechanism is still the subject of intensive investigation.
Light-driven dynamic controls in nano-hybrid materials
387
Fig. 10: Chemical structure of the components of the nano-hybrid materials for SRG production.
All attempts for SRG experiments to date deal with single-component Az polymers including the side-chain and main-chain types. A novel approach using a binary component hybrid host–guest system for the formation of SRG is introduced here [31,32]. The present film comprise of two materials, 6Az10-PVA and a typical liquid crystal (LC) molecule, 4 -pentyl-4-cyanobiphenyl (5CB) (Fig. 10). 3.1.2. SRG Formation The spin-cast films (50–100 nm thickness) are prepared from chloroform solutions dissolving 6Az10-PVA and 5CB at f = 0.5, f being the molar fraction of 5CB ([5CB]/([Az unit] + [5CB])). The hybrid films are irradiated with non-polarized UV (365 nm) light in advance to attain a cis-rich photoequilibrated state (UV light treatment). The interference Ar ion laser beam can be obtained simply by mixing a direct irradiation beam and one reflected by a mirror (Fig. 11). Fig. 12a displays
Fig. 11: Experimental setup for irradiation of interference Ar ion laser beam onto the nano-hybrid film.
388
T. Seki
Fig. 12: (a) First-order diffraction efficiency of the spin-cast films of pure 6Az10-PVA (diamonds) and an equimolar 6Az10-PVA/5CB hybrid film ( f = 0.5) with (circles) and without (triangles) UV light treatment as a function of total irradiation energy. (b) and (c) AFM images of the 6Az10-PVA/5CB hybrid film ( f = 0.5) after irradiation with Ar ion interferometric beam for 0.5 s (b) and 5.0 s (c) at 50 mW cm−2 .
the growth profiles of the first-order diffraction efficiency evaluated with an reflected He–Ne laser beam (488 nm) as a function of the total exposure energy. This figure contains the data obtained with the pure 6Az10-PVA film (diamonds), the hybrid film of 6Az10-PVA/5CB with UV light treatment (circles) and without treatment (triangles). For the hybrid film with UV light treatment, a sharp growth of diffraction efficiency is observed at an early exposure stage (circles). The diffraction efficiency reaches a maximum in 250 mJ cm−2 , corresponding to ca. 5.0 s exposure at 50 mW cm−2 . The increase in the diffraction efficiency synchronizes with the growth of the surface geometrical modulation. Fig. 12b and c show AFM images of the hybrid films exposed with the interferometric Ar ion laser beams at 50 mJ cm−2 for 0.5 and 5.0 s, respectively. The initial film surface before exposure to the writing beams is smooth within 5 nm depth fluctuations with no regular periodicity. The SRG structure is clearly generated
Light-driven dynamic controls in nano-hybrid materials
389
even at a strikingly short period of 0.5 s with a spatial frequency 500 mm−1 in exact coincidence with the interference pattern of the light intensity. It is noteworthy that such quick topological induction proceeds at room temperature. At 5.0 s exposure, the surface modulation becomes more clear and provides a regularly spaced sinusoidal structure (12c). The depth from the peak to the bottom (∆h) is ca. 70 nm, which is comparable to the initial film thickness. The photo-modulated structure is stable and unchanged at least for a month at room temperature. The surface undulation can be erased by non-polarized UV light irradiation at 200 mJ cm−2 that is nearly sufficient for the attainment of the cis-rich photoequilibrated state at room temperature. Heating at 100°C (isotropic phase of the hybrid film) for 30 min also deletes the surface structure. For the pure 6Az10-PVA film in the same procedure with UV light treatment, no significant surface undulation is formed as confirmed by AFM. This striking enhancement of the SRG efficiency in the hybrid film is attributable to (i) the sufficient overlapping of the n–π∗ absorption band of cis-Az with the 488 nm Ar-ion laser beam and (ii) the plasticization of the film in the cis-Az form. The trans-tocis photoisomerization of Az leads to an increase in the film fluidity of 6Az10-PVA as shown by the microscopic observation [8]. Judging from the general knowledge that the SRG is formed below Tg of polymer films [30], the surface undulation should be formed along with stiffening of the hybrid film due to the back isomerization of Az to the trans form by 488 nm illumination. Fig. 13 indicates the most essential feature of the binary system. This figure displays the diffraction efficiency (closed circles) and the surface modulation depth (open circles, ∆h) of the hybrid films at various molar fractions performed after UV light treatment. The diffraction efficiency shows a sudden increase (to 2.3%) at f = 0.67, which corresponds to the stoichiometry of two 5CB molecules per one Az unit. Below and above this ratio the diffraction efficiency rapidly decreases. The profile of ∆h almost follows that of the diffraction efficiency to give the maximum depth (100 nm) at f = 0.67.
Fig. 13: The first-order diffraction efficiency (closed) and surface modulation depth (open) of 6Az10PVA/5CB hybrid film at various molar fraction of 5CB and 6Az10-PVA.
390
T. Seki
Fig. 14: DSC diagrams of pure 6Az10-PVA and 5CB and their binary mixtures.
3.1.3. Thermal Properties and SRG Stability Differential scanning calorimetric (DSC) profiles for the pure materials and the mixtures are shown in Fig. 14 [32]. The DSC trace of pure 6Az10-PVA ( f = 0.0) contains two endothermic peaks at 71 and 98°C. The higher temperature endotherm is associated with the clearing transition. On cooling from the isotropic phase, pure 6Az10-PVA developed focal conic and schlieren textures which are characteristic of smectic phase. The DSC profile of pure 5CB ( f = 1.0) shows two endothermic peaks at 23 and 35°C which correspond to the crystal–nematic and nematic–isotropic phase transitions, respectively. On the other hand, DSC diagrams of the binary mixtures of 6Az10-PVA and 5CB ( f = 0.5–0.8) exhibit a new strong endothermic peak near 46–49°C. This indicates that the mixtures behave like a newly formed hybridized liquid crystalline polymer. At higher 5CB molar ratios ( f > 0.67), DSC diagrams provide additional two peaks in the low temperature region (indicated by arrows), which are enlarged with increasing the 5CB content. These endothermic peaks should arise from pure 5CB, and thus the phase separation is suggested above f = 0.67. The Langmuir monolayer experiments also show that the co-spreading of these materials on water provides the lateral phase separation above f = 0.67 as proven from the surface pressure–area isotherms, UVvisible spectroscopic data, and Brewster angle microscopic observation [16]. In both the bulk and monolayer, the 6Az10-PVA can accommodate two 5CB molecules per Az side chain unit without phase separation, and above this criterion, the phase separation starts to occur. The film transport seems to be strongly assisted by the cooperative self-assembling process.
Light-driven dynamic controls in nano-hybrid materials
391
Upon heating the SRG inscribed film ( f = 0.5), the diffraction efficiency increased gradually until 55°C, and then shows a sudden decrease. The critical temperature in the Az/LC hybrid film is nearly in accord with the strong endothermic peak at 47°C in the DSC curve (Fig. 14). Elevating temperature up to 114°C almost completely diminished the light diffraction. After heating to this level, the surface becomes highly flat with some small hollows with a depth comparable to the initial film thickness. The topological modulation should be erased at the stage of either the mesophasic (47–90°C) or isotropic (>90°C) state. 3.2. Soft Crosslinkable Liquid Crystalline Polymer 3.2.1. Motivation Another important requirement for SRG material is the shape stability in terms of longterm storage and durability at higher temperatures. The stability can be improved when one employs amorphous polymers with high Tg [33,34] or liquid crystalline polymers having high Ti (transition temperature to isotropic state) [35]. However such materials markedly reduce the mass mobility. A unique soft and crosslinkable SRG polymer is developed by Zettsu et al. [36] The polymer involves an oligo(ethylene oxide) (EO) side chain (6Az10-PE4.5, Fig. 15) instead of incorporation with LC molecule. After the surface relief structure is formed, the polymer is then subjected to chemical crosslinking via formalization (acetal formation with formaldehyde) between the hydroxyl group at the terminus of EO. 3.2.2. SRG formation 6Az10-PE4.5 adopts a liquid crystalline state at room temperature where light exposure experiments are performed. The SRG structure is formed also at very low dose levels around 100 mJ cm−2 . The photo-modulated structure is adequately stable as far as it is kept at room temperature. The surface undulation could be erased by heating up to the isotropization temperature above 80°C (see below), and regenerated at the same position many times.
Fig. 15: Chemical structure of soft and crosslinkable liquid crystalline Az polymer (6Az10-PE4.5) that shows rapid photoinduced mass transfer.
392
T. Seki
Fig. 16: Schematic illustration of chemical crosslinking using formaldehyde vapor.
Fig. 17: Changes in the first-order diffraction efficiency (normalized) upon heating. Circles and squares indicate data for non-crosslinked and crosslinked films, respectively. The initial first-order diffraction efficiency was ca. 3%.
3.2.3. Crosslinking When the inscribed film is exposed to a mixed vapor of formaldehyde and hydrogen chloride for 24 h at room temperature, the formalization reaction (acetal formation) proceeds. This procedure chemically links two hydroxyl groups located at the terminus of EO unit to yield a crosslinked polymer network (Fig. 16). Fig. 17 shows the changes in diffraction efficiency (normalized) on heating for the inscribed 6Az10-PE4.5 film before and after formalization. Before formalization, the diffraction efficiency reduces drastically around 80°C, which is in exact accord with the isotropization temperature of this liquid crystalline polymer. In sharp contrast, the formalized SRG film retains
Light-driven dynamic controls in nano-hybrid materials
393
the diffraction efficiency up to 240°C without appreciable reduction. The 6Az10-PE4.5 film after formalization reaction becomes insoluble in tetrahydrofuran or chloroform both of which are good dissolving solvents for the untreated 6Az10-PE4.5. The SRG structure disappears by heating at 100°C for the untreated film whereas the formalized film exactly preserves the surface modulation pattern even at 250°C without any damage. The 6Az10-PE4.5 is in an isotropic state above ca. 100°C, however, this does not cause a reduction in diffraction efficiency. The process employed here can be compared with a simple approach using Az polymers of high Tg . Fukuda et al. [33,34] employed maleimide-based high Tg amorphous polymer (Tg = 170–279°C). In their polymer systems, the thermal stability is considerably improved, however, in compensation for requirement of vast amounts of exposure energy. Light doses required for SRG inscription for such high Tg polymer typically range some hundred J cm−2 . This is at least ca. 103 fold larger than that needed for the 6Az10-PE4.5 film. The soft and crosslinked 6Az10-PE4.5 holds a comparable thermal stability as such high Tg polymers. Since the discovery of the photoinduced migration the typical light dose required for SRG generation ranges in the order of tens to hundreds J cm−2 , which require long irradiation time spans reaching to several ten minutes with an Ar ion laser beam of moderate intensity (10 mW cm−2 level). Due to the requirement of vast light doses, the application of this process is limited to attain static resulting functions such as holographic optical recording, waveguide formation, liquid crystal aligning etc. [30]. On the other hand, for the rapid migrating polymer systems described here, utilization of motional functions may be possible, for example, application to soft micro actuators and micro-patterning of other guest materials by means of mass transfer.
4. Summary This chapter described two sets of light-driven dynamic material systems. In the orientational transfer system from the Az monolayer to the polysilane film, nanohybridization at the two-dimensional interface should play important roles for the effective control. Here, the monolayer modifications at nanometer levels such as subtle lateral density and tail length crucially alter the control behavior. Probably the cooperative interactions between the alkyl tail of the Az unit in the monolayer and the hexyl substituent in the polysilane are essentially involved in this process. The cooperative effect is quite obvious in the photoinduced mass transfer process observed for the hybrid films. The effective transport is attained only when the LC molecule or a soft segment is incorporated. It is emphasized that the nano-hybridization provides new functions in soft materials. For the present cases, high segmental flexibility and molecular mobility is realized while molecular orientation is retained in the nanohybridized states. This situation can be found typically in biological systems, where we find a number of processes and structures that inspire us for materials science and technology.
394
T. Seki
Acknowledgements The data described in this chapter were obtained by Dr. K. Fukuda (Section 2, present address: Mitsui Chemical Co.), Dr. T. Ubukata (Section 3, present address: RIKEN Frontier Program), and Mr. N. Zettsu (Section 3) of our laboratory with great cooperation and aid of Prof. K. Ichimura (Present address: Science University of Tokyo) and Dr. M. Nakagawa (CRL Tokyo Institute of Technology). I am grateful to all of them. The work was supported by the Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology of Japan.
References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34.
K. Ichimura, Chem. Rev. 100, 1847 (2000). M. O’Neill and S.M. Kelly, J. Phys. D: Appl. Phys. 33, R67 (2000). T. Seki, K. Fukuda, and K. Ichimura, Langmuir 15, 5098 (1999). K. Fukuda, T. Seki, and K. Ichimura, Macromolecules 35, 2177 (2002). K. Fukuda, T. Seki, and K. Ichimura, Macromolecules 35, 1951 (2002). T. Todorov, N. Tomova, and L. Nikolva, Opt. Commun. 47, 123 (1983). K. Anderle, R. Birenheide, M.J.A. Werner, and J.H. Wendorff, Liquid Cryst. 9, 691 (1991). T. Seki, H. Sekizawa. S. Morino, and K. Ichimura, J. Phys. Chem. B 102, 5313 (1998). M. Barnik, V.M. Kozenkov, N.M. Shtykov, S.P. Palto, and S.G. Yudin, J. Mol. Electron. 5, 53 (1989). S. Yokoyama, M. Kakimoto, and Y. Imai, Langmuir 10, 4594 (1994). M. Schönhoff, L.F. Chi, H. Fuchs, and M. Löshe, Langmuir 11, 163 (1995). R. Wang, L. Jiang, T. Iyoda, D.A. Tryk, K. Hashimoto, and A. Fujishima, Langmuir 12, 2052 (1996). T. Seki, Supramol. Sci. 3, 25 (1996). T. Seki, K. Ichimura, R. Fukuda, and Y. Tamaki, Kobunshi Ronbunshu 52, 599 (1995). T. Seki, R. Fukuda, T. Tamaki, and K. Ichimura, Thin Solid Films, 243, 675 (1994). T. Ubukata, T. Seki, and K. Ichimura, J. Phys. Chem. B 104, 4141 (2000). T. Ubukata, T. Seki, S. Morino, and K. Ichimura, J. Phys. Chem. B 104, 4148 (2000). M. Eich, J.H. Wendorff, B. Reck, and H. Ringsdorf, Makromol. Chem. Rapid. Commun. 8, 59 (1987). S. Ivanov, I. Yakovlev, S. Kostromin, V. Shibaev, L. Läsker, J. Stumpe, and D. Kreysig, Makromol. Chem. Rapid. Commun. 12, 709 (1991). J. Stumpe, L. Müller, and D. Kreysig, Makromol. Chem. Rapid. Commun. 12, 81 (1991). A. Natansohn, P. Rochon, J. Gosselin, and S. Xie, Macromolecules 25, 2268 (1992). T. Ikeda and O. Tsutsumi, Science 268, 1873 (1995). P. Rochon, E. Batalla, and A. Natansohn, Appl. Phys. Lett. 66, 136 (1995). D.Y. Kim, S.K. Tripathy, L. Li, and J. Kumar, Appl. Phys. Lett. 66, 1166 (1995). P.S. Ramanujam, N.C.R Holme, and S. Hvilsted, Appl. Phys. Lett. 68, 1329 (1996). J. Kumar, L. Li, X.L. Jiang, D.Y. Kim, T.S. Lee, and S.K. Tripathy, Appl. Phys. Lett. 72, 2096 (1998). C.J. Barrett, P. Rochon, and A. Natansohn, J. Chem. Phys. 109, 1505 (1998). P.S. Ramanujam, M. Pedersen, and S. Hvilsted, Appl. Phys. Lett. 74, 3227 (1999). K. Sumaru, T. Yamanaka, T. Fukuda, and H. Matsuda, Appl. Phys. Lett. 75, 1878 (1999). N.K. Viswanathan, D.Y. Kim, S. Bian, J. Williams, W. Liu, L. Li, L. Samuelson, J. Kumar, and S.K. Tripathy, J. Mater. Chem. 9, 1941 (1999). T. Ubukata, T. Seki, and K. Ichimura, Adv. Mater. 12, 1675 (2000). T. Ubukata, T. Seki, and K. Ichimura, Colloids Surfaces A 113, 198 (2002). T. Fukuda, H. Matsuda, N. Viswanathan, S.K. Tripathy, J. Kumar, T. Shiraga, M. Kato, and H. Nakanishi, Synth. Met. 102, 1435 (1999). T. Fukuda, H. Matsuda, T. Shiraga, T. Kimura, M. Kato, N. Viswanathan, J. Kumar, and S.K. Tripathy, Macromolecules 33, 420 (2000).
Light-driven dynamic controls in nano-hybrid materials 35. 36.
395
L. Andruzzi, A. Altomare, F. Ciardelli, R. Solaro, S. Hvilsted, and P.S. Ramanujam, Macromolecules 32, 448 (1999). N. Zettsu, T. Ubukata, T. Seki, and K. Ichimura, Adv. Mater. 13, 1693 (2001).
Part E
Fabrication and Characterization Technology
Nanotechnology and Nano-Interface Controlled Electronic Devices Editors: M. Iwamoto, K. Kaneto and S. Mashiko © 2003 Elsevier Science B.V. All rights reserved
CHAPTER 19
Solvent-induced morphology in nano-structures Bin Cheng a , Hongtao Cui b , Brian R. Stoner b , and Edward T. Samulski a a Department
of Chemistry, CB# 3290, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599-3290, USA b Department of Physics and Astronomy, CB# 3255, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599-3255, USA
1.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1. Carbon nanotube brushes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2. Templated semiconductor oxide nanotubes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Preparation of honeycomb structure in CNT–polymer composite films . . . . . . 2.2. Preparation of aligned semiconductor oxide nanotubes . . . . . . . . . . . . . . . . . . . . . 2.3. Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. CNT–polymer composite films . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Templated oxide nanotubes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
399 400 401 401 401 401 401 402 402 405 407 409 410
1. Introduction As devices grow smaller and approach the nanoscopic scale, conventional fabrication techniques (pick-and-place, wafer-to-wafer transfer, etc.) will become obsolete. Forces such as gravity become virtually irrelevant as other interactions (surface tension, van der Waals forces, etc.) grow in importance. Even on the mesoscopic scale, wetting, capillary, and adhesion forces may dominate the interactions between components, and these forces can be employed to locate and assemble ∼100 µm-size components [1]. Hence, it is not surprising that when nano-objects are synthesized, the ultimate morphology in condensed phases of such objects – the relative spatial positioning of the objects – can be manipulated by controlling the attributes of the solvents
400
B. Cheng et al.
Fig. 1: SEM image of an aligned multi-walled carbon nanotube “brush” prepared via microwave plasmaenhanced chemical vapor deposition (from Ref. [14]; scale bar 3 µm).
from which they are harvested. Herein we will show that the morphology of two seemingly unrelated nano-structures – arrays of multi-walled carbon nanotubes and sol–gel-templated semiconductor-oxide nanotubes – share strikingly similar responses to a high-surface-tension solvent such as water. 1.1. Carbon nanotube brushes Chemical-vapor deposition (CVD) can be used to grow multi-walled carbon nanotubes (CNTs) in a close-packed (<100 nm spacing), vertical orientation from the surface of silicon impregnated with a dispersion of catalyst particles [2]. The resulting array of upright CNTs, each anchored to the silicon substrate at a catalyst site and each approximately the same height (∼30 µm), is referred to herein as a CNT “brush”. (Fig. 1 shows a scanning electron microscope image of the bare silicon substrate and edge of the CNT brush.) We affect a dramatic change in the morphology of the brush from the original upright, parallel array of CNTs into a honeycomb nano-structure. This is done by evaporating a dilute, volatile polymer solution deposited on the surface of the CNT brush. The thickness of the final dry polymer film is designed to be less than the height of the brush “bristles” – the length of the CNTs. The presence of water vapor during the polymer solution evaporation causes micro-droplets of water to condense on the cold evaporating solution surface, and the water surface tension causes severe structural changes in the array of CNTs – namely, coalescence of the CNTs into crop-circle-like defects as these water micro-droplets dry. The mechanical properties of this honeycomb reinforcement in thin films impregnated with polymer are examined. Also, field emission data from the honeycomb morphology was contrasted with data from the original CNT brush.
Solvent-induced morphology in nano-structures
401
1.2. Templated semiconductor oxide nanotubes The sol–gel porous alumina templating method pioneered by Martin and coworkers [3] was used to fabricate arrays of aligned nanotubes of In2 O3 and Ga2 O3 [4]. After the synthesis of the nanotubes the excess oxide is ground off the exposed surface of the template – a microporous alumina substrate comprised of nearly hexagonally packed cylindrical pores (200 nm i.d.) spanning the template thickness (50 µm). The other surface of the template is bonded to a glass substrate before the template is dissolved. The resulting dried, anchored, array of oxide nanotubes exhibits a collapsed or bundled morphology due to the high surface tension of the water and the associated forces during the drying process. This morphology can be affected by exchanging the aqueous washing solvent and drying the array of nanotubes in a low surface tension solvent (supercritical carbon dioxide).
2. Experimental 2.1. Preparation of honeycomb structure in CNT–polymer composite films Aligned multi-walled CNTs were deposited via microwave plasma enhanced CVD on oxidized silicon substrates in advance [5]. Dilute polymer solutions (e.g., polystyrene in toluene, polystyrene in benzene etc.) were slowly evaporated on the top surface of the aligned CNT brush. The residual solvent was removed by heating at 50°C in vacuum. A honeycomb morphology was induced in the brush when the evaporation was carried out in the presence of a moist airflow. The moisture content of the airflow was controlled by passing the air stream through a KNO3 saturated water solution (relative humidity, 94%). 2.2. Preparation of aligned semiconductor oxide nanotubes Aligned semiconductor oxide nanotubes were synthesized via the sol–gel porous alumina templating method [4]. In a typical experiment, a porous alumina membrane (purchased from Whatman, 1.3 cm diameter, 50 µm thickness) was used as a template by immersing in a sol of metal ions for a specific time. The sol-containing membrane was dried and annealed in air at elevated temperatures: 973 K for 12 h (In2 O3 ), and 773 K for 12 h (Ga2 O3 ). The oxide-containing membrane was bonded to a substrate and the exposed surface was polished with a 13 micron sandpaper. The array of quasiparallel semiconductor-oxide nanotubes was isolated by dissolving the alumina template with a 6 M aqueous NaOH solution. The surplus of NaOH can be removed by washing with water or ethanol prior to drying the array. 2.3. Measurements Field emission measurements were carried out in a vacuum chamber (10−7 Torr) at room temperature. The experimental setup is a point-plane type [6]. CNT brushes with
402
B. Cheng et al.
and without the honeycomb morphology were used as the cathode. The brushes were attached to a stainless steel substrate by silver paste. In this configuration, the CNTs comprising a brush were placed with their long axes perpendicular to the substrate. A stainless rod was used as the anode. The distance between the emitting brush surface and the end of the rod was determined by first lowering the rod to the sample until electrical contact was observed, then lifting the rod to a desired distance from the cathode surface. SEM measurements were conducted using a JEOL JEM-6300 scanning electron microscope. The samples were attached to a SEM stub with conductive tape, and a very thin layer of Au was sputtered on the surface. Mechanical measurements were investigated using a dynamic mechanical analyzer (Perkin-Elmer DMA7e) system using the oscillatory mode (1 Hz) in a temperature range from −120°C to 100°C. 3. Results and discussion 3.1. CNT–polymer composite films The nucleation and growth of aligned multi-walled CNTs via microwave plasmaenhanced CVD has been described in detail [5]. Fig. 1 shows a typical SEM image of aligned multi-walled CNTs prepared by this method wherein the height of the CNT “brush” is about 30 µm. We explored different polymer-solvents casting solutions (e.g. polystyrene in benzene, polystyrene in toluene, and polystyrene in chloroform) before choosing dilute (0.5%) polystyrene in toluene. The final thickness of polymer film can be designed to be less than the height of the CNTs by controlling the amount and the concentration of the polymer casting solution. Our initial goal was to slowly evaporate the polymer solution in order to produce a thin (∼1 µm) polymer film at the substrate–CNT brush interface. The objective of adhering a polymer to the CNT brush was originally pursued in an attempt to merely produce a stronger bond between the CNT “bristles” of the brush and the silicon substrate in an attempt to produce more robust arrays for field emission applications. However, after the solvent evaporated, we discovered that the entire brush could be lifted from the substrate intact in the form of a flexible thin film with the CNTs normal to the film surface. Fig. 2 shows a part of a partially peeled composite polymer film with the CNT brush embedded in it. The film appears contiguous and probably results from the evaporating polymer solution going to the substrate and wetting the substrate via capillary aided transport between the ∼100 nm-spaced CNTs. Since the CNTs are embedded in the polymer film with their axes normal to the film surface, it may be possible to pattern the underside of such a flexible CNT-containing polymer film with a conductor and thereby achieve patterned emission from a flexible composite. This fabrication methodology may provide an alterative to patterned growth of CNT nanostructures from patterned catalyst sites on rigid substrates [7]. When a polystyrene–toluene solution is evaporated on the top surface of the aligned CNT brush in the presence of moisture, crop-circle-like defects formed among the brush bristles (see Fig. 3). The defect distribution is not homogeneous. This result shows
Solvent-induced morphology in nano-structures
403
Fig. 2: SEM image of part of a peeled-up, flexible polystyrene composite film with the CNT brush morphology where the CNT bristles are oriented normal to the film surface (scale bar 100 µm).
Fig. 3: SEM image of crop-circle-like defects formed in the CNT brush after a dilute polystyrene solution in toluene is evaporated on the brush surface (scale bar 1 mm).
404
B. Cheng et al.
that moisture may play a key role to the formation of defects. Currently there is a considerable effort underway to understand the mechanism of “breath-figure” formation – the two-dimensional ordered microdroplets of water that form on the surface of evaporating organic liquids in the presence of water vapor. These two-dimensional structures can be “imprinted” in the solid state if one evaporates polymer solutions. In a recent report [8] this phenomenon is illustrated with evaporative cooling of a polystyrene–toluene solution. The resulting ordered array of holes in the polymer film was formed by evaporating in the presence of moisture with a forced airflow across the solution surface. A hexagonally packed array of holes – replicas of the ordered 2D array of condensed water microdroplets which sinks into the less-dense, evaporating polymer solution – is evident in the resulting polymer film [8]. Such observations prompted us to evaporate polystyrene–toluene solutions on the CNT brush surface in the presence of water vapor. The final result is the formation of honeycomb nano-structure in the brush with the honeycomb cell dimensions related to the humidity and velocity of moist airflow during the evaporation process. The honeycomb morphology of a CNTs–polystyrene composite film on its silicon substrate is shown in Fig. 4. At high magnification, the SEM pictures show that many CNTs comprise the walls of the honeycomb cells. We speculate that the high surface tension of water droplets (that have condensed on the brush surface during the polymer casting procedure) causes the CNT brush bristles to coalesce and collapse into this unique ultra-structure when the water dries. We have studied the field emission and mechanical properties of this novel polymer composite. Fig. 5 displays typical current–voltage (I –V ) characteristics corresponding to Fowler–Nordheim behavior for the original CNT brush (Fig. 5a) and the honeycomb composite (Fig. 5b). The maximum current we measured in the original CNT brush is about 20 µA whereas for composite honeycomb structure it is about 24 µA. No significant difference was observed in the field emission results between these two samples. Carbon fibers are routinely used as reinforcements for polymeric matrices in the aircraft and sports industries because of their high modulus and low density. For similar reasons – a high Young modulus (over 1 TPa) and high strength (tensile strength, 200 GPa) [9] – carbon nanotubes may have potential for applications requiring highmodulus and high-strength composite materials. In fact, some research groups [10,11] have reported that load transfer from CNTs to the matrix played a key role in the mechanical properties of composites. We have contrasted the Young modulus of a thin polystyrene film reinforced with the CNT honeycomb nanostructure with the modulus of pure polystyrene films of comparable dimensions (Fig. 6). The reinforced composite film is prepared by evaporating a second more concentrate polystyrene solution on the honeycomb morphology. SEM pictures of the honeycomb nanostructure impregnated with polymer show the distribution of CNTs in the polystyrene matrix. The CNTs are not parallel to one another within in the honeycomb wall; the solidifying polystyrene matrix during the second solution evaporation does not destroy the integrity of the tubes. Good interfacial bonding between polystyrene and nanotubes appears to be present in the SEM image and may account for the good load transfer from the polystyrene matrix to the nanotubes via the interfacial shear stresses and thereby explain the enhanced modulus of the reinforced film.
