nanotechnology and nano-interface controlled electronic devices

529

Upload: rafik-mnassri

Post on 10-Mar-2015

219 views

Category:

Documents


5 download

TRANSCRIPT

Page 1: Nanotechnology and Nano-Interface Controlled Electronic Devices
Page 2: Nanotechnology and Nano-Interface Controlled Electronic Devices

Nanotechnology and Nano-InterfaceControlled Electronic Devices

Page 3: Nanotechnology and Nano-Interface Controlled Electronic Devices

This Page Intentionally Left Blank

Page 4: Nanotechnology and Nano-Interface Controlled Electronic Devices

Nanotechnology and Nano-InterfaceControlled Electronic Devices

Edited by

M. IwamotoK. KanetoS. Mashiko

2003ELSEVIERAmsterdam – Boston – London – New York – Oxford – ParisSan Diego – San Francisco – Singapore – Sydney – Tokyo

Page 5: Nanotechnology and Nano-Interface Controlled Electronic Devices

ELSEVIER SCIENCE B.V.Sara Burgerhartstraat 25P.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:

PhotocopyingSingle photocopies of single chapters may be made for personal use as allowed by national copyright laws. Permis-sion of the Publisher and payment of a fee is required for all other photocopying, including multiple or systematiccopying, copying for advertising or promotional purposes, resale, and all forms of document delivery. Special ratesare 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 completeyour 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., 222Rosewood Drive, Danvers, MA 01923, USA; phone: (+1) 978 7508400, fax: (+1) 978 7504744, and in the UKthrough the Copyright Licensing Agency Rapid Clearance Service (CLARCS), 90 Tottenham Court Road, LondonW1P 0LP, UK; phone: (+44) 207 631 5555, fax: (+44) 207 631 5500. Other countries may have a local reprographicrights agency for payments.

Derivative WorksTables of contents may be reproduced for internal circulation, but permission of Elsevier Science is required forresale or distribution of such material.Permission of the Publisher is required for all other derivative works, including compilations and translations.

Electronic Storage or UsagePermission of the Publisher is required to store or use electronically any material contained in this work, includingany 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 anyform or by any means, electronic, mechanical, photocopying, recording or otherwise, without prior written permis-sion of the Publisher.Address permissions requests to: Elsevier’s Science & Technology Rights Department, at the phone, fax ande-mail addresses noted above.

NoticeNo responsibility is assumed by the Publisher for any injury and/or damage to persons or property as a matter ofproducts liability, negligence or otherwise, or from any use or operation of any methods, products, instructions orideas contained in the material herein. Because of rapid advances in the medical sciences, in particular, independentverification of diagnoses and drugs dosages should be made.

First edition 2003

Library of Congress Cataloging in Publication DataA catalog record from the Library of Congress has been applied for.

British Library Cataloguing in Publication DataA 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.

Page 6: Nanotechnology and Nano-Interface Controlled Electronic Devices

Contents

Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix

List of Contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi

A. Single Molecular Electronics and Photonics

1. Nanostructure fabrication using electron and ion beamsShinji Matsui (Himeji Institute of Technology) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2. Information storage using a scanning probeKiyoshi Takimoto (Canon Co. Ltd.) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

3. Single electron tunneling organic devicesTohru Kubota, Shiyoshi Yokoyama, Tatsuo Nakahama, Shinro Mashiko(Communications Research Laboratory), Yutaka Noguchi, and MitsumasaIwamoto (Tokyo Institute of Technology) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

4. Spatial light confinement and laser emission from a gain medium containingdendrimerShiyoshi Yokoyama and Shinro Mashiko (Communications Research Labo-ratory) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

5. Control of molecular selective-assembling on metal surfaceTakashi Yokoyama (National Institute for Materials Science), ToshiyaKamikado, Shiyoshi Yokoyama, Yoshishige Okuno, and Shinro Mashiko(Communications Research Laboratory) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49

B. NICE Devices

6. Polymer optoelectronics – towards nanometer dimensionsOlle Inganäs and Fengling Zhang (Linköping University) . . . . . . . . . . . . . . . . . . . 65

7. Control of charge transfer and interface structures in nano-structured dye-sensitized solar cellsShozo Yanagida, Takayuki Kitamura, and Yuji Wada (Osaka University) . . . . 83

8. Materials and devices for ultrafast molecular photonicsToshihiko Nagamura (Shizuoka University) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105

Page 7: Nanotechnology and Nano-Interface Controlled Electronic Devices

vi Contents

9. Carrier transport behavior in OLEDTatsuo Mori and Teruyoshi Mizutani (Nagoya University) . . . . . . . . . . . . . . . . . . 133

10. Electrical characterization of organic semiconductor films by in situ field-effect measurementsKazuhiro Kudo (Chiba University) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157

C. Smart Soft Materials

11. Introducing ruber into the Langmuir–Blodgett techniqueH. Xu, R. Heger, F. Mallwitz, M. Blankenhagel, C. Peyratout, and WernerA. Goedel (University of Ulm and Max-Planck-Institut für Kolloid- &Grenzflächenforschung) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183

12. Design of functional interface between living systems and semiconductornano-structuresMotomu Tanaka (Technische Universität München) . . . . . . . . . . . . . . . . . . . . . . . . 191

13. Structural color forming system composed of polypeptide-based LB filmsTakatoshi Kinoshita (Nagoya Institute of Technology), Shujiro Hayashi,Yoshiyuki Yokogawa (National Institute of Advanced Industrial Science andTechnology), and Shintaro Washizu (Fuji Photo Film Co. Ltd.) . . . . . . . . . . . . . . 233

14. Generation of a strong dipole layer and its function by using helical peptidemolecular assembliesShunsaku Kimura, Tomoyuki Morita, and Kazuya Kitagawa (Kyoto Univer-sity) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253

D. Interfacial Dynamic Technology

15. Guided mode studies of liquid crystal layersFuzi Yang (Tsinghua University) and J.R. Sambles (University of Exeter) . . . 271

16. Explanation of the static and dynamic director orientation in thin nematicliquid crystal films using deuterium NMR spectroscopyAkihiko Sugimura (Osaka Sangyo University) and Geoffrey R. Luckhurst(University of Southampton) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313

17. MDC–SHG spectroscopy of organic monolayer filmAtsushi 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 materialsTakahiro Seki (Tokyo Institute of Technology) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 377

Page 8: Nanotechnology and Nano-Interface Controlled Electronic Devices

Contents vii

E. Fabrication and Characterization Technology

19. Solvent-induced morphology in nano-structuresBin Cheng, Hongtao Cui, Brian R. Stoner, and Edward T. Samulski (Univer-sity of North Carolina at Chapel Hill) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 399

20. Polarons in conjugated polymer and its composite with fullereneKazuhiro Marumoto and Shin-ichi Kuroda (Nagoya University) . . . . . . . . . . . . . 411

21. Characterization of semiconductor surfaces with noncontact atomic forcemicroscopySeizo Morita and Yasuhiro Sugawara (Osaka University) . . . . . . . . . . . . . . . . . . . 429

22. Transport and photocarrier generation in poly(3-alkylthiophene) and metaljunctionsKeiichi Kaneto, Koichi Rikitake, and Wataru Takashima (Kyushu Instituteof Technology) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 455

23. Thermochromic behavior in novel conducting polymers at the solid–liquidphase transitionMitsuyoshi Onoda and Kazuya Tada (Himeji Institute of Technology) . . . . . . . 479

Subject Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 509

Page 9: Nanotechnology and Nano-Interface Controlled Electronic Devices

This Page Intentionally Left Blank

Page 10: Nanotechnology and Nano-Interface Controlled Electronic Devices

Preface

The International Workshop on Nanotechnology and NICE Devices (IWNND) washeld on March 19 and 20, 2002 at Nagoya Congress Center, Nagoya, Japan. Thisinternational workshop was organized as one of the important events of the NagoyaNanotechnology International Forum 2002 (NNIF2002), from March 18 through 20,2002. Needless to say, Nano-Technology and Materials, Bio-Technology, InformationTechnology and others are recognized as the important key technologies in the 21stcentury, and these techniques are believed to make a significant contribution to our dailyand fruitful life. The purpose of the IWNND workshop was to provide opportunitiesto 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 distinguishedscientists from overseas and Japan, working in universities, laboratories and companies.All the participants could benefit from the presentations and discussions. This book is acollection of papers based on the invited talks at this workshop.

As you know, many organic materials that are interesting in terms of electronicshave been synthesized and discovered during the past few decades. We can see oneof the most remarkable achievements in the Nobel Prize for Chemistry 2000 awardedto Heeger, MacDiarmid and Shirakawa, for their contribution to the discovery anddevelopment of conducting polymers. In the hope of observing novel and usefulelectrical and optical properties, many investigations have been carried out to buildup organic devices. Plastic solar cells, flexible-type field effect transistors (FETs),electroluminescent (EL) devices and so on have been developed, along with thedevelopment of new organic materials. Insightful ideas have also been proposed toopen-up new methods in electronics. However these are no longer sufficient. One mustdevelop techniques to catch the specific properties of organic materials, molecules,biological materials and so on, and then to create a novel method to benefit fromthe specific functions of these materials in electronic devices. One way would beto use nano-interfaces and related nanometric interfacial phenomena, although ourunderstanding of nano-interfacial phenomena is far way from the viewpoint of scienceand technology. In the IWNND workshop, five topics, i.e., Single Molecular Electronicsand Photonics, NICE Devices, Smart Soft Materials, Interfacial Dynamic Technology,and Fabrication and Characterization Technology, are selected in association with nano-interfacial phenomena and their electronic applications. This book covers these fivetopics. It will be very much appreciated to hear comments, critiques, and suggestionsfrom any readers for the benefit of Nano-Interface Controlled Electronic devices.

Finally, we would like to express our sincere thanks to the organizing members ofthe IWNND, the NNIF2002, the staffs of Nano-Device Group, National Institute for

Page 11: Nanotechnology and Nano-Interface Controlled Electronic Devices

x Preface

Materials Science (NIMS), and the staffs of Nanotechnology Section, Kansai AdvancedResearch Center (KARC), Communications Research Laboratory (CRL) for their kindhelp and enthusiastic effort. Thanks are also given to the City of Nagoya, NIMS andKARC for their support.