Solvent-induced morphology in nano-structures
405
Fig. 4: SEM images of honeycomb structures of polystyrene composite films with CNTs; (a) the top surface view of large area of honeycomb structure (scale bar 10 µm); (b) view of part of the area of he honeycomb structure (10 µm); (c) one enlarged hole with CNTs clearly visible within the walls (1 µm); (d) a tilted view of the honeycomb morphology (10 µm).
3.2. Templated oxide nanotubes When using the sol–gel templating method to prepare semiconductor oxides in the form of an array of aligned nanotubes, the synthesis includes purification steps wherein water or ethanol is used to wash away the surplus of NaOH employed used to dissolve the alumina template. After the NaOH is completely removed, the resulting array of tubular oxides is dried in air at room temperature. Our findings show that it is difficult to get a very good aligned array of individual, free-standing nanotubes. Fig. 7 shows the usual solvent-induced morphology of semiconductor nanotubes. Most nanotubes are either collapsed entirely (Fig. 7a), or bundled together (Fig. 7b). Other groups have reported similar results [12]. This morphology appears to be the result of solvent inducedforces (a surface tension effect) when the nanotube array is dried in air following the dissolution of the template. In order to avoid this collapsed/bundled morphology, a low surface tension solvent must be used in the final stage of processing. The surface tension of liquid CO2 (0.005 N/m) is much less than that of H2 O (0.072 N/m) and ethanol (0.024 N/m), and critical
406
B. Cheng et al.
-20
238 µm
-22
-24
2
ln(I/V )
20
-26
Z
-28 -30 -32 -34 -36
16
-38
Z
Current / µA
-40 0.000
0.002
0.004
0.006
0.008
0.010
1 / V
14 µm
12
8
4
0
(a) 200
400
600
800
1000
1200
1400
Voltage / V 238 µm
-24
24 2
ln(I / V )
-28
20
-32
Z
-36
Current / µA
-40
d e cr 0.002
0.004
0.006
e a si 0.008
ng Z
0.010
0.012
1/V
16
12
14 µm 8
4
0
(b) 200
400
600
800
1000
1200
1400
Voltage / V Fig. 5: Current–voltage characteristics of the original CNT brush and the polystyrene composite film; with the corresponding Fowler–Nordheim plots (insets); (a) the original CNT brush; (b) the polystyrene composite film with a honeycomb morphology.
Solvent-induced morphology in nano-structures
407
CNT and polymer composites Pure polymer
8
3.0x10
Young's Modulus(Pa)
8
2.5x10
8
2.0x10
8
1.5x10
8
1.0x10
7
5.0x10
0.0 -120 -100
-80
-60
-40
-20
0
20
40
60
80
100
o
Temperature( C) Fig. 6: A comparison of the Young modulus for the CNT-reinforced honeycomb morphology of polystyrene composite film with that of a comparable pure polystyrene film.
point drying is a well-established technique [13]. Comparisons of the solubility of some reagents in liquid CO2 , suggest that if ethanol is used to wash away the surplus of NaOH, the ethanol can be readily exchanged with liquid carbon dioxide. After the NaOH was completely removed after the template dissolution step, the templated nanotube array, wet with ethanol, was transferred to a high-pressure stainless cell for exchange with CO2 . Several exchanges with liquid CO2 insured that the array was eventually immersed in pure liquid CO2 . A reduction of the pressure of CO2 yielded a dry array wherein the morphology of the nanotubes exhibited little of the collapsed or bundled morphology. Fig. 8 shows the morphology of nanotubes harvested from liquid CO2 . Fig. 8a shows the SEM image of the top surface of the aligned array of nanotubes and Fig. 8b shows the SEM image of a cross section of the array. Individual nanotubes are evident confirming that the morphology of these arrays can be dramatically altered by the solvent from which they are dried (compare with Fig. 7).
4. Concluding remarks Solvent-induced phenomena sometimes have undesirable influences on the morphology of carefully designed nanomaterials. We have shown that the CNT brush morphology can be converted into a honeycomb nanostructure on silicon substrates by evaporating
408
B. Cheng et al.
Fig. 7: SEM images of semiconductor nanotubes developed in water or ethanol; (a) collapsed nanotubes (scale bar 10 µm); (b) bundled nanotubes (scale bar 10 µm).
a dilute polymer solution in a volatile organic solvent on the brush surface. This composite CNT honeycomb morphology exhibits a combination of good electrical and mechanical properties and may have potential as a robust field emitter or other thin-film, functional polymer composites. The honeycomb morphology appears to be the result of water-induced capillary forces within the brush during drying. The same forces appear to be the origin of the collapsed or bundled morphology exhibited by sol–gel-templated processing of semiconductor nanotube arrays. Liquid and supercritical CO2 drying of nanostructures can eliminate the effect of capillary (or interfacial tension) forces on the
Solvent-induced morphology in nano-structures
409
Fig. 8: SEM images of semiconductor nanotubes developed in liquid CO2 ; (a) the top surface view (scale bar 10 µm); (b) a cross section view (scale bar 10 µm); neither views show collapsed nor bundled morphologies.
final morphology and enable the unique properties and potential applications of these novel materials to be realized.
Acknowledgements We thank Professor Joe DeSimone, and Professor Otto Zhou for their help and advice, and thank Dr. Guozhen Yue, Yuan Cheng for their assistance in the field emission
410
B. Cheng et al.
measurements. These findings are based upon work supported in part by the NASA Grant NAC-1-2301, the STC Program of the National Science Foundation under Agreement No.CHE-9876674, and instrumentation used in this research is supported by the Office of Naval Research MURI Grant N00014-98-1-0597.
References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14.
H.O. Jacobs, A.R. Tao, A. Schwartz, D.H. Gracias, and G.W. Whitesides, Science 296, 323 (2002). C.J. Lee, J. Park, S.Y. Kang, and J.H. Lee, Chem. Phys. Lett. 323, 554 (2000). J.C. Hulteen and C.R. Martin, J. Mater. Chem. 7, 1075 (1997). B. Cheng and E.T. Samulski, J. Mater. Chem. 11, 2901 (2001). H. Cui, O. Zhou, and B.R. Stoner, J. Appl. Phys. 88, 6072 (2000). W. Zhu, L. Bower, O. Zhou, G. Kochanski, and S. Jin, Appl. Phys. Lett. 75, 873 (1999). B.Q. Wei, R. Vajtai, Y. Jung, J. Ward, R. Zhang, G. Ramanth, and P.M. Ajayan, Nature 416, 495 (2002). M. Srinivasarao, D. Collings, A. Philips, and S. Patel, Science 292, 79 (2001). J.P. Salvetat, J.M. Bonard, N.H. Thomson, A.J. Kulik, L. Forro, W. Benoit, and L. Zuppiroli, Appl. Phys. A 69, 255 (1999). H.D. Wagner, O. Lourie, Y. Feldman, and R. Tenne, Appl. Phys. Lett. 72, 188 (1998). L.S. Schadler, S.C. Giannaris, and P.M. Ajayan, Appl. Phys. Lett. 73, 3842 (1998). B.A. Hernandez, K.-S. Chang, E.R. Fisher, and P.K. Dorhout, Chem. Mater. 14, 480 (2002). D.L. Goldfarb, J.J. de Pablo, and P.F. Nealey, J. Vac. Sci. Technol. B 18, 3313 (2000). C. Bower, W. Zhu, S. Jin, and O. Zhou, Appl. Phys. Lett. 77, 830 (2000).
Nanotechnology and Nano-Interface Controlled Electronic Devices Editors: M. Iwamoto, K. Kaneto and S. Mashiko © 2003 Elsevier Science B.V. All rights reserved
CHAPTER 20
Polarons in conjugated polymer and its composite with fullerene Kazuhiro Marumoto and Shin-ichi Kuroda Department of Applied Physics, Nagoya University, Chikusa-ku, Nagoya 464-8603, Japan
1. 2. 3.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Experimental details . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Light-induced electron spin resonance (LESR) and excitation spectra of photogenerated polarons in regioregular poly(3-alkylthiophene) (PAT) . . . . . . 3.2. LESR and excitation spectra of photogenerated polarons in regioregular PAT–C60 composite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3. Microwave saturation of LESR in regioregular PAT–C60 composite: spin– lattice and transverse relaxation time of photogenerated polarons . . . . . . . . . . . 3.4. Transient response of LESR in regioregular PAT–C60 composite . . . . . . . . . . . . 3.5. Dependence of LESR in regioregular PAT–C60 composite on the light excitation intensity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
411 413 414 414 416 420 423 425 426 427 427
1. Introduction Quasi-one-dimensional conducting nondegenerate conjugated polymers have been investigated extensively because of a wide variety of interesting physical properties such as electroluminescence, nonlinear optical effect, semiconductor–metal transition, etc. [1–3]. The charge carriers in the conjugated polymers are considered to be polarons [1], which attract much attention as nonlinear excitations in one-dimensional electronic systems. Polarons carry spin and charge and are also expected to play important roles in manifesting the functions such as high conductivity, electroluminescence, etc. of these materials. However, the nature of the polaronic state is not yet completely clear. Photoinduced charge separation in polymeric semiconductors is important because its study not only contributes to understanding of the basic photoexcited states in these one-dimensional semiconductors, but also contributes to the development of efficient
412
K. Marumoto and S.-i. Kuroda
nonlinear optical and photovoltaic devices [4]. Composites of conducting polymers, such as poly(3-alkylthiophene) (PAT) or poly( p-phenylenevinylene) (PPV) derivatives, and a high electron-affinity species of fullerene (C60 ) have been investigated extensively [5–14], because highly efficient photoinduced charge separation occurs in the polymer– C60 composites upon C60 doping, which is useful to develop efficient nonlinear optical and photovoltaic devices. This charge separation is attributed to photoinduced electron transfer between polymer and C60 , which forms positive and negative polarons on the polymer chain and C60 molecule, respectively. Recently, it has also been shown that such efficient charge separation occurs in blends of PPV derivatives and cadmium selenide nanocrystals [15,16]. Light-induced electron spin resonance (LESR) is a direct microscopic method for detecting and studying the photogenerated polarons, and has been employed to clarify the polaronic state in the typical electroluminescent polymer PPV and its derivatives [17–26] and oligothiophenes [27]. For the PAT–C60 composites using regiorandom PATs, the LESR spectra and microwave power dependence of the LESR intensity have been studied [9,11], in addition to observations of the photoluminescence quenching and the remarkable enhancement of the photoconductivity [5,7,8]. Recently, regioregular PATs have been synthesized, and electronic absorption, X-ray diffraction, and crossed polarizing micrograph studies show that the cast films of the regioregular PATs are self-organized, crystalline, flexible, and bronze-colored films with a metallic luster, while those of the regiorandom PATs are amorphous and orange-colored films [28]. The regioregular PATs exhibit a small exciton bandgap (1.7 eV) which is 0.4 eV lower than that of the regiorandom PATs (2.1 eV) [28]. The typical conductivity and charge carrier mobility of the regioregular PATs in thin films is greatly improved and are three orders of magnitude larger than those of the regiorandom PATs, which is due to self organization of the regioregular PATs, having a lamella structure with two-dimensional conjugated sheets formed by interchain stacking [29–32]. Moreover, gate-induced superconductivity in the regioregular PAT film of a field-effect device has been observed very recently, and it is reported that the appearance of superconductivity seems to be closely related to the self-assembly properties of the polymer [33]. The charge separation process in the PAT–C60 composites is expected to be greatly affected by such regioregularity of PATs, however, PAT–C60 composites using regioregular PATs have not been studied. Moreover, excitation spectra, transient response upon irradiation by light, and excitation-light intensity dependence of the LESR signals of the PAT–C60 composites have not been reported so far, which would provide important information concerning the mechanism of the charge separation and recombination of the photogenerated polarons. In the sections that follow, we report on LESR studies of photogenerated polarons in a regioregular PAT–C60 composite, in addition to a pure regioregular PAT, using variable photoexcitation energy [34,35]. A remarkable enhancement of the LESR signals in the excitation spectrum is observed compared with the case of the pure regioregular PAT, which is consistent with the enhancement of the photoconductivity at around 1.8 eV where an optically forbidden transition of C60 occurs. Prompt and persistent LESR signal components are observed, and the excitation-light intensity dependence shows that the prompt contribution increases monotonically with increasing excitation-
Polarons in conjugated polymer and its composite with fullerene
413
light intensity while the persistent contribution is found to be excitation intensity independent. The comparison of the present LESR results with those of PPV and its derivatives, oligothiophenes, the regiorandom PAT–C60 composites, and the composites of PPV derivative and C60 derivative is also discussed.
2. Experimental details Regioregular poly(3-octylthiophene) (PAT8) was used to prepare the PAT–C60 composite (Fig. 1). The concentration of C60 to PAT8 was 5 mol%. Ultrasonic treatment of PAT8– C60 toluene solution was carried out with an ultrasonic disintegrator for better uniform mixing. Cast films of the pure regioregular PAT8 and the PAT–C60 composite with each a thickness of approximately 6 µm were fabricated inside ESR sample tubes. ESR measurements were performed with a Bruker E500 X-band spectrometer with a microwave cavity with optical windows down to liquid-He temperature using an Oxford ESR900 gas-flow cryostat. The absolute magnitude of the g value was calibrated utilizing an NMR gaussmeter for obtaining the static magnetic-field strength and a microwave frequency counter. A JASCO SM-5 light source with a 300 W xenon lamp was used to provide excitation in the range 300–1100 nm (1.1–4.1 eV) at power levels up to 2 mW/cm2 with a spectral width of 10 nm. The light was delivered by an optical fiber to the quartz sample tube. For the excitation spectrum, the light intensity was adjusted to give the same photon flux at each wavelength. The LESR experimental procedure consisted of the following sequence [14]: (i) scan the ESR spectrum of the nonilluminated sample; (ii) scan the ESR spectrum under light illumination; (iii) turn off the illumination and scan the ESR spectrum; and (iv) warm the sample up to room temperature, cool it down to working temperature, and scan
Fig. 1: Structural formula of the (a) poly(3-octylthiophene) (PAT8) and (b) fullerene (C60 ).
414
K. Marumoto and S.-i. Kuroda
the ESR spectrum again. These ESR signals will be referred to as: “dark”, “light-on”, “light-off”, and “annealed” signals, respectively. As discussed below, the switching off of the excitation light does not lead to the disappearance of the ESR signals at low temperature. To eliminate the ESR spectrum completely, the sample had to be warmed up to room temperature (annealing). This step was performed every time while measuring the dependence of the temperature, microwave power, excitation-light energy, and excitation-light intensity. The dark and annealed ESR signals were confirmed to be the same every time. As default definition of the term LESR we choose the light-on signal corrected by subtracting the dark signal. Finally, we distinguish between the prompt LESR signal (light-on minus light-off) and the persistent one (light-off minus dark).
3. Results and discussion 3.1. Light-induced electron spin resonance (LESR) and excitation spectra of photogenerated polarons in regioregular poly(3-alkylthiophene) (PAT) The upper curves in Fig. 2 show the observed first derivative ESR spectra of the PAT8 under dark condition (dotted line) and 700 nm illumination (solid line) at 60 K with a microwave power of 0.06 mW. The lower curve in Fig. 2 shows the LESR spectrum obtained by the subtraction of the two curves mentioned above. The g value is obtained as g = 2.002. The observed spins in the dark condition might be attributed
Fig. 2: Upper curves: ESR spectra of the PAT8 under dark condition (dotted line) and 700 nm illumination (solid line) at 60 K. Lower curve: LESR spectrum of the PAT8 obtained by subtracting the dark spectrum from that under 700 nm illumination at 60 K.
Polarons in conjugated polymer and its composite with fullerene
415
to trapped polarons, possibly generated by oxygen doping or thermal excitation during the synthesis, as observed in PPV [17] and oligothiophenes [27]. The spin concentration in the dark condition is obtained as 1 spin per 3.5 × 105 PAT-molecular unit, which is approximately 14 times lower than that of the regiorandom poly(3-hexylthiophene) (PAT6) [36]. This low spin concentration of the regioregular PAT8 is probably due to less disorder sites in the polymer chains than in the case of the regiorandom PATs, because disorder sites are related to traps for the polarons. This result is consistent with the great improvement of the typical conductivity and charge carrier mobility of the regioregular PATs in thin films [29–32]. In the cases of two dialkoxy derivatives of PPV (CN-PPV and MEH-PPV), the dark signals were almost negligible [23–25], indicating almost no disorder sites in the polymer chains mentioned above. The ESR intensity of the PAT8 is enhanced considerably upon irradiation by light, and the LESR intensity is approximately 14 times as large as the dark ESR intensity. The dark ESR and LESR spectra are almost the same and have asymmetric line shapes. The peak-to-peak linewidth (∆Hpp) of the ESR signal of PAT8 is approximately 3.2 G. The LESR signal of the PAT8 becomes undetectable above approximately 240 K due to the higher recombination rate, confirming that the LESR signal is transient in nature [22,23,25]. The temperature dependence of the LESR intensity is similar to that of the PAT–C60 composite discussed later. The ∆Hpp of the LESR signal does not depend on the temperature, except for the low temperature region where the saturation occurs. The excitation spectrum of the LESR signal provides important information concerning the mechanism of the charge separation. Fig. 3 shows the variation of the normalized peak-to-peak LESR intensity (Ipp ) with the photon energy of the incident light for the PAT8. The measurements were performed with a microwave power of 0.06 mW at 60 K. For the comparison, the absorption spectrum of the PAT8 is also shown in Fig. 3.
Fig. 3: Excitation spectrum of the LESR signal of the PAT8 at 60 K. The solid line shows the absorption spectrum of the PAT8.
416
K. Marumoto and S.-i. Kuroda
The absorption measurement was performed at room temperature using a thin film cast on a glass substrate. The absorption spectrum is qualitatively consistent with that of poly(3-hexylthiophene) [28]. The excitation spectrum of the LESR signal shows a monotonic and rapid increase above 2.5 eV up to 4.1 eV, despite of the lower optical density, as observed in the PPV and its derivatives [20–25]. This behavior indicates that the observed spins are photogenerated polarons whose generation efficiency increases above the optical absorption peak due to excitons, because weakly bound electrons and holes created can be more readily dissociated into polarons. In the cases of PPV derivatives (CN-PPV and MEH-PPV), the excitation spectrum has a clear threshold at an energy of 2.7–2.8 eV in CN-PPV and approximately 3 eV in MEH-PPV, which is higher than the peak energy of the optical absorption of approximately 2.5eV in both polymers [23–25]. The difference in the energy of the optical absorption peak and the threshold of the LESR excitation spectrum provides the exciton binding energy, and the value of the binding energy becomes approximately 0.3–0.4 eV in these PPV derivatives [25]. For other energy region below 2.5 eV, there is an increase of the LESR intensity at around 1.77 eV (700 nm) as shown in Fig. 3. The LESR spectral line shape at 700 nm almost agrees with that at 300 nm, suggesting that the light-induced spin species are also polarons. Even with low photon energy, charge separation such as electron transfer between the molecule and oxygen or bipolaron dissociation may occur due to the existence of defects. Similar increases have been observed in PPV [20–22] and oligothiophenes [27], and a correlation between such peak and the existence of the spin concentration in the dark condition is suggested. Another explanation for the peak at around 1.77 eV may be given by the DeVore theory, which presents a similar peak in the shape of photoconductivity spectral distribution curves by a theoretical analysis based upon the effects of surface and volume recombination of the charge carriers liberated by the light [37]. Measurements of the photoconductivity spectra and comparison with the LESR excitation spectra are needed in order to study this peak in more detail, which are now in progress and will be published elsewhere [38]. 3.2. LESR and excitation spectra of photogenerated polarons in regioregular PAT–C60 composite The upper curves in Fig. 4 show the observed first derivative ESR spectra of the PAT–C60 composite under dark condition (dotted line) and 700 nm illumination (solid line) at 60 K with a microwave power of 0.06 mW. The lower curve in Fig. 4 shows the LESR spectrum obtained by the subtraction of the two curves mentioned above. Two LESR signals due to the photoinduced electron transfer between PAT8 and C60 are observed. The obtained g value of g1 = 2.002 and g2 = 1.999 correspond to positive and negative polarons on PAT8 and C60 , respectively, which is consistent with those of the regiorandom PAT–C60 composites [9,11]. The composites of PPV derivative and C60 derivative or C60 have been also reported to show two LESR signals with g values of g = 2.0000–2.0025 and g = 1.9955–1.9995 for low- and high-field ESR signals, which were assigned to the positive polaron P + on the conjugated polymer backbone and the C− 60 radical, respectively [6,9,14]. The observed polaron spins in the dark condition are considered to be attributed to the ground-state electron transfer in the
Polarons in conjugated polymer and its composite with fullerene
417
Fig. 4: Upper curves: ESR spectra of the PAT–C60 composite under dark condition (dotted line) and 700 nm illumination (solid line) at 60 K. Lower curve: LESR spectrum of the PAT–C60 composite obtained by subtracting the dark spectrum from that under 700 nm illumination at 60 K.
PAT–C60 composite, as discussed in the case of the regiorandom PAT–C60 composites [11], in addition to the trapped polarons mentioned above in the discussion of the pure PAT8. The spin concentration in the dark condition is obtained as 1 spin per 2.2 × 105 PAT-molecular unit, in other words, 1.4 × 1016 spin/g, which is more than 17 times lower than that of the regiorandom PAT–C60 composites [11]. In the case of the regiorandom PAT–C60 composites, the enhancement of spin density upon C60 doping − was more modest than that by using conventional dopants such as BF− 4 and AsF6 [11,39]. The ESR intensity of the PAT–C60 composite is enhanced considerably upon irradiation by light, and the LESR intensity is approximately 10 times as large as the dark ESR intensity. The dark ESR and LESR spectra are almost the same and have asymmetric line shapes. Since the LESR signals of PAT8 (g1 ) and C60 (g2 ) overlap with each other, it is difficult to obtain a precise half-amplitude linewidth (∆H1/2) from the experimental results directly. Hereafter, we define the peak-to-peak linewidths (∆Hpp) of the LESR signals of PAT8 and C60 as an interval of the magnetic field between a maximum and a minimum of the first derivative LESR spectrum around g1 and g2 , respectively. The ∆Hpp of the LESR signals of PAT8 and C60 are approximately 3.1 G and 2.4 G, respectively. Fig. 5 shows the temperature dependence of the LESR intensity Ipp in the PAT–C60 composite. The filled circles and open squares denote the signal intensity of PAT8 (g1 ) and C60 (g2 ), respectively. The data were recorded with a microwave power of 0.06 mW under 300 nm illumination. Similar temperature dependence of the LESR signals was also observed under 700 nm illumination. The ∆Hpp of the LESR signals does not depend on the temperature, except for the low temperature region where saturation occurs. The LESR signals become undetectable above approximately 200 K due to the higher recombination rate, confirming that the LESR signals are transient in nature [22,23,25]. The decrease of the signals at low temperature is caused by saturation of the ESR signals due to the longer spin–lattice relaxation rate. The LESR signal intensity
418
K. Marumoto and S.-i. Kuroda
Fig. 5: Temperature dependence of the LESR intensity of the PAT–C60 composite. The filled circles and open squares denote the signal intensity of PAT8 and C60 , respectively.
of PPV and its derivatives also shows a similar temperature dependence as mentioned above [22,23,25]. The LESR signal of PAT8 tends to saturate at higher temperature than that of C60 , indicating that the spin–lattice relaxation time of the polaron spins of PAT8 is longer than that of C60 , which is consistent with the results of the microwave-power dependence of the LESR intensity as discussed later. The excitation spectrum of the LESR signals in the PAT–C60 composite is shown in Fig. 6, which presents the variation of the normalized Ipp with the photon energy of the incident light for the PAT–C60 composite. The measurements were performed with a microwave power of 0.06 mW at 60 K. The data are plotted by using the signal of PAT8 (g1 ). For the comparison, the previously reported absorption spectrum (solid line) [39] and photocurrent spectrum (dotted line) [8] of the PAT–C60 composites are also shown in Fig. 6. The samples utilized for the absorption and photocurrent measurements are regiorandom poly(3-hexylthiophene) (PAT6)–C60 composite (C60 : 5 mol%) and regiorandom poly(3-octadecylthiophene) (PAT18)–C60 composite (C60 : 5 mol%), respectively. Although the data of the composites using regiorandom PAT6 and PAT18 are presented here, other composites using regiorandom PATs with different alkyl-chain length have been reported to exhibit similar characteristics [7]. The photoconductivity of the PAT–C60 composites is remarkably enhanced upon C60 doping due to the photoinduced electron transfer at around 1.8 eV where the optically forbidden transition of C60 (h u → t1u ) occurs in the composites. It has been reported that the enhancement of the photoconductivity in the regiorandom PAT18–C60 composites increases monotonically with increasing the concentration of C60 up to 10 mol% [8]. The excitation spectrum of the LESR signal of PAT8 (g1 ) shows a remarkable enhancement at around 1.77 eV (700 nm) and a monotonic increase above 2.5 eV up to 4.1 eV. The excitation spectrum of the LESR signal of C60 (g2 ) shows almost the same behavior as that of PAT8. There is almost no excitation-energy dependence of the LESR line shapes. The
Polarons in conjugated polymer and its composite with fullerene
419
Fig. 6: Excitation spectrum of the LESR signals of the PAT–C60 composite at 60 K. The solid and dotted lines show the absorption spectrum of PAT6–C60 composite (C60 : 5 mol%) and photocurrent spectrum of PAT18–C60 composite (C60 : 5 mol%) for the comparison, respectively.
enhancement at around 1.8 eV is clearly shown by comparing the LESR excitation spectrum of the PAT–C60 composite with that of the pure PAT8 shown in Fig. 3. If the magnitude of the LESR data of the PAT8 and PAT–C60 composite films is normalized using the same photon flux for the light intensity, the same sample weight for the PAT8 polymer, and the same film dimension, the LESR magnitude of the PAT–C60 composite is enhanced by a factor of 3.9 compared with that of the pure PAT8 at 700 nm, although it is enhanced by a factor of 1.1 at 300 nm. The enhancement of the LESR signals at around 1.8 eV is qualitatively similar to the enhancement of the photocurrent spectrum on the whole, which is consistent with the photogeneration of the polarons due to the photoinduced electron transfer. However, the detailed structures of the excitation spectra of the LESR and photocurrent around the peaks are quantitatively different from each other. The peak energy of the excitation spectrum of the LESR is slightly lower than that of the photocurrent, and considerable photogenerated signals are observed in the lower photon-energy region even at around 1 eV. One possible explanation for the differences of the excitation spectra between the LESR and photocurrent is given by considering aggregation effects of the polymer chain. The studies of PPV derivative (MEH-PPV) films have presented strong evidence that interchain species do form in conjugated polymer films, and that the degree of interchain interactions can be controlled by varying the solvent and polymer concentration of the solution from which the films are cast [40]. In the case of PATs films, as descried in a previous paper [41], the regioregular PATs have a planar conformation and exhibit greatly improved field-effect mobility [29]. This planarity leads to extension in the conjugation length having strong electron–phonon coupling leading to a significant correlation between the conformation and electro-optical properties [42]. The photocurrent spectrum shown in
420
K. Marumoto and S.-i. Kuroda
Fig. 6 has been measured using the regiorandom PAT–C60 composite, and the planar conformation in such regiorandom PAT–C60 composite may be different from that in the regioregular PAT–C60 composite. Therefore, the regioregularity of PATs may affect the interaction between PAT and C60 and cause the differences of the excitation spectra between the LESR and photocurrent mentioned above. For another possibility to explain the differences of the excitation spectra, the difference of the measurement conditions such as the effect of electric field may be considered, as mentioned in the discussion of the transient response later. Anyway, these differences on the aggregation effects and/or measurement conditions may affect the charge-separation processes between PAT and C60 in the PAT–C60 composite. Further measurements of the absorption, LESR and photoconductivity using the same sample seem to be needed to clarify the charge-separation processes and the photoexcitation of the polarons in more detail. 3.3. Microwave saturation of LESR in regioregular PAT–C60 composite: spin–lattice and transverse relaxation time of photogenerated polarons To obtain more information about the photogenerated polarons at around 1.8 eV where the enhancement occurs, the microwave-power dependence, the transient response upon irradiation by light, and the excitation-light intensity dependence of the LESR signals due to the photogenerated polarons have been studied. In this section, the microwave saturation of the LESR in regioregular PAT–C60 composite and spin–lattice and transverse relaxation time of photogenerated polarons are presented below. Fig. 7 shows the microwave-power dependence of the LESR signals at 60 K under 700 nm illumination. In the low microwave-power region, the signal intensity of PAT8 is large compared with that of C60 . At a microwave power of 200 mW, however, the signal of PAT8 has almost disappeared due to the saturation and almost only the signal of C60 appears. Similar microwave-power dependence of the LESR signals has been observed in the composites of PPV derivative and C60 derivative or C60 [14]. The ∆Hpp of the signal of C60 increases from 2.1 G to 5.3 G as the microwave power increases from 2 µW to 200 mW for the saturation effect. The definition of the ∆Hpp used here is mentioned above. The ∆Hpp of the signal of PAT8 seems to decrease from 3.4 G to 1.3 G as the microwave power increases from 2 µW to 60 mW because of the overlap between the LESR signals of PAT8 and C60 , as reported in the regiorandom PAT18–C60 composite [11]. Actually, the upper peak of the PAT8 signal shifts to the lower magnetic field as the microwave power increases, and therefore the true ESR linewidth of the signal of PAT8 is considered to increase as the microwave power increases for the saturation effect. Since the ESR linewidth considerably depends on the microwave power, the double integrated LESR intensity is estimated for the individual ESR signals and used in the following discussion of the microwave-power dependence. Fig. 8 shows the microwave-power dependence of the double integrated LESR intensity of the PAT–C60 composite at 60 K under 700 nm illumination. The filled circles and open squares denote the signal intensity of PAT8 (g1 ) and C60 (g2 ), respectively. When the microwave power increases, the LESR intensity of PAT8 saturates, showing a maximum at around 0.6 mW, and then decreases above 2 mW. On the other hand, the LESR intensity of C60 increases monotonically and does not saturate within the
Polarons in conjugated polymer and its composite with fullerene
421
Fig. 7: Microwave-power dependence of the LESR signals of the PAT–C60 composite at 60 K under 700 nm illumination.
experimentally available microwave-power range up to 200 mW. These features are similar to those of the regiorandom PAT–C60 composites [11] and the composites of PPV derivative and C60 derivative [14]. In the latter case, the temperature variation of the saturation curve for the low-field LESR signal due to the positive polarons has been measured between 90 K and 200 K. The microwave power corresponding to the saturation maximum increases from around 0.2 mW to 0.7 mW as temperature increases from 90 K to 120 K, and finally the saturation maximum cannot be reached with the microwave power available above 150 K [14]. The field for the saturation maximum 2 ). Here H1max is related to the spin–lattice relaxation time T1 by T1 ∝ 1/(T2 γe2 H1max is the microwave-field amplitude at the sample where the maximum in the saturation curve occurs, T2 is the transverse relaxation time, and γe is the electronic magnetogyric ratio, respectively. The LESR signal of PAT8 can be saturated at around 0.6 mW (60 K), whereas that of C60 does not saturate under the same conditions up to a power which is 330 times higher. Therefore, the T1 of the polaron spins of PAT8 is at least approximately 18 times longer than that of C60 . As discussed in the previous paper [14], the high-spin–lattice relaxation rate of the polaron spins of C60 is an intrinsic property.
422
K. Marumoto and S.-i. Kuroda
Fig. 8: Microwave-power dependence of the double integrated LESR intensity of the PAT–C60 composite at 60 K under 700 nm illumination. The filled circles and open squares denote the signal intensity of PAT8 and C60 , respectively.
A Jahn–Teller-type distortion on the fullerene ball would split the T1u level, and result in two closely spaced energy levels available for the polaron spins of C60 [43,44]. Thermal averaging over such states would provide a dominant relaxation channel for the spin, and would account for the nonsaturation of the C60 signal in the microwave-power dependence. The ESR linewidth of PAT8 signal of the PAT–C60 composite is governed by inhomogeneous broadening due to the hyperfine interaction between the polarons and protons bonded to the carbon in the system. In the case of an inhomogeneously broadened ESR line, the T2 cannot be determined directly by using the ∆Hpp. Hereafter, the continuous-wave saturation method considering the inhomogeneous broadening of the ESR linewidth is utilized to determine the T2 and T1 of the present system according to the standard procedures by Castner [45]. The concept of a spin packet is needed to describe the inhomogeneous broadening. A spin packet is composed of all spins having the same static field to within a field interval determined by the dephasing time T2 . The inhomogeneous broadening producing the static width is larger than the individual spin-packet width. The ratio ∆HL /∆HG of the Lorentzian spin-packet width to the inhomogeneous Gaussian width measures the degree of inhomogeneous broadening, and is expressed using a parameter a as a = ∆HL /∆HG . The quantity a is determined by using the ratio H1 (V R1/2 Upper )/H1(V R1/2 Lower ), where V R1/2 Upper and V R1/2 Lower are 1/2 the maxima of the ESR intensity V R for a given saturation curve, and the H1 (V R1/2 Upper ) and H1(V R1/2 Lower ) are corresponding microwave fields. The ratio H1 (V R1/2 Upper )/H1(V R1/2 Lower ) ≈ 50 is obtained from the experimental result shown in Fig. 8, and then the a is estimated as a ≈ 0.045 with the help of figure 3 in Ref. [45]. The T2 of the homogeneous Lorentzian line is defined as T2 = √ 1/(γe ∆HL ) = 1/(aγe ∆HG ). The ∆HG is estimated approximately by using ∆HG = 2 ln 2 ∆Hpp/2, and therefore
Polarons in conjugated polymer and its composite with fullerene
423
the T2 is estimated as T2 ∼ = 0.7 µs. The T1 is related to the T2 by T1 =
1 2 T2 γe2 H1/2
,
(1)
where H1/2 is H1 at the onset of saturation. The intersection of the linear signal below saturation with the horizontal line drawn tangent to the maximum of the saturation curve locates what we shall call H1/2 Uncorrected . The multiplicative correction factor which relates it to the true H1/2 is a known function of a and ranges from 1 for a equal zero to 2 for the completely homogenous case (for the homogenous saturation case H1/2 corresponds to V R max ) [45]. Using the values of T2 and H1/2 Uncorrected , the T1 is estimated as T1 ∼ 100 µs. 3.4. Transient response of LESR in regioregular PAT–C60 composite Fig. 9 shows the transient response of the LESR signal in the PAT–C60 composite upon irradiation by repeated light (700 nm) pulses with different time widths. The data were recorded using the lower field peak of the PAT8 signal (g1 ) with a microwave power of 0.06 mW at 100 K. The photoresponse is found to have two parts: a prompt part and a persistent one. That is, upon irradiation by light, the LESR signal increases stepwise then almost saturates. When irradiation ceases, the LESR signal decreases by some amount stepwise, and then the remaining part decreases very slowly. The same behavior is also observed for the data using the higher field peak of the C60 signal (g2 ). The prompt and persistent components are considered to be caused by shallow
Fig. 9: Transient response of the LESR signal of the PAT–C60 composite upon irradiation by repeated light (700 nm) pulses with different time widths. The data were recorded using the lower field peak of the PAT8 signal (g1 ) at 100 K.
424
K. Marumoto and S.-i. Kuroda
and deep traps in the PAT–C60 composite, respectively. The time response of the prompt component is considerably faster than that of PPV and its derivative [22,23]. The decay time of the remanent component is of the order of several hours, which is extremely long compared with that of the prompt component. It is difficult to fit these decay processes by double exponential functions. These features of the transient response are similar to those of the photoconduction of the regiorandom poly(3-octadecylthiophene) (PAT18)–C60 composite [46]. However, there are some differences between the transient responses of the LESR and photoconduction. In the LESR transient response, almost no slowly rising component after the stepwise increase upon irradiation by light is observed, and the magnitude of the remanent component is almost independent of the pulse width (duration) of the irradiation light. These features are inconsistent with those of the photoconduction of the PAT18–C60 composite, where the slowly rising component increases as the pulse width increases and so does the persistent photoconductivity. It is considered that these features in the photoconduction may originate in the accumulation of carriers contributing to the persistent photoconductivity [46]. The accumulation of carriers may be caused by the effect of the electric field that is not applied in the LESR measurements, which may cause the differences mentioned above. Fig. 10 shows the transient responses of the prompt LESR component for the PAT8 signal at the temperatures 60 K, 100 K and 120 K. As temperature increases, the magnitude of the prompt LESR component as well as the persistent one decreases. The decay time to half value of the prompt LESR component also decreases from 18 s (60 K) to 2.1 s (120 K). Similar tendency is also observed in the transient response of the prompt LESR component for the C60 signal. These features indicate that the recombination rate of the
Fig. 10: Transient responses of the prompt LESR signal of the PAT–C60 composite upon irradiation by light (700 nm). The data were recorded using the lower field peak of the PAT8 signal (g1 ) at 60 K, 100 K and 120 K.
Polarons in conjugated polymer and its composite with fullerene
425
polarons caught on the shallow and deep trap sites increases as temperature increases. For PPV and its derivative, similar results have been obtained from the LESR transient response, that is, the rise and decay time constants decrease as temperature increases [22,23]. 3.5. Dependence of LESR in regioregular PAT–C60 composite on the light excitation intensity The excitation-light intensity dependence of the double integrated LESR intensity for the prompt and persistent LESR signals in the PAT–C60 composite is shown in Fig. 11a and b, respectively. The filled circles and open squares denote the signal intensity of PAT8 (g1 ) and C60 (g2 ), respectively. These measurements were performed under 700 nm illumination at 60 K with a microwave power of 0.06 mW. The prompt component shows almost equal amounts of spins (i.e., almost equal doubly
Fig. 11: Dependence of the magnitude (in log10 base) of the LESR of the PAT–C60 composite on the intensity of the exciting light at 60 K: the prompt components (a), and the persistent components (b). The filled circles and open squares denote the signal intensity of PAT8 and C60 , respectively.
426
K. Marumoto and S.-i. Kuroda
integrated ESR intensity) and increases monotonically as excitation-light intensity increases for both the low- and high-field LESR lines of PAT8 (g1 ) and C60 (g2 ), 0.27 power dependence on the respectively. Both prompt components show a ∼ Iexc. intensity of excitation light. Almost the same behavior for the excitation-light intensity dependence is also observed at 100 K. This dependence shows a saturation behavior 0.5 ), with respect to the excitation-light intensity rather than the bimolecular type (Iexc. which is inconsistent with the results of the composites of PPV derivative and C60 derivative [14]. The present results suggest quadrumolecular-type recombination among the photogenerated positive and negative polarons, because the excitation-light intensity 0.25 , which is consistent with dependence of the quadrumolecular-type recombination is Iexc. the experimental results. The LESR transient responses shown in Fig. 10 have been also explained preliminary by using calculated curves based on the quadrumolecular-type recombination at various temperatures, which shows that the recombination rate of the photogenerated polarons obtained by the fitting calculation increases monotonically as temperature increases. Further experimental studies on the sample dependence, the C60 concentration dependence, and the effects of the regioregularity of PATs using regiorandom PATs are needed in order to study this power dependence on the excitationlight intensity in more detail. These studies are now in progress. Anyway, the prompt LESR data clearly shows the monotonic increase on excitation-light intensity, which means that the photogeneration of the polarons due to the photoinduced electron transfer between PAT8 and C60 increases as the excitation-light intensity increases. For the persistent component, it is nearly independent of the intensity of the previously applied light within the light-power range 0.038–0.61 mW/cm2. Similar excitation-light intensity dependence is also observed at 100 K. This result is consistent with that of the composites of PPV derivative and C60 derivative [14], which is related to the deep traps due to defect sites or disorder in the PAT–C60 composite.
4. Conclusion The cast films of the regioregular poly(3-octylthiophene) (PAT8) and its composite with fullerene (C60 ) having 5 mol% C60 concentration have been investigated by means of the light-induced ESR (LESR) method using variable photoexcitation energy up to 4.1 eV. For the pure regioregular PAT8, the excitation spectrum of the transient LESR signal due to the photogenerated polarons shows a monotonic increase above 2.5 eV, despite of the lower optical density, because of the dissociation of weakly bound electrons and holes created into polarons. For the PAT–C60 composite, the transient two LESR signals are observed, which come from the photogenerated positive and negative polarons on PAT8 and C60 due to the photoinduced electron transfer between PAT8 and C60 , respectively. A remarkable enhancement of the LESR signals in the excitation spectrum at around 1.8 eV is observed, which is consistent with the enhancement of the photoconductivity at around 1.8 eV where the optically forbidden transition of C60 occurs. Spin–lattice relaxation time of the polaron spins of PAT8 is obtained considering inhomogeneous broadening of the ESR linewidth, which is at least approximately 18 times longer than that of C60 . The prompt and persistent LESR signal components are observed, and
Polarons in conjugated polymer and its composite with fullerene
427
the excitation-light intensity dependence shows that the prompt contribution increases monotonically as the excitation-light intensity increases while the persistent contribution is found to be excitation-light intensity independent due to deep traps.
Acknowledgements The authors would like to thank Professor H. Ito and Dr. N.C. Greenham for valuable discussions and comments. We also would like to thank N. Takeuchi and T. Ozaki for their cooperation in the LESR measurements. This work is supported by NEDO International Joint Research Program, 99MB1 ‘Nonlinear Excitations in Molecular Electronic Materials: Detection, Control and Device Application’.
References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20.
21. 22.
A.J. Heeger, S. Kivelson, J.R. Schrieffer, and W.-P. Su, Rev. Mod. Phys. 60, 781 (1988). H.G. Kiess (Ed.), Conjugated Conducting Polymers, Springer Series in Solid-State Sciences 102, (Springer, Berlin-Heidelberg, 1992). F.H. Friend, R.W. Gymer, A.B. Holmes, J.H. Burroughes, R.N. Marks, C. Taliani, D.D.C. Bradley, D.A. Dos Santos, J.L. Brédas, M. Lögdlund, and W.R. Salaneck, Nature 397, 121 (1999). J.J.M. Halls, C.A. Walsh, N.C. Greenham, E.A. Marseglia, R.H. Friend, S.C. Moratti, and A.B. Holmes, Nature 376, 498 (1995). S. Morita, A.A. Zakhidov, and K. Yoshino, Solid State Commun. 82, 249 (1992). N.S. Sariciftci, L. Smilowitz, A.J. Heeger, and F. Wudl, Science 258, 1474 (1992). K. Yoshino, S. Morita, T. Kawai, H. Araki, X.H. Yin, and A.A. Zakhidov, Synth. Met. 55–57, 2991 (1993). K. Yoshino, X.H. Yin, S. Morita, T. Kawai, and A.A. Zakhidov, Solid State Commun. 85, 85 (1993). L. Smilowitz, N.S. Sariciftci, R. Wu, C. Gettinger, A.J. Heeger, and F. Wudl, Phys. Rev. B 47, 13835 (1993). G. Yu, J. Gao, J.C. Hummelen, F. Wudl, and A.J. Heeger, Science 270, 1789 (1995). S.B. Lee, A.A. Zakhidov, I.I. Khairullin, V.Yu. Sokolov, P.K. Khabibullaev, K. Tada, K. Yoshimoto, and K. Yoshino, Synth. Met. 77, 155 (1996). E.S. Maniloff, D. Vacar, D.W. McBranch, H.-L. Wang, B.R. Mattes, J. Gao, and A.J. Heeger, Opt. Commun. 141, 243 (1997). E.K. Miller, K. Lee, K. Hasharoni, J.C. Hummelen, F. Wudl, and A.J. Heeger, J. Chem. Phys. 108, 1390 (1998). V. Dyakonov, G. Zoriniants, M. Scharber, C.J. Brabec, R.A.J. Janssen, J.C. Hummelen, and N.S. Sariciftci, Phys. Rev. B 59, 8019 (1999). N.C. Greenham, X. Peng, and A.P. Alivisatos, Phys. Rev. B 54, 17628 (1996). D.S. Ginger and N.C. Greenham, Phys. Rev. B 59, 10622 (1999). S. Kuroda, T. Noguchi, and T. Ohnishi, Phys. Rev. Lett. 72, 286 (1994). S. Kuroda, K. Murata, T. Noguchi, and T. Ohnishi, J. Phys. Soc. Jpn. 64, 1363 (1995). S. Kuroda, Int. J. Mod. Phys. B 9, 221 (1995). S. Kuroda, K. Murata, Y. Shimoi, S. Abe, T. Noguchi, and T. Ohnishi, in: Materials and Measurements in Molecular Electronics, Springer Proceedings in Physics 81, edited by K. Kajimura and S. Kuroda (Springer, Tokyo, 1996), p. 256. K. Murata, S. Kuroda, Y. Shimoi, S. Abe, T. Noguchi, and T. Ohnishi, J. Phys. Soc. Jpn. 65, 3743 (1996). K. Murata, Y. Shimoi, S. Abe, S. Kuroda, T. Noguchi, and T. Ohnishi, Chem. Phys. 227, 191 (1998).
428 23. 24. 25. 26. 27. 28. 29. 30. 31.
32. 33. 34.
35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46.
K. Marumoto and S.-i. Kuroda S. Kuroda, K. Marumoto, H. Ito, N.C. Greenham, R.H. Friend, Y. Shimoi, and S. Abe, Chem. Phys. Lett. 325, 183 (2000). S. Kuroda, K. Marumoto, N.C. Greenham, R.H. Friend, Y. Shimoi, and S. Abe, Synth. Met. 119, 655 (2001). S. Kuroda, K. Marumoto, Y. Shimoi, and S. Abe, Thin Solid Films 393, 304 (2001). S. Kuroda, K. Marumoto, N.C. Greenham, R.H. Friend, Y. Shimoi, and S. Abe, Mol. Cryst. Liq. Cryst. 371, 159 (2001). K. Marumoto, N. Takeuchi, S. Kuroda, R. Azumi, and M. Matsumoto, Synth. Met. 119, 549 (2001). T.-A. Chen, X. Wu, and R.D. Rieke, J. Am. Chem. Soc. 117, 233 (1995). Z. Bao, A. Dodabalapur, and A.J. Lovinger, Appl. Phys. Lett. 69, 4108 (1996). H. Sirringhaus, N. Tessler, and R.H. Friend, Science 280, 1741 (1998). H. Sirringhaus, P.J. Brown, R.H. Friend, M.M. Nielsen, K. Bechgaard, B.M.M. Langeveld-Voss, A.J.H. Spiering, R.A.J. Janssen, E.W. Meijer, P. Herwig, and D.M. de Leeuw, Nature 401, 685 (1999). K. Kaneto, W.Y. Lim, W. Takashima, T. Endo, and M. Rikukawa, Jpn. J. Appl. Phys. 39, L872 (2000). J.H. Schön, A. Dodabalapur, Z. Bao, Ch. Kloc, O. Schenker, and B. Batlogg, Nature 410, 189 (2001). K. Marumoto, N. Takeuchi, and S. Kuroda, in: EPR in the 21st Century: Basics and Applications to Material, Life and Earth Sciences, edited by A. Kawamori, J. Yamauchi and H. Ohta (Elsevier, Amsterdam, 2002), p. 367. K. Marumoto, N. Takeuchi, T. Ozaki, and S. Kuroda, Synth. Met. 129, 239 (2002). M. Onoda, Y. Manda, S. Morita, and K. Yoshino, Jpn. J. Appl. Phys. 31, 2265 (1992). H.B. DeVore, Phys. Rev. 102, 86 (1956). H. Ito, K. Hayashi, T. Ozaki, K. Marumoto, and S. Kuroda, unpublished. S. Morita, A.A. Zakhidov, T. Kawai, H. Araki, and K. Yoshino, Jpn. J. Appl. Phys. 31, L890 (1992). T.-Q. Nguyen, I.B. Martini, J. Liu, and B.J. Schwartz, J. Phys. Chem. B 104, 237 (2000). S.S. Pandey, S. Nagamatsu, W. Takashima, and K. Kaneto, Jpn. J. Appl. Phys. 39, 6309 (2000). P. Barta, F. Cacialli, R.H. Friend, W.R. Salaneck, M. Zagorska, and A. Pro´n, Synth. Met. 101, 296 (1999). D. Dubois, M.T. Jones, and K.M. Kadish, J. Am. Chem. Soc. 114, 6446 (1992). J. Stinchcombe, A. Pénicaud, P. Bhyrappa, P.D.W. Boyd, and C.A. Reed, J. Am. Chem. Soc. 115, 5212 (1993). T. G. Castner, Phys. Rev. 115, 1506 (1959). K. Yoshino, T. Akashi, K. Yoshimoto, S. Morita, T. Kawai, and A.A. Zakhidov, Jpn. J. Appl. Phys. 34, L127 (1995).
Nanotechnology and Nano-Interface Controlled Electronic Devices Editors: M. Iwamoto, K. Kaneto and S. Mashiko © 2003 Elsevier Science B.V. All rights reserved
CHAPTER 21
Characterization of semiconductor surfaces with noncontact atomic force microscopy Seizo Morita and Yasuhiro Sugawara Department of Electronic Engineering, Graduate School of Engineering, Osaka University, Yamada-Oka 2-1, Suita, Osaka, Japan
1.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1. History of AFM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2. Principle of NC-AFM Measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3. Principles of spatial resolutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3.1. Expected vertical resolution of NC-AFM . . . . . . . . . . . . . . . . . . . . . . . 1.3.2. Expected lateral resolution of NC-AFM . . . . . . . . . . . . . . . . . . . . . . . . 1.4. Contact point and noncontact region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. Spatial resolution of high performance NC-AFM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Vertical resolution of home-built NC-AFM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. Lateral resolution of home-built NC-AFM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3. Atomic displacement around missing-dimer defects in Si(100)2×1 surface . 3. Functions of NC-AFM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Confirmed functions of high performance NC-AFM . . . . . . . . . . . . . . . . . . . . . . . 3.2. Three-dimensional mapping of atomic force . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3. Discrimination of atomic force mechanisms and atom species . . . . . . . . . . . . . . 3.4. Control of atomic force and atom position . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5. Electrostatic force imaging of local charges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.6. Atomic imaging of contact potential difference and capacitance . . . . . . . . . . . . 3.7. Atom manipulation using NC-AFM based on a mechanical method . . . . . . . . 3.7.1. Vertical manipulation based on a mechanical method . . . . . . . . . . . . 3.7.2. Lateral manipulation based on a mechanical method . . . . . . . . . . . . 4. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
429 430 431 433 433 433 434 436 437 438 439 442 442 443 444 447 448 449 449 450 451 452 452
1. Introduction The scanning tunneling microscope (STM) is an atomic tool based on an electric method that measures the tunneling current between a conductive tip and a conductive
430
S. Morita and Y. Sugawara
surface. On the other hand, the atomic force microscope (AFM) is a unique atomic tool based on a mechanical method that measures the deflection of a cantilever due to the force between the cantilever tip and a sample surface. Since the invention of AFM in 1986 by G. Binnig et al. [1] the AFM has rapidly developed into a powerful surface analysis technique on micro/nanoscales and even on atomic/molecular scales, because the AFM has the following characteristics: (1) it has true atomic resolution, (2) it can measure atomic force (so-called atomic force spectroscopy), (3) it can observe even insulators, and (4) it can measure mechanical responses such as elastic deformation. The noncontact AFM (NC-AFM) can fully satisfy these characteristics as follows. 1.1. History of AFM In 1986 Binnig et al. [1] invented the AFM and in 1987 Binnig et al. [2] succeeded to obtain the lattice image of a graphite surface in the contact mode. After that, the AFM was believed to be an atomic resolution microscope which could measure even in ambient atmosphere. Nevertheless, Pethica and Oliver [3] pointed out that the contact AFM could not observe atomic defects such as atom vacancy and also that the normal load for usual AFM measurement awfully exceeded the load limit of a single atom because of the strong repulsive force between a tip apex and a sample surface. Therefore the contact area for AFM measurement should be larger than a single atom, as shown in Fig. 1a, and so the contact AFM has not true atomic resolution. In connection with this suggestion, in 1992 Giessibl and Binnig [4,5] showed that KBr in UHV was
Fig. 1: Schematic models of AFM. (a) Contact AFM is destructive and has a large contact area because of the strong repulsive force. (b) Noncontact AFM (NC-AFM) is nondestructive and can observe even an atomic point defect if weak attractive force can be detected as a frequency shift of the mechanical oscillation of the cantilever.
Characterization of semiconductor surfaces with noncontact atomic force microscopy
431
destroyed by scanning from the monoatomic step lines at the loading force of only approximately 1 nN, although they succeeded to take a step line image with atomic resolution in UHV at 4 K using the contact AFM. Their best results were obtained when the total force acting on the cantilever was approximately −1 nN (attractive). Furthermore, Ohnesorge and Binnig [6] succeeded to obtain true atomic resolution of calcite in an aqueous environment by the AFM through repulsive and attractive forces and resolved atomic-scale kinks, representing point-like defects, along monoatomic step lines. Atomic resolution that showed STM-like reliability was achieved, particularly in the attractive-force regime in an aqueous environment. Nevertheless, to demonstrate true atomic resolution of AFM more clearly, atomically resolved imaging of both Si(111)7×7 and atomic defects with STM-like reliability have been claimed. In 1995 Giessibl [7] and Kitamura and Iwatsuki [8] succeeded in NC-AFM imaging of Si(111)7×7 surface with atomic resolution in UHV at room temperature (RT) using a frequency modulation (FM) detection method [9]. However, their images could not achieve STM-like reliability at that time. In the same year, Ueyama et al. [10] attained NC-AFM imaging of atomic vacancy with STM-like reliability, and then Sugawara et al. [11] finally achieved observation of defect motion of atomic vacancy on an InP(110) cleaved surface. In 1997 Bammerlin et al. [12] reported how true atomic resolution on NaCl(001) in UHV can be systematically obtained using NC-AFM. Thus using the FM detection method, the NC-AFM in the attractive regime in UHV clearly demonstrated to have true atomic resolution even for insulators at RT. 1.2. Principle of NC-AFM Measurement The NC-AFM using the FM detection method detects the frequency shift ∆ν of mechanical resonant oscillation of a cantilever due to the weak attractive force between a tip apex atom and a sample surface atom, as shown in Figs. 1b and 2. Under the conditions of small oscillation amplitude (A0 force interaction distance) and weak
Fig. 2: Oscillation amplitude as a function of mechanical oscillation frequency of the cantilever. ν0 , ν, ∆ν and A0 are the mechanically free oscillation frequency, the mechanical oscillation frequency of the cantilever under weak attractive force, frequency shift ∆ν = ν − ν0 , and oscillation amplitude, respectively.
432
S. Morita and Y. Sugawara
Fig. 3: Schematic diagram of the NC-AFM using the FM detection method. This system has three feedback loops.
force interaction (|∂ f/∂ z| k), the frequency shift ∆ν can be approximated as
∂f ∆ν = (ν0 /2k) − . ∂z
(1)
Here k, f and −∂ f /∂ z are the spring constant, the tip–sample force and the force gradient, respectively. However, most of the measurements were executed under the condition of large oscillation amplitude. Then, ∆ν is approximated as the geometric mean of the interaction energy V and tip–sample force f = −∂ V /∂ z, i.e., ∆ν ∝
Vf.
(2)
The NC-AFM uses three feedback loops as shown in Fig. 3. In contrast, the STM measurement under constant tunneling current uses only one feedback loop for STM topography. The first feedback loop of the NC-AFM is used to oscillate the cantilever at its mechanical resonant oscillation frequency ν under weak attractive force. The second feedback loop is used to maintain oscillation amplitude A0 (switch 1) or the excitation voltage to oscillate (switch 2) constant. The last feedback loop that includes the FM demodulator circuit is used to obtain NC-AFM topography under constant frequency shift condition (∆ν = constant). Thus the electronic circuit for the NC-AFM measurement is very complicated compared with that of the STM. Besides, high sensitive detection of very small deflection of the NC-AFM cantilever is a very difficult task. On the other hand, the STM can easily detect very small tunneling current with a high signal(S)-to-noise(N) ratio (S/N) using a simple current/voltage converter. Furthermore, the cantilever for the NC-AFM is not so stiff compared with the metal tip for the STM. Hence building/floor vibration will easily perturb the tip–sample distance of the NC-AFM and increase noise in the signal. As a result, achievement of high-S/N measurement in the NC-AFM is very difficult compared with the STM.