MITSUMASA IWAMOTO

KEIICHI KANETO

SHIRO MASHIKO

Editors

Page 12: Nanotechnology and Nano-Interface Controlled Electronic Devices

List of Contributors

M. BlankenhagelMax-Planck-Institut für Kolloid- & Grenzflächenforschung, Berlin, Germany

Bin ChengDepartment of Chemistry, University of North Carolina at Chapel Hill, Chapel Hill,NC 27599-3290, USA

Hongtao CuiDepartment of Physics and Astronomy, University of North Carolina at Chapel Hill,Chapel Hill, NC 27599-3255, USA

Werner A. GoedelOrganic and Macromolecular Chemistry, OC3, University of Ulm, Albert-Einstein-Allee 11, D-89069 Ulm, Germany

Shujiro HayashiNational Institute of Advanced Industrial Science and Technology, Hirate-cho 1-1,Kita-ku, Nagoya 462-8510, Japan

R. HegerMax-Planck-Institut für Kolloid- & Grenzflächenforschung, Berlin, Germany

Ryouhei HiyoshiDepartment of Physical Electronics, Tokyo Institute of Technology, 2-12-1 O-oka-yama, Meguro-ku, Tokyo 152-8552, Japan

Olle InganäsBiomolecular and organic electronics, Department of Physics and MeasurementTechnology, Linköping University, SE - 581 83 Linköping, Sweden

Mitsumasa IwamotoDepartment of Physical Electronics, Tokyo Institute of Technology, 2-12-1 O-oka-yama, Meguro-ku, Tokyo, 152-8552, Japan

Toshiya KamikadoCommunications Research Laboratory, 588-1 Iwaoka, Nishi-ku, Kobe 651-2401,Japan

Page 13: Nanotechnology and Nano-Interface Controlled Electronic Devices

xii List of Contributors

Keiichi KanetoLife Science & Systems Engineering, Kyusyu Institute of Technology, Iizuka,Fukuoka, 820-8502, Japan

Shunsaku KimuraDepartment of Material Chemistry, Graduate School of Engineering, Kyoto Univer-sity, Yoshida Honmachi, Sakyo-ku, Kyoto, 606-8501, Japan

Takatoshi KinoshitaDeparment of Materials Science & Engineering, Nagoya Institute of Technology,Gokiso-cho, Showa-ku, Nagoya, 466-8555, Japan

Kazuya KitagawaDepartment of Material Chemistry, Graduate School of Engineering, Kyoto Univer-sity, Yoshida Honmachi, Sakyo-ku, Kyoto 606-8501, Japan

Takayuki KitamuraDepartment of Material and Life Science, Graduate School of Engineering, OsakaUniversity, Yamada-oka 2-1, Suita, Osaka 565-0871, Japan

Tohru KubotaKansai Advanced Research Center, Communications Research Laboratory, 588-2Iwaoka, Nishi-ku, Kobe, Hyogo, 651-2492, Japan

Kazuhiro KudoFaculty of Engineering, Chiba University, 1-33 Yayoi-cho, Inake-ku, Chiba, Chiba263-8522, Japan

Shin-ichi KurodaDepartment of Applied Physics, Graduate School of Engineering, Nagoya University,Furo-cho, Chikusa-ku, Nagoya, 464-8603, Japan

Geoffrey R. LuckhurstDepartment of Chemistry and Southampton Liquid Crystal Institute, University ofSouthampton, Highfield, Southampton, SO17 1BJ, UK

F. MallwitzOrganic and Macromolecular Chemistry, OC3, University of Ulm, Albert-Einstein-Allee 11, D-89069 Ulm, Germany

Takaaki ManakaDepartment of Physical Electronics, Tokyo Institute of Technology, 2-12-1 O-oka-yama, Meguro-ku, Tokyo 152-8552, Japan

Kazuhiro MarumotoDepartment of Applied Physics, Nagoya University, Chikusa-ku, Nagoya 464-8603,Japan

Page 14: Nanotechnology and Nano-Interface Controlled Electronic Devices

List of Contributors xiii

Shinro MashikoKansai Advanced Research Center, Communications Research Laboratory, 588-2Iwaoka, Nishi-ku, Kobe, Hyogo, 651-2492, Japan

Shinji MatsuiLaboratory of Advanced Science and Technology for Industry, Himeji Institute ofTechnology, 3-1-2 Koto, Kamigori, Ako, Hyogo, 678-1201, Japan

Teruyoshi MizutaniDepartment of Electrical Engineering, Graduate School of Engineering, NagoyaUniversity, Furo-cho, Chikusa-ku, Nagoya, 464-8603, Japan

Tatsuo MoriDepartment of Electrical Engineering, Graduate School of Engineering, NagoyaUniversity, Furo-cho, Chikusa-ku, Nagoya 464-8603, Japan

Seizo MoritaDepartment of Electronic Engineering, Graduate School of Engineering, OsakaUniversity, Yamada-Oka 2-1, Suita, Osaka, 565-0871, Japan

Tomoyuki MoritaDepartment of Material Chemistry, Graduate School of Engineering, Kyoto Univer-sity, Yoshida Honmachi, Sakyo-ku, Kyoto 606-8501, Japan

Toshihiko NagamuraMolecular Photonics Laboratory, Research Institute of Electronics, Shizuoka Univer-sity, 3-5-1 Johoku, Hamamatsu, 432-8011, Japan

Tatsuo NakahamaKansai Advanced Research Center, Communications Research Laboratory, 588-2Iwaoka, Nishi-ku, Kobe, Hyogo, 651-2492, Japan

Yutaka NoguchiDepartment of Physical Electronics, Tokyo Institute of Technology, 2-12-1 O-oka-yama, Meguro-ku, Tokyo 152-8552, Japan

Yoshishige OkunoCommunications Research Laboratory, 588-1 Iwaoka, Nishi-ku, Kobe 651-2401,Japan

Mitsuyoshi OnodaGraduate School of Engineering, Himeji Institute of Technology, 2167 Shosha,Himeji, Hyogo, 671-2201, Japan

C. PeyratoutMax-Planck-Institut für Kolloid- & Grenzflächenforschung, Berlin, Germany

Page 15: Nanotechnology and Nano-Interface Controlled Electronic Devices

xiv List of Contributors

Koichi RikitakeLife Science & Systems Engineering, Kyusyu Institute of Technology, Iizuka,Fukuoka, 820-8502, Japan

J.R. SamblesThin Film Photonics, School of Physics, University of Exeter, Exeter, EX4 4QL, UK

Edward T. SamulskiDepartment of Chemistry, University of North Carolina, Chapel Hill, NC 27599-3290, USA

Takahiro SekiPhotofunctional Chemistry Devision, Chemical Resources Laboratory, Tokyo Insti-tute of Technology, 4259 Nagatsuta-cho, Midori-ku, Yokohama, 226-8503, Japan

Brian R. StonerDepartment of Physics and Astronomy, University of North Carolina at Chapel Hill,Chapel Hill, NC 27599-3255, USA

Yasuhiro SugawaraDepartment of Electronic Engineering, Graduate School of Engineering, OsakaUniversity, Yamada-Oka 2-1, Suita, Osaka, Japan

Akihiko SugimuraDepartment of Information Systems Engineering, Osaka Sangyo University, 3-1-1Nakagaito, Daito, Osaka, 574-0013, Japan

Kazuya TadaGraduate School of Engineering, Himeji Institute of Technology, 2167 Shosha,Himeji, Hyogo 671-2201, Japan

Wataru TakashimaLife Science & Systems Engineering, Kyusyu Institute of Technology, Iizuka,Fukuoka, 820-8502, Japan

Kiyoshi TakimotoAdvanced Devices Division, Canon Research Center, Canon Inc., 5-1, Morinosato-Wakamiya, Atsugi, Kanagawa, 243-0193, Japan

Motomu TanakaLehrstuhl fur Biophysik E22, Technische Universität München, D 85748, Garching,Germany

Atsushi TojimaDepartment of Physical Electronics, Tokyo Institute of Technology, 2-12-1 O-oka-yama Meguro-ku, Tokyo 152-8552, Japan

Page 16: Nanotechnology and Nano-Interface Controlled Electronic Devices

List of Contributors xv

H. XuOrganic and Macromolecular Chemistry, OC3, University of Ulm, Albert-Einstein-Allee 11, D-89069 Ulm, Germany

Shozo YanagidaDepartment of Material and Life Science, Graduate School of Engineering, OsakaUniversity, Yamada-oka 2-1, Suita, Osaka 565-0871, Japan

Fuzi YangLiquid Crystal Research Center, Department of Chemistry, Tsinghua University,Beijing 100084, China

and

School of Physics, University of Exeter, Exeter Ex4 4QL, UK

Yoshiyuki YokogawaNational Institute of Advanced Industrial Science and Technology, Hirate-cho 1-1,Kita-ku, Nagoya 462-8510, Japan

Shiyoshi YokoyamaKansai Advanced Research Center, Communications Research Laboratory, 588-2Iwaoka, Nishi-ku, Kobe, Hyogo, 651-2492, Japan

Takashi YokoyamaNanomaterial Laboratory, National Institute for Material Science, Shidami HumanScience Park, 2268-1 Anagahora, shimo-Shidami, Moriyama-Ku, Nagoya, 463-0003,Japan

Yuji WadaDepartment of Material and Life Science, Graduate School of Engineering, OsakaUniversity, Yamada-oka 2-1, Suita, Osaka 565-0871, Japan

Shintaro WashizuFujinomiya Research Laboratories, Fuji Photo Film co., LTD., Fujinomiya Shizuoka418-8666, Japan

Fengling ZhangDepartment of Physics and Measurement Technology, Laboratory of AppliedPhysics, IFM Linköping University, S-581 83, Linköping, Sweden

Ou-Yang ZhongcanInstitute of Theoretical Physics, Academia Sinica, P.O. Box 2735, Beijing, 100080,China

Page 17: Nanotechnology and Nano-Interface Controlled Electronic Devices

This Page Intentionally Left Blank

Page 18: Nanotechnology and Nano-Interface Controlled Electronic Devices

Part A

Single Molecular Electronicsand Photonics

Page 19: Nanotechnology and Nano-Interface Controlled Electronic Devices

This Page Intentionally Left Blank

Page 20: Nanotechnology and Nano-Interface Controlled Electronic Devices

Nanotechnology and Nano-Interface Controlled Electronic DevicesEditors: M. Iwamoto, K. Kaneto and S. Mashiko© 2003 Elsevier Science B.V. All rights reserved

CHAPTER 1

Nanostructure fabrication using electronand 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. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32. Electron beam nanolithography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43. Room temperature nanoimprint technology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

3.1. Desktop compact imprint apparatus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83.2. Room-temperature nanoimprint into HSQ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

4. Three-dimensional nanostructure fabrication by focused-ion-beam . . . . . . . . . . . . . . . . . 124.1. Fabrication process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134.2. Micro-system parts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

5. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

1. Introduction

Recent years have witnessed a number of investigations concerning nanostructuretechnology. The objective of research on nanostructure technology is to explore thebasic physics, technology, and applications of ultra-small structures and devices withdimensions in the sub-100-nm regime. Today, the minimum size of Si and GaAsproduction devices is down to 0.15 μm or less. Nanostructure devices are now beingfabricated in many laboratories to explore various effects, such as those created bydownscaling existing devices, quantum effects in mesoscopic devices, or tunnelingeffects in superconductors, etc. In addition, new phenomena are being explored in anattempt to build switching devices with dimensions down to the molecular level.

Fig. 1 summarizes the resolution capabilities of several lithography processes that useelectrons, ions, and photons. It includes the narrowest line width of feature size obtainedwith 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). A256-Mb dynamic random-access memory (DRAM) Si ULSI of 0.25-μm dimensionscan be fabricated by using an i-line stepper with a phase shift mask, or an excimer laserstepper. An excimer laser can be applied to a 1-Gb DRAM with 0.15 μm feature size.

Page 21: Nanotechnology and Nano-Interface Controlled Electronic Devices

4 S. Matsui

Fig. 1: Microfabrication using electrons, ions, and photons.