Characterization of semiconductor surfaces with noncontact atomic force microscopy
433
1.3. Principles of spatial resolutions Spatial resolution is a very fundamental value that determines the basic performance of the NC-AFM. Here, we will simply deduce approximated equations of the vertical and lateral resolutions of the NC-AFM. 1.3.1. Expected vertical resolution of NC-AFM Here the value g(z) measured with the NC-AFM was assumed to be proportional to exp(−z/L), where z and L are the tip–sample distance and the decay length of the measured value, respectively. This means that, by increasing the tip–sample distance from z to z + δz, the measured value will decrease from g(z) to g(z + δz) = g(z) exp(−δz/L). Then the S/N of g(z) was defined as h = S/N. This means that the measurable smallest change δg(z) of the signal g(z) is given by δg(z) = g(z)/ h, which is equivalent to the noise level. If the smallest change of the tip–sample distance controllable by the feedback loop is δz, the vertical resolution can be defined as δz. This also means that δg(z) is equivalent to g(z) − g(z + δz). Thus we can obtain the relation δg(z) = g(z)/ h = g(z) − g(z + δz) = g(z)[1 − exp(−δz/L)]. Then from 1/ h = [1 − exp(−δz/L)], we obtained the equation of vertical resolution: δz = L ln[h/(h − 1)]
(3)
as a function of the decay length L and the signal-to-noise ratio h. By assuming δz/L 1, that is, h 1 Eq. 3 is approximated as ∼ L/ h. δz = (4) Eq. 4 suggests that the scanning probe microscope (SPM) as well as the NC-AFM can obtain better vertical resolution by decreasing the decay length L, i.e., by making the tip–sample distance dependence of the measured value stronger, and by increasing the signal-to-noise ratio h [13]. 1.3.2. Expected lateral resolution of NC-AFM By assuming that the radius R of tip curvature is much larger than the tip–sample distance z 0 , i.e., R z 0 , the lateral resolution δx is approximately determined by the vertical resolution δz as shown in Fig. 4a, because the ambiguity of tip–sample distance is δz. From Fig. 4a, we can obtain the relation δz = R(1 − cos θ) and δx = R sin θ. From these equations, we can derive the equation of lateral resolution ∼ [δz(2R − δz)]1/2 . (5) δx = By assuming R δz, we can rewrite Eq. 5 as δx ∼ = [2Rδz]1/2 .
(6)
Eq. 6 suggests that we can obtain better lateral resolution by decreasing the radius R of tip curvature, and by improving the vertical resolution δz, i.e., by decreasing the decay length L and by increasing the signal-to-noise ratio h [13]. By assuming that the radius R of tip curvature is much smaller than the tip–sample distance z 0 , i.e., R z 0 , the lateral resolution δx is approximately determined by the
434
S. Morita and Y. Sugawara
Fig. 4: (a) Schematic simple one-dimensional model of tip–sample configuration in case of R z 0 [13]. (b) Schematic simple one-dimensional model of tip–sample configuration in case of R z 0 [13].
vertical resolution δz as shown in Fig. 4b, because the ambiguity of tip–sample distance is δz. From Fig. 4b, we can obtain the relation (z 0 + δz)2 = z 02 + δx 2 . From this relation, we can derive the equation of lateral resolution δx ∼ = [δz(2z 0 + δz)]1/2 .
(7)
By assuming z 0 δz, we can rewrite Eq. 7 as δx ∼ = [2z 0 δx]1/2 .
(8)
Eq. 8 suggests that we can obtain better lateral resolution by decreasing the tip–sample distance z 0 , and by improving the vertical resolution δz, i.e., by decreasing the decay length L and by increasing the signal-to-noise ratio h [13]. 1.4. Contact point and noncontact region Now we will show the experimentally obtained definitions of contact point and noncontact regions. Fig. 5 shows simultaneously measured approaching curves of the frequency shift and the oscillation amplitude in the constant-excitation mode (switch 2 in Fig. 3). From the oscillation amplitude curve in Fig. 5, we determined the contact point z = 0 nm where the oscillation amplitude begins to decrease due to cyclic contact by decreasing the tip–sample distance [14]. Then we assigned the contact and
Characterization of semiconductor surfaces with noncontact atomic force microscopy
435
Fig. 5: Simultaneously measured frequency shift and oscillation amplitude curves. Sample is Zn-doped p-GaAs(110) cleaved surface [15].
noncontact regions to the regions below and above the contact point as shown in Fig. 5 [15]. It should be noted that the absolute value of the frequency shift in the noncontact region at first increases slowly then quickly by decreasing the tip–sample distance. The smallest decay length was 0.16 nm. In general, the decay length takes the smallest value just before the contact point as shown in Fig. 5. This phenomenon also suggests that the rather strong repulsive force due to cyclic contact works below the contact point and suppresses the increase of the absolute value of the frequency shift by decreasing the tip–sample distance. Then we will show where we can obtain true atomic resolution using the NC-AFM. We obtained the NC-AFM images on Zn-doped p-GaAs(110) cleaved surface at three different tip–sample distances: (a) z ∼ 0.4 nm, (b) z ∼ 0.1 nm and (c) z ∼ 0.08 nm before the contact point as shown in Fig. 5. Fig. 6a, b and c show measured NC-AFM images and line profiles along the white lines in the corresponding NC-AFM images. The variable frequency shift mode which measures the frequency shift image (see Fig. 3) was used under weak feedback condition to suppress the thermal drift. From Fig. 6a measured at z ∼ 0.4 nm, we found that the NC-AFM image at the rather far distance shows only a large-scale vague contrast, perhaps, due to defects, but no atomic-scale contrast. On the other hand, from Fig. 6b measured at z ∼ 0.1 nm, we found that the NC-AFM image shows atomic-scale point defects as well as periodic lattice structures, although the image seems a little vague. Furthermore, from Fig. 6c measured at z ∼ 0.08 nm, we found that the NC-AFM image just before contact becomes very clear. Thus the distance dependence of the NC-AFM image is very strong. This result qualitatively agrees with the expectation predicted by the tip–sample distance dependence of the lateral resolution given by Eq. 8. On the other hand, in the contact region shown in Fig. 5, we could not stably obtain the atomic point-defect image because of the strong contact between the tip and sample surfaces. Therefore, we conjectured that true atomic resolution could be experimentally achievable only just before the contact point.
436
S. Morita and Y. Sugawara
Fig. 6: Distance dependence of the frequency shift image and cross-sectional line profile along white lines indicated in NC-AFM images measured at (a) z ∼ 0.4 nm, (b) z ∼ 0.1 nm and (c) z ∼ 0.08 nm [15]. Measurements were done in the constant excitation-voltage mode. Scan area is 10 nm × 10 nm.
2. Spatial resolution of high performance NC-AFM Here, we will demonstrate spatial resolutions obtained with our home-built NC-AFM. Table 1 roughly summarizes vertical and lateral resolutions of both high performance STM and NC-AFM for commercial, home-built and developing home-built SPM, respectively. Our home-built NC-AFM roughly has vertical resolution of 1 pm and lateral resolution of 10 pm. And it can observe metal atoms as follows.
Table 1 Spatial resolutions of STM and NC-AFM (2001) High performance STM Spatial resolution (rms) Vertical
Lateral
Commercial SPM
1 pm 0.01 Å
10 pm 0.1 Å
Home-built SPM
100 fm 0.1 pm 0.001 Å
Developing home-built SPM
10 fm 0.01 pm
High performance NC-AFM Applications
Spatial resolution (rms)
Applications
Vertical
Lateral
STS/metal atom observation
10 pm 0.1 Å
100 pm 1Å
semiconductor atom/ionic atom
1 pm 0.01 Å
atom manipulation/ electron waves on metal surface /inelastic STM
1 pm 0.01 Å
10 pm 0.1 Å
observation of metal atom/3D-measurement and control of atomic force, etc.
100 fm 0.1 pm 0.001 Å
true observation of metal atom?
100 fm 0.1 pm 0.001 Å
1 pm 0.01 Å
atom and molecule manipulation?
Characterization of semiconductor surfaces with noncontact atomic force microscopy
437
Fig. 7: (a) Model of Ag(111) film evaporated on Si(111)7×7 surface. (b) NC-AFM image of Ag(111) film evaporated on Si(111)7×7 surface [16]. Scan area is 80 nm × 80 nm. Cross-sectional line profile obtained along the white broken line shows two terraces with height difference of an atomic layer of Ag(111) surface. (c) NC-AFM image with two Ag(111) terraces separated by an atomic step [16]. Scan area is 2 nm × 2 nm. (d) NC-AFM image of the atomically flat surface of Ag(111) [16]. Cross-sectional line profile obtained along the white line of the NC-AFM image clearly shows lattice spacing of 0.28 ± 0.01 nm of Ag(111) surface.
2.1. Vertical resolution of home-built NC-AFM Fig. 7a and b show a model of Ag(111) film evaporated on Si(111)7×7 surface and a wide-area NC-AFM image, respectively [16]. The cross-sectional line profile of Fig. 7b obtained along the white broken line shows two terraces with height difference of an atomic layer of Ag(111) surface. Fig. 7c shows a high-resolution NC-AFM image with two Ag(111) terraces separated by an atomic step [16]. The trigonal pattern that corresponds to the Ag(111) lattice can be clearly observed on the upper terrace. However, the image obtained near the atomic step shows poor resolution. This may
438
S. Morita and Y. Sugawara
Fig. 8: NC-AFM images and cross-sectional line profiles of (a) Si(100)2×1 clean surface and (b) hydrogen passivated Si(100)2×1:H monohydride surface [17]. Cross-sectional line profiles are obtained along the white dotted lines of NC-AFM images. Scan areas are 6.9 nm × 4.6 nm.
mean that Ag atoms near the atomic step move around even at room temperature (RT). Fig. 7d shows an NC-AFM image of the atomically flat surface of Ag(111) [16]. The cross-sectional line profile of Fig. 7d obtained along the white line clearly shows lattice spacing of 0.28 ± 0.01 nm of Ag(111) surface. Thus we can observe even metal atoms using atomic force, in spite of the nearly homogeneous and hence site-independent surface on atomic force due to free-electron screening. It should be noted that the cross-sectional line-profile of Fig. 7d shows nearly 10 pm (= 0.01 nm) corrugation. Therefore, our home-built NC-AFM has nearly 1 pm vertical resolution. 2.2. Lateral resolution of home-built NC-AFM Here we measured the NC-AFM images of a Si(100)2×1 clean surface and a Si(100)2×1:H monohydride surface. Si(100)2×1 surface has tilted dangling bonds and hence is reactive, while Si(100)2×1:H surface is terminated by hydrogen atoms and hence inactive. Furthermore, these surfaces have nearly the same structure. Therefore, the tilted dangling bond effect on the NC-AFM imaging of Si(100)2×1 surface and the hydrogen termination effect on the NC-AFM imaging of Si(100)2×1:H surface can be clarified by comparing both NC-AFM images with the same Si tip. Fig. 8a shows the atomic resolution image of Si(100)2×1 surface measured at ∆ν = −30 Hz [17]. Paired bright spots (imaged dimer) constituting rows with a 2×1 symmetry were clearly observed. Further, the distance between paired bright spots is 0.32 ± 0.01 nm. As shown in the cross-sectional line profile of Fig. 8a, the change of the frequency shift along a white dotted line in the NC-AFM image of Fig. 8a was estimated
Characterization of semiconductor surfaces with noncontact atomic force microscopy
439
to be about 9 Hz. Fig. 8b shows the atomic resolution image of hydrogen terminated silicon [Si(100)2×1:H monohydride] surface measured at ∆ν = −11 Hz. Paired bright spots (imaged dimer) constituting rows with a 2×1 symmetry were observed. Further, the distance between paired bright spots is 0.35 ± 0.01 nm, which approximately agrees with the distance between hydrogen atoms on monohydride surface, i.e., 0.352 nm. As shown in the cross-sectional line profile of Fig. 8b, the averaged change of the frequency shift along a white dotted line in the NC-AFM image of Fig. 8b was estimated to be ca. 3 Hz. Thus our home-built NC-AFM can measure the distances between paired bright spots with an accuracy of ±0.01 nm (= 10 pm). This result roughly shows that our home-built NC-AFM has nearly 10 pm lateral resolution. From such precise measurements of the distances between paired bright spots, we can discuss the origin of the imaged dimers in NC-AFM images on Si(100)2×1 surface (Fig. 8a) and Si(100)2×1:H surface (Fig. 8b) by comparing with corresponding surface structure models. In the case of Si(100)2×1 surface, we found that the bright spots of Fig. 8a do not image the silicon atom site, because the distance 0.32 ± 0.01 nm between bright spots forming the dimer structure is larger than the distance between paired silicon atoms of a dimer structure model (maximum distance between alternating upper Si atoms on an asymmetric dimer structure is 0.292 nm). This result suggests that chemical bonding interaction strongly works between the tilted dangling bond out of the silicon dimer and the dangling bond out of silicon tip apex. As a result, a dimer structure with a larger distance than that between paired silicon atoms of a dimer is obtained. Besides, as shown in the cross-sectional line profile of Fig. 8a, the frequency shift rapidly increases around imaged dimers. This also suggests that a strong chemical interaction occurs between dangling bonds. On the other hand, the bright spots of Fig. 8b seem to be located at hydrogen atom sites on Si(100)2×1:H surface, because the distance between bright spots 0.35 ± 0.01 nm forming a dimer structure approximately agrees with the distance between the paired hydrogen atoms, i.e., 0.352 nm. As a result, each bright spot of the NC-AFM image corresponds to the individual hydrogen atom site on the top most layer. On Si(100)2×1:H surface, the dangling bond on the silicon surface is terminated by a hydrogen atom, and the hydrogen atoms on the top most layer do not have a strong chemical reactivity as a silicon atom on Si(100)−2×1 surface. Therefore, the interaction between the hydrogen atom on the top most layer and the silicon tip apex may be weak chemical or physical interaction without strong directional dependence, in contrast to the Si(100)2×1 surface. 2.3. Atomic displacement around missing-dimer defects in Si(100)2×1 surface Fig. 9 presents direct NC-AFM observations of the atomic displacements of Si dimers around a missing-dimer defect in a Si(100)2×1 surface. As shown in Fig. 9 indicated by the white arrow, the NC-AFM clearly images the missing-dimer defect. The thin broken lines indicate the centerlines of the dimers. By careful investigation, we found that the lateral position of the dimer is displaced toward the missing-dimer defect along the dimer row. In general, the dimers adjacent to missing-dimer defects exhibit subtle variations in height, length and lateral position depending on the size and structure
440
S. Morita and Y. Sugawara
Fig. 9: NC-AFM image of Si(100)2×1 surface. Thin broken lines indicate centerlines of dimers along the dimer bonds. White arrow shows a missing-dimer defect.
Fig. 10: Averaged lateral displacement of a dimer along a dimer row in Si(100)2×1 and Si(100)2×1:H surfaces as a function of distance from a defect. Noise level is ±0.01 nm. The regular distance between adjacent dimers (lattice constant) is 0.384 nm.
of the defect. Some of these varying properties, such as the height, are related to the scanning direction of the NC-AFM, and the variation can be explained as overshoot due to feedback error. However, most of the dimers adjacent to missing-dimer defects exhibited nearly identical lateral displacement along the dimer row, irrespective of the scanning direction. As shown in Fig. 10, the first and second dimers adjacent to the missing-dimer defects in Si(100)2×1 surface are displaced. The figure shows the average lateral displacement of dimers with respect to the regular distance between adjacent dimers of 0.384 nm, which corresponds to the lattice constant along the dimer
Characterization of semiconductor surfaces with noncontact atomic force microscopy
441
Fig. 11: Schematic models of atomic structure around a missing-dimer defect (a) on a Si(100)2×1 clean surface and (b) on a Si(100)2×1:H monohydride surface.
row. The first neighbouring dimer is displaced by ca. 20% of the lattice constant, and the second neighbouring dimer is displaced by ca. 10%. The third and fourth dimers are displaced only within the experimental error (±0.01 nm). In contrast, as shown in Fig. 10, the dimers adjacent to missing-dimer defects in the Si(100)2×1:H monohydride surface are not displaced. The corresponding models of the plan structures of the Si(100)2×1 and Si(100):H surfaces with a single missing-dimer defect are shown in Fig. 11a and b. In Si(100)2×1, a single missing-dimer defect results in 4 dangling bonds in the second Si layer, cutting 4 covalent bonds between 2 Si atoms to form one Si dimer in the first layer and leaving 4 Si atoms in the second layer. These 4 dangling bonds then form 2 Si dimers in the second layer to reduce the number of dangling bonds, yielding the rebonded structure for a single missing-dimer defect ((1) in Fig. 11a) [18]. The bonds of these second-layer (lower) dimers are orientated normal to those of the first-layer (upper) dimers ((2) in Fig. 11a), parallel to the dimer row of the first layer. The stress created by the attractive force between paired Si atoms in the lower dimers pulls the upper dimers toward the missing-dimer defect. If the length of the lower dimer is the same as that of the upper dimers (asymmetric, 0.230 nm; symmetric, 0.234 nm), the first neighbouring upper dimer will be displaced by approximately 0.077 nm (asymmetric) or 0.075 nm (symmetric) toward the missing-dimer defect ((3) in Fig. 11a). This value is consistent with the experimentally observed value of 0.08 nm for the first adjacent dimer. The proposed model of the rebonded dimer structure for a single missing-dimer defect agrees well with the previous results obtained by STM at low negative sample bias and by local density approximation (LDA) computer modeling [18]. Although the LDA model and STM image calculated for high negative sample bias clearly predicted the lateral displacement of the first neighbouring Si dimer toward the missing-dimer defect in agreement with the present NC-AFM results, the practical spatial resolution of STM was insufficient to verify the predictions [18]. Further LDA simulation taking the lateral displacement of the second neighbouring dimer into consideration will be required in order to explain the lateral displacement of the second neighbouring dimer
442
S. Morita and Y. Sugawara
((4) in Fig. 3(a)) in terms of atomic stress and atomic relaxation around the dimer. In the case of the Si(100)2×1:H monohydride surface, 4 dangling bonds in the second Si layer left by the missing-dimer defect are each terminated by hydrogen atoms (Fig. 11b). This termination appears to eliminate the stress that would otherwise result in displacement of the Si dimers around the missing dimer. This is the first NC-AFM measurement that has provided a direct visualization of the atomic displacement around missing-dimer defects. Our home-built NC-AFM has been shown to provide resolution sufficient to directly observe distortions in crystal lattices, and was used successfully in the present study to show that the atomic stress and strain induced by the missing-dimer defects extends up to two dimers from the defect, with direct observations of the displacement of the first and second dimers adjacent to a missing-dimer defect. NC-AFM is therefore a highly useful method for making direct measurements of atomic displacements and strain with subatomic resolution, allowing the atomic structure around various kinds of defects to be investigated in detail.
3. Functions of NC-AFM The NC-AFM is not only a mere microscope but has various kinds of functions. As shown in Fig. 12a, a sample surface atom B exerts atomic force on a tip apex atom A. Then by changing the relative position (x, y, z) of the tip apex atom A to the sample surface atom B, the NC-AFM can measure atomic force between the tip apex atom A and the sample surface atom B three-dimensionally. From three-dimensional mapping of atomic force, we can deduce the strength, distance dependence and angular dependence of atomic force between A and B. Then we can investigate the mechanism of the atomic force. If there are two kinds of atoms B and C on the sample surface, as shown in Fig. 12b, we can discriminate atom species B and C from three-dimensional mapping of atomic force, which clarifies the difference of the strength, distance dependence and angular dependence of atomic forces between A and B, and A and C. Besides, by placing a suitable atom D at the tip apex instead of the atom A, we can control atomic force between the tip apex atom and the surface atom B. Furthermore, by moving the tip apex toward the sample surface, we can increase the strength of the attractive force and we can pull up the surface atom. In principle, using such a strong attractive force or moderately weak repulsive force, we can mechanically manipulate individual atoms/molecules and can assemble novel materials and devices from individual atoms/molecules. 3.1. Confirmed functions of high performance NC-AFM Table 2 shows functions of high performance NC-AFM and samples measured by our group. Existence of measured samples means confirmation of corresponding functions. Table 2 shows that our home-built NC-AFM can observe compound semiconductor surfaces such as InP(110) [10,11] and GaAs(110) [15] cleaved surfaces, silicon surfaces such as Si(111)7×7 [19] and Si(100)2×1 [17] clean surfaces, metal surfaces such as Ag(111) [16] and Cu(111) [20], metal oxide surfaces such as TiO2 [21] with atomic resolution. Our home-built NC-AFM can observe molecules such as adenine [22] and
Characterization of semiconductor surfaces with noncontact atomic force microscopy
443
Fig. 12: Models of functions of the NC-AFM. (a) Three-dimensional mapping of atomic force between a tip apex atom A and a sample surface atom B. (b) Discriminations of force mechanisms between A and B, and A and C from three-dimensional mapping of atomic force, which enables us to discriminate atoms B and C. (c) Control of atomic force between tip apex atom and surface atom B by placing a suitable atom D on the tip apex instead of atom A.
thymine [23], self-assembled monolayers such as alkanethiol [23], and DNA with high resolution [24]. 3.2. Three-dimensional mapping of atomic force As shown in Table 2, our NC-AFM can map atomic force three-dimensionally. There are two methods for three-dimensional mapping of atomic force as shown in Fig. 13. One method measures the tip–sample distance dependence of the NC-AFM images such as A, B and C as shown in Fig. 13a. This kind of three-dimensional (3D) mapping √ √ of atomic force on an atomic scale was achieved on Si(111) 3 × 3–Ag surface [25] and on oxygen adsorbed Si(111)7×7 surface. The other method measures the sitedependence of frequency shift curves such as A, B and C as shown in Fig. 13b. This kind of three-dimensional mapping of atomic force on an atomic scale was achieved on Si(111)7×7 surface [26]. Fig. 14 shows one example of three-dimensional mapping of atomic force obtained by measuring the tip–sample distance dependence of the NC-AFM images on oxygen adsorbed Si(111)7×7 surface. At relatively far distance, only several bright spots indicated by white open circles can be seen clearly, as shown in Fig. 14a. It should be noted that at this distance the 7×7 structure cannot be observed clearly. On the
444
S. Morita and Y. Sugawara
Table 2 Functions of high performance NC-AFM Measured samples by NC-AFM (Osaka University) Observation (atom/molecule)
InP(110) Si(111)7×7 GaAs(110) Si(100)2×1
Ag(111) Cu(111)*
TiO2 * Adenine* Alkanethiol* Si(100)2×1:H Thymine* DNA*
3D-Measurement Si(111) (atomic force) 7×7
Si(111) √ √ 3 × 3–Ag
Discrimination (atomic force)
Si(111) Si(111) oxygen √ √ √ √ 3 × 3–Ag 5 3 × 5 3–Sb adsorbed Si(111)7×7
Si(111) 7×7
Discrimination (atom) Control (atomic force)
oxidation of Si tip
Ag adsorbed Si tip
oxygen adsorbed Si(111)7×7
Si(111) oxygen √ √ 5 3 × 5 3–Sb adsorbed Si(111)7×7
Si(111) 7×7–Ge
Sb adsorbed Si tip
Si(100) 1×1:2H
Si(100) 1×1:2D
Si(100) 1×1:2H
Si(100) 1×1:2D
Control (atom position) Manipulation (atom)
Si(111) 7×7
Si(100) 2×1–Ge
Si(111) Si(111) √ √ √ √ 3 × 3–Ag 5 3 × 5 3–Sb
Assembly (atom by atom) Point charge imaging
n+ -GaAs (110)
CPD and dC/dz image
Si(111) 7×7
Si(111) √ √ 5 3 × 5 3–Sb
* = Collaboration with JRCAT.
other hand, at near distance, the 7×7 structure can be seen clearly, as shown in Fig. 14b. These experimental results directly show that three-dimensional mapping of atomic force shows clear difference in distance dependence of atomic forces exerted from oxygen and Si atoms. 3.3. Discrimination of atomic force mechanisms and atom species From three-dimensional mapping of atomic force such as Fig. 14a and b, we can conclude that two force mechanisms work between the Si tip and oxygen adsorbed Si(111)7×7 surface. One is a kind of long-range force, which works well even at relatively far distance. The other is a kind of short-range force, which works strongly only at near distance. The short-range force contributes to NC-AFM imaging of the 7×7 structure, that is, Si adatoms, as shown in Fig. 14b. Hence the short-range force is the covalent bonding force between dangling bonds of Si adatoms and the Si tip apex atom, and has a critical distance dc , below which the covalent bonding force works strongly, as shown in Fig. 15a and b. On the other hand, at relatively far distance,
Characterization of semiconductor surfaces with noncontact atomic force microscopy
445
Fig. 13: Two methods for three-dimensional mapping of atomic force. (a) 3D-mapping of atomic force using vertical mapping (tip–sample distance dependence) of the NC-AFM image such as A, B and C. (b) 3D-mapping of atomic force using lateral mapping of site-dependent frequency shift curve such as A, B and C.
different atoms from Si adatoms seem to be imaged by the long-range force. Oxygen adsorbed Si(111)7×7 surface has only two kinds of atoms, that is, Si atoms and oxygen atoms. Therefore, we conjectured that, at relatively far distance, oxygen atoms were imaged by the long-range force. Oxygen atoms will become negatively charged ions, because silicon substrate is n-type. Therefore, the origin of the long-range force may be electrostatic force between O− ion and Si tip. Thus using three-dimensional mapping of atomic force, we can discriminate mechanisms of atomic forces between the tip and the sample surface, and also we can discriminate atom species on the sample surface. We will show another example of discrimination of atom species. We fabricated Si(111)7×7–Ge mixed surface as shown in Fig. 16. On this surface, Si adatoms and Ge adatoms coexist. As a result, using the clean Si tip, covalent bonding force works between the Si tip apex atom and adatoms of Si and Ge. We found two kinds of bright spots forming a 7×7 structure, as shown in Fig. 16. One kind of bright spots is brighter than the other kind of bright spots. These two kinds of bright spots seem to correspond
446
S. Morita and Y. Sugawara
Fig. 14: NC-AFM images of oxygen adsorbed Si(111)7×7 at (a) relatively far distance and at (b) near distance. Scan areas are 15 nm × 15 nm.
Fig. 15: Models of NC-AFM imaging of oxygen adsorbed Si(111)7×7 at (a) relatively far distance and at (b) near distance.
to Si adatoms and Ge adatoms. Then, by increasing Ge atoms from 0.06 ML, 0.29 ML and 0.52 ML, we found that the number of brighter spots decreases, as shown in Fig. 16a, b and c. Hence we attributed the brighter spots to Si adatoms and the dim spots to Ge adatoms. Covalent bond energies for Si–Si bond and Si–Ge bond are 2.32 eV and 2.12 eV, respectively. Therefore, the covalent bonding force for Si adatoms will be larger than that of Ge adatoms. This speculation qualitatively agrees with the experimental results. Thus, our home-built NC-AFM can clearly discriminate Si adatoms and Ge
Characterization of semiconductor surfaces with noncontact atomic force microscopy
447
Fig. 16: NC-AFM images of Si(111)7×7–Ge mixed surfaces obtained for Ge depositions of (a) 0.06 ML, (b) 0.29 ML and (c) 0.52 ML.
adatoms, although the difference of covalent bond energies for Si–Si bond and Si–Ge bond is only ca. 10%. We also succeeded to discriminate the difference of strength of the covalent bonding force for Si√adatoms √ with a dangling bond and for Sb adatoms with a lone pair on Si(111)5 3 × 5 3–Sb surface using a clean Si tip with a dangling bond. Such difference of strengths of the covalent bonding force depends on the bond order and enables us to discriminate Si adatoms and Sb adatoms. 3.4. Control of atomic force and atom position By placing a suitable atom at the tip apex, we can control atomic force and force mechanism between the tip apex atom and the surface atom. This kind of atomic force control was demonstrated on Si(111)7×7 surface by using an oxidized Si tip instead of √ √ 3√ × 3–Ag surface by picking up an Ag atom on the clean Si tip [19,26], on Si(111) √ the Si tip [27], and on Si(111)5 3 × 5 3–Sb surface by picking up an Sb atom on the Si tip. Furthermore, by moving the tip apex toward the sample surface, we can increase the strength of the attractive force and we can pull up the surface atom. This is another kind of atomic force control, which also enables us to control atom position. This kind of atomic force control was demonstrated on Si(100)1×1:2H dihydride surface [28], and on Si(100)1×1:2D dideuteride surface. In case of Si(100)1×1:2H surface [or Si(100)1×1:2D surface], a repulsive force works between adjacent hydrogen [or deuterium] atoms due to the proximity effect and originates a 1×1 canted structure whihc increases the distance between adjacent hydrogen atoms, i.e., reduces the repulsive force strength. Fig. 17a, b and c show NC-AFM images of Si(100)1×1:2D surface obtained using a clean Si tip and decreasing the tip–sample distance from (a) 0.12 nm, through (b) 0.09 nm, to (c) 0.07 nm before the contact point. At the
448
S. Morita and Y. Sugawara
Fig. 17: Tip–sample distance dependence of the NC-AFM image on Si(100)1×1:2D obtained using a clean Si tip and decreasing the tip–sample distance from (a) 0.12 nm, through (b) 0.09 nm, to (c) 0.07 nm.
relatively far distance of Fig. 17a, the NC-AFM showed a 1×1 pattern. But at the near distance of Fig. 17b, the NC-AFM showed a 2×1 pattern. Moreover, at the close distance of Fig. 17c, the NC-AFM again showed a 1×1 pattern. Thus, the NC-AFM on Si(100)1×1:2D surface showed a pattern change from 1×1, through 2×1, to 1×1 similar to Si(100)1×1:2H surface [28] by decreasing clean Si tip–sample distance. To explain such tip–sample distance dependence of NC-AFM image, we considered the effect of attractive force between the tip apex atom and the nearest deuterium atom on the surface structure. At the relatively far distance, the magnitude of the attractive force between the tip apex atom and the nearest deuterium atom is smaller than that of the repulsive force between adjacent deuterium atoms. In this case, the original 1×1 canted structure is stable even under NC-AFM measurement. Then NC-AFM will observe the original 1×1 canted structure at the relatively far distance as shown in Fig. 17a. However, at the near distance, the magnitude of the attractive force becomes nearly equivalent to that of the repulsive force. As a result, the attractive force will pull up the nearest deuterium atom, and will play the role of trigger to induce self-organized 2×1 novel structure. Thus NC-AFM will observe the tip-induced self-organized 2×1 novel structure as shown in Fig. 17b. Further, at the close distance, the magnitude of the attractive force becomes larger than that of the repulsive force. As a result, the attractive force will pull up the nearest deuterium atoms one by one. This vertical motion of each deuterium atom due to pulling up seems large enough compared with the corrugation of the self-organized 2×1 novel structure, and hence tip-induced 1×1 novel structure will be observed by the NC-AFM. Thus NC-AFM can control atomic force and atom position by moving the tip apex toward the sample surface atom. 3.5. Electrostatic force imaging of local charges To simultaneously image the topography due to the van der Waals force and the electrostatic force using the noncontact AFM, at first we eliminated the contribution due to the electrostatic force by applying a sample voltage which compensates the
Characterization of semiconductor surfaces with noncontact atomic force microscopy
449
averaged contact potential difference (CPD) between the Si tip and n+ -GaAs(110) cleaved surface. Therefore, under this sample voltage, the van der Waals force will be the dominant force mechanism, because the contribution due to the electrostatic force will be eliminated. Then, to image the noncontact AFM topography due to the van der Waals force, we closed the feedback loop to maintain a constant frequency shift under this sample voltage and imaged the feedback voltage by scanning with the tip. On the other hand, by applying a large positive sample voltage or a large negative sample voltage, we enhanced the electrostatic force between tip and sample surface. For the electrostatic force imaging, we opened the feedback loop under a large positive sample voltage or a large negative sample voltage, and imaged the change of the frequency shift between two alternate sample voltages by scanning with the tip. This method enables us to simultaneously image the atomic corrugation and the charge of atomic point defects on n+ -GaAs(110) cleaved surface [29]. It made clear that there are atomic point defects with and without point charge, and that the sign of the point charge is positive. Further, electron screening clouds were observed [15]. 3.6. Atomic imaging of contact potential difference and capacitance An electrostatic force microscope (EFM) combined with NC-AFM (EFM/NC-AFM) achieved charge imaging on an atomic scale. However, the contact potential difference (CPD) will be locally different at the site of the point charge. This effect will change the NC-AFM topography, and hence change the magnitude of the electrostatic force. Therefore, we should compensate the contact potential difference with atomic resolution. This means that we should combine Kelvin probe force microscopy (KFM) with NC-AFM (KFM/NC-AFM) and that we should simultaneously observe the NCAFM topography and the CPD image of the point charge. We succeeded in simultaneous imaging of the NC-AFM topography, the CPD image and the z-derivative image of capacitance (dC/dz) of Si(111)7×7 sample surface [30]. Both CPD image and z-derivative image of capacitance as well as NC-AFM topography clearly showed a Si(111)7×7 pattern. Thus charge distribution and capacitance change are important even on an atomic scale. KFM measurement compensates contact potential difference and hence electrostatic force on an atomic scale, while it originates excess 2 /4 in the DC term and a 2ω component in the frequency shift. Here, terms of (dC/dz)VAC VAC is the modulation amplitude of the alternating bias potential VAC sin ωt between the tip and sample. Therefore, it should be noted that the NC-AFM topography using this 2 /4 which should be subtracted. method still includes an excess term of (dC/dz)VAC Using√ KFM/NC-AFM, we also succeeded in atomically resolved imaging of CPD on √ Si(111)5 3 × 5 3–Sb surface. The CPD image clearly showed the difference of CPD of Si adatoms and Sb adatoms on an atomic scale, and enabled us to discriminate Si adatoms and Sb adatoms. 3.7. Atom manipulation using NC-AFM based on a mechanical method STM is the first generation of the atom manipulation tool. However, STM is a microscope based on the electric method that measures the tunneling current between the tip
450
S. Morita and Y. Sugawara
apex atom and surface atom. Therefore the STM has inherent limitations such as STM cannot observe an insulator surface, cannot manipulate atoms and molecules on insulator surfaces, and cannot directly measure atomic force or force related information. On the other hand, AFM is the microscope based on the mechanical method that measures the atomic force or force related information between the tip apex atom and the surface atom, and has potentials for insulator surface observation with atomic resolution, manipulation of atoms and molecules on insulator surface, and measurement of atomic force or force related information on an atomic scale. 3.7.1. Vertical manipulation based on a mechanical method NC-AFM succeeded in a kind of atom manipulation √ at room temperature such as √ 3 × 3–Ag sample surface [27], and extraction of a single Ag atom from Si(111) √ √ extraction of Sb atoms from Si(111)5 3 × 5 3–Sb sample surface. We also tried extraction of single Si atom from Si(111)7×7 using low-temperature NC-AFM (LTNC-AFM), as shown in Fig. 18. After NC-AFM imaging, such as in Fig. 18a, we mechanically moved the Si(111)7×7 sample surface toward the tip apex Si atom up to some distance z. Then by retracting the sample surface down to the initial distance, we again imaged the sample surface using NC-AFM. When we cannot observe signs of atom extraction, we again moved the sample surface toward the tip apex Si atom further. By repeating such procedure, we succeeded to extract a single Si adatom on the Si(111)7×7 surface mechanically. Fig. 18a and b are NC-AFM images before extraction and after extraction, respectively. White open circles show a missing Si adatom, which is a marker of the site. By comparing Fig. 18b with Fig. 18a, it is clear that a corner Si adatom indicated by white arrows was mechanically extracted. In the case of Ag extraction, a tip apex Si atom with a dangling √ bond √ will cut an Ag– Si covalent bond and will pick up the Ag atom from the Si(111) 3 × 3–Ag sample surface. This is a kind of rearrangement of the Ag–Si covalent bond and the number
Fig. 18: NC-AFM images of Si(111)7×7 at 9.3 K (a) before Si adatom extraction and (b) after Si adatom extraction.
Characterization of semiconductor surfaces with noncontact atomic force microscopy
451
Fig. 19: Models of Si adatom extraction. By (a) moving the Si(111)7×7 surface toward the Si tip, (b) at first an attractive force works between Si tip apex atom and Si adatom, (c) then a repulsive force works and cuts three covalent bonds, and (d) finally a repulsive force pushes out an Si adatom from the Si(111)7×7 surface.
of the dangling bonds does not change. On the other hand, in the case of Si adatom extraction shown in Fig. 19, the tip apex Si atom with a dangling bond will cut the three Si–Si covalent bonds between the Si adatom and the lower three Si atoms. However, the extraction process such as in Fig. 19 will increase the number of dangling bonds. Thus the Si extraction process also seems to be complicated. One possible explanation is that cutting of three Si–Si covalent bonds happened under the repulsive force as shown in Fig. 19c. This means that the Si adatom was not extracted due to the pull-up process indicated in Fig. 19b, but due to the push-out process as indicated in Fig. 19c and d. In this process, mechanical energy due to atomic indentation under weak repulsive force will increase the atom potential, and may finally cut the three Si–Si covalent bonds and push out a single Si adatom. Then, the pushed out Si atom may adsorb at the top of the tip apex, the side of the tip or on the sample surface. 3.7.2. Lateral manipulation based on a mechanical method Using NC-AFM based on the mechanical method, we sequentially extracted Si adatoms on a Si(111)7×7 sample surface up to 5 Si adatoms. During this procedure, we accidentally succeeded to laterally manipulate Si adatoms. Fig. 20a and b are NC-AFM images before lateral manipulation and after lateral manipulation, respectively. White open circles indicate a manipulated Si adatom. By comparing Fig. 20b with a, it is clear that a Si adatom, indicated by white open circles, was laterally manipulated by the mechanical method. This lateral manipulation may occur if two covalent bonds between the manipulated Si adatom and lower two Si atoms were cut, and then this Si adatom formed two covalent bonds with two different lower Si atoms with dangling bonds.
452
S. Morita and Y. Sugawara
Fig. 20: NC-AFM images of Si(111)7×7 at 8.6 K and models of 7×7 structure (a) before lateral manipulation of a Si adatom and (b) after lateral manipulation of a Si adatom. Scan areas are 9.2 nm × 9.2 nm.
4. Summary Our home-built NC-AFM roughly has a vertical resolution of 1 pm and a lateral resolution of 10 pm. Using this NC-AFM, we can observe metal atoms and can measure atomic forces three-dimensionally on an atomic scale. Besides, we can discriminate mechanisms of atomic force and species of atoms. Further, we can control atomic force and atom position. Moreover, recently, we succeeded to manipulate a single atom by mechanical contact. In this process, mechanical energy due to atomic indentation under weak repulsive force will increase the atom potential and may finally push out a single atom. We should explore near contact and close contact phenomena in more detail. Then this will enable us to assembly atom by atom, which may lead to construction of novel nanomaterials and nanodevices.
References 1. 2. 3. 4. 5. 6. 7. 8.
G. Binnig, C.F. Quate, and Ch. Gerber, Phys. Rev. Lett. 56, 930 (1986). G. Binnig, Ch. Gerber, E. Stoll, T.R. Albrecht, and C.F.Quate, Europhys. Lett. 3, 1281 (1987). J.B. Pethica and W.C. Oliver, Phys. Scri. T 19, 61 (1987). G. Binnig, Ultramicroscopy 42–44, 7 (1992). F.J. Giessibl and G. Binnig, Ultramicroscopy 42–44, 281 (1992). F. Ohnesorge and G. Binnig, Science 260, 1451 (1993). F.J. Giessibl, Science 267, 68 (1995). S. Kitamura and M. Iwatsuki, Jpn. J. Appl. Phys. 34, L145 (1995).
Characterization of semiconductor surfaces with noncontact atomic force microscopy 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30.
453
T.R. Albrecht, P. Grütter, D. Horne, and D.Rugar, J. Appl. Phys. 69, 668 (1991). H. Ueyama, M. Ohta, Y. Sugawara, and S. Morita, Jpn. J. Appl. Phys. 34, L1086 (1995). Y. Sugawara, M. Ohta, H. Ueyama, and S. Morita, Science 270, 1646 (1995). M. Bammerlin, R. Lüthi, E. Meyer, A. Baratoff, J. Lü, M. Guggisberg, Ch. Gerber, L. Howald, and H.-J. Güntherodt, Probe Microsc. 1, 3 (1997). S. Morita and Y. Sugawara, Appl. Surf. Sci. 140, 406 (1999). H. Ueyama, Y. Sugawara, and S. Morita, Appl. Phys. A 66, S295 (1998). S. Morita, M. Abe, K. Yokoyama, and Y. Sugawara, J. Cryst. Growth 210, 408 (2000). S. Orisaka, T. Minobe, T. Uchihashi, Y. Sugawara, and S.Morita, Appl. Surf. Sci. 140, 243 (1999). K. Yokoyama, T. Ochi, A. Yoshimoto, Y. Sugawara, and S. Morita, Jpn. J. Appl. Phys. 39, L113 (2000). J.H.G. Owen, D.R. Bowler, C.M. Goringe, K. Miki, and G.A.D. Briggs, Surf. Sci. 341, L1042 (1995). T. Uchihashi, Y. Sugawara, T. Tsukamoto, M. Ohta, and S. Morita, Phys. Rev. B 56, 9834 (1997). K. Yokoyama, T. Ochi, T. Uchihashi, M. Ashino, Y. Sugawara, N. Suehira, and S. Morita, Rev. Sci. Instrum. 71, 128 (2000). M. Ashino, T. Uchihashi, K. Yokoyama, Y. Sugawara, S. Morita, and M.Ishikawa, Appl. Surf. Sci. 157, 212 (2000). T. Uchihashi, T. Okada, Y. Sugawara, K. Yokoyama, and S. Morita, Phys. Rev. B 60, 8309 (1999). T. Uchihashi, T. Ishida, M. Komiyama, M. Ashino, Y. Sugawara, W. Mizutani, K. Yokoyama, S. Morita, H. Tokumoto, and M. Ishikawa, Appl. Surf. Sci. 157, 244 (2000). T. Uchihashi, M. Tanigawa, M. Ashino, Y. Sugawara, K. Yokoyama, S. Morita, and M. Ishikawa, Langmuir 16, 1349 (2000). T. Minobe, T. Uchihashi, T. Tsukamoto, S. Orisaka, Y. Sugawara, and S. Morita, Appl. Surf. Sci. 140, 298 (1999). S. Morita, Y. Sugawara, K. Yokoyama, and T. Uchihashi, Nanotechnology 11, 120 (2000). K. Yokoyama, T. Ochi, Y. Sugawara, and S. Morita, Phys. Rev. Lett. 83, 5023 (1999). S. Araragi, A. Yoshimoto, N. Nakata, Y. Sugawara, and S. Morita, Appl. Surf. Sci. 188, 272 (2002). Y. Sugawara, T. Uchihashi, M. Abe, and S. Morita, Appl. Surf. Sci. 140, 371 (1999). K. Okamoto, Y. Sugawara, and S. Morita, Appl. Surf. Sci. 188, 381 (2002).
Nanotechnology and Nano-Interface Controlled Electronic Devices Editors: M. Iwamoto, K. Kaneto and S. Mashiko © 2003 Published by Elsevier Science B.V.
CHAPTER 22
Transport and photocarrier generation in poly(3-alkylthiophene) and metal junctions Keiichi Kaneto, Koichi Rikitake, and Wataru Takashima Graduate School of Life Science and Systems Engineering, Kyushu Institute of Technology, Iizuka, Fukuoka 820-8502, Japan E-mail:
[email protected],
[email protected]
1. 2. 3. 4.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sample preparation and experimental setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Optical absorption in HT- and NR-PATn . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . J –V Characteristics of Al/PATn/Au diodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1. Rectification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2. Depletion layer in the Schottky junction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3. New estimation method of the thickness of the depletion layer . . . . . . . . . . . . . 4.4. Potential profile of diodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5. Carrier mobilities in PATn . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1. TOF mobilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2. Negative field dependence of photocarrier mobility due to non-uniform fields 5.3. FE mobilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6. Photovoltaic effects and mechanism of carrier generation . . . . . . . . . . . . . . . . . . . . . . . . . . 7. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
455 458 460 461 461 462 463 464 466 466 468 471 474 476 477 477
1. Introduction Since the discovery of conducting polymers such as polyacetylene [1,2], polypyrrole [3] and polythiophene [4,5], the concept of organic materials and polymers has been turned into active elements from passive materials in electronic devices. The conducting polymers were novel materials until they became the Nobel polymers in 2000. They demonstrate a variety of functions such as semiconductors, metals and even switching property between semiconductor and metal upon oxidation and reduction [6]. The unique features, which originate from the one dimensional π-electron systems along the polymer chain, are the high anisotropy in carrier transport and the excellent photore-
456
K. Kaneto et al.
Fig. 1: Variety of conducting polymers and functional polymers.
sponse [7]. The metallic properties due to the delocalized π-electrons in the oxidized state can be utilized as a lightweight electrical conductor, antielectrostatic shield and capacitor. The semiconductive property due to the localized π-electrons in the neutral state is of noting from the view point of high potential for use in diodes with efficient light emission [8], in solar cells [9] and also thin film transistors. Other interesting features of organic and polymer devices are flexibility of the laminate structure and low cost of the fabrication, which cannot be achieved by conventional semiconductors. The varieties of conducting polymers as shown in Fig. 1 are categorized for their
Transport and photocarrier generation in poly(3-alkylthiophene) and metal junctions
457
Fig. 2: Schematic structures HT (head to tail) and NR (non-regiocontrolled) PATn.
stability into two groups, which are (1) stable at the neutral state and (2) those that oxidize or decompose easily at ambient conditions. The former and its derivatives are poly(alkylthiophene) (PAT), emeraldine at the neutral form of polyanilines (PANI), and poly-paraphenylene vinylenes (PPV). The latter is polyacetylene (PA), polypyrrole (PPy) and leucoemeraldine at the reduced state of PANI. Most of the conducting polymers are insoluble in common organic solvents, therefore it is very difficult to purify them by recrystallization and process them into desired forms. However, substitution of long alkyl chains or alkoxy groups to the ring increases the solubility in organic solvents such as chloroform [10], dichloromethane, etc. Polythiophene and its derivatives are studied intensively for active layers of photoand electronic devices [7,10–12], since they are stable against oxidation in ambient conditions. The substitution at 3 position of the thiophene ring introduces regioregularity, because of the possibility for coupling positions of head to tail (HT), head to head and tail to tail. When the coupling position is precisely controlled at the HT position, the regioregularity is enhanced as shown in Fig. 2. It has been shown that the HT-PATs exhibit larger carrier mobility compared to those of NR-PATns by one or two orders of magnitude [11,12]. The photovoltaic effect, which is the reversed phenomenon to light emitting, is another important function in optical and electrical conversion devices. The photovoltaic effect results from photocarrier generation and migration toward the respective electrode. In inorganic semiconductors, the mechanism of photocarrier generation is well accounted for by the internal built-in field at the Schottky junction or pn-junction, which is based on the band theory and frame works of the difference of the work functions between the contacting materials. On the other hand, in organic materials
458
K. Kaneto et al.
the band theory is not fully applicable. This is due to the fact that organic materials are discontinuously assembled with inert and isolated molecules and condensed with the weak Van der Waals force. Namely, the carrier transport in organic materials is governed by intermolecular hopping or conduction in a narrow band. Furthermore, photocarrier generation also is not favorable in organic materials. However, it is interesting to note that quite high intrinsic conversion efficiencies as high as 25% in organic solar cells are demonstrated. For instance, the dye sensitized solar cell or Grätzel type cell shows an overall efficiency of close to 10% and the solid state of a poly( p-phenylene vinylenes)/fullerene system is exceeding 3.2%. Taking the purity and crystallinity of organic materials in comparison with that of inorganic materials into consideration, the intrinsic efficiency of organic photocells should be quite large. The potential of organic materials for electronic devices is high and, therefore, it is worthwhile to study the mechanisms of photocarrier generation and transport in the near-future devices. In this chapter, studies of electrical transport and photovoltaic properties of diodes with the structure Al/PAT films/Au are presented. These characteristics are examined as a function of alkyl chain length and regioregularity. The results show that the carrier transport and photocarrier generation are strongly influenced by the regioregularity as well as alkyl chain length, and are discussed with regard to the role of the depletion layer and transport mechanisms.
2. Sample preparation and experimental setup HT-PATn (n = 4, 6, 8, 10, 12 and 18) were synthesized based on the modified method of Chen and Rieke et al. [13,14], as reported in our previous papers [10–12,15]. Also were they obtained commercially and purified further as described below. The ratios of HT coupling contents are more than 98% as determined by 13 C NMR measurement [16]. The NR-PATn were synthesized with FeCl3 (an oxidizing agent) in chloroform solution [17]. They were purified thoroughly with ammonia and ethylenediamine tetraacetic acid to remove chloride and ferric ions followed by Soxhlet extraction of low-weight impurities and oligomers with methanol and hexane [18]. Thus obtained HT- and NR-PATn were dissolved in chloroform at concentrations of 0.7 ∼ 4.0 wt% depending on the film thickness. The solutions were used for drop cast or spin-coating on glass substrates. For the measurement of optical absorption spectra, the films on glass substrates were prepared by spin coating. The chloroform solution was dropped onto the glass substrate, which was spun at 1000 rpm for 30 s and then switched to 4000 rpm for 10 s. The film on the substrate was annealed at 60°C for 24 h before measurements. The absorption spectra of ultraviolet–visible light were measured using a JASCO V-570 spectrophotometer. Diodes [10,11,15] with a structure of indium tin oxide (ITO) or Au/PATn/Al, as shown in Fig. 3, were fabricated. A 18 mm2 ITO glass was washed with acetone and ethanol for 10 min with sonication and etched to leave electrodes with a width of 1 mm using a solution of hydrochloric acid and Zn powder for 1 min followed by rinsing with toluene and ethanol. The ITO on the substrate was polished with 0.05
Transport and photocarrier generation in poly(3-alkylthiophene) and metal junctions
459
Fig. 3: Diode fabricated for the measurement of I –V characteristics, TOF mobility and photovoltaic effects.
µm colloidal silica (Musashinno Densi) for 5 min, then sonicated with Semico Clean 56 (Furuuchi Chemical) for 10 min, followed by washing with running distilled water. The thickness of ITO electrodes was about 150 nm. After thermal treatment at 200°C in vacuum for 24 h, the ITO substrates were subjected to spin-coating with PATn. Al and Au deposited on a glass substrate in vacuum were also used as the bottom electrode. The chloroform solution of PATn was spin-coated on the ITO substrate, followed by annealing in a similar manner to the sample prepared for the absorption measurement. The typical thickness () of PATn films prepared by spin coating was approximately 500 nm, except those for the measurement of thickness dependence. The top Au or Al electrodes were also deposited on PATn film with a thickness of 30 ∼ 40 nm in vacuum. The transparency of Al electrode with a thickness of 40 nm was approximately 25% [15]. Current–voltage (I –V ) characteristics of the cells were measured using a Keithley 6517 electrometer in air atmosphere or in vacuum. Temperature of the diode was controlled with a He gas refrigerator. The carrier mobility was measured by the time of flight (TOF) and field effect (FE) methods. The TOF signal was generated by N2 laser excitation with a pulse width of 0.6 ns and energy of 70 µJ/pulse under the external bias (Va ) across the electrodes of sandwich cells. The transit times (Ttr ) of photocarriers during the migration in PATn film were evaluated from the inflection point of the Log(i) vs. Log(t) plot of TOF signals, where i and t are photocurrent and time, respectively. The mobility was calculated by µ = 2 /Va Ttr .
(1)
The FE mobility was obtained by fabricating an FE transistor [12]. An n+ -Si wafer with a conductivity of approximately 100 S/cm and (100) surface was used as the substrate and gate electrode. The substrate was oxidized to form SiO2 gate insulator with a thickness of 400 nm. The capacitance (Ci ) of the gate insulator was approximately 8.4 nF/cm2 . Comb shaped Au–Cr electrodes on the SiO2 surface served as the source and drain electrodes. The channel length (L) and width (W ) were 25 µm and 10 mm, respectively. PAT solution was cased on the top of the electrodes to form the active channel and the thickness of the PAT films was typically 1–2 µm. The I –V characteristics of FE devices were measured with the HP 4145B parameter analyzer at ambient atmosphere. The gate (VG ) and drain (VD ) voltages were measured
460
K. Kaneto et al.
against the source potential at the earth potential of 0 V. The general formula of drain current, ID , µCi W V2 (VG − VT )VD − D , (2) ID = L 2 was employed [18] to analyze the data, where µ and VT are FET mobility and the threshold voltages. To evaluate FET mobilities, µCi W ∆ID = VD , ∆VG L
(3)
the gradient of the ID –VG curve at the linear region or small VD was used. Excitation spectra of the photovoltaic effect (photoaction spectra) were measured using a lock-in amplifier by the integrated constant photon source system. Monochromatic light of a Xe lamp with the constant photon energy of 50 µW was irradiated to the cell, and the photocurrent due to the photovoltaic effect was monitored with a load resistance of 1 k via a lock-in amplifier at a chopping frequency of 82 Hz.
3. Optical absorption in HT- and NR-PATn Absorption spectra of HT-PATn and NR-PATn films are shown in Fig. 4a and b, respectively. In HT-PAT4, 6, 10 and 12, the spectra are almost the same in their peaks and shoulders except for HT-PAT18, which shows blue shift in the peak and no structure in comparison with others. This fact indicates that the electronic structures of PAT4∼12 films are nearly the same in the π-conjugation length and interchain interaction. In
Fig. 4: Absorption spectra of (a) HT-PATn and (b) NR-PATn.
Transport and photocarrier generation in poly(3-alkylthiophene) and metal junctions
461
other words, the interchain interaction of the π-electron system even in PAT4 is not so strong as to shift the electronic energy level by the optical detection. On the other hand, NR-PATn basically show blue shifting in the absorption peaks with increasing n, indicating that the effective conjugation length of π-electron decreases due to the more randomness with longer side chains. It is noted that the absorption coefficients decrease with increasing alkyl chain length [12], indicating that the absorption coefficients are determined with the density of π-electrons. Namely, the side chain simply separates main chains and makes the film bulky. These results are important to account for the alkyl chain length dependence of the mobilities in PATn film.
4. J –V Characteristics of Al/PATn/Au diodes 4.1. Rectification Typical current–voltage (J –V ) curves in semi-log plots for sandwich cells of ITO/HTPATn/Al, n = 4, 6, 10, 12 and 18 films are shown in Fig. 5. The positive bias to the ITO side against Al is the forward bias. Excellent rectification with a rectification ratio of more than 103 is obtained. These J –V characteristics of the diode is given by Eq. 4 based on the model of thermo-ionic emission, J = Js [exp(eV /nkT ) − 1],
(4)
where Js , e, V , n, k and T are the saturated current at reversed bias, electron charge, bias, ideality factor, Boltzmann constant and temperature, respectively. The ideality factor n is the index of diode quality and estimated from the slope of curves at the steep
Fig. 5: Typical semi-log plots of current–voltage (J –V ) curves for HT-PAT4, 6, 10, 12 and 18 films in sandwich cells of ITO/HT-PATn/Al.
462
K. Kaneto et al.
increase of current at forward bias. The best is unity. In our diodes n was 1.7 ∼ 3.2 depending on the preparation procedure. The result in Fig. 5 clearly indicates that performance of the diodes is better for shorter alkyl chain length. The smallest one of n = 1.7 was obtained when Al was first deposited on glass substrate, immediately followed by spin coating with HT-PAT4 and then Au deposition. The detailed survey of fabrication techniques for better performance of PATn/Al diodes is published elsewhere [7]. The magnitude of the current density for the forward bias at higher voltages is determined by the bulk conductivity (mobility and concentration of carriers) of the PAT films and also the contact resistance between Au and PATn. The shoulder observed at low forward bias is conjectured to result from defects of the Schottky junction such as pin holes and impurities. 4.2. Depletion layer in the Schottky junction The depletion layer in Schottky diodes plays the most important role in photocarrier generation as well as rectification. The thickness and the potential profile of the depletion layer are subjects to investigate. Taking the fact that an Au (or ITO)/PAT contact shows ohmic behavior, the rectification results from the Schottky junction between the PAT film and Al electrode [7,16,17]. Therefore, PAT films are conjectured to be a p-type semiconductor due to oxidation with a trace of acceptors or oxygen in air. According to the semiconductor theory, the depletion layer is formed by electron transfer from Al (n-type metal with low work function) to acceptors in PATn film with a large work function. This results in negative space charge and a strong electric field at the depletion layer. It should be also noted that the resistance of the depletion layer increases to that of an insulator. The number of electrons transfered from Al to PATn film is equilibrated by the difference of the work functions (Vw or the diffusion potential). The thickness of depletion layer (d) depends [27] on the reciprocal density of acceptors (Na ) and is given by Eq. 5,
1/2 2εr ε0 d= (Vw − Va ) , eNa
(5)
where a homogeneous distribution of acceptors is assumed. Eq. 5 indicates that the depletion layer vanishes when Va reaches Vw , followed by the onset of forward current. On the other hand, the depletion layer expands at reversed bias. Therefore, the rectification ratio (at the bias higher than the onset bias) is approximately the ratio of conductance of the bulk region (p-type PATn film) to that of the depletion layer. Schottky diodes using NR-PATn were fabricated also and showed similar rectification characteristics in J –V curves. However, the performance of NR-PATn diodes was inferior to that of HT-PATn diodes in the rectification ratio and ideality factors. Since the depletion layer is an insulator, the junction capacitance (C) of C = εr ε0 /d per unit area appears and can be expressed by VW − Va 1 =2 2 C εr ε0 eNA
[cm4 /F2 ].