Electron beam (EB) lithography is the most widely used and versatile lithography toolused in fabricating nanostructure devices. Because of the availability of high-qualityelectron sources and optics, EB can be focused to diameters of less than 10 nm.The minimum beam diameters of scanning electron microscopes (SEM) and scanningtransmission microscopes (STEM) are 1.5 and 0.5 nm, respectively. While focused-ionbeam (FIB) can be focused close to 5 nm. EB and FIB can be used to make nano-scalefeatures in the 100- to 1-nm regime. Scanning tunneling microscopy (STM) is used foratomic 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 (NipponZeopn Co.) positive resists [3], 20 nm for SAL601 (Shipley Co.) [4], and 10 nm forCALIXARENE 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 patternsof 8 nm have been fabricated with SEM [9], and 8-nm PMMA patterns have beendemonstrated 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 accuracyis improved due to less beam drift on the nanometer scale. Moreover, it enables theuse of a highly sensitive resist such as ZEP520 [11], which has sufficient resolution

Page 22: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 wasobtained using ZEP520 positive resist. The ZEP520 resist was spin-coated onto a Siwafer to a thickness of 50 nm, and prebaked at 200°C. After EB exposure, the ZEP520was developed with hexyl acetate for 2 min and rinsed with 2-propanol. Fig. 3 shows

Page 23: Nanotechnology and Nano-Interface Controlled Electronic Devices

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-nmand it works as an ultrahigh-resolution negative EB resist. CALIXARENE is a singlemolecule 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 aresolution limit. The molecular uniformity of CALIXARENE and its small molecularsize is the origin of surface smoothness and the resulting ultrahigh resolution.

Such characteristics seem to be convenient for a nanodevice fabrication process. Thebasic component of CALIXARENE is a phenol derivative which seems to have highdurability and stability, originating from the strong chemical coupling of the benzenering. The threshold of sensitivity was about 800 μC/cm2, which is almost 20 timeshigher than that of PMMA. CALIXARENE negative resist exposure was carried out. A30-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 andwas rinsed in IPA for 1 min.

The etching durability of CALIXARENE was tested using a DEM-451 (ANELVACorp.) plasma dry-etching system with CF4 gas. The etching rate of CALIXARENE isalmost comparable with that of Si, and the durability is about four times higher than thatof PMMA. This durability seems to be sufficient to make a semiconductor or a metalnanostructure.

Nanodot arrays are useful not only for quantum devices but also for studyingexposure properties. In this experiment, the EB current was fixed to 100 pA at 50 kVaccelerating voltage, for which the spot size is estimated to be about 5 nm. All thedot 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 nmdiameter with 35 nm pitch.

Germanium pattern transfer is shown in Fig. 5. The 20-nm-thick Ge layer requires atleast a 5-nm-thick CALIXARENE layer to be etched down, and the resist thickness was30 nm. Fig. 5(a) shows the line patterns of the resist on Ge film exposed at a line doseof 20 nC/cm. Delineation was done using the S-5000 (Hitachi Corp.) SEM with a beamcurrent of 100 pA at a 30-kV acceleration voltage. A 10-nm line width and a smooth lineedge were clearly observed. This smoothness is the key point in fabricating quantumnanowires by etching processes. Fig. 5(b) shows the transferred pattern treated by 1min 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 byoveretching is a standard technique to obtain a fine line, however, side-wall roughnesslimits the line width. The smoothness of the CALIXARENE side wall enables the linewidth 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 bydeforming the physical shape of the resist by embossing with a mold, is a very useful

Page 24: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 aquantized magnetic disk have, been demonstrated [19–22]. The technique has excellentfeatures that are sub-10 nm in size over a large area and have high throughput and arelow in cost.

However, a conventional NIL process has to heat a resist above the glass transitiontemperature to deform the physical shape of the resist with a mold. Consequently, aconventional NIL process (system) must require a thermal cycle of the resist. Thisheating process causes serious problems in replicated pattern accuracy, and a reductionin 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-NILprocess 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 processsteps without heating and cooling are shorter compared with those in conventional NILas shown in Fig. 6(a) [12,13].

First, we have demonstrated RT-NIL using Spin-on-Glass (SOG), as the materialreplicated and obtained excellent replicated SOG patterns and Si etched patterns bytransferring replicated SOG patterns using CF4 RIE [23]. However, SOG has theimportant technological shortcoming that SOG hardens gradually by reaction with waterin the air at room temperature. To overcome the disadvantage, we have proposed RT-NIL using a hydrogen silsequioxane (HSQ) as the replicated material. Further, we havecarried out the step-and-repeat imprinting using HSQ, and evaluated the uniformity ofthe replicated HSQ patterns.

Page 25: Nanotechnology and Nano-Interface Controlled Electronic Devices

8 S. Matsui

Fig. 5: Pattern transfer to Ge. (a) 10-nm line width CALIXARENE pattern and (b) transferred 7-nm linewidth Ge pattern.

3.1. Desktop compact imprint apparatus

The desktop compact imprint apparatus which uses a stepping motor as driving powerthat we have developed is shown in Fig. 7(a). The apparatus is 17 cm in width and30 cm in height, and has 10 × 10 mm2 mold-mask holder and 2 inch wafer stage. Thez-positioning accuracy of the stepping motor is 2 μm per pulse. A heater is buried in thewafer stage to heat resist coated on a wafer to above the glass temperature. Therefore, byusing this system, conventional NIL and RT-NIL can be performed. A 2 in. wafer can beimprinted on the x–y step-and-repeat stage. The z-axis of the mold holder and the x- andy-axis of the wafer stage are controlled by three stepping motors receiving pulse signalsfrom a computer. The wafer temperature can be varied from room temperature up to200°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 measurepressures when pressing a mold-mask into the resist on the substrate.

Page 26: Nanotechnology and Nano-Interface Controlled Electronic Devices

Nanostructure fabrication using electron and ion beams 9

Fig. 6: Schematic of a NIL process: (a) conventional NIL using PMMA, and (b) room temperature NILusing SOG or HSQ.

Fig. 7: (a) Desktop compact imprint system which uses a stepping motor as driving power. (b) Imprintsystem with a mold-mask holder and an imprint stage.

Page 27: Nanotechnology and Nano-Interface Controlled Electronic Devices

10 S. Matsui

Fig. 7(b) illustrates the system of the mold-mask holder and the imprint stage. Amagnet is inserted between a mold-mask and a mold wafer, and a magnetic sheet and aniron plate are inserted between the imprint stage and a wafer. The purpose of using amagnet is to make hard contact between the mold-mask and the wafer, and can bring amold-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 formulaH2(SiO)3 was used. Infrared (IR) spectra confirmed that there was no hydrocarbon inthe HSQ material. Therefore, this HSQ resin contains no organic groups, such as vinylgroups. And then, HSQ has a favorable etching durability [24]. Since HSQ has a highviscosity without pre-baking, pre-baking must be performed, in contrast to SOG, beforethe imprinting process. The effect of pre-baking for the HSQ is to remove the solventfrom the content of HSQ, and to make the viscosity of the HSQ moderate so thatphysical transformation of the HSQ with a mold pattern is possible.

Fig. 8 shows the pre-baking dependence of the imprint depth when using HSQ. Wemeasured 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 werevaried from 50 to 200°C. The results indicate that the imprinting depth decreasessuddenly around 150°C. It is suggested that the hardness of HSQ increases around150°C. Therefore we used a pre-baking temperature in the range from 50°C to 100°C toobtain 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 wasspin-coated on an Si substrate. Then a mold and an HSQ coated substrate with pre-baking were pressed together for 1 min at a set press-pressure in the range from 2.5 to4.5 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 45degrees of imprinted HSQ patterns with 0.8-μm line widths and 6-μm pitches after 1min pressing using RT-NIL. In this experiment, an imprinting pressure of 2.0 MPa and

Fig. 8: Imprinting characteristics using HSQ.

Page 28: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 patternswith 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 moldpatterns 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 depthwas 150 nm and the residual depth was 150 nm from Fig. 9(c), which nearly correspondsto the imprinting characteristics using HSQ as shown in Fig. 8. Fig. 10(a) and (b) showsSEM photographs of a top view and at a tilt angle of 45 degrees of SiO2/Si pillarsmold patterns with 90 nm diameter, 600 nm period and which are 0.4 μm in height. Byusing the mold, holes with 90 nm diameter and 600 nm pitch were obtained after 1 minimpress time by RT-NIL, as shown in Fig. 11(a) and (b). The pre-baking temperatureand pressure were 60°C and 4.0 MPa. This indicates that the mold pillar patterns were

Page 29: Nanotechnology and Nano-Interface Controlled Electronic Devices

12 S. Matsui

Fig. 11: SEM micrographs of a top view of holes with 90 nm diameter and 600 nm pitch imprinted after 1min press time by RT-NIL.

Fig. 12: SEM micrographs of patterns with 50 nm line width and 200 nm pitch replicated in HSQ usingRT-NIL.

imprinted with high precision. Fig. 12 (a) and (b) show SEM photographs of patternswith 50 nm line width and 200 nm pitch replicated in HSQ using RT-NIL. These resultsdemonstrate 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-ion-beam (FIB) has been achieved, and has been applied to make various nanostructure

Page 30: Nanotechnology and Nano-Interface Controlled Electronic Devices

Nanostructure fabrication using electron and ion beams 13

devices such as single-electron transistors and MOS transistors with nanometer gate-length. Ten-nm structures can be formed by using a commercially available EB or FIBsystem 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-dimensionalfabrication using laser-CVD, FIB- and EB-CVD are superior [26] in the points of spatialresolution and beam-scan control. Koops et al. demonstrated some applications such asan AFM tip and field emitter by using EB-CVD [27]. Blauner et al. demonstrated pillarsand 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 factorssuch as the difference of mass between electrons and ions. Furthermore, a smallerpenetration-depth of ions compared to electrons allows to make complicated three-dimensional nanostructures. For example, when we make a coil nanostructure with 100nm line width, electrons with 10–50 keV pass the ring of the coil and reach the substratebecause of the large electron range (over a few μm), so it may be difficult to make a coilnanostructure by EB-CVD. On the other hand, as the ion range is less than a few tennm, ions stop inside the ring. So far complicated nanostructures using FIB-CVD havenot been reported. This paper presents a description of the fabrication of a complicatedthree-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 aspot 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 intothe vacuum chamber by means of a nozzle, which is located at a height of 500 μmabove the sample surface, at an angle of about 45 degrees with respect to the samplesurface. 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 afterintroducing the source gas is 5×10−5 Pa. The FIB is controlled by a computer to writethe 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 trans-mission electron microscope (TEM) and measuring of Raman spectra. A carbon thinfilm with 200 nm thickness was deposited on a silicon substrate by 30 keV Ga+ FIBusing a phenanthrene precursor gas. The cross-section structures and electron diffractionpatterns were observed by using a 300-kV TEM. The result was that there were nocrystal structures in the TEM images and diffraction patterns. It is concluded that thedeposited film is amorphous carbon (a-C). Raman spectra of a-C films were measured atroom temperature with the 514.5 nm line of an argon ion laser. The Raman spectra wererecorded by a monochromator equipped with a CCD multi-channel detector. Ramanspectra 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

Page 31: Nanotechnology and Nano-Interface Controlled Electronic Devices

14 S. Matsui

Fig. 13: Raman spectra of a diamond-like amorphous carbon film obtained at 0.1 mW by 514.4 nmexcitation 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 areobserved in the spectra excited by a 514.5 nm line. Two Raman bands were plotted afterGaussian line shape analysis. The Raman bands at 1550 cm−1 and 1400 cm−1 originatefrom the trigonal (sp2) bonding structure of graphite and tetrahedral (sp3) bond structureof diamond. This result indicates that a-C film deposited by FIB-CVD is diamond-likeamorphous carbon, which has attracted attention because of the hardness, chemicalinertness, 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 formedon the substrate by fixing a beam-position (position 1). After that, the beam-positionis moved within the diameter of the pillar (position 2) and is then fixed until thedeposited terrace thickness exceeds the ion range which is a few ten nm. This process isrepeated to make three-dimensional structures. The key point to make three-dimensionalstructures is to adjust the beam-scan speed so that the ion-beam remains within thedeposited terrace, which means that the terrace thickness exceeds an ion range. Thegrowth conditions in the x and y directions are controlled by the beam-deflectors. Thegrowth in the z direction is determined by the deposition rate, that is, the height of astructure 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 pillarwas made by fixing the beam-position for 120 s at 16 pA beam-current. And then, thegrowth of a branch was carried out using the process explained above. The diameterof 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 befabricated by FIB-CVD.

Page 32: Nanotechnology and Nano-Interface Controlled Electronic Devices

Nanostructure fabrication using electron and ion beams 15

Fig. 14: Fabrication process for three-dimensional nanostructures by FIB-CVD.

Fig. 15: Branch structure with 0.08 μm diameter.

Page 33: Nanotechnology and Nano-Interface Controlled Electronic Devices

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-beakerswith 1.0 μm height, and 1.0 and 1.5 μm diameters. Total exposure-time of two beakerswas 600 s at 16 pA beam-current. Some applications are considered, such as the study ofmicro-crystal growth or micro-chemical reactions by filling a beaker with the examinedmaterial.

Fig. 17: Micro-coil with 0.6 μm coil-diameter, 0.7 μm coil-pitch, and 0.08 μm line width.

Page 34: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 0.08μm line width. Exposure time was 40 s at 0.4 pA beam current. The coil-pitch can easilybe 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 themicro-coil are 0.25, 0.20, and 3.8 μm, respectively. Exposure time was 60 s at 0.4 pAbeam-current.

A bellows is one of the important parts in a mechanical system, just like a coil and adrill. Fig. 19 shows a micro-bellows with 0.8 μm pitch, 0.1 μm thickness, 2.75 μm ex-ternal diameter, and 6.1 μm height. Exposure time was 300 s at 16 pA beam-current. Theresults show that FIB-CVD is one of the promising techniques to make parts of micro-systems, 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. Todemonstrate the possibility, a micro-wineglass was created as a work of microstructureplastic arts. A micro-wineglass with 2.75 μm external diameter and 12 μm height wasformed on Si substrate and a human hair, as shown in Fig. 20(a) and (b). Fabrication timewas 600 s at 16 pA beam-current. The beautiful micro-wineglass gives us expectationsof opening up microstructure plastic arts.

Page 35: Nanotechnology and Nano-Interface Controlled Electronic Devices

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.

Page 36: Nanotechnology and Nano-Interface Controlled Electronic Devices

Nanostructure fabrication using electron and ion beams 19

5. Summary

Ten-nanometer structure fabrication has been achieved by using a commercial availableelectron beam (EB) apparatus and resists. Nanoimprint lithography (NIL) is a veryuseful technique to overcome the low throughput and high cost in EB lithography. Roomtemperature NIL using HSQ as the replicated material has been proposed to achievehighly precise replication and to perform step-and-repeat imprinting. As a result, wedemonstrated 50 nm line width and 90 nm in diameter hole replicated HSQ patterns, andthe possibility of step-and-repeat imprinting in HSQ to 1.5 in. wafers. The results revealthat the room-temperature NIL process using HSQ as the replicated material is a veryuseful technique to achieve a highly precise nanoimprinting process and this techniquecan be applied to make nanostructure devices.

Three-dimensional structures with spatial resolution in the nanometer range can begenerated 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 spectrathat the deposited film is a diamond-like amorphous carbon. Micro-coil, drill, andbellows with 0.1 μm dimension were fabricated as parts of micro-systems. We proposemicrostructure plastic arts as a new field using micro-beam technology. A micro-wineglass 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 maybecome interesting tools for the generation of micro- and nano-systems in the field ofelectronics, mechanics, optics and biology.

References

1. F. Emoto, K. Gamo, S. Namba, N. Samoto, and R. Shimizu, Jpn. J. Appl. Phys. 24, L809 (1985).2. W. Chen and H. Ahmed, Appl. Phys. Lett. 63, 1116 (1993).3. K. Kurihara, K. Iwadate, H. Namatsu, M. Nagase, H. Takenaka, and K. Murase, Jpn. J. Appl. Phys.

34, 6940 (1995).4. T. Yoshimura, Y. Nakayama, and S. Okazaki, J. Vac. Sci. Technol. B10, 2615 (1992).5. J. Fujita, Y. Ohnishi, Y. Ochiai, and S. Matsui, Appl. Phys. Lett. 68, 1297 (1996).6. M. Isaacson and A. Murray, J. Vac. Sci. Technol. 19, 1117 (1981).7. D.R. Allee and A.N. Broers, Appl. Phys. Lett. 57, 2271 (1990).8. J. Fujita, H. Watanabe, Y. Ochiai, S. Manako, J.S. Tsai, and S. Matsui, Appl. Phys. Lett. 66, 3065

(1995).9. A.N. Broers, W.W. Molzen, J.J. Cuomo, and N.D. Wittles, Appl. Phys. Lett. 29, 596 (1976).

10. R.L. Kubena, J.W. Ward, F.P. Stratton, R.J. Joyce, and G.M. Atkinson, J. Vac. Sci. Technol. B9, 3079(1991).

11. T. Nishida, M. Notomi, R. Iga, and T. Tamamura, Jpn. J. Appl. Phys. 31, 4508 (1992).12. S.Y. Chou, P.R. Krauss, and P.J. Renstrom, Appl. Phys. Lett. 67, 3114 (1995); Science 272, 85

(1996).13. S.Y. Chou, P.R. Krauss, W. Zhang, L. Guo, and L. Zhuang, J. Vac. Sci. Technol. B15, 2897 (1997).14. X. Sun, L. Zhuang, W. Zhang, and S.Y. Chou, J. Vac. Sci. Technol. B16, 3922 (1998).15. B. Heidari, I. Maximov, and L. Montelius, J. Vac. Sci. Technol. B18, 3557 (2000).16. H. Schift, R.W. Jaszewski, C. David, and J. Gobrecht, Microelectronic Eng. 46, 121 (1999).

Page 37: Nanotechnology and Nano-Interface Controlled Electronic Devices

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).

18. M. Komuro, J. Taniguchi, S. Inoue, N. Kimura, Y. Tokano, H. Hiroshima, and S. Matsui, Jpn. J. Appl.Phys. 39, 7075 (2000).

19. W. Wu, B.C.X. Sun, W. Zhang, L. Zhuang, L. Kong, and S.Y. Chou, J. Vac. Sci. Technol. B16, 3825(1998).

20. L. Zhuang, L. Guo, and S.Y. Chou, Appl. Phys. Lett. 72, 1205 (1998).21. J. Wang, A. Schablitsky, Z. Yu, W. Wu, and S.Y. Chou, J. vac. Sci. Technol. B17, 2957 (1999).22. I. Martini, S. Kuhn, M. Kamp, L. Worschech, A. Forchel, D. Eisert, J. Koeth, and R. Sijbesma, J.

Vac. Sci. Technol. B18, 3561 (2000).23. S. Matsui, Y. Igaku, H. Ishigaki, J. Fujita, M. Ishida, Y. Ochiai, M. Komuro, and H. Hiroshima, J.

Vac. Sci. Technol. B19, 2801 (2001).24. H. Namatsu, Y. Takahashi, K. Yamazaki, T. Yamaguchi, M. Nagase, and K. Kurihara, J. Vac. Sci.

Technol. B16, 69 (1998).25. S. Matsui, Proc. IEEE 85, 629 (1997).26. O. Lehmann, F. Foulon, and M. Stuke, NATO ASI Ser. E: Appl. Sci., 265, 91–102 (1994).27. H.W. Koops, Jpn. J. Appl. Phys. 33, 7099 (1994).28. A. Wargner, J.P. Levin, J.L. Mauer, P.G. Blauner, S.J. Kirch, and P. Long, J. Vac. Sci. Technol. B8,

1557 (1990).29. P.G. Blauner, Proc. 1991 Int. MicroProcess Conf., p. 309 (1991).30. S. Matsui, T. Kaito, J. Fujita, M. Komuro, K. Kanda, and Y. Haruyama, J. Vac. Sci. Technol. B18,

3181 (2000).31. J. Fujita, M. Ishida, T. Sakamoto, Y. ochiai, T. Kaito, and S. Matsui, J. Vac. Sci. Technol. B19, 2834

(2001).

Page 38: Nanotechnology and Nano-Interface Controlled Electronic Devices

Nanotechnology and Nano-Interface Controlled Electronic DevicesEditors: 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 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212. Restriction in transfer rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223. Formation of an ideal metal–insulator–metal junction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224. Size of data bits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245. Rate of reading and writing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256. Error rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 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 anatomic manipulation has been achieved [2,3]. A surface modification according to agiven pattern and its observation with a scanning probe can also be seen as a writing andreading procedure in an information storage device. Then the pattern produced by thesurface modification is regarded as a set of data bits. The recording density estimatedfrom the typical size of each modification becomes much higher than that of any presentstorage device. If the size is 10 nm, for example, the recording density can be roughlyestimated to be 1 Tbit/cm2 (= 1012 bit/cm2). Therefore the SPM-technology is thoughtto make it possible to establish a high-density and huge-capacity information storagesystem. There have been many reports concerning information storage using an SPM[4–12].

However, the recording density is only one of the performances characterizingan 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 bitsat a rate comparable to that of the present storage devices in order to realize a practicalsystem.

Page 39: Nanotechnology and Nano-Interface Controlled Electronic Devices

22 K. Takimoto

2. Restriction in transfer rate

In the SPM, the surface image is based on some interaction between a probe and thesurface of a sample when they are placed in close proximity. The surface topography isobserved as the trajectory of the probe motion when the spacing between the probe andthe surface is precisely controlled so as to maintain the interaction between them to beconstant during scanning of the probe. The operation speed in the SPM is limited by thecharacteristic rate of the feedback loop, and it is generally low. The highest operationspeed in the SPM is achieved in an atomic force microscope (AFM) operated in contactmode, where the tip of a probe is in contact with the surface of a sample. Then, theoperation speed is limited to the mechanical resonance frequency of the cantilever forsupporting the probe, which is limited to several mega-Hertz in practice.

When the data bits are formed with topographical modification of the recordingmedium, the reading rate of information storage with a scanning probe is restrictedby the operation speed in the SPM. One of the ways to overcome this restriction isthe use of a cantilever with high mechanical resonance frequency. In this case, precisefabrication of a cantilever with a small size and a small weight is necessary. Another wayis the use of a recording medium in which data bits can be written without topographicalmodification of this recording medium. In this procedure, a scanning probe must beused for simultaneously detecting two kinds of interactions, independent of each other.The spacing between the probe and the medium is controlled based on one interactionbetween them, and the data bits are written and read based on the other interaction. Thenthe former interaction does not influence the reading rate directly, since it is not affectedby the presence of the data bits. The reading rate is determined by the rate for detectionof the data bits and the rate for scanning of the probe on the medium. For high-speedscanning of the probe, reducing the surface roughness of the medium and controlling thespacing 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 recordingprocedure was introduced which is not based on direct surface modification of arecording medium with a scanning probe. In this procedure, a scanning probe is usedas a tool in order to form an ideal metal–insulator–metal (MIM) junction. That is, oneelectrode in the MIM junction is replaced by the scanning probe. For this purpose, anAFM with an electrically conducting probe is used. In practice, the conducting probe isobtained by coating a conventional cantilever for AFM with a metal film. The probe canbe in contact with the sample surface using a weak force controlled by AFM so as tokeep 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 becharacterized with sufficient resolution as shown in Fig. 1. If the electronic properties ofthe film can be modified by application of a pulse voltage to the MIM junction formedwith the probe, it can be considered as another procedure to form data bits. The areawith modified electronic properties can be observed using the AFM based apparatus as

Page 40: Nanotechnology and Nano-Interface Controlled Electronic Devices

Information storage using a scanning probe 23

Fig. 1: Schematic drawing of an AFM based apparatus with an electrically conducting probe. Surfacetopography and electronic properties of a recording medium are simultaneously characterized using thisapparatus.

shown in Fig. 1. If the size of the modified area is small enough, it can be regarded as abit in high-density information storage. One expects to realize high-speed reading andwriting if the modification is not associated with a change in surface morphology. Onealso 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 recordingprocedure is a polyimide Langmuir–Blodgett (LB) film. Sakai et al. reported a switchingand memory phenomenon in a MIM junction with LB film as insulating layer [13]. Itshows a reversible transition between high and low conductance states by application ofpulse voltages. Each state has a corresponding threshold, and the state is maintained forapplication of voltage below the threshold, even 0 V. That is, this MIM junction shows anon-volatile memory effect. This phenomenon shows a clear material dependence, anddoes not depend on junction area [14]. Conventional polyimide is one of the typicalmaterials showing this phenomenon.