(6)
Transport and photocarrier generation in poly(3-alkylthiophene) and metal junctions
463
Fig. 6: Reversed bias dependence of capacitors. Curves (a) and (b) are the data from the cell configurations Au/HT-PAT6/Al and Al/HT-PAT6/Au.
Although the model is derived from a conventional semiconductor, the experimental data agree for the first-order approximation with Eq. 6 as shown in Fig. 6. Basically, the result indicates that the thickness of the depletion layer increases with reversed bias according to Eq. 6 and can be estimated approximately from the intercept at zero bias. Curves (a) and (b) in Fig. 6 are the data from the cell configurations Au/HT-PAT6/Al and Al/HT-PAT6/Au. The difference of the samples is that the bottom electrode, which is represented by the left hand side, was first deposited, then followed by spin coating and deposition of the top electrode. Namely, in the latter the Al electrode might be oxidized during the spin coating of PAT and covered with insulating Al2 O3 . Using the relative dielectric constant [20] of εr = 2.8 estimated from the capacitance measurement in Al/HT-PATn/Al cell, the d is estimated to be 170 and 90 nm for the devices of (a) Au /HT-PHT/Al and (b) Al/HT-PHT/Au, respectively. The thicker depletion layer was obtained in the (a) configuration, probably it results from the activity of Al. The Na is estimated to be 1.5 × 1016 and 3.0 × 1016 cm−3 for devices of (a) Au/HT-PHT/Al and (b) Al/HT-PHT/Au, respectively, from the gradients of the curves using Eq. 6. The density of acceptors in HT-PHT is in the order of 1016 cm−3 and coincides well with those estimated from the measurement of field effect transistor [12]. The bias voltages at the intercept with the horizontal axis give rise to the VW being approximately 0.8 ± 0.1 V, which coincides with the difference of work functions between Al and PATn. 4.3. New estimation method of the thickness of the depletion layer Fig. 7 shows the change of resistance in a HT-PAT6 cell as shown by the inset upon deposition of Al on the top of the film with a deposition rate of 0.5–1.0 nm/min. The increase of resistance is due to the formation of an insulating depletion layer in the HT-PAT6 film at the interface with Al. The response curve consists of two components, which are an initial abrupt increase followed by a gradual increase. Assuming the resistance of depletion layer is sufficiently larger compared to the bulk region, the
464
K. Kaneto et al.
Fig. 7: The change of conductance in PAT6 upon deposition of Al. The inset shows the structure of the test cell.
thickness of the depletion layer (d ) is given by simply d = d0 (1 − R/R ),
(7)
where d0 is the thickness of PATn film, R and R are the resistances before and after deposition of Al. It is supposed that the first increase in resistance is due to the formation of a depletion layer and the second gradual increase is due to the diffusion or migration of Al into the PATn film. Typically the first abrupt increase of resistance gives the thickness of insulating depletion layer as 20–30 nm and the level-off value is 100 nm. The experimental results suggest that the deposited Al delivers electrons to acceptors in the PATn film immediately and gradually migrates inside the film. 4.4. Potential profile of diodes A bias voltage between two terminals of a diode distributes differently in the semiconductor region depending on the resistances. Therefore, it is supposed that the bias is applied mainly to the depletion layer since the resistance of the depletion layer is much larger than that of the bulk region. The potential profile along a cross section of the Al/HT-PAT6/Au diode has been measured directly using probing tips mounted on a micromanipulator with the accuracy of 3 nm. Fig. 8a shows a schematic drawing of the measurement system for the potential profile. A planer type of Schottky diode was fabricated, because a sandwich type cell is spatially too thin to map the potential profile. The Pt/Ir tip prepared for scanning tunneling microscopy was used as the potential probing tip and touched the surface of the PAT film between the Al and Au
Transport and photocarrier generation in poly(3-alkylthiophene) and metal junctions
465
Fig. 8: (a) Experimental setup for the direct measurement of the potential profile in a Al/HT-PATn/Au diode. (b) Potential profile of a Al/HT-PATn/Au diode with as parameter the bias voltage.
electrodes. The Al electrode was connected to earth and various Va were applied to the Au electrode. The potential at each position in the PATn film was picked up with the probing tip. The curves in Fig. 8b show the potential profile along a cross section of the Al/HTPAT6 film/Au diode. A steep jump was observed within 3 µm from the Al electrode and the potential gradually increased reaching to the Au electrode. For reversed bias, the bias was mainly applied to the depletion layer and the magnitude was approximately proportional to the bias voltage. On the other hand, the potential at the depletion layer under forward bias did not increase proportional to the bias and the potential gradient at the bulk region increases with increasing bias. For forward bias, however, a large voltage drop at the depletion layer, which should vanish at a bias of Va > Vw , is observed and could be due to some barrier originating from Al oxide or inherent contact resistance between metal and PAT film. In fact, there is appreciable contact resistance at the interface between PAT film and Au electrode also. For the detailed carrier transport between organic materials and metals not only Schottky type but also ohmic type contact is an important factor in fabricating better performance devices.
466
K. Kaneto et al.
5. Carrier mobilities in PATn 5.1. TOF mobilities The responses of photocurrents were measured under laser irradiation onto the ITO, Au and Al side of the diode at various bias fields and polarities. The carrier sign, place of carrier generation and carrier mobilities were investigated. The largest photocurrent was observed when the laser was irradiating the positively biased Al electrode against the ITO electrode [10,15,19,20]. The results indicate that the carrier sign is positive (holes) and is consistent with the results of the I –V characteristics. Furthermore, the photocarriers are generated effectively in the PAT film near the Al electrode or the depletion layer with the help of the strong built-in field at the Schottky junction. Fig. 9a and b show the typical photoresponses in logi–logt plots obtained in cast and spin coated HT-PAT10 films, respectively. In the photoresponse, a clearer kink associated with the transit time of photocarriers was observed in cast film than in spin coated film, indicating that the carrier migration in cast film is less dispersive than that of spin coated film. The dispersive transport results from photocarriers drifting to the opposite electrode by hopping via multiple traps distributed energetically and/or spatially. Generally the HT-PATn films with ordered structure show a clearer inflection point due to less multiple trapping levels [21]. The log–log plot of the photoresponse has been suggested by Scher–Montroll [22] for dispersive transport. The transit time (Ttr ) is analytically estimated from the time at the intercept of the two tangential lines before and after the inflection point of the photocurrent. It is found that the inflection points change to obscure with increasing alkyl chain length. The field dependence of transit times is shown in Fig. 10 for various spin coated PATn films, in which the field in the film is assumed to be uniform along the film direction. As shown in Fig. 10, it is noted that the transit time comes to be constant
Fig. 9: Typical photoresponses in logi–logt plots observed in (a) spin coated and (b) cast HT-PAT10 films.
Transport and photocarrier generation in poly(3-alkylthiophene) and metal junctions
467
Fig. 10: The field dependence of transit times as a function of bias field in logarithmic plots in spin coat PATn films.
Fig. 11: The field dependence of photocarrier mobilities as a function of bias field.
at lower fields, indicating the negative field dependence of the mobility. Photocarrier mobilities µ evaluated from Eq. 1 are shown in Fig. 11 as function of bias. As will be mentioned later, it is noted that the mobilities at the lower fields below 5 × 105 V/cm show a negative field dependence, then come to be constant or increase at higher fields. The mobility of HT-PAT4 is 2.8 × 10−2 cm2 /V s at the lowest field of around 2 × 103 V/cm and has the largest value [10,15,19,20]. The HT-PAT18 shows the lowest mobility in HT-PAT films and similar behavior to RD-PAT12 film. The field dependence of mobilities has been discussed using the Poole–Frenkel [23], the polaron hopping [24] and the off-diagonal order and disorder [21] model. The Poole–Frenkel model originally accounts for the positive field dependence in the manner of ln µ ∝ E 1/2 , which is based on the increased probability of carriers detrapping from ionic impurities at the higher field. And also the polaron hopping model explains the positive field dependence with
468
K. Kaneto et al.
the relationship of ln µ ∝ E. Especially, the negative field dependence of the mobilities in organic materials has been discussed using the modified Poole–Frenkel model by Gill [25] and Schein et al. [26]. The models are based on the relationship given by E 1/2 E0 exp , (8) µ = µ0 exp − kT β/T − γ where µ0 is the pre-exponential factor of mobility, E 0 is the activation energy, β is the constant given by the Poole–Frenkel model and γ is an empirical parameter. When T > β/γ , the field dependence of mobilities is negative. Bässler [21] proposed a model to explain the negative field dependence of the mobility by a Monte Carlo simulation of random walk of carriers in a Gaussian distribution of the density of states in trapping sites, when the off-diagonal disorder is superior to the energetic disorder. Namely, carriers hop to lower barrier sites, which may not be the field direction, or they take a detour at lower fields, resulting in the negative field dependence of the mobility. Analytically, these models account well for the experimental data for the negative field dependence of the mobility by choosing parameters appropriately. However, the physical meaning of the chosen parameter is not clear. The positive field dependences of mobilities are accounted satisfactorily by both the Poole–Frenkel and the Bässler model, since the range of fields which fit to the theoretical prediction is not sufficiently large to distinguish them. The temperature dependence of the TOF mobilities [15] gives as activation energies of mobilities 72, 88, 121 and 135 meV for HT-PAT4, 6, 12 and 18 films, respectively. The activation energies are larger for the films with longer alkyl chains. The result indicates that the activation energy results from the hopping between the π-conjugated main chains and is unlikely to be due to the intrachain π-conjugation, since the HTPATn is supposed to be extended to the whole length of the individual chain. It is also found that the activation energy scarcely depends on the field, which contradicts the prediction of Eq. 8 and suggests another transport mechanism. 5.2. Negative field dependence of photocarrier mobility due to non-uniform fields In the Schottky junction of semiconductors the external bias is mainly applied to the depletion layer [27], because of the larger resistance compared to that in the bulk region as discussed in the previous section. A simplified model of the non-uniform field distribution for a cell with a depletion layer is shown in Fig. 12, where the space charges at the depletion layer are ignored because of the first-order approximation. The conductivities of the depletion layer and bulk region are assumed to be constant as σd and σb , respectively. Hence, the fields at the depletion layer (E d ) and the bulk region (E b ) are constant and given by Ed =
σb Va , σb d + σd ( − d)
(9)
Eb =
σd Va . σb d + σd ( − d)
(10)
Transport and photocarrier generation in poly(3-alkylthiophene) and metal junctions
469
Fig. 12: A model of the non-uniform field distribution for a cell with a Schottky junction.
The velocity, v, of carriers is assumed to be proportional to the field, v = µE, and µ is a constant, the transit times of carriers at the depletion layer (Td ) and the bulk region (Tb ) are given by Td = d/µE d and Tb = ( − d)/µE b , respectively. Eventually, the transit time (Tr ) is expressed by Eq. 11, Tr = Td + Tb =
d 2 + (σd /σb + σb /σd )( − d)d + ( − d)2 . µVa
(11)
One may remind that the rectification ratio above the onset is roughly proportional to the conductance ratio of depletion to bulk region. This fact indicates σd / σb /( − d), and the thinner depletion layer compared with the film thickness, d , indicating that Eq. 11 can be simplified as σb d 2 . (12) Ttr 1 + σd µVa The transit time dominantly originates from the carrier transit at the bulk region. At −1/2 1/2 1 < (σb /σd )(d/), the Ttr should be roughly proportional to Va , since d ∝ Va , 1/2 as predicted in Fig. 10 at lower fields. Plots of Ttr Va vs. Va are shown in Fig. 13, indicating a fairly good linear relationship. The straight lines are fitted to the data with a 1/2 linear least-squares fitting. The intercepts of the lines, Ttr Va at Va → 0 are 2.0 × 10−7 , 6.2 × 10−7 , 3.7 × 10−7 and 5.3 × 10−5 V s from extrapolation of the linear region, hence the mobilities are obtained to be 1.8 × 10−2 , 4.7 × 10−3 , 1.3 × 10−3 and 1.4 × 10−3 cm2 /V s for HT-PAT4 (0.65 µm), PAT4 (1.7 µm), PAT6 (6.9 µm) and HT-PAT12 (2.7 µm) films, respectively. The evaluation of TOF mobilities for the negative field dependence is based on the Schottky junction with the conditions σb σd , a field independent mobility and, more importantly, that the carrier generation occurs at the depletion layer. It has already been mentioned that the carrier generation takes place at the depletion layer with the help of the large built-in field and reversed bias in our previous paper [8,17]. This can also be confirmed by plotting of the number of generated charges with the laser light against 1/2 the thickness of the depletion layer (d or Va ). The number of generated charges is
470
K. Kaneto et al.
1/2
Fig. 13: Plots of Ttr Va vs. Va junction cell.
under the assumption of a non-uniform field distribution in a Schottky
Fig. 14: Dependence of number of photocarriers Itr Ttr vs. Va1/2 .
obtained by integrating the current against time in the i ph vs. t curve of Fig. 9, or approximately evaluated by Itr Ttr , where Itr is the current at the transit time. Fig. 14 1/2 shows fairly good dependence of Itr Ttr vs. Va , indicating that the photocarriers are 1/2 generated at the depletion layer, expanding with increasing the reversed bias, Va . It is 1/2 interesting to note that the Itr Ttr vs. Va curves of all PAT films fall on approximately a single line. It is confirmed that the photocarriers are actually generated at the depletion layer that is expanding with the reversed bias. It should be reminded again that the proposed model of the non-uniform field distribution is based on σb σd , or the larger conductivity. In fact, the negative field dependence has been scarcely observed in regiorandom PAT films with lower conductivity [11,20].
Transport and photocarrier generation in poly(3-alkylthiophene) and metal junctions
471
Fig. 15: ID –VD characteristics at various VG in the HT-PAT4 FE transistor.
5.3. FE mobilities Fig. 15 shows ID –VD characteristics at various VG in the FE transistor using HT-PAT4 film at room temperature. The curves are the typical characteristics [10,18] of an FE transistor, showing signs of saturation of ID at high VD due to the pinch off of the channel. The enhancement of ID by the negative gate bias indicates that a positive carrier (holes) is favorably induced in the PATs channel and coincides with the results of TOF measurements. The on/off ratio of 15 at VG = 30 V and −30 V for VD = −10 V is relatively small, which could result from the rather thick PAT film employed. The channel conductivities (σ ) of PAT films were evaluated from the curve at VG = 0 and at low VD and are summarized in Table 1 with the data of NR-PAT6 and NR-PAT12 for comparison. The typical conductivity of HT-PAT4 was 1.4 × 10−6 S/cm, which is larger than those of HT-PAT18 and NR-PATn by three orders of magnitude. ID –VG curves for various VD are shown in Fig. 16, where the ID was obtained by subtracting ID at VG = 0 from the raw data of ID . The mobilities are estimated from these curves according to Eq. 3 from the gradient at the linear part of the curves and relatively large VG . The carrier densities (N ) were estimated from the relationship Table 1 Conductivities, mobilities and carrier densities in PATn films Material
Conductivity (S/m)
Mobility (cm2 /V s)
Carrier density (cm−3 )
HT-PAT4 HT-PAT6 HT-PAT10 HT-PAT12 HT-PAT18 NR-PAT6 NR-PAT12
1.4 × 10−6 1.2 × 10−6 1.2 × 10−6 9.1 × 10−7 6.1 × 10−9 4.8 × 10−9 1.6 × 10−9
1.0 × 10−2 2.9 × 10−3 9.1 × 10−4 7.9 × 10−4 5.0 × 10−6 4.3 × 10−6 2.8 × 10−7
8.9 × 1014 2.6 × 1015 8.2 × 1015 7.2 × 1015 7.5 × 1015 7.9 × 1015 3.6 × 1016
472
K. Kaneto et al.
Fig. 16: ID –VG curves for various VD , where the ID was obtained by subtracting ID at VG = 0 from the raw data of ID .
Fig. 17: (a) Mobilities of PATn films and (b) conductivities and carrier densities as a function of the alkyl chain length.
N = σ/eµ. The dependencies of conductivity and carrier density on alkyl chain length are shown in Fig. 17. It is apparent that the mobilities are larger for the films with shorter alkyl chain length. The mobility in HT-PAT4 of 1.0 × 10−2 cm2 /V s is noticeably larger than those of the other HT-PAT films and about three times larger than that of NR-PAT4 film reported by Ohmori et al. [28]. This fact is explained by the assumption that carrier migration is consisting of two components, namely, of the transport in main chains within the effective π-conjugation and of hopping between main chains. The carrier transport in the main chain is supposed to be much faster than that of inter-chain hopping. Therefore the carrier transport is predominantly controlled by the interchain hopping, hence PATn with the shorter the alkyl chain show the larger mobility. It is interesting to note that carrier densities are approximately the same for all PAT films as shown by the curves in Fig. 17 and the conductivities are primary determined
Transport and photocarrier generation in poly(3-alkylthiophene) and metal junctions
473
Fig. 18: The energetic diagram for the randomness of the main chain in (a) HT-PATn and (b) NR-PATn.
by the mobilities. The reason for slightly lower residual carrier density for PATs with shorter alkyl chains is not known, however, may be related to impurities in films. The exceptionally low conductivity due to low mobility in HT-PAT18, NR-PAT6 and NRPAT12 results from the randomness of main chains. The small mobility of HT-PAT18,
474
K. Kaneto et al.
Fig. 19: FE and TOF mobilities in HT-PATn.
which deviates from the extrapolation line, is conjectured being due to the entanglement of too long side chains. The energetic diagram for the randomness of the main chain is schematically shown in Fig. 18. In HT-PATn the π-conjugation extends to the whole length of the polymer chain and the carriers transport as if in a single band. On the other hand, in NR-PATn the π-conjugated region plays the role of the potential barriers for carrier transport within a main chain. Recently, Bao et al. [29] and Sirrighaus et al. [30] have reported FE mobilities in HT-PAT4 of 0.045 cm2 /V s and 0.1 cm2 /V s, respectively, which were attained by careful purification and surface treatment of SiO2 . It is noted, as depicted in Fig. 19, that the FE mobilities generally agree with the TOF mobilities [15]. In fact the FE mobility represents the carrier transport along the film surface, whereas the TOF mobility is the transport perpendicular to film surface. Therefore, the agreement of FE and TOF mobilities indicates isotropic carrier transport in PAT films.
6. Photovoltaic effects and mechanism of carrier generation The curves in Fig. 20 show the action spectra of quantum efficiency (Q.E.) of photon to carrier estimated from photocurrent in Al/HT-PHT/Au diode for the illumination at the Al and Au side. The action spectrum for the light illuminated at Al side is similar to the absorption spectrum at the absorption maximum around 2.5 eV. On the other hand, the action spectrum for the illumination at the Au side is quite different from the absorption spectrum. The results indicate that photocarriers are generated dominantly at the junction of Al and HT-PATn films. This idea is supported by taking the penetration depth of light (120 nm: inverse of absorbance at the peak of HT-PATs films [7]) and the film thickness of 200 nm into account. Namely, upon the illumination at the Au side
Transport and photocarrier generation in poly(3-alkylthiophene) and metal junctions
475
Fig. 20: Action spectra of quantum efficiency (Q.E.) of carrier to photon estimated from photocurrent in Al/HT-PAT6/Au diode for illumination at the (a) Al and (b) Au side, and (c) is the absorption spectrum.
the light at the absorption maximum is absorbed in the bulk region before reaching the depletion layer, resulting in opposite spectra to that of Al side illumination. It should be noted that a large Q.E. is observed at a photon energy above 3 eV, indicating that a different carrier generation path is taking place at higher electronic energy levels. The dependencies of film thickness on Q.E. are shown for illuminations at the Au and Al sides by the curves in Fig. 21a and b, respectively. For the illumination at the
Fig. 21: Dependencies of film thickness on Q.E. for the illumination at the (a) Au and (b) Al side.
476
K. Kaneto et al.
Fig. 22: Schematic drawing of the carrier generation mechanism for illumination at the Al side.
Au side, the larger Q.E. and peak shifting to shorter wavelengths were observed for thinner films, and the Q.E. spectra came to be similar to those for illumination at the Al side. The result indicates that the thickness of the photocarrier generation layer is less than 80 nm. The thickness dependence of the Q.E. for the illumination at the Al side was that the Q.E. did not markedly depend on the thickness of 80 ∼ 160 nm. This fact supports the hypothesis that the bulk region does not contribute to the carrier generation, and the thickness of the photocarrier generation layer is less than 80 nm or equal to the depletion layer. The carrier generation mechanism is summarized as shown by the schematic drawing in Fig. 22 for illumination at the Al side. The light generates excitons, which dissociate into a electron and hole pair by the assistance of the strong internal field. Electrons transfer to the Al electrode and holes migrate to the Au electrode through the bulk region, resulting in a photovoltaic effect. The migration of holes in the bulk region is probably diffusion by the concentration gradient of holes.
7. Summary Rectification and photovoltaic effects in diodes fabricated with the structure of Al/ poly(3-alkylthiophene) film/Au have been mentioned in view of carrier transports and photocarrier generation at the Schottky junction. These functions of optoelectronic
Transport and photocarrier generation in poly(3-alkylthiophene) and metal junctions
477
devices are found to originate from the depletion layer with nanometric thickness and space charge. The thickness of the depletion layer is estimated by a couple of methods to be several 10 nm. It is mentioned that the photocarriers are generated most effectively in the depletion layer with the assistance of the strong built-in field. Carrier mobilities which have been evaluated by means of TOF and FE are discussed from the viewpoint of regioregularity and effect of alkylchain length. The authors are planning to fabricate a diode with high performance of the photovoltaic effect or solar cells based on poly(3-alkylthiophene).
Acknowledgements The authors express their sincere thanks to Dr. S.S. Pandey, Mr. S. Nagamatsu for assistance in carrying out experimental works and discussion. This study is supported by a Grant-in-Aid for Scientific Research from the Ministry of Education, Science and Sport and Culture, Japan. The authors express their sincere thanks to Fukuoka IST for use of the experimental facility.
References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22.
H. Shirakawa, T. Ito, and S. Ikeda, Makromol. Chem. 179, 1565 (1978). C. Chiang, M. Druy, S. Gau, A. Heeger, E. Louis, A. MacDiarmid, Y. Park, and H. Shirakawa, J. Am. Chem. Soc. 100, 1013 (1978). A.F. Diaz, K. Kanazawa, and G.P. Gardini, J. Chem. Soc. 100, 1013 (1978). K. Kaneto, K. Yoshino, and Y. Inuishi, Jpn. J. Appl. Phys. 21, L567 (1982). K. Kaneto, K. Yoshino, and Y. Inuishi, Jpn. J. Appl. Phys. 22, L412 (1983). C. Chiang, C.R. Fincher, Jr., Y.W. Park, A.J. Heeger, H. Shirakawa, E.J. Louis, S.C. Gau, and A.G. MacDiarmid, Phys. Rev. Lett. 39, 1098 (1977). K. Kaneto and W. Takashima, Curr. Appl. Phys. 1, 355 (2001). P. Barta, F. Cacialli, R.H. Friend, and M. Zagorska, J. Appl. Phys. 84, 6279 (1998). M. Granstrom, K. Petritsch, A. Arias, A. Lux, M. Anderson, and R. Friend, Nature 395, 257 (1998). S. Pandey, W. Takashima, S. Nagamatsu, and K. Kaneto, IEICE Trans. Electron. E83-C, 1088 (2000). W. Takashima, S.S. Pandey, T. Endo, M. Rikukawa, and K. Kaneto, Curr. Appl. Phys. 1, 90 (2001). K. Kaneto, W. Lim, W. Takashima, T. Endo, and M. Rikukawa, Jpn. J. Appl. Phys. 39, L872 (2000). T. Chen, X. Wu, and R.D. Rieke, J. Am. Chem. Soc. 117, 233 (1995). R.S. Loewe, S.M. Khersonsky, and R.D. McCullough, Adv. Mat. 11, 250 (1999). K. Kaneto, K. Hatae, S. Nagamatsu, W. Takashima, S. Pandey, K. Endo, and M. Rikukawa, Jpn. J. Appl. Phys. 38, L1188 (1999). M. Rikukawa, Y. Tabuchi, K. Ochiai, K. Sanui, and N. Ogata, Thin Solid Films 327, 469 (1998). M. Ahlskog, J. Paloheimo, H. Stubb, P. Dyrklev, M. Fahlman, O.Inganas, and M. Anderson, J. Appl. Phys. 76, 893 (1994). D.K. Schroeder, Semiconductor Material and Device Characterization (John Wiley and Sons, New York, 1990). S.S. Pandey, W. Takashima, S. Nagamatsu, T. Endo, M. Rikukawa, and K. Kaneto, Jpn. J. Appl. Phys. 39, 94 (2000). S.S. Pandey, S. Nagamatsu, W. Takashima, and K. Kaneto, Jpn. J. Appl. Phys. 39, 6309 (2000). H. Bässler, Phys. Rev. 175, 15 (1993). H. Scher and E.W. Montroll, Phys. Rev. B 12, 2455 (1975).
478 23. 24. 25. 26. 27. 28. 29. 30.
K. Kaneto et al. J. Frenkel, Phys. Rev. 54, 647 (1938). E. Lebedev, T. Dittrich, V. Petrova-Koch, S. Karg, and W. Brutting, Appl. Phys. Lett. 71, 2686 (1997). W. Gill, J. Appl. Phys. 43, 5033 (1972). L. Schein, A. Rosenberg, and D. Glatz, J. Appl. Phys. 60, 25 (1986). M. Goodman, J. Appl. Phys. 34, 329 (1963). Y. Ohmori, H. Takahashi, K. Muro, M. Uchida, T. Kawai, and K. Yoshino, Jpn. J. Appl. Phys. 30, L610 (1991). Z. Bao, A. Dodabalapur, and A.J. Lovinger, Appl. Phys. Lett. 69, 4108 (1996). H. Sirringhaus, N. Tessler, and R. Friend, Science 280, 1741 (1998).