Formation of an MIM junction using a scanning probe and a polyimide LB filmhas 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 junctionusing an AFM based apparatus has also been achieved [16,17]. The transition fromlow conductance state to high conductance state has been induced by application ofpulse voltage to the polyimide LB film through the scanning probe. The area wherethe transition in conductance occurred has been observed as the conducting spot in thecurrent image obtained using the conducting AFM probe. On the other hand, no changehas 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 ascanning probe, a set of data bits composed of conducting spots can be formed withoutchange in surface morphology. The reading rate depends on the rate in current detection

Page 41: Nanotechnology and Nano-Interface Controlled Electronic Devices

24 K. Takimoto

Fig. 2: Schematic drawing of an information storage system using an AFM based apparatus and a polyimideLB 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. Ifthe sample surface is very flat, the reading rate can be expected to be much higher thanand not restricted by the mechanical resonance frequency of the cantilever. The size ofthe conducting spot is about 10 nm in diameter. The recording density can be estimatedto be 1 Tbit/cm2 from this spot size when the conducting spots are considered to be databits.

Fig. 2 shows a schematic drawing of the information storage system using the AFMbased apparatus and the polyimide LB film as the recording medium. The probe isscanned on the recording medium in contact. In the recording procedure, the array ofpulse voltages according to a set of binary data is applied to the medium through theprobe during scanning, and high-conductance regions are formed in the medium. Afterrecording, current flow through the medium is detected with the probe during scanningand the obtained scan profile of current is converted to the set of binary data. In sucha configuration, a pattern of encoded binary data consisting of more than a thousandconducting spots could be formed in a 2×2 μm2 area. And then reading the informationback could be accomplished by converting the line scan profiles of the current image tobit patterns [18].

4. Size of data bits

The size of the conducting spots was about 10–20 nm in diameter, independent ofvarious 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 coatedwith Pt. The Pt film has grain structures, which are several 10 nm in diameter. The tip

Page 42: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 probefabricated 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 maybe determined by the size of the grains. This seems to be a reason why the size of aconducting spot does not depend on the probes. To form a smaller conducting spot, acurvature in the tip of the probe has to be formed less than that in a grain of the coatedmetal. Yagi et al. fabricated a probe with a quite sharp tip by replicating a Si mold withpyramidal etch pits [19]. Fig. 3(b) shows an SEM image of a tip of a probe fabricatedby 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 thereplica probe [19]. Conducting spots observed in the current image were induced usingthe same probe. The sizes of the conducting spots are 10 nm or less. In addition, amonoatomic step structure in the Au(111) surface can be seen in the AFM image, whichshows an increase in resolution of the AFM due to the decrease in the curvature in thetip of the probe. This shows that conducting spots with small sizes of 10 nm or less canbe formed stably using the replica probe, and also shows that no degradation of the tipoccurs 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 voltagepulse 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 aroundthe MIM junction with the scanning probe. Actually, the writing with a 2 μs pulse was

Page 43: Nanotechnology and Nano-Interface Controlled Electronic Devices

26 K. Takimoto

Fig. 4: Current image (a) and AFM image (b) observed simultaneously using the replica probe. Conductingspots in the current image were induced by applying 13 V, 0.2 μs, rectangular voltage pulses using the sameprobe. Reproduced from Ref. [19] with permission (© 1997 IEEE).

achieved by a reduction of the stray capacitance. Furthermore, the transient responseof the current observed during the spot formation by applying a 2 μs voltage pulseindicated that further reduction of the stray capacitance would make it possible to forma conducting spot with a voltage pulse shorter than 1 μs. The increase of the current bythe 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 thana 2 μs pulse. The conducting spots observed in Fig. 4(a) are formed by application of0.2 μs voltage pulses [19]. The formation of conducting spots could be carried out bythe application of pulse voltages of 25 ns in width under further optimization in straycapacitance, though such a result is not shown here. This indicates that a reading rate ofabout 40 Mbps may be achieved.

For reading at a fast rate, high-speed scanning of the probe and high-speed detectionof the current are necessary. To detect a low current at high scanning rate, a currentamplifier is designed which has a small input capacitance and little gain for lowfrequencies. Using the current amplifier, it was demonstrated that the edge of theconducting spot could be detected within about 10 μs during scanning at a rate of 8μm/s. This indicated that achieving a reading rate of about 100 kbps could be expectedif the probe could be scanned with sufficient speed, typically 2 mm/s for an array of 10nm spots [18].

High-speed scanning at a rate of 2 mm/s has been actually performed withoutdamage to the medium. In this procedure, the probe was scanned so that the trajectoryof its tip drew a circle. Although the scanning rate of 2 mm/s is much higher comparedwith that in a conventional AFM observation, the surface roughness of the polyimideLB film deposited on Au(111) is so small that high-speed scanning of the probe on it

Page 44: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 fromRef. [20] with permission.)

seems to be possible. When both the detection rate of 10 μs per bit and the scanningrate 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 ofthe cantilever are not necessary.

6. Error rate

Stable formation of about one thousand data bits without the degradation of the tip ofthe 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, anderror rate estimation is required. Yano et al. demonstrated 1 Mbit recording in an areaof 40×80 μm2 without tip degradation [20]. Fig. 5 shows a current image of a part ofthe recorded 1 Mbits. The area is about 3.5 × 0.7 μm2. An image similar to Fig. 5 canbe acquired at any area where the voltage pulses were applied. In 1 Mbit recording, atransient response in current was monitored for each pulse application. A transition tothe high-conductance state was confirmed when the current exceeded a predeterminedvalue within a predetermined period of voltage application. That is, the case that thecurrent exceeding this value was not observed within the period was regarded as afailure in the formation of the bit. Thus, the error rate was estimated to be 1.7×10−4 for1 Mbit recording [20].

7. Conclusion

The concept of an information storage system was demonstrated based on the formationof ideal MIM junctions using an AFM with a conducting probe. In practice, polyimideLB film is used as an insulating layer in an MIM junction, and a local conductingregion in the polyimide LB film induced by applying a pulse voltage through the probeis considered to a recording bit. The reduction of bit size, the possibilities of fast ratereading and writing, and further stable bit writing were also shown. In this system, thesurface topography of the medium is maintained even after writing data bits, whichmakes possible to read and write data bits at fast rates without severe requirementsfor the mechanical resonance frequency of a cantilever. Then an extremely flat surfaceof the recording medium is required over a wide range. It is essential to fabricate asufficiently flat and large medium. LB film seems to be a suitable recording medium,

Page 45: Nanotechnology and Nano-Interface Controlled Electronic Devices

28 K. Takimoto

since the thickness of the film can be precisely controlled on a molecular scale. So, it isimportant to fabricate a flat and large substrate.

Today, the recording density in some present storage devices reaches several tensgigabit per square inch. And it is growing at a rate of 60% every year. If today’sgrowing rate has to be kept, it is predicted that the recording density will be equal tothe atomic density of the solid surface within twenty years. Molecular memory willalso be realistic. Then, the storage devices will have to read and write the data bitswith atomic resolution at a rate further exceeding that of present storages. The scanningprobe method has enough potential concerning the resolution. However, the higher therecording density, the lower the reading rate becomes actually. The use of multipleprobes and the parallel operation of them is an effective way to achieve a faster read andwrite rate[21–24]. However, essential breakthroughs on the problem of the reading ratealso seems to be necessary.

References

1. C.F. Quate, NATO ASI Ser. E: Appl. Sci. 184, 281 (1990).2. D.M. Eigler and E.K. Schweizer, Nature 344, 524 (1990); Appl. Phys. Lett. 68, 34 (1996).3. I.W. Lyo and P. Avouris, Science, 253, 173 (1991).4. H.J. Mamin, P.H. Guethner, and D. Ruger, Phys. Rev. Lett. 65, 2418 (1990) .5. H.J. Mamin and D. Ruger, Appl. Phys. Lett. 61, 1003 (1992).6. 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).7. 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).8. R.C. Barrett and C.F. Quate, J. Appl. Phys. 70, 2725 (1993).9. H. Kado and T. Tohda, Appl. Phys. Lett. 66. 2961 (1995).

10. B.W. Chui, H.J. Mamin, B.D. Terris, T.D. Stowe, D. Rugar, and T.W. Kenny, Appl. Phys. Lett. 69,2767 (1996).

11. 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).

12. 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).

13. K. Sakai, H. Matsuda, H. Kawada, K. Eguchi, and T. Nakagiri, Appl. Phys, Lett. 53, 1274 (1988).14. K. Takimoto, K. Yano, K. Hatanaka, K. Eguchi, and T. Nakagiri, Oyo Butsuri 63, 470 (1994)15. K. Takimoto, H. Kawade, E. Kishi, K. Yano, K. Sakai, K. Hatanaka, K. Eguchi, and T. Nakagiri,

Appl. Phys. Lett. 61, 3032 (1992).16. 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).17. 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).18. K. Takimoto, R. Kuroda, S. Shido, S. Yasuda, H. Matsuda, K. Eguchi, and T. Nakagiri, J. Vac. Sci.

Technol. B 15, 1429 (1997).19. 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).20. K. Yano and T. Ikeda, Appl. Phys. Lett. 80, 1067 (2002).21. 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).22. S.C. Minne, G. Yaralioglu, S.R. Manalis, J.D. Adams, J. Zesch, A. Atalar, and C.F. Quate, Appl.

Page 46: Nanotechnology and Nano-Interface Controlled Electronic Devices

Information storage using a scanning probe 29

Phys. Lett. 72, 2340 (1998).23. S.C. Minne, J.D. Adams, G. Yaralioglu, S.R. Manalis, A. Atalar, and C.F. Quate, Appl. Phys. Lett.

73, 1742 (1998).24. Y. Shimada, T. Yagi, T. Yamazaki, S. Shido, H. Matsuda, and K. Takimoto, Technical Digest of the

16th Sensor Symposium, p. 273 (1998).

Page 47: Nanotechnology and Nano-Interface Controlled Electronic Devices

This Page Intentionally Left Blank

Page 48: Nanotechnology and Nano-Interface Controlled Electronic Devices

Nanotechnology and Nano-Interface Controlled Electronic DevicesEditors: 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. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 312. Molecules and samples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

2.1. Molecules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 322.2. Sample structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 322.3. I –V measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

3. Current–voltage characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 343.1. Single electron tunneling characteristic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 343.2. Organic molecule as Coulomb island . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 373.3. Light-irradiated single electron tunneling characteristics . . . . . . . . . . . . . . . . . . . 38

4. Future prospect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

1. Introduction

In recent years, many research studies have been done along with the rapid progressof nano-fabrication technology in the field of nano-electronics [1]. Until now, mucheffort has been done on the fabrication of novel devices based on physics of quantummechanics, principally using the nano-fabrication technique developed in semiconductordevice technology [2]. As a result, single electron tunneling (SET) devices using smallparticles in their systems have been successfully prepared. Nano-fabrication technologydeveloped in the field of semiconductor device technology may lead to a new way toelectronics and many novel electronic devices such as high-density memory devices,high-speed low-power switching devices, high-sensitive electrometer devices and otherswill be produced in the near future. As such, it is needless to say that the research alongwith this trend is important. However, this is not sufficient. The study of observing

Page 49: Nanotechnology and Nano-Interface Controlled Electronic Devices

32 T. Kubota et al.

specific functions of organic molecules and applying these functions to electronic andoptical molecular devices is of crucial help, because one can realize novel functionaldevices only by using organic molecules, without using the maturing nano-fabricationtechnology in the field of semiconductor device technology. In this chapter, singleelectron 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 sothat 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 beintroduced into the molecular systems to be used. However, there are many difficultiesto do this, e.g., synthesis and others. Overcoming the difficulties, an attempt to use anorganic mono-molecule as a so-called Coulomb island has been successfully made bythe present authors [3–6]. In the following, single electron tunneling devices preparedusing organic molecules is described.

2. Molecules and samples

2.1. Molecules

As the molecule used for fabricating the device, dendrimer molecules (Rh-G2), whichhave called much attention in the field of organic synthesis, were used as the Coulombisland. Polyimide LB film, which shows excellent insulating properties [7], was usedas tunneling barrier in the device. The chemical structures of Rh-G2 and PI are shownin Fig. 1. The polyimide LB film functions as electron tunneling barrier, where themonomolecular film thickness of 0.4 nm can be controlled by the LB method. Thedendrimer molecule Rh-G2 used as the Coulomb island has a spherical shell molecularstructure, i.e. electrically insulating CH chains enclose rhodamine dye molecules [8].That is, the dye molecules located at the center are electrically isolated from theirsurroundings. The use of such organic molecules facilitates the preparation of singleelectron 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 Coulombisland. Au was evaporated on glass substrate with a thickness of about 100 nm, andthe evaporated electrode was used as a bottom electrode of the SET device. Onto thisevaporated electrode, 13 to 31 layers of polyimide LB film were deposited by theLB technique to form an electron tunneling layer, in the same manner as that in ourprevious study [7]. Then, polyimide LB film mixed with rhodamine dendrite moleculeswas deposited by mono-layer deposition to form a single layer working as a Coulombisland. The mixing ratio of PI : Rh-G2 was 500 : 1 in molar ratio. From the experimentalsurface pressure–area (F–A) isotherm of the mixture monolayer film, a number ofabout 1000 molecules of Rh-G2 was estimated to be present within an area of about

Page 50: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 LBfilms were again deposited as an upper electron tunneling layer. Finally, an Au electrodewas evaporated with a thickness of 50–100 nm to form a top electrode. The workingarea of the resulting junction was about 50×100 μm (see Fig. 2). Furthermore, for themeasurement of I–V characteristics under photoillumination, junctions with transparentindium–tin-oxide (ITO) electrodes were also fabricated.

2.3. I–V measurement

The electrical resistance of the prepared devices was several hundreds of M� toseveral tens of G�. Thus, current–voltage (I–V ) measurement was performed using a2-terminal method by applying a step voltage under constant temperature in a cryostat(Cryokelvin CG308SCPR; Nagase Electronics). The measurement temperature wasbetween room temperature and 5 K.

Page 51: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 ata temperature of 5.2 K. The voltage step with constant spacing is clearly seen. Thisis the characteristic of SET devices, and the step spacing is given by e/C . Here C isthe capacitance between the Coulomb island and the electrode [2]. The step spacinge/C is about 100 mV. In order to further clarify the step structure observed in the I–Vcharacteristic, the dV /dI–V characteristic was plotted in Fig. 3(b). For both positiveand negative voltage, peaks of dV /dI are seen at 50, 150 and 250 mV with an equalstep voltage spacing of 100 mV.

Page 52: Nanotechnology and Nano-Interface Controlled Electronic Devices

Single electron tunneling organic devices 35

Fig. 3: (a) Typical I–V characteristics of metal/organic SET layer/metal junctions. (b) Typical dV /dI–Vcharacteristics of metal/organic SET layer/metal junctions.

Page 53: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 fromroom temperature to 50 K. This indicates that a thermally activated conduction typecurrent flows through the junction. The thermal activation energy at a temperaturehigher than 50 K is estimated as less than 30 meV. On the other hand, the electricalcurrent density is nearly constant at a temperature lower than 50 K. This indicatesthat a tunneling-conduction type current flows through the junction in this temperatureregion. These results reveal that both thermally activated conduction and tunneling-conduction currents are allowed to flow in this device. It is also suggested that thethermally activated conduction type current dominates at temperatures higher than roomtemperature, whereas this type of current decreases as temperature is getting lower andlower. Eventually, the tunneling-conduction type current becomes the main contributor.Of interest is that a single electron tunneling characteristic is found at lower temperatureas shown in Fig. 3.

Fig. 5 shows the I–V characteristics at various temperatures. At a temperature lowerthan 50 K, tunneling-conduction type current is observed, and the step voltage can beseen. The voltage step width is the same as that observed at 5.2 K. The position of thestep voltage is not dependent on the temperature. As described above, in the junctionsusing Rh-G2 dendrite molecules, single electron tunneling behavior is observed. Fromthe theoretical side, the charging energy e2/2C associated with one electron tunnelingmust be greater than the thermal energy kT [2]. A voltage step due to single electron

Page 54: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 junctionsshowing 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 junctionactually functions as a Coulomb island or not. For this purpose, the size of the Coulombisland is estimated. The size of the Coulomb island is a dominant factor to explain theI–V characteristic. The size can be estimated from the spacing of step voltage of e/Cin the I–V characteristic. The capacitance C of a spherical conductor with radius r thatis separated by a distance d from a planar electrode is given by

C = 4πεrε0 F with F ≈ (1/r −1/2d)−1, (1)

under the assumption r � d. Therefore, assuming the observed voltage step spacing isgiven by �V , the radius r of the Coulomb island can be expressed as

r =(

4πεrε0

e�V + 1

2d

)−1

. (2)

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). Thesize of an Rh-G2 molecule is speculated to be about 1–2 nm in radius from computer

Page 55: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 sizeestimated from I–V characteristics and that from the computer simulation, possibly dueto the assumptions made in the calculation. This discrepancy is within the constraintsof our estimation. As mentioned above, using dendrite molecules designed by molecularensemble, 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 thedye 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 spacecharge layer. In other words, it is possible to control the single electron tunnelingprocess 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 positionand the spacing of step voltages. These are specific characteristics seen in the junctionsusing dendrimer molecules. These results suggest that the changes induced by lightirradiation did not originate from the Rh-G2 molecules working as the Coulomb islandbut from the PI tunnel barrier [9].

Page 56: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 asCoulomb 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 ofthe molecule will be added to the single electron tunnel characteristics. By choosingmolecules, double type SET devices and others that can be used in e.g. detectionof electromagnetic waves, light detection, etc. [10] will be produced. Furthermore,arranging the above double tunneling type device in another organic molecular matrixas shown in Fig. 7, the tunneling characteristics will be controllable by the field of thematrix with external stimuli, such as optical light or electrical field. As described above,by utilizing a molecular ensemble, building up organic electronics and molecular fieldcontrol type electronics will be possible.

References

1. K.K. Likharev, Proc. IEEE 87, 606 (1999).2. J.H. Fendler, Nanoparticles and Nanostructured Fulms, (WILEY-VCH, Weinheim) Ch. 15 (1998).3. Y. Noguch, Y. Majima, M. Iwamoto, T. Kubota, S. Yokoyama, T. Nakahama, and S. Mashiko, IEICE

Trans. Electron. E83-C, 1076 (2000).4. T. Kubota, S. Yokoyama, T. Nakahama, S. Mashiko, Y. Noguch, Y. Majima, and M. Iwamoto, Thin

Solid Films 393, 379 (2001).5. Y. Noguchi, Y. Majima, and M. Iwamoto, J. Appl. Phys. 90, 1368 (2001).6. Y. Noguchi, M. Iwamoto, T. Kubota, and S. Mashiko, J. Appl. Phys. 92, 1174 (2002).7. M. Iwamoto, T. Kubota, and M. Sekine, J. Phys. D 23, 575 (1990).8. S. Yokoyama, T. Nakahama, A. Otomo, and S. Mashiko, Chem. Lett. 11, 1137 (1997).9. E. Itoh, Y. Niwa, and M. Iwamoto, Thin Solid Films 284, 545 (1996).

10. T. Fujisawa and S. Tarucha, Jpn. J. Appl. Phys. 36, 4000 (1997).

Page 57: Nanotechnology and Nano-Interface Controlled Electronic Devices

This Page Intentionally Left Blank

Page 58: Nanotechnology and Nano-Interface Controlled Electronic Devices

Nanotechnology and Nano-Interface Controlled Electronic DevicesEditors: M. Iwamoto, K. Kaneto and S. Mashiko© 2003 Elsevier Science B.V. All rights reserved

CHAPTER 4

Spatial light confinement and laser emissionfrom a gain medium containing dendrimer

Shiyoshi Yokoyama a,b and Shinro Mashiko a

a Communications Research Laboratory and b PRESTO, Japan Science and TechnologyCorporation (JST), 588-2, Iwaoka, Nishi-ku, Kobe 651-2492, Japan

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 412. Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 433. Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 434. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

1. Introduction

Since the first reports on organic and polymeric materials that showed high opticalgain and stimulated emission properties, there has been growing interest in exploitinglaser applications. This growing interest is because of their wide wavelength tunabilityand processing flexibility in solutions [1–3], films [4–7], and fibers [8,9]. Dendriticmacromolecules, called dendrimers, are a new category of hyper-structured material[10]. Their long branching chain and the high degree of control over regular molecularweight created a three-dimensional structure that is roughly spherical or globular[11]. In optical applications, radiative action from the high-gain medium, containingsmall particles such as dendrimers, may be altered significantly under coherent opticalexcitation [12]. In the study reported here, we found that using a homogeneousgain medium containing dendrimers increases the stimulated emission efficiency andfacilitates fine-tuning the laser modes. We also identified an optical input–outputthreshold behavior above which laser emission with a linewidth of less than 0.1 nmwas observed, even though our optical system lacked a real optical cavity. The inputthreshold energy from the gain medium was much smaller than the energy from the puredye solution.

Organic laser dyes typically show a large fluorescence yield ranging from about 0.6to near the optimum 1.0. In spite of this large yield, the dye concentration in the dye

Page 59: Nanotechnology and Nano-Interface Controlled Electronic Devices

42S.Y

okoyama

andS.M

ashiko

Fig. 1: Details of materials and optical experiment. (A) Chemical structure of dendrimers, 1 and 2, and DCM. (B) Schematic of stimulated emission experimentsetup. (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.

Page 60: Nanotechnology and Nano-Interface Controlled Electronic Devices

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. Athigher concentrations, molecular aggregation, which forms dimers or higher aggregates,almost completely suppresses the fluorescence [13]. This is in contrast with the generaltendency of π-electron-conjugated chromophores, which easily aggregate to formcomplex structures. Therefore, a dye concentration of less than 10−3 mol/l is generallyused in laser operations.