Nanotechnology and Nano-Interface Controlled Electronic Devices Editors: M. Iwamoto, K. Kaneto and S. Mashiko © 2003 Elsevier Science B.V. All rights reserved
CHAPTER 23
Thermochromic behavior in novel conducting polymers at the solid–liquid phase transition Mitsuyoshi Onoda and Kazuya Tada Graduate School of Engineering, Himeji Institute of Technology, 2167 Shosha, Himeji, Hyogo 671-2201, Japan E-mail:
[email protected]
1. 2. 3. 4.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Preparation of samples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Experimental details . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1. Large change of electrical conductivity of poly(3-alkylthiophene) at the solid–liquid phase transition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2. Large change of the absorption spectrum of poly(3-alkylthiophene) at the solid–liquid phase transition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3. Anomalous dependencies of photoluminescence of poly(3-alkylthiophene) on temperature and alkyl chain length . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4. Solvatochromism in poly(3-alkylthiophene) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.5. Influence of hydrostatic pressure on electrical and optical properties of poly(3-alkylthiophene) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.6. Thermochromism in poly( p-phenylene vinylene) derivatives . . . . . . . . . . . . . . . 5. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
479 480 482 483 483 485 488 491 495 501 507 507
1. Introduction The Nobel prize in chemistry 2000 was awarded for the discovery and development of conducting polymers by Professors Hideki Shirakawa, Alan G. MacDiarmid and Alan J. Heeger. Conducting polymers with highly extended π-conjugated electron systems in their main chains have been considered to be insoluble and infusible materials because of their stiff main chain. The difficulty of processing them has restricted the practical applications of conducting polymers. However, to shape conducting polymers into films
480
M. Onoda and K. Tada
or fibers, preparation methods, such as using solution-processable precursors [1] or introducing long side chains [2,3], have been determined. In particular, the introduction of long side chains (e.g., alkyl or alkoxy) in the molecular structures makes it possible for the conducting polymers to be soluble in some solvents such as chloroform, and fusible by heating. By utilizing the solubility [2,4] and fusibility [5,6], these polymers can be easily shaped into thin films by wet processes such as casting, spin-coating and hot-press techniques. This is a great advantage from the viewpoint of application to electronic devices [7]. Their mechanical, electrical and optical properties depend highly on their side chain length, moreover they exhibit unique characteristics such as thermochromism [8–10]. Because of this fusibility of poly(3-alkylthiophene), PAT, and poly(2,5-dialkoxy- p-phenylene vinylene), ROPPV, any shape of samples can be prepared by molding methods. For example, films and fibers can be prepared easily by a hot-press method and also by a melt-spinning method. These films and fibers indicated low conductivity as a characteristics of insulators but turns into metal upon exposure to iodine vapor and a conductivity as high as 10–100 S/cm can be obtained just in the case of non-substituted samples. The fusibility can be also applied to connect or stick two conducting polymer films, as conducting binder, by sandwiching powder or a film of fusible polymer between two non-fusible conducting polymer films and pressing at high temperature. This connected sample also demonstrated an insulator–metal transition upon doping. In this paper, electrical and optical properties of poly(3-alkylthiophene) in the liquid state, and their dramatic change at the phase transition are reported. Moreover, remarkable effect of hydrostatic pressure on electrical and optical properties of poly(3alkylthiophene) and dependence of absorption spectra on solvent are discussed in detail. On the other hand, poly(2,5-alkoxy- p-phenylene vinylene), ROPPV [11] is one example of soluble and fusible conducting polymers. The ionization potential of this conducting polymer is lowered by the donor alkoxy chain, and hence its doped-state stability is extremely high [12]. Therefore, ROPPVs are expected to be promising candidates for organic electronic devices. Therefore, we report on the thermochromic behavior of ROPPVs. The variation of properties with side chain lengths is also mentioned. The results will be discussed in comparison with those for poly(3-alkylthiophene)s, PATs.
2. Preparation of samples Fig. 1a and b shows schematic diagrams of the synthetic route to PATn and ROPPVn, respectively. PATn were prepared from corresponding 3-alkylthiophene monomers using FeCl3 as a catalyst as shown in Fig. 1a [2,5,13]. That is, 3-alkylthiophene monomer and FeCl3 were dissolved in anhydrous chloroform at −10°C. After reaction the mixture was kept at this temperature for 12 h. The polymer was precipitated by pouring the whole reaction mixture into a large excess of methanol and isolated by filtration. The raw product was washed in a Soxhlet apparatus with methanol to remove FeCl3 . The PATn are soluble in conventional organic solvents such as chloroform, tetrahydrofuran, dichloromethane, etc. and red in color.
Thermochromic behavior in novel conducting polymers at the solid–liquid phase transition
481
Fig. 1: Schematic diagram of the synthetic route to (a) PAT and (b) ROPPV.
On the other hand, ROPPVn were prepared from 1,4-bis(chloromethyl)-2,5-dialkoxybenzene monomer using KOt-Bu (potassium t-butoxide) as a catalyst as shown in Fig. 1b [14]. That is, 2,5-dialkoxybenzene was prepared through the use of the Williamson ether synthesis from phenol and alkylbromide. 1,4-bis(chloromethyl)-2,5dialkoxybenzene monomer was obtained according to the procedure which is an adaptation of the method described by Wood and Gibson. A solution of KOt-Bu in t-butanol was quickly added at room temperature under N2 atmosphere to a stirred solution of 1,4-bis(chloromethyl)-2,5-dialkoxybenzene in p-xylene/t-butanol. The reaction mixture was refluxed for 22 h. The red solution was concentrated and the polymer was precipitated by pouring the whole reaction mixture into a large excess of methanol and isolated by filtration. The raw product was washed in a Soxhlet apparatus with methanol to remove KOt-Bu and dried under vacuum. The ROPPVn is soluble in conventional organic solvents such as chloroform, tetrahydrofuran, dichloromethane, etc., and reddish orange in color. Here, the n of ROPPVn and PATn denotes the number of carbon atoms in the alkoxy and alkyl chains. The side chain length dependence of melting points of ROPPVs and PATs is shown in Fig. 2. It is observed that the melting points of ROPPVs and PATs with longer side chains is lower. The spin-coated films (film thickness: about 200 nm) and casted films (about 1 µm) of these conducting polymers were fabricated on glass substrates from chloroform solution.
482
M. Onoda and K. Tada
Fig. 2: Dependence of the melting point of PAT and ROPPV on the side chain length.
3. Experimental details Electrical conductivity under vacuum was measured utilizing a Keithley 616 electrometer. The optical absorption spectra of the spin-coated films were measured using a spectrophotometer (Hitachi U-3410). The bandgap energies of these conducting polymers were evaluated from the absorption edge of the optical absorption spectrum by a (αhν)2 vs. hν plot. Here, α indicates the absorption coefficient at the photon energy hν. The photoluminescence (PL) spectra of the casted films were obtained with a monochromator and a photomultiplier tube. A Xe lamp combined with monochromator was used as an excitation source. Each sample was set in the pressure chamber with optical windows and brought up to a pressure of 10 kbar by an oil-press method. The ionization potentials of these conducting polymers were measured using an atmospheric low-energy photoelectron spectroscopy instrument (Rikenkeiki AC-1) [15]. Fig. 3 shows the schematic diagram of this instrument. Monochromatic ultraviolet light in the energy range of 3.4–6.2 eV is irradiated onto the sample in air. The photoelectrons emitted from the sample surface stimulate the ionization of the atmospheric oxygen, and the number of ionized oxygen molecules is counted. These measurements were carried out at temperatures ranging from room temperature to 200°C. For the measurements of the optical absorption spectra and the PL spectra, the surface of the polymers spin-coated or casted on a glass plate were covered with another glass plate in order to avoid oxidation during heating. Prior to carrying out the measurements, the polymers were heated to higher temperatures than their melting points to adhere them to the glass surfaces.
Thermochromic behavior in novel conducting polymers at the solid–liquid phase transition
483
Fig. 3: Schematic diagram of low-energy photoelectron spectroscopy apparatus.
4. Results and discussion 4.1. Large change of electrical conductivity of poly(3-alkylthiophene) at the solid–liquid phase transition Findings of solubility in some solvents and even fusibility at relatively low temperature of different types of PATn have stimulated the extensive study of these series of polymers. The study of electronic processes in the liquid state has been highly limited. That is, the electron mobility has been observed in various simple liquids such as hydrocarbon and rare gas liquids, but their electronic band scheme has not been clarified. Especially, the electronic conduction in polymer liquids has not been studied. Fig. 4 indicates the temperature dependence of the electrical conductivity of PAT12 in both the solid and the liquid state. In the solid state near the premelting region with increasing temperature, some anomalous temperature dependence of the electrical conductivity was observed. In particular, at the phase transition from solid to liquid state, the conductivity decreased dramatically in a step-wise manner, and again increased with an activation energy in the liquid state with increasing temperature. The step-wise decrease could be related to a rapid change of either electron mobility or carrier density at the phase transition. By contrast, during the cooling cycle, the conductivity increased in a step-wise manner at the phase transition by the same magnitude as that during the heating stage. However, there existed a hysteresis in the conductivity observed in the heating and cooling cycles. That is, the temperature at which the conductivity increased step-wise in the cooling stage was lower than that in the heating stage. This could be related to a super-cooling effect when the temperature was lowered at a relatively high speed (0.5°C/min). The activation energy of the conductivity in liquid state was calculated to be about 0.7 eV. A similar unique temperature dependence of the conductivity was also observed in other fusible PATn such as poly(3-docosylthiophene), PAT22, shown in Fig. 5 as an
484
M. Onoda and K. Tada
Fig. 4: Temperature dependence of the electrical conductivity of PAT12 in both the solid and the liquid state.
Fig. 5: Temperature dependence of electrical conductivity of PAT22 in both the solid and the liquid state.
example. The activation energy of the conductivity in the liquid state was evaluated to be about 0.9 eV in this case.
Thermochromic behavior in novel conducting polymers at the solid–liquid phase transition
485
4.2. Large change of the absorption spectrum of poly(3-alkylthiophene) at the solid–liquid phase transition The PATn films show a reversible large color change with temperature (thermochromism) and are red in color at low temperature and yellow in color at high temperature. Figs. 6 and 7 show the absorption spectra of PAT12 and PAT22 at various temperatures ranging from solid to liquid phase, respectively. Thermochromism in PATn films was previously reported [8]. A shift of the absorption peak to a higher energy with increasing temperature was found. However, the shift of the absorption edge was relatively small. In these figures, a large shift of the absorption peak but a small shift of the edge are also confirmed in the solid state. Figs. 8 and 9 show the temperature dependence of the absorption peak and the edge energies of PAT12 and PAT22 as evaluated from Figs. 6 and 7, respectively. In these figures it is clearly indicated that the absorption peak in the solid phase shifts with the temperature shift, but not the absorption edge. At the phase transition from solid to liquid, a very large absorption change was observed. That is, the absorption edge shifted remarkably to a higher energy in the liquid phase. The temperature dependence of the absorption peak energy in the liquid state was small compared to that in the solid state. The dramatic change at the phase transition could be due to the large conformation
Fig. 6: The change of absorption spectra of PAT12 at various temperatures ranging from solid to liquid phase.
486
M. Onoda and K. Tada
Fig. 7: The change of absorption spectra of PAT22 at various temperatures ranging from solid to liquid phase.
changes with melting. In the liquid state, the co-planarity of the neighboring thiophene rings is expected to be minimized. The resultant decreased conjugation length should give rise to an increase of the bandgap. In the case of PAT12, the bandgap in the liquid phase was evaluated to be about 2.5 eV from the absorption spectrum. This is larger by about 0.3 eV compared with that in the solid phase. Two modifications of the polymer main chain seem to be candidates as conformation changes which reduce co-planarity: bending and/or twisting (torsion) of neighboring thiophene rings along the main chains. However, the latter may be more probable, because of the rigidity of the double bond system.
Fig. 8: Temperature dependence of the absorption peak and edge energy of PAT12 obtained from Fig. 6.
Thermochromic behavior in novel conducting polymers at the solid–liquid phase transition
487
Fig. 9: Temperature dependence of the absorption peak and edge energy of PAT22 obtained from Fig. 7.
The shift of the absorption peak in the solid state can also be explained in terms of the influence of some conformation change with temperature triggered by the change of conformation of alkyl group. That is, in the solid state at high temperatures of the premelting region, the torsion of the neighboring thiophene rings should be initiated at several points along a polymer main chain. The separation (number of thiophene rings) between two neighboring torsioned points which corresponds to the conjugation length should decrease at a high temperature and be randomly distributed. This results in the formation of regions of short conjugation whose lengths are widely distributed. The decrease of longer conjugated chains, and formation of shorter conjugation sections, can explain the blue shift of the absorption peak. The fact that the absorption edge does not change remarkably means that some original long conjugation region still remains. It should also be noted that in the low energy section of the absorption peak, a structure is observed. This may indicate the existence of some shorter conjugation lengths. However, this can be explained by a contribution of vibrational subbands because electron–phonon interaction seems to be strong in the one-dimensional system. The existence of a long alkyl chain may enhance the one-dimensionality. The increase of the bandgap also explains the step-wise decrease of conductivity at the phase transition. That is, in this case, the decrease of the carrier density in the liquid state due to the increased bandgap may be the cause of the lower conductivity in the liquid state. However, electron scattering should also be enhanced in the premelting and melting regions. The anomalous temperature dependence of conductivity observed in the premelting region could be related to electron scattering and correlated with the anomalous absorption peak shift in this range. Fluctuation of the bandgap due to the existence of regions of various conjugation lengths and molecular motion of alkyl side chains in the premelting region could be the cause of strong carrier scattering. The electronic band scheme in the liquid state has not been explained in detail so far. In general, the bandgap of liquids so far studied is expected to be extremely large. Experimental study is fairly difficult due to the necessity of using vacuum ultraviolet light. There are only a few papers in which the bandgap has been estimated in liquids
488
M. Onoda and K. Tada
and it has not been completely clarified whether the band model is still applicable in liquids or not. The present observation of a clear absorption edge, even in the liquid state, and the similarity of the shape of the absorption spectra support the concept of the applicability of electronic bands in liquids. Optical studies of liquid conducting polymers are much easier compared with conventional liquids due to the small bandgap which corresponds to the energy of visible light. The electrical study is also simplified because the electronic carrier density is relatively high due to the small band gap. The authors propose the study of the properties of conjugated polymers in the liquid state (liquid conducting polymer) for the understanding of fundamental electronic processes in liquids. 4.3. Anomalous dependencies of photoluminescence of poly(3-alkylthiophene) on temperature and alkyl chain length Some of conducting polymers such as PATn have been found to be soluble in some solvents and even fusible. In these new type of substituted conducting polymers various interesting phenomena such as thermochromism and anomalous temperature dependence of electrical conductivity were found in a relatively low temperature range. In this section, we report detailed characteristics of anomalous temperature dependence of photoluminescence of PATn film both below and above melting point. It is also indicated that photoluminescence intensity depends strongly on alkyl chain length. Fig. 10 indicates photoluminescence spectra of PAT12 at various temperatures. Similar spectra and their temperature dependencies were also found in other PATn. It should be noted in this figure that with increasing temperature the emission peak shifts to higher energy slightly and its intensity increases in the solid phase. Above the melting point, it again decreased in the liquid phase.
Fig. 10: Photoluminescence of PAT12 at various temperatures.
Thermochromic behavior in novel conducting polymers at the solid–liquid phase transition
489
Fig. 11: Temperature dependency of the photoluminescence intensity for PATs.
The temperature dependence of the photoluminescence intensity is more clearly indicated in Fig. 11. That is, with increasing temperature the emission intensity increases in the solid phase and after showing a maximum at the melting point it again decreases in the liquid phase. Corresponding to the lower melting point, the photoluminescence maximum intensity appears at lower temperature in PATn with longer alkyl chain length. These temperature dependencies of the photoluminescence are quite anomalous compared with those of ordinary inorganic and organic semiconductors and insulators, in which the photoluminescence intensity decreases with increasing temperatures remarkably due to the increased probability of non-radiative recombination. This anomalous behavior in PATn can be explained qualitatively in terms of conformation change of polymer main chain with temperature and its influence on the dynamics of photoexcited species as follows. By the excitation with the Ar ion laser of photon energy 2.54 eV, electrons and holes are excited in conduction and valence bands, and some of them should turn into negative and positive polarons, respectively, in a very short time [16]. The photoluminescence in these conducting polymers is considered to be due to the radiative recombination of these excited species. Therefore, when positive and negative excited species move out from the excited region and separate, escaping from the recombination, the photoluminescence intensity should be suppressed. We have already reported that with increasing temperature co-planarity of neighboring thiophene rings decreases due to the introduction of torsion at some bondings between thiophene rings, resulting in the decrease of effective conjugation length [17]. This can explain the thermochromism in PATn. Therefore, at higher temperature due to the decreased effective conjugation length the mobility of the excited species and their escape probability from the excited region should decrease, which can result in the enhancement of the probability of recombination and the emission of the observed stronger photoluminescence. When the effect of enhancement of recombination probability is larger than the increase of non-radiative recombination at higher temperature, the enhancement of photoluminescence should be
490
M. Onoda and K. Tada
Fig. 12: Dependence of peak photoluminescence intensity on the alkyl chain length.
observed as in the present results. The activation energy of the photoluminescence intensity in the solid phase was estimated to be around 0.1–0.2 eV. This should correspond to the energy necessary for the introduction of the torsion angle with the activation energy of non-radiative recombination subtracted. We already reported that the torsion angle does not change remarkably with temperature in the liquid phase contrary to in the solid state [18], though it is larger than that in the solid phase. Therefore, the effective conjugation length should not change remarkably with temperature in the liquid state and the probability of confinement of excited species in the excited region should not change drastically with temperature. In such a case, the increased non-radiative recombination at higher temperature will result in the suppression of photoluminescence intensity in the liquid phase. It should be also pointed out that the existence of similar emission spectra in both the liquid and solid phase clearly supports the applicability of the concept of electronic bands in the liquid phase. This observation of stronger photoluminescence in the system with shorter conjugation length reminds of the case of cis- and transpolyacetylene films. That is, in cis-polyacetylene with a non-degenerated ground state in which the formation and migration of solitons is highly limited showed much stronger photoluminescence compared with the case of trans-polyacetylene in which solitons can be excited and migrate freely [19,20]. It should be also noticed in Fig. 11 that PATn with longer alkyl chains shows stronger photoluminescence. Fig. 12 shows the dependence of peak photoluminescence intensity on the alkyl chain length. This can be also tentatively interpreted in terms of dynamics of photoexcited species as follows. In PATn with longer alkyl chains, because of the bulky side chain effect, the interaction of neighboring polymer main chains should be reduced and the probability of the transfer of photoexcited species from a polymer main chain to the neighboring one should be suppressed. In such a case the density of positive and negative excited species in a chain will increase, resulting in the enhancement of recombination photoluminescence. That is, the interchain transfer of excited species should also reduce on the quantum efficiency of photoluminescence. This also supports
Thermochromic behavior in novel conducting polymers at the solid–liquid phase transition
491
the idea that the photoluminescence originates in the recombination of excited species on a same polymer chain. 4.4. Solvatochromism in poly(3-alkylthiophene) PATs with long alkyl chain have been found to be soluble in several solvents. Fusibility of these polymers has also been observed. The color of the solution of these PATs is strongly dependent on temperature, which has been explained in terms of large conformation change with solvent and temperature [17]. For example, the color of the solution of poly(3-docosylthiophene), PAT22, in anisole is red at 30°C (low temperature) and yellow at 50°C (high temperature). The blue shift of the absorption at high temperature has been interpreted in terms of the decrease of the effective conjugation length due to the decrease of the co-planarity of thiophene rings triggered by the conformation change of alkyl side chains. However, it is not clear even why PAT is soluble in some solvents but not in other solvents. In this session, the dependence of absorption spectra on solvent is discussed in detail. That is, solubility was checked by dissolving 300 mg of polymer in 1 l of solvent at room temperature. The color of the solution of PAT varies with some solvents even at identical temperature [21]. As an example, the color of the solution of poly(3-hexylthiophene), PAT6, is red in anisole, orange in p-xylene and yellow in n-heptane. It is confirmed that the color of the solution depends on the solvents delicately. Table 1 shows the solubility of PAT6 and PAT-22 in various solvents. There are lots of good solvents which are common to both, but it is also clear that a nice distinction lies in the solubility. Figs. 13 and 14 show the absorption spectra of PAT6 and PAT22 in several solvents, respectively. It should be noted that the absorption spectral shape and maximum λmax are strongly dependent on the solvent. As already reported in our previous paper [17], the absorption spectra of the solutions of PATn can be divided into two groups; red and yellow forms. In the present case, PAT6 in anisole and PAT22 in anisole, dichloromethane and di-n-butylether showed red as color. This remarkable difference in color has been explained in terms of difference of effective conjugation length with solvent and also temperature due to the change of co-planarity of thiophene rings. However, it should be also noted that even in the same yellow group solutions, the absorption maximum depends on the solvent. Here, we would like to discuss this spectral difference in these yellow solutions. However, PAT6 in anisole was also included because the spectral width of the main peak is just similar to those of yellow solutions. The relatively small spectral differences with solvent are discussed in terms of Taft’s solvatochromic parameter π ∗ , which is a measure of solute chromophore–solvent interaction effects [22]. When the absorption maximum λmax is a linear function of π ∗ , the existence of interaction between solute chromophore and solvent is evidenced. The Taft plot (λmax vs. π ∗ ) for PAT6 is shown in Fig. 15. A reasonable correlation of λmax with π ∗ was confirmed in PAT6, as evident in Fig. 15. This fact may suggest that the interaction between solvent and polymer main chain exists in PATn with shorter alkyl chain. Solvation of the substituted alkyl group is considered, gradually, to be essential for
492
M. Onoda and K. Tada
Table 1 Solubility of PAT6 and PAT22 in various solvents Solvent
n-Heptane n-Hexane Di-n-butyl ether Carbon tetrachloride p-Xylene 1-Butanol Tetrahydropyran Trichloroethylene Toluene 1,4-Dioxane 2-Butanone Anisole Chloroform Dichloromethane N,N-Dimethylaniline
PAT6
PAT22
RT
50°C
RT*
RT
50°C
RT*
◦ ◦a ∗ ◦ ◦ ◦a ◦a ◦ ◦a ◦
◦ ◦ ∗ ◦ ◦ ◦ ◦ a,b ◦ ◦ ◦
◦ ◦ ∗ ◦ ◦ ◦ ◦a ◦ ◦ ◦
∗ ∗ ∗ ◦ ◦ ∗ ◦a ◦ ◦a ∗ ∗ ∗ ◦ ∗ ◦
◦ ◦ ◦ ◦ ◦ ∗ ◦ ◦ ◦ ∗ ∗ ◦ a,b ◦ ◦ ◦
◦a ◦a ◦a ◦ ◦ ∗ ◦ ◦ ◦ ∗ ∗ ◦a ◦ ◦ ◦
◦ = soluble; ∗ = non-soluble; = slightly soluble. RT = room temperature; RT* = the sample dissolved at 50°C was left at room temperature. a A little precipitation. b The sample dissolved uniformly at 80°C.
Fig. 13: Absorption spectra of PAT6 in various solvents.
Thermochromic behavior in novel conducting polymers at the solid–liquid phase transition
493
Fig. 14: Absorption spectra of PAT22 in various solvents.
Fig. 15: Taft plot for PAT6.
solubility. However, as evident from Table 1, n-hexane and n-heptane (n-hydrocarbons) are poor solvents for PAT6. This indicates that some other interaction also plays a role for solubility.
494
M. Onoda and K. Tada
Fig. 16: Taft plot for PAT22.
From a detailed examination of the data in Table 1 and Fig. 15, we can conclude that the solubility of PAT6 is not a simple function of either solvent dipolarity or polarizability. It is also clear that solubility is higher in electron-rich solvents which have non-bonded electron pairs and low in electron-poor solvents. For example, the halogenated compounds such as carbon tetrachloride (CCl4 ), chloroform (CHCl3 ), dichloromethane (CH2 Cl2 ), which have six or more non-bonded electron pairs per molecule, are good solvents, but hydrocarbons without such electrons are poor solvents. That is, solvent electron donation is one of the important parameters for solubility of PATn with relatively short alkyl chain length such as PAT6. On the other hand, the corresponding Taft plot for PAT22 is a scatter diagram as shown in Fig. 16. The large difference in these two Taft plots suggests that solute–solvent interaction effects for PAT6 and PAT22 are very different. In PAT22, the interaction of alkyl side chain and solvent molecule becomes more important. Therefore, PAT22 becomes soluble in linear hydrocarbon liquids. Also the interaction between polymer main chain and solvent becomes smaller in PAT22. In PAT22, long alkyl groups act as molecular sieve and the polymer main chain is shielding the interaction with solvent molecules. Only solvent molecules which are small in size and also can penetrate through the molecular sieve of the alkyl group can interact with the polymer main chain, resulting in an influence on solubility and spectral shift. Solubility of PATn is determined not only by alkyl group solvation but also solvent (donor)–polymer chain (acceptor) interaction. The spectral shift with solvent is determined by both change of co-planarity of thiophene rings and solvent–polymer chain interaction. That is, the solvatochromic shifts of λmax in the spectra of PAT6 and PAT22 solutions have been analyzed in terms of Taft linear solvation energy relationships.
Thermochromic behavior in novel conducting polymers at the solid–liquid phase transition
495
Fig. 17: Temperature dependence of the electrical conductivity in PAT12 under hydrostatic pressure.
4.5. Influence of hydrostatic pressure on electrical and optical properties of poly(3-alkylthiophene) Remarkable changes of electrical and optical properties have been observed in PATn with the hydrostatic pressure. We have already reported that the conductivity decreases step-wise at the phase transition from the solid to the liquid state during a heating stage at 1 bar (ambient pressure) [17,18]. It was also confirmed that during a cooling stage, the conductivity again increases step-wise at the phase transition temperature. However, there exists some hysteresis perhaps due to super-cooling in the latter case. That is the phase transition temperature with lower by 25°C in the cooling stage. As also evident from Fig. 17, under hydrostatic pressure similar temperature dependencies of the conductivity in the solid and the liquid state and a step-wise change at the phase transition were observed. It is clear from this figure that the phase transition temperature increases with hydrostatic pressure. In this case, measurements were carried out during the cooling stage. The pressure dependence of the phase transition temperature Tm is more clearly indicated in Fig. 18. Tm increases with pressure with a rate of 20°C/kbar. The step-wise change of conductivity at the phase transition can be explained by the increase of the bandgap and/or increase of carrier scattering. Electrical conductivity in the solid phase increased with increasing pressure, as shown in Fig. 17. This is more clearly indicated in Fig. 19. This increase of conductivity with pressure can be tentatively explained in terms of the decrease of carrier scattering under pressure, as is discussed later. On the other hand, in the liquid phase, electrical conductivity and its activation energy were not influenced remarkably. If the conduction in the liquid phase originated in ionic transport, the conductivity should be much suppressed due to the decreased free volume under hydrostatic pressure. The observed small pressure dependence of conductivity seems to support the interpretation of
496
M. Onoda and K. Tada
Fig. 18: Hydrostatic pressure dependence of the phase transition temperature (Tm ) in PAT22.
Fig. 19: Hydrostatic pressure dependence of electrical conductivity in PAT12 at 70°C.
electronic transport even in the liquid phase. This result also suggests that in the liquid state both the bandgap and the carrier scattering are not influenced markedly by pressures in this range. Fig. 20 shows the absorption spectral change of PAT12 film in the pressure cell with temperature at 1 bar. As already reported by us, in the pre-melting region, the blue shift of the absorption peak is more remarkable than the shift of the absorption edge [8]. This can be explained by the introduction of torsioning of the bonding between thiophene rings at several points on the polymer main chain, with which shorter conjugation chains are created and the number of original long conjugated chain decreases. At higher temperature, the number of torsioned bonds increases. This results in the shift of the absorption peak to the higher energy side (blue shift) at high temperature. We have also
Thermochromic behavior in novel conducting polymers at the solid–liquid phase transition
497
Fig. 20: Absorption spectral change of PAT12 measured at various temperatures under atmospheric pressure.