2. Experiment

In order to obtain a higher gain medium for stimulated emission, we used a den-drimer, which encapsulates the laser dye inside and increases the dye concen-tration with little fluorescence quenching. The dendrimer used in this study waspoly(amidoamine) with 64 hydroxyl-terminated groups (Starburst® PAMAM-OH den-drimer, Dendritech, Inc.) 1 [14], and the laser dye used was 4-(dicyanomethylene)-2-methyl-6-(4-dimethylaminostyril)-4H-pyrane (DCM) (Fig. 1A). The DCM-doped den-drimer was obtained by mixing DCM and a dendrimer in a methanol solution. The DCMconcentration was varied between 2.0 and 12.0 mM, whereas the DCM/dendrimer ratiowas kept constant. As long as the DCM/dendrimer ratio was kept at 2.0, the fluores-cence 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 dendrimerwas thus a good host for the DCM, increasing its solubility and yielding high emissionefficiency.

We used a nitrogen laser (337 nm, pulse duration 4.0 ns, repetition rate 10 Hz) as theexcitation source for stimulated emission experiments. The excitation intensities werevaried between 0 and 20 μJ/pulse. A cylindrical lens focused the excitation beam intoa stripe, 200 μm × 5 mm on a quartz cuvette, which contained either the DCM anddendrimer mixture or a pure DCM solution in methanol (Fig. 1B). The emissions guidedalong 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 afunction of the excitation intensity gradually narrowed from relatively low excitationintensities up to 15 μJ/pulse (Fig. 2). The emission intensity, Ise, grew exponentiallywith 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 isthe 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 fittedin Fig. 2, inset. When the excitation intensity became high, the emission spectrum

Page 61: Nanotechnology and Nano-Interface Controlled Electronic Devices

44 S. Yokoyama and S. Mashiko

Fig. 2: Emission spectra from DCM/dendrimer (= 2/1, [DCM] = 2.0 mM) solution in methanol atincreasing 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 increasedlinearly as the excitation intensity increased.

A clear threshold behavior in the Ise vs. I plot (Fig. 4A) and a second decreasein the linewidth at higher excitation intensity (Fig. 4B) indicated the onset of laseraction. The strongly modulated spectrum, with numerous peaks that were evenly spaced,clearly indicated the resonant cavity modes. The resonant peaks had a linewidth thatwas 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 emissionintensities with polarization in longitudinal and lateral directions. The output beam waseasily visible, as shown in Fig. 1C. The interference pattern is clearly projected afterpassing 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/(2nL), where nis the refractive index and L is the optical length of the resonator or cavity [16]. Usingthe 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, andthe cuvette sides may have become a reflecting mirror. However, such a reflection (10mm separation) was inconsistent with the optical length estimated from the emissionspectrum. We attributed the resonant mode of the output beam to the spatial confinementof the emitted light in the slab laser. To clarify this spatial confinement, the near-field

Page 62: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 inmethanol at excitation intensity above the threshold. Inset: beam-intensity dependence of near-field patternmeasured at facet of cuvette in lateral a direction.

pattern of the laser beam was measured using side-imaging spectroscopy. We placed a10 μm pinhole near the face of the cuvette to monitor the emission intensity at a givenposition (Fig. 3, inset). The laser intensities were concentrated on the face of the cuvetteapproximately 140 μm in the lateral section. Assuming that the excitation stripe can actas a waveguide on the cuvette, forming a 140 μm-long slab laser, this optical length is inexcellent agreement with the resonant mode separation of L = 142 μm estimated fromthe spectrum. This indicates that the emitted light was confined by gain guiding withinthe stripe, resulting in laser feedback.

The dendrimer is a good host to encapsulate DCM; its concentration is increasedup to 12 mmol/l. Since the DCM/dendrimer has a high emission efficiency at variousconcentrations, laser emission intensities increased as the DCM/dendrimer concentra-tion was increased as shown in Fig. 4. More importantly, the lasing threshold intensitybecame much lower as the concentration of the DCM/dendrimer was increased.

DCM/dendrimer was found to be a high-gain medium for laser emission. Theexplanation of how the dendrimer acts as a small particle for the laser feedback isnot obvious. However, optical gain in the homogeneous medium may, in part, haveoccurred when dendrimer behaves as a small particle. The surface unit of dendrimer2 was modified with a hydrogen-bonding unit, such as a N-tert-butoxycarbonyl-L-phenylalanine. Because of this modification, dendrimer 2 achieved a highly densehydrogen-bond shell with solid-state characteristics [17], becoming a molecular particlewith a diameter of about 5 nm. Lasing actions from the DCM solution with or without

Page 63: Nanotechnology and Nano-Interface Controlled Electronic Devices

46 S. Yokoyama and S. Mashiko

Fig. 4: (A) Dependence of emission intensity on excitation intensity for DCM/dendrimer (= 2/1) solutionsat 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 DCMinside because of its hard shell structure. The threshold intensity of lasing becamevery small when the DCM solution contained dendrimer. It seems that dendrimerbehaves as a scattering center, increasing the optical gain. In fact, a demonstration ofstimulated emission from an inhomogeneous scattering medium, such as a microcavity

Page 64: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 opticaltechnologies can be used for emitting materials to make the optical devices simpleand small [2,19]. These systems focused on the random laser, where the feedbackmechanism of the laser emission is attributed to multiple scattering by the particleswhich keeps the light inside the scattering media for an extended period [19]. In thesestudies, though a submicrometer particle became a strong optical scattering center, itmay also have induced a large optical loss. Since, in our experiment, the diameter ofthe dendrimer was much smaller than the optical wavelength, the optical gain was largein the DCM/dendrimer media, while the optical loss due to the passive scattering wasinhibited by gain guiding. The significantly lower threshold of the laser action fromthe 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 insidethe gain medium. It is obvious that dendrimer produced multiple light scattering inthe homogeneous gain medium. We are tempted to attribute the phenomena to photonlocalization providing an optical feedback for the high-gain laser dye.

4. Conclusion

In conclusion, we have described the spatial light confinement in DCM/dendrimermedia, which generated a resonant mode in lasing action. The laser emission wascharacterized 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 polarizationabove the threshold; and (iv) increased coherency. Encapsulating the laser dye into

Page 65: Nanotechnology and Nano-Interface Controlled Electronic Devices

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. Theresults showed that a random optical system consisting of emitting materials, opticalexcitation, and cavity modes, can be used to fine-tune mirrorless optical devices, even insmall-device applications.

References

1. S. Qian, J.B. Snow, H. Tzeng, and R.K. Chang, Science 231, 486 (1986).2. N.M. Lawandy, R.M. Balachandran, A.S.L Gomes, and E. Sauvain, Nature 368, 436 (1994).3. S.V. Frolov, Z.V. Vardeny, and K. Yoshino, Phys. Rev. B 57, 9141 (1988).4. F. Hide, M.A.Díaz-García, B.J. Schwartz, M.R. Andersson, Q. Pei, and A. J. Heeger, Science 273,

1833 (1996).5. V.G.K. Fozlov, V. Bulovic, P.E. Burrows, and S.R. Forrest, Nature 389, 362 (1997).6. N. Tessler, G.J. Denton, and H. Frend, Nature 382, 695 (1996).7. J.H. Schon, Ch. Kloc, A. Dodabalapur, and B. Batlogg, Science 289, 599 (2000).8. S.V. Frolov, A. Fujii, D. Chinn, Z.V. Vardeny, K. Yoshino, and R.V. Gregory, Appl. Phys. Lett. 72,

2811 (1998).9. F. Marlow, M.D. Mcgehee, D. Zhao, B.F. Chmlka, and G.D. Stucky, Adv. Mater. 11, 632 (1999).

10. D.A. Tomalia, A.M. Naylor, and W.A. Goddard III, Angew. Chem. Int. Ed. 29, 138 (1990); inAdvances in Dendrite Macromolecules, edited by: G.E. Newkome (JAI Press, Greenwich, CT, 1994);J.M. Fréchet, Science 263, 1710 (1994).

11. 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).

12. A. Otomo, S. Yokoyama, T. Nakahama, and S. Mashiko, Appl. Phys. Lett. 77, 3881 (2000).13. G. Jones II, in Dye Laser Principle, edited by: F.J. Duarte and L.W. Hillman (Academic Press, San

Diego, 1990) Ch. 7.14. D.A. Tomalia, V. Berry, M. Hall, and D.M. Hedstrand, Macromolecules 20, 1164 (1987).15. A.E. Siegman, in Laser (Univ. Sci. Book, CA, 1986) Ch. 13.16. 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.

17. J.F.G.A. Jansen, E.M.M. de Brabander-van den Berg, and E.W. Meijer, Science 266, 1226 (1994).18. 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).

19. 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).

Page 66: Nanotechnology and Nano-Interface Controlled Electronic Devices

Nanotechnology and Nano-Interface Controlled Electronic DevicesEditors: M. Iwamoto, K. Kaneto and S. Mashiko© 2003 Elsevier Science B.V. All rights reserved

CHAPTER 5

Control of molecular selective-assemblingon 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 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 492. H2-TBPP and Au(111) as basic molecule and substrate . . . . . . . . . . . . . . . . . . . . . . . . . . . . 503. Nonplanar conformation and orientational ordering of H2-TBPP on Au(111) . . . . . . . 504. Selective aggregation of cyanophenyl-substituted porphyrins on Au(111) . . . . . . . . . . . 555. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60

1. Introduction

The realization of molecular nanodevices with advanced functions requires the de-velopment of new approaches to construct desired molecular nanostructures [1,2].Supramolecular approach starting from molecular building blocks can lead to controlledstructures [3], which is achieved by selective and directional intermolecular interac-tions. When non-covalent intermolecular interactions such as hydrogen bonding areintroduced into functional molecules, the selective intermolecular interaction results inthe controlled formation of molecular nanostructures, which have yielded exclusivelycrystals or dissolved structures [3]. To adapt the functional supramolecular structures tonanodevices, it should be necessary for the supramolecular structures to be supported onsuitable substrates and at suitable positions.

On substrate surfaces, atomic-scale investigation of adsorbed molecules has been en-abled by using scanning probe microscopy, particularly scanning tunneling microscopy(STM) [4]. In particular, recent advance of high-resolution STM imaging allows oneto directly determine their arrangement, configuration, and conformation of individuallargish molecules on surfaces [5–9]. Based on the submolecular-resolution STM studies

Page 67: Nanotechnology and Nano-Interface Controlled Electronic Devices

50 T. Yokoyama et al.

and on theoretical calculations, several groups have been reported specific interactionsof adsorbed molecules on metal surfaces [10–12], revealing self-assembled aggregationon surfaces.

Whereas these surface-supported supramolecular structures have been directly ob-served using STM, the further control of their size and shape on surfaces should becomea next step for realizing molecular nanodevices. In this work, we demonstrate selectiveassembly of supramolecular aggregates with controlled size and shape on a gold surfaceby 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-tertiarybutyl-phenyl (tBP) substituents, as shown in Fig. 1(a). An ideal shape of H2-TBPP exhibits aplanar macrocyclic conformation of the central porphyrin through 60°–90° rotation ofthe phenyl rings with respect to the phenyl mean plane, as shown in Fig. 1(b), obtainedfrom semi-empirical molecular orbital calculations. By the bulky tBP substituents,the aromatic π system of the central porphyrin should be sterically decoupled to thesubstrate surface even after adsorption, fulfilling the requirements for the molecularnanoelectronic or optoelectronic devices [5,6,13]. To understand the conformation andarrangement of the H2-TBPP molecules, all experiments were performed in an ultrahigh-vacuum (UHV) chamber with a low-temperature STM. A Au(111) surface was used asa substrate because of its inertness and its properties of reconstruction. The atomicallyclean surface of Au(111), which was formed by deposition of Au on mica in UHV, wasprepared by repeated cycles of Ar+ sputtering and annealing at 700 K. Fig. 2 shows anSTM 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, andthe preferential nucleation of adsorbates at the elbows has been observed in varioussystems [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 temperatureby sublimation from a Knudsen cell in UHV. The sample was then subsequentlytransferred to the cooled STM stage for direct observation. Fig. 3(a) shows an STMimage of the Au(111) surface at 63 K after a small amount of the H2-TBPP deposition. Inthis image, most of the molecules are located at the elbows of the surface reconstructedpatterns, revealing regular arrays of isolated single molecules. In this system, it shouldbe noted that the selective molecular positioning allows direct imaging of isolated singlemolecules without intermolecular interactions. Fig. 3(b) shows the high-resolution STM

Page 68: Nanotechnology and Nano-Interface Controlled Electronic Devices

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. Theabout 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 exhibitsfour 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

Page 69: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 associatedwith domain walls between fcc and hcp stackings, as indicated by dashed lines, and the herringbone patternsare formed by periodic rotations of the uniaxial domains.

the paired lobes suggests that the phenyl rings are oriented close to the porphyrinmacrocyclic 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 anddarker ones, and the lateral distance is estimated to be about 4.5 Å. By comparing theSTM results with the molecular model, we have derived the dihedral angle between theporphyrin and phenyl rings to be about 20°, where the four phenyl rings are alternatelyrotated with respect to the porphyrin mean plane. The rotations of the phenyl–porphyrinbonds 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 rotationalflexibility of the phenyl–porphyrin bonds allows the tertiary-butyl substituents to fit intothe surface geometry, leading to the conformational changes.

In the STM image of Fig. 3(b), the most distinguishing feature is that the internalstructure of the central porphyrin has been resolved in the STM image, which iscomposed of the two oblong protrusions. We observed that the STM images wereindependent of the bias polarity, suggesting that the atomic structure was mainlycontributed in the STM image, compared with the electronic structure. Thus, theoblong protrusions should be associated with the nonplanar deformation of the centralporphyrin, induced by the rotations of the phenyl-based substituents. To confirmthe nonplanar macrocyclic conformation, we performed the semi-empirical molecularorbital calculations with the AM1 hamiltonian [18]. Fig. 3(c) shows the calculatedconformation of H2-TBPP with fixed 20° rotations of the four phenyl rings with respectto the central porphyrin. The relaxed structure shows that the 20° alternate rotations

Page 70: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 theH2-TBPP deposition. A regular array of single molecules results from preferential adsorption at the elbowsof the herringbone patterns. (b) High-resolution STM image (2.1 nm × 2.1 nm) at 63 K of a single H2-TBPPmolecule, which is composed of four paired lobes surrounding two oblong protrusions. (c) Top view ofcalculated macrocyclic conformation of the H2-TBPP molecule with 20° alternate rotations of the phenylrings, obtained using semi-empirical molecular orbital calculations. A saddle-shaped nonplanar deformationof 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, whilethe 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

Page 71: Nanotechnology and Nano-Interface Controlled Electronic Devices

54 T. Yokoyama et al.

porphyrin is characterized by alternately tilting of pyrrole rings above and below themean 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 estimatedto 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 beassociated with the tilt-up pyrrole rings of the nonplanar porphyrin macrocycle, whereasthe tilt-down pyrrole rings weakly appear as bridges between two oblong protrusions. Inaddition, 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 resultshould exclude a possibility that the two oblong protrusions are related to the electronicasymmetry of the central H2-porphyrin.

With increasing coverage of H2-TBPP onto the Au(111) surface, two-dimensionalislands are formed through self-assembled aggregation. As shown in Fig. 4(a), we haveobserved larger islands even without thermal annealing, suggesting a low diffusion bar-rier of the adsorbed molecules on the surface. In the islands, the nonplanar macrocyclicconformation remains, and the molecules exhibit a close-packed arrangement on thesurface [Fig.4(b)]. A detailed analysis of the STM images indicates that the moleculararrangement 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 formedon the Au(111)-1 × 1 structure, where intermolecular interactions should be governedby 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 differentorientations of the nonplanar porphyrins are randomly distributed within the island. Wefind 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 (for1 min at about 470 K). Inside the island, the orientations of neighboring molecules arethe same in the [112] direction and rotated by 90° in the [211] direction, as illustratedin Fig. 5(d). It should be noted that the orientational ordering is accompanied withouta change of the molecular arrangement, and such the orientations have been observedover the surface. Because it appears that the degree of the orientational orderingdepends on the annealing temperature and time, the transformation should be relatedto rapid molecular diffusion on the surface, promoting breaking and rearrangement ofthe molecular islands. This kinetic process might allow the molecular islands to berearranged in a stable manner.

The orientational ordering should be associated with steric intermolecular interactionsbetween the tBP substituents. Because of the 20° rotations of the phenyl rings, twopossible steric interactions are expected between tBP substituents; cross- and parallel-type interactions of the tBP substituents between neighboring molecules in both the[11 0] and the [112] direction. The ordered orientations indicate that the cross-typeinteraction (up–down connections of each tertiary-butyl substituent) should be morestable than the parallel-type one (up–up and down–down connections), although theenergy difference might be extremely small.

Page 72: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 an11 × 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 intermolecularinteractions between tBP groups. This result indicates that the molecular assemblyformed on a surface can be controlled by changing substituents. In supramolecularchemistry, a large number of different selective and directional intermolecular interac-tions has been developed to control molecular aggregation, although it has been focusedmainly on dissolved structures [3]. In this work, we have used a cyanophenyl substituent

Page 73: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 annealingat about 470 K for 1 min. (c) and (d) Schematic illustrations of the molecular arrangement and orientationswithin the island [17].

to control the molecular aggregation [19,20], because it has a simple and symmetricstructure as well as an asymmetric charge distribution at the cyano group that shouldintroduce dipole–dipole interactions between neighboring cyanophenyl substituents. Anillustration of this interaction is given in Fig. 6, which shows the optimized arrangementof a cyanobenzen dimer and trimer obtained in ab initio molecular orbital calculationsat the MP2/6-31G* level [20,21]. In the dimer, the cyano groups have an antiparallelconfiguration, whereas the trimer structure results from a cyclic arrangement of thecyano groups. The length of the CH. . .NC contacts is about 2.39 Å and 2.65 Å forthe dimer and trimer, respectively, and thus shorter than the van der Waals distance ofabout 2.7 Å. The interaction energies are estimated to be −7.12 kcal/mol and −12.40kcal/mol for the dimer and trimer structures, respectively, and thus comparable to theenergy of hydrogen-bonding interactions. The relative orientation of molecules in theseaggregates 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 synthe-sized 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 shownin Fig. 7(a). At low coverage, most CTBPP molecules assembled into triangular clusters

Page 74: Nanotechnology and Nano-Interface Controlled Electronic Devices

Control of molecular selective-assembling on metal surface 57

Fig. 6: Calculated molecular aggregations of (a) a cyanobenzen dimer and (b) trimer, which are obtainedfrom ab initio molecular orbital calculations [19].

on the Au(111) surface. As shown in Fig. 8(a), identical clusters are located separatelyat 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 apaired lobe corresponds to one di-tertiarybutyl substituent. The high-resolution STM

Page 75: Nanotechnology and Nano-Interface Controlled Electronic Devices

58 T. Yokoyama et al.

Fig. 7: Structural formula of the cyanophenyl-substituted porphyrins (a) CTBPP, (b) cis-BCTBPP, and (c)trans-BCTBPP [19].

Page 76: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 cyanophenylsubstituents 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 thecharacteristic aggregation of CTBPP, we have not observed such supramolecular clus-ters for phenyl-tris(3,5-di-tertiarybutylphenyl) porphyrins, where the cyano substituentswere removed from the CTBPP molecule. In this case, a random arrangement of themolecules was formed on the surface, confirming that the supramolecular aggregation isdominated by the cyano groups.

Page 77: Nanotechnology and Nano-Interface Controlled Electronic Devices

60 T. Yokoyama et al.

To probe the supramolecular aggregation in our system further, we substitutedone 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 tothe 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 cyanophenylsubstituents lead to a macrocyclic arrangement of the porphyrin molecules, forming amolecular ring [Fig. 8(f)].

In contrast to the macrocyclic clusters of CTBPP and cis-BCTBPP, sequentialaggregation was achieved for the trans-BCTBPP molecules, where two cyanophenylgroups were substituted at the trans positions. Fig. 8(g) shows the STM image of theAu(111) surface after deposition of the trans-BCTBPP molecules. In this structure,the antiparallel configuration between the cyanophenyl substituents results in a lineararrangement 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 ofthe herringbone patterns, because the molecules were initially nucleated at the elbows.Furthermore, it is remarkable that the maximum length of the straight wire was above100 nm, although some branches are also formed due to the three-fold symmetry ofAu(111).

5. Summary

The controlled aggregation of porphyrins has succeeded on a gold surface, which wasvisualized by low-temperature STM. On this surface, monomer, trimer, tetramer, or wire-like arrangements were controlled by local substituent interactions, and these structureswere spontaneously and selectively formed by changing substituents. We believe theselective aggregation approach should become a general strategy for the rational designand construction of desired molecular architectures on substrate surfaces.

References

1. Y. Wada, M. Tsukada, M. Fujihira, K. Matsushige, T. Ogawa, M. Haga, and S. Tanaka, Jpn. J. Appl.Phys. 39, 3825 (2000).

2. C. Joachim, J. K. Gimzewski, and A. Aviram, Nature 408, 541 (2000).3. J.-M. Lehn, Supramolecular Chemistry: Concept and Perspectives (VCH, Weinheim, 1995).4. H.-J. Guntherodt and R. Wisendanger (editors), Scanning Tunneling Microscopy I (Springer-Verlag,

1994).5. T.A. Jung, R.R. Schlittler, J.K. Gimzewski, H. Tang, and C. Joachim, Science 271, 181 (1996).6. T.A. Jung, R.R. Schlittler, and J.K. Gimzewski, Nature 386, 696 (1997).7. G.P. Lopinski, D.J. Mofatt, D.D.M. Wayner, and R.A. Wolkow, Nature 392, 909 (1998).8. M.O. Lorenzo, C.J. Baddeley, C. Muryn, and R. Raval, Nature 404, 376 (2000).9. F. Moresco, G. Meyer, K.-H. Rieder, H. Tang, A. Gourdon, and C. Joachim, Phys. Rev. Lett. 86, 672

(2001).10. M. Frukawa, H. Tanaka, K. Sugiura, Y. Sakata, and T. Kawai, Surf. Sci. Lett. 445, L58 (2000).

Page 78: Nanotechnology and Nano-Interface Controlled Electronic Devices

Control of molecular selective-assembling on metal surface 61

11. J.V. Barth, J. Weckesser, P. Gunter, L. Burgi, O. Jeandepeux, and K. Kern, Angew. Chem. Int. Ed. 39,1230 (2000).

12. M. Bohringer, K. Morgenstern, W.-D. Schneider, R. Berndt, F. Mauri, A.D. Vita, and