Fig. 21: Absorption spectra of PAT12 measured under various hydrostatic pressures at high temperature (116°C).
reported that the absorption spectrum in the liquid phase does not change remarkably with temperature [17,18]. Fig. 21 shows the effect of hydrostatic pressure on the absorption spectrum at 116°C. As evident from this figure, with applying hydrostatic pressure the absorption peak at high temperature which has blue shifted, again shifts to the lower energy side and comes
498
M. Onoda and K. Tada
Fig. 22: Absorption spectra of PAT12 measured under various hydrostatic pressures at room temperature (32°C).
back to that at room temperature. On the other hand, as shown in Fig. 22, the absorption peak at room temperature does not shift remarkably with application of hydrostatic pressure up to 3 kbar, in which range the remarkable change was observed at high temperature. Only at high pressures, the absorption peak shows slight red shift. These results indicate that the torsioning of bonds on the polymer main chain and the resultant decrease of co-planarity of conjugations can be suppressed by the applied hydrostatic pressure. That is, under high hydrostatic pressure the conjugation length again becomes long even at high temperature. The slight red shift of the absorption peak at room temperature when applying extremely high pressure may suggest that even at room temperature some torsioned bonds exist at 1 bar. The probability also exists that the one-dimensionality can be enhanced under high pressure. The enhancement of electrical conductivity by pressure shown in Figs. 17 and 19 can be interpreted as follows. The existence of various lengths of conjugated chains due to the introduction of torsioning of bonds may result in the deformation of the electronic band scheme. That is, fluctuation of conduction and valence bands and bandgaps can be created, which results in a strong carrier scattering in the pre-melting region. The suppression of torsioning of bonds at high pressure as observed by spectroscopic measurements can suppress the fluctuation of the energy band and carrier scattering. Therefore, electrical conductivity in the pre-melting region can be enhanced under hydrostatic pressure as observed. The shift of absorption spectrum with temperatures is much less in liquid phase as already mentioned. Therefore, the effect of hydrostatic pressure on electronic states should not be so remarkable compared with that in the pre-melting region. This may
Thermochromic behavior in novel conducting polymers at the solid–liquid phase transition
499
Fig. 23: Temperature dependence of photoluminescence in PAT12 at ambient pressure (1 bar) in a range covering both the solid and the liquid phase.
explain the small influence of hydrostatic pressure on the conductivity in the liquid phase, shown in Fig. 17. Just the similar effect of hydrostatic pressure on the melting point, electrical conductivity and absorption spectra were also observed in other fusible PATn. Fig. 23 shows the temperature dependence of the photoluminescence in PAT12 at an ambient pressure (1 bar) in a range covering both the solid and the liquid phase. As is evident from this figure, the luminescence peak shifted to higher energy with increasing temperature. The luminescence intensity in the solid phase increased with increasing temperature; however, at higher temperatures beyond the melting point (117°C) the photoluminescence intensity is again suppressed. A blue shift to the higher-energy side of the absorption peak had already been reported for PATn [8,23] and has been explained in terms of an increased bandgap at higher temperature due to the decrease of effective conjugation length on account of the introduction of torsions at bonds between neighboring thiophene rings. The increase of photoluminescence in the solid phase with temperature can also be explained by the introduction of torsion. That is, photoexcited species (such as electrons and holes, or negative and positive polarons) should be confined in the narrow excited region due to the decrease of effective conjugation length, resulting in the enhancement of recombination photoluminescence. On the other hand, the increased suppression of photoluminescence with increasing temperature in the liquid phase in which the torsion angle is considered to be markedly independent of temperature can be explained by the increased non-radiative decay of excited species. Fig. 24 shows emission spectra of PAT12 under various pressures at 150°C. It is clearly observable in this figure that the emission peak does not shift very noticeably but its intensity increases markedly with pressure up to 3 kbar. However, above 3 kbar
500
M. Onoda and K. Tada
Fig. 24: Photoluminescence spectra of PAT12 under various pressures at 150°C.
the photoluminescence intensity again decreases and the emission peak shifts to a lower energy. These results can be tentatively explained as follows by taking the change of molecular conformation with pressure into consideration. With increasing pressure at 150°C, a transition from the liquid phase into the solid phase was clearly observed at around 2.5 kbar [24]. This means that the enhancement of photoluminescence with pressure is a characteristic in the liquid phase. If the torsion angle decreased markedly with pressure in the liquid phase, the photoluminescence intensity should be enhanced contrary to observation. However, it should be noted that at higher pressure the difference of temperature (Td = 150 − Tm ) from the melting point Tm markedly decreased. Molecular motion is expected to become more violent at larger Td even if the mean torsion angle is kept nearly constant. In such a case, non-radiative decay of photoexcited species should be more effective. Therefore, with increasing pressure the photoluminescence can be enhanced. On the other hand, Fig. 25 shows the effect of hydrostatic pressure on the absorption spectrum of PAT12 at 32°C. Only at high pressures does the absorption peak show a slight red shift. We have also confirmed that the effective conjugation length becomes longer with hydrostatic pressure as shown in Fig. 25 in the solid phase. This is considered to be due to the decrease of the torsion angle. This is also consistent with the present result. That is, even at 150°C, above 2.5 kbar PAT12 is in the solid phase. Therefore, with increasing pressure, the effective conjugation length should become shorter, which restricts the escape of the excited species from the excited region, resulting in the enhancement of the recombination photoluminescence. The shift of the absorption peak to lower energy can also be satisfactorily explained by the increased conjugation length. Exactly the same behavior was also observed in other PATn with an alkyl chain
Thermochromic behavior in novel conducting polymers at the solid–liquid phase transition
501
Fig. 25: Absorption spectra of PAT12 measured under various hydrostatic pressures at 45°C.
Fig. 26: Optical absorption spectra of (a) ROPPV16 and (b) PAT16 at various temperatures.
longer than the butyl unit, although the temperature at which such effects become noticeable increases in the sample with shorter alkyl chains. The changes of the torsion angle between neighboring thiophene rings are expected to be triggered by the conformation change of the side alkyl group such as a gauche–trans change. 4.6. Thermochromism in poly( p-phenylene vinylene) derivatives Fig. 26a shows the optical absorption spectrum of ROPPV16 measured at various temperatures. The absorption peak and the absorption edge shift slowly to higher energy
502
M. Onoda and K. Tada
Fig. 27: Temperature dependence of the bandgap energies of ROPPVs and PATs.
with increasing temperature. As a result of this spectral change, the color of the sample varies from orange to yellow, i.e. thermochromism occurs. This temperature dependence of the optical absorption spectrum of ROPPV16 is different from that of PAT16 shown in Fig. 26b, in particular, at the shift of the absorption edge. The bandgap energy of PAT16 increases suddenly from about 2.0 eV to about 2.4 eV around the melting point of 160°C. On the other hand, the bandgap energy of ROPPV16 increases gradually from about 2.2 eV at room temperature to about 2.4 eV at 200°C. This difference of bandgap energy shift is more clearly indicated in Fig. 27, which shows the temperature dependence of bandgap energies of various ROPPVs and PATs. The side chain length dependence of the increase of bandgap energy is clear in PATs, but not in ROPPVs. In other words, the bandgap energies of PATs change suddenly at the melting point, which is lowered with longer alkyl chains. This clearly shows a correlation between the thermochromism in PATs with the solid–liquid phase transition. On the other hand, the increase of bandgap energies of all ROPPVs shows a similar tendency, and the side chain length dependence is not observed distinctly. This means that although the bandgap energies of ROPPVs depend on the temperature, they are not correlated with the solid–liquid phase transition. The mechanism of thermochromism in a conducting polymer was proposed to be as follows [25]. With increasing temperature, the side chain changes its conformation due to the trans–gauche transition and becomes bulky. The steric hindrance between the bulky side chain and the polymer main chain forces the main chain to twist. Therefore, the π-electron systems in the main chain of the polymer tend to lose their planarity with increasing temperature, resulting in deceased effective π-conjugation length and increased bandgap energy. The difference in the thermochromic behavior between ROPPVs and PATs can be tentatively interpreted in terms of the main chain structure which is schematically shown in Fig. 28 [10]. The phenyl rings composing the PPV-type main chain can rotate freely with respect to the main chain without significant deformation of the main chain, as
Thermochromic behavior in novel conducting polymers at the solid–liquid phase transition
503
Fig. 28: Relationship between polymer main chain structure and twist.
they are connected at para-positions. Therefore, the planarity of π-electron systems can be gradually decreased with increasing temperature. On the other hand, the thiophene rings in PATs cannot rotate without causing significant deformation of the main chain, because the bonds connecting a thiophene ring and the main chain are not in a straight line. This indicates that the planarity of the π-electron systems in PATs changes not in the solid state but in the liquid state. Therefore, the π-electron systems in PATs lose their planarity not gradually with increasing temperature at the solid state, but suddenly at the solid–liquid phase transition. Fig. 29a shows the PL spectrum of ROPPV16 measured at various temperatures. The PL peak shifts slowly to higher energy corresponding to the increase of bandgap energy with increasing temperature. However, the PL intensity decreases monotonically with increasing temperature. This is opposite to the tendency of PATs shown in Fig. 29b, which indicates that the PL intensity of PAT16 increases with increasing temperature. The origin of the increase of PL intensity of PATs with increasing temperature in the solid state is explained as follows [8,25]. With the increase in temperature, the polymer main chain twists and the effective π-conjugation length decreases, resulting in enhancement of the confinement effect of the radiative-excited species such as excitons and/or exciton polarons. Therefore, the probability of radiative recombination is increased and the PL intensity is enhanced. However, the decrease of PL intensities of ROPPVs with increasing temperature cannot be explained by this mechanism. As schematically shown in Fig. 30a, the distance between polymer main chains is enlarged by the side chains. The side chains take trans conformation at lower temperature, and gauche conformation near and above the melting point. The side chain is more bulky and shorter when it is in gauche than in trans. Therefore, with increasing
504
M. Onoda and K. Tada
Fig. 29: Photoluminescence spectra of (a) ROPPV16 and (b) PAT16 at various temperatures.
Fig. 30: Separation of the excited species due to the reduction of the distance between polymer main chains.
Thermochromic behavior in novel conducting polymers at the solid–liquid phase transition
505
Fig. 31: Enhancement of the overlap of π-electrons due to the polymer main chain twist.
temperature, the distance between polymer main chains becomes shorter, as shown in Fig. 30b. In this case, the dissociation probability of the excited species such as excitons and/or exciton polarons through interchain charge transfer increases and the probability of radiative recombination decreases. On the other hand, the twist of polymer main chain would enhance the overlap of π-electrons, as schematically shown in Fig. 31. Assume that there is a pair of neighboring polymer main chains whose phenyl rings are perpendicular to each other and therefore the π-electrons do not overlap each other, as shown in Fig. 31a. In such a case, interchain charge transfer does not occur. However, with increasing temperature the polymer main chain starts to twist and the π-electron orbitals of the pair become more parallel to each other, as shown in Fig. 31b. Consequently, the dissociation probability of the excited species through the interchain charge transfer increases and the probability of radiative recombination decreases. Thermally excited phonons would also increase the exciton dissociation probability. In the above discussion, we only concentrated on the change of interaction between the main chains. However, the interaction between the main chain and the oxygen atoms of alkoxy side chains would also play an important role, because the interaction essentially contributes to the electronic energy structure of ROPPV [1].
506
M. Onoda and K. Tada
Fig. 32: Temperature dependence of ionization potentials of ROPPVs and PATs.
Fig. 33: Schematic energy diagrams of ROPPV16 and PAT16 in the solid and the liquid state.
Fig. 32 shows the temperature dependence of ionization potentials of ROPPVs and PATs. The ionization potentials of PATs change steeply from 4.7 eV to about 5.4 eV near the melting point. Thus, the electronic states of PATs change significantly with the solid–liquid phase transition. However, in ROPPVs the ionization potentials decrease slightly with increasing temperature and the degree of decrease becomes larger with longer alkoxy chains. For example, the ionization potential of ROPPV16 decreases from about 5.4 eV at room temperature to about 5.1 eV at 200°C, while the ionization potential of ROPPV5 is maintained at about 5.2 eV from room temperature to 200°C. The origin of the temperature dependence of the ionization potentials of ROPPVs is not clear at this stage. From the results mentioned above, the schematic energy diagrams of ROPPV16 and PAT16 are estimated as shown in Fig. 33. In ROPPV16 with increasing temperature from room temperature to 200°C the bandgap energy increases gradually from about 2.2
Thermochromic behavior in novel conducting polymers at the solid–liquid phase transition
507
eV to about 2.4 eV, and the ionization potential decreases from about 5.4 eV to about 5.1 eV. The electronic state of PAT16 changes markedly at the melting point. That is, the bandgap energy increases from about 2.0 eV to about 2.4 eV, and the ionization potential increases from about 4.7 eV to about 5.4 eV.
5. Conclusion A reversible absorption spectrum change of poly(3-alkylthiophene), PAT, films with temperature and anomalous temperature dependence of photoluminescence have been observed. By the application of hydrostatic pressure the spectrum at high temperature turns into that of low temperature. Drastic change of electrical conductivity and optical properties have been also found at the melting point. These results were explained in terms of modification of polymer conformation and conjugation system at high temperature, i.e. introduction of torsioning of some bonds between thiophene rings and its suppression under hydrostatic pressure. To confirm this interpretation, time dependence of the photoluminescence in the picosecond range, Raman and infrared spectra measurements are necessary. The solvatochromic shifts of λmax of spectra of poly(|3-hexylthiophene) and poly(3docosylthiophene) solutions have been analyzed in terms of Taft linear solvation energy relationships. Solubility of PAT is determined not only by alkyl group solvation but also solvent (donor)–polymer chain (acceptor) interaction. The spectral shift with solvent is determined by both change of co-planarity of thiophene rings and solvent–polymer chain interaction. We have observed the thermochromic behaviors of poly(2,5-dialkoxy- p-phenylene vinylene)s and compared the results with the properties of poly(3-alkylthiophene)s. The results are summarized as follows. The temperature dependencies of the optical absorption spectra of ROPPVs were different from those of PATs, in particular, at the shift of the absorption edge. In other words, a difference in temperature dependence of the bandgap energy was observed. This difference in the thermochromic behavior can be tentatively interpreted in terms of the main chain structure. The PL intensities of ROPPVs decreased monotonically with increasing temperature. This was opposite to the tendencies of PATs. As the mechanism of the decrease of PL intensities of ROPPVs, we speculate on the separation of the excited species due to the reduction of the distance between polymer main chains and the enhancement of the overlap of π-electrons with the twist of polymer main chain, in addition to thermally excited phonons. The ionization potentials of ROPPVs decrease slightly with increasing temperature and the degree of decrease becomes larger with longer alkoxy chains, although the origin of this tendency is not clear at this stage.
References 1. 2.
I. Murase, T. Ohnishi, T. Noguchi, and M. Hirooka, Polym. Commun. 25, 327 (1984). R. Sugimoto, S. Takeda, H.B. Gu, and K. Yoshino, Chem. Express 1, 635 (1986).
508 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25.
M. Onoda and K. Tada M. Sato, S. Tanaka, and K. Kaeriyama, J. Chem. Soc. Chem. Commun., 635 (1986). D.L. Elsenbaumer, K.Y. Jen, and R. Oboodi, Synth. Met. 15, 169 (1986). K. Yoshino, S. Nakajima, and R. Sugimoto, Jpn. J. Appl. Phys. 26, L1038 (1987). K. Yoshino, S. Nakajima, M. Fijii, and R. Sugimoto, Polym. Commun. 28, 309 (1987). D. Braun and A.J. Heeger, Appl. Phys. Lett. 58, 1982 (1991). K. Yoshino, S. Nakajima, D.H. Park, and R. Sugimoto, Jpn. J. Appl. Phys. 27, L716 (1988). M. Hamaguchi and K. Yoshino, Jpn. J. Appl. Phys. 33, L1478 (1994). K. Tada, M. Onoda, and K. Yoshino, J. Phys. D Appl. Phys. 30, 2063 (1997). I. Murase, T. Ohnishi, T. Noguchi, and M. Hirooka, Polym. Commun. 26, 362 (1985). K. Yoshino and M. Onoda, Kobunshi Electronics (Polymer Electronics), (Corona, Tokyo, 1996), pp. 18 (in Japanese). K. Yoshino, S. Nakajima, H.B. Gu, and R. Sugimoto, Jpn. J. Appl. Phys. 26, L2046 (1987). K. Yoshino and M. Onoda, Kobunshi Electronics (Polymer Electronics), (Corona, Tokyo, 1996), pp. 117 (in Japanese). M. Onoda, Y. Manda, M. Yokoyama, R. Sugimoto, and K. Yoshino, J. Phys. Condens. Matter 1, 3859 (1989). K. Kaneto, F. Uesugi, and K. Yoshino, J. Phys. Soc. Jpn. 57, 747 (1988). K. Yoshino, D.H. Park, B.K. Park, M. Onoda, and R. Sugimoto, Jpn. J. Appl. Phys. 27, L1612 (1988). K. Yoshino, D.H. Park, B.K. Park, M. Onoda, and R. Sugimoto, Solid State Commun. 67, 1119 (1988). K. Yoshino, S. Hayashi, Y. Inuishi, H. Kato, and Y. Watanabe, Jpn. J. Appl. Phys. 21, L653 (1982). K. Yoshino, S. Hayashi, Y. Inuishi, K. Hattori, and Y. Watanabe, Solid State Commun. 46, 583 (1983). K. Yoshino and M. Onoda, Chemistry 44, 1989 (in Japanese). M.J. Kamlet, J.L. Abboud, M.H. Abraham, and R.W. Taft, J. Org. Chem. 48, 2877 (1983). K. Yoshino, S. Nakajima, D.H. Park, and R. Sugimoto, Jpn. J. Appl. Phys. 27, L454 (1988). K. Yoshino, K. Nakao, M. Onoda, and R. Sugimoto, Solid State Commun. 68, 513 (1988). K. Yoshino, Y. Manda, K. Sawada, M. Onoda, and R. Sugimoto, Solid State Commun. 69, 143 (1989).
Subject Index 1×1 canted structure, 447 Adenine, 442 Adsorption, 50, 53 Ag(111) film, 437 Alkanethiol, 443 Alkylsiloxane monolayer, 193 Ambipolar diffusion, 86, 92, 93 Amplified spontaneous emission (ASE), 43, 44 Anatase, 90 Antibiotic peptide, 217–220 Atom extraction, 450 Atom manipulation, 449 Atomic displacement, 439 Atomic force control, 447 Atomic force microscopy (AFM), 200, 204, 214, 430 Atomic imaging of contact potential difference and capacitance, 449 Atomic relaxation, 442 Atomic stress, 442 Au(111) surface, 50–55, 57, 60 Azobenzene, 378 Azobenzene monolayer, 378, 382 Bond order, 447 Breath-figure, 404 Brookite, 90 C60 , 101 Calixarene, 6, 7 Carbon dioxide, 401 Carbon nanotubes, 400 Cell membranes on semiconductors, 215 Cell surface glycocalix, 193, 201, 209, 215, 226 Charge resonance, 113 Chemical “switching” of cell/surface interactions, 207 Chemical bonding interaction, 439
Composite, 402, 412, 416 Computer simulation, 144, 145, 149 Conducting polymer, 455, 456, 479, 482, 488, 502 Conduction, 134–136, 139, 140, 145 Conduction mechanism, 135, 136, 139, 154 Conduction model, 135, 136, 140 Constant-excitation mode, 434 Contact point, 434 Contact potential difference, 449 Control of atomic force and atom position, 447 Coulomb island, 32, 37 CPD image of the point charge, 449 Critical point drying, 407 Cu(111), 442 Dangling bond, 438 Decay length, 433 Dendrimer, 41, 42 Dendrite molecules, 32, 36, 38 Dendrite polymer, 33 Deuterium NMR, 313–316, 321, 324, 325, 329, 333, 336, 337, 342, 346 Dimer structure, 439 Dipole, 264 Director distribution, 314, 316, 318–322, 325–327, 329, 331, 344, 346 Director dynamics, 314, 315, 331, 335, 342, 346, 347 Discrimination of atom species, 445 Discrimination of atomic force mechanisms and atom species, 444 DNA, 443 Dye-sensitized solar cells, 83 EFM/NC-AFM, 449 Elastomeric membranes, 183, 186, 187 Electrical transport, 458 Electron-beam (EB), 12 Electron beam (EB) lithography, 4
510 Electron beam (EB) nanolithography, 4 Electron hopping, 97 Electron spin resonance (ESR) measurements, 413 Electron-transfer, 85 Electron transport, 86 Electrostatic force imaging, 448 Electrostatic force microscope (EFM), 449 Extracellular matrices (ECMs), 201 F-doped SnO2 , 88 Fabrication process, 13 Fick’s diffusion model, 88 Field emission, 400 Fill factor, 84 Fluorine-doped TiO2 , 95 Fluorine doping, 95 Focused-ion-beam (FIB), 12 Focused-ion-beam chemical vapor deposition (FIB-CVD), 13, 14 Fowler–Nordheim behavior, 404 Frequency shift of mechanical resonant oscillation, 431 Fullerene (C60 ), 100 ,412, 413, 422, 426 Fully guided mode, 274, 275, 288, 289 Fully leaky guided mode (FLGM), 274, 293, 301–305, 306, 307–309 Fully leaky waveguide, 276, 309 Functions of NC-AFM, 442 GaAs(110) cleaved surfaces, 442 Gallium arsenide (GaAs), 194 Glycocalix model, 209, 214 Glycolipid monolayers, 209, 210 Grating-coupling, 283–286, 310 Guided mode technique, 273, 274, 279, 289, 290, 301, 306–308, 310 Guided wave mode, 117–119, 124 Half-leaky guided mode (HLGM), 274, 293– 301, 299, 300 Half-leaky waveguide, 275, 309 Helical peptide, 253, 254, 263 Helical peptide on gold, 264 Helical peptide SAM (self-assembled monolayer), 256, 265 Helical peptides on gold, 257 Helical peptides on subphase, 258 Helical peptides on water, 258
Subject Index High performance NC-AFM, 436, 442 High performance STM, 436 Hole transporting material, 168, 172 Home-built NC-AFM, 436 Honeycomb nano-structure, 400 Hopping conduction, 137, 139 Hopping conduction model, 153 HSQ, 10 Imidazolium, 91, 97 In situ field-effect measurement, 157, 158, 162, 164, 176 Incident photon to current efficiency (IPCE), 85 Indium arsenide quantum dots (InAs QDs), 198 Information storage, 21–24 InP(110) cleaved surface, 431 Interfacial stress rheometer (ISR), 214 Iodine exchange (Grotthuss-type) mechanism, 97 Ionic head groups, 183–185, 187 Ionic liquids, 98 Kelvin probe force microscopy (KFM), 449 KFM/NC-AFM, 449 Langmuir–Blodgett (LB) film, 23, 24 Langmuir–Blodgett technique, 183 Langmuir film, 352 Langmuir-monolayer, 353, 366 Laser dye, 41, 43, 47 Laser emission, 41, 43–45 Lateral manipulation, 451 Lateral resolution of NC-AFM, 433, 438 Light-induced electron spin resonance (LESR), 412–414 Limiting molar conductivity, 97 Lipid bilayer, 207, 209, 215, 218, 220 Lipid membrane, 202, 216–218, 220 Lipid monolayer, 215, 223 Liquid crystal, 313–315, 347 Liquid crystal layer, 279, 280, 286, 288, 289, 291–294, 304, 306, 307 Living systems, 191, 192, 196 Lone pair, 447 Long-range force, 444 Low-molecular-weight gelators, 96
Subject Index Low-temperature NC-AFM (LT-NC-AFM), 450 Maxwell displacement current (MDC), 353, 355, 356, 359, 362, 365 Mechanical method, 449 Mechanical resonant oscillation frequency, 432 Membrane charge sensor, 221, 222 Mesoporous, 85 Metal oxide surfaces, 442 Metal surfaces, 442 Micro-system parts, 16 Missing-dimer defects, 439 Model cell membranes, 192, 193, 199, 203, 205, 215 Molecular assembly, 253, 254 Molecular devices, 32 Molecular orientation, 258 Molecular photonics, 105–107 Molecular reorientation, 379 Molecular systems, 32 Molten salts, 98 Monolayer, 353, 356, 357, 359, 364, 365 Morphology of glycolipids, 211 n+ -GaAs(110) cleaved surface, 449 Nanoimprint-lithography (NIL), 6–9 Nanometer, 65, 66, 80 Nanophase-separated structure, 239, 242 Nanostructure, 49 Native cell membranes, 193, 203, 205, 215, 218, 223, 226 Nernst equation, 92 Noncontact atomic force microscopy (NCAFM), 429 Noncontact region, 434 Open-circuit photovoltage, 84 Optical guided mode, 277 Optical second harmonic generation (SHG), 353, 355, 357, 359, 360, 362, 364, 365 Optical waveguide, 273, 274, 279, 282, 286, 287 Optoelectronic devices, 65, 76 Optoelectronics, 65, 76 Organic electronic device, 158 Organic light-emitting diode (OLED), 133– 136, 139, 144, 145, 153
511 Organic semiconductor films, 157, 167 Organic semiconductor materials, 159 Organic semiconductors, 158 Organic thin film, 159 Orientation of helical peptides, 254 Orientational control, 378 Orientational order, 381, 382, 384 Oscillation amplitude, 432 Oxygen adsorbed Si(111)7×7, 443 Oxygen vacancies, 95 Passivation, 194, 196, 197 Photoalignment, 378, 379, 382, 384 Photoalignment procedure, 379 Photocarrier generation, 455, 457–459, 461– 463, 465, 467, 469, 471, 473, 475–477 Photoconversion efficiency, 84 Photodiodes, 66–68, 72–76, 79, 80 Photoexcitation, 412, 420, 426 Photoinduced anisotropy, 383 Photoinduced electrochromism, 106, 109, 111 Photoinduced mass migration, 378, 385 Photoinduced mass transfer, 391, 393 Photoinduced migration, 393 Photoinduced orientation, 380 Photoinduced surface relief grating, 386 Photon localization, 47 Photoorientation, 383 Photoorientation behavior, 382 Photovoltaic effect, 457, 460, 474, 476, 477 Photovoltaic properties, 458 Physical interaction, 439 Polarons, 411, 412, 415, 416, 420, 422, 425, 426 Poly(3-alkylthiophene) (PAT), 412, 414, 455 Poly(3-hexylthiophene) (PAT6), 415, 418 Poly(3-octadecylthiophene) (PAT18), 418, 424 Poly(3-octylthiophene) (PAT8), 413, 426 Poly(methylsiloxane), 100 Poly(phenylene oxide-co-2-allylphenylene oxide), 100 Polyimide, 23, 24, 32 Polyiodide, 96 Polymer, 65–80 Polymer brushes, 205, 207 Polypeptide-based LB films, 233, 236, 242, 244, 249
512 Polypyrrole (PPy), 100, 101 Polysaccharide films, 202–205, 226 Polysilane, 378 Polysilane chain, 378 Polysilane film, 378, 393 Prism-coupling, 282, 283, 286, 287, 303 Pull-up process, 451 Push-out process, 451 Radius of tip curvature, 433 Rebonded dimer structure, 441 Rebonded structure, 441 Reflection absorption spectroscopy (RAS), 257, 258, 261, 264 Regiorandom, 470 Regioregularity, 457, 458, 477 Room temperature nanoimprint-lithography (RT-NIL), 7, 8 Roughness factor, 85 Rutile, 91 Scanning probe microscope (SPM), 21 Scanning tunneling microscope (STM), 429 Schottky junction, 457, 462, 466, 468–470, 476 Self-assembled monolayer (SAM), 193, 194, 256, 258, 265, 443 Self-organized two-dimensional patterning, 235, 236, 242 Semiconductor surfaces, 429 Semiconductor-oxide nanotubes, 400 Short circuit photocurrent density, 84 Short-range force, 444 Si(100)1×1:2D dideuteride, 447 Si(100)1×1:2H dihydride surface, 447 Si(100)2×1 clean surface, 438 Si(100)2×1:H monohydride surface, 438 √ √ Si(111)5 3 × 5 3–Sb surface, 447 Si(111)7×7, √ √ 431 Si(111) 3 × 3–Ag surface, 443 Signal-to-noise ratio, 432 Single electron tunneling (SET), 31, 32, 35, 38 Single electron tunneling characteristic, 34, 38 Sol–gel, 401 Solid–liquid phase transition, 483, 485, 495, 502, 506
Subject Index Solvatochromism, 491 Space charge limited current (SCLC), 136, 140, 145 Spatial light modulation, 107, 116, 117 Spatial resolution of NC-AFM, 433 Spectroscopy, 362, 365 Stimulus–response coupling, 234, 249, 252 Structural color formation, 233, 236, 242, 244, 249 Supramolecular structure, 49, 50 Surface anchoring, 314, 315, 318–320, 322, 323, 326, 327, 331, 335, 345, 347 Surface anchoring energy, 320 Surface engineering of GaAs, 199 Surface relief grating, 385 Surface relief structure, 391 Surface states, 95 Surface tension, 405 Synthetic polymer brushes, 205 Template, 401 Thermochromism, 480, 485, 488, 489, 502 Thin-film transistor (TFT), 159, 160, 168 Three-dimensional mapping of atomic force, 443 Three-dimensional nanostructure, 12, 14 Thymine, 443 TiO2 , 83, 442 Tip–sample distance, 433 Tip-induced 1×1 novel structure, 448 Tip-induced self-organized 2×1 novel structure, 448 Transparent conducting glass, 88 Trapping model, 86 Vacuum evaporation, 158 Van der Waals force, 449 Vertical manipulation, 450 Vertical resolution of NC-AFM, 433, 437 Viscoelasticity of oligosaccharides, 214 X-ray photoelectron spectroscopy (XPS), 196, 197 Young modulus, 404 z-derivative image of capacitance, 449 ZEP520, 4–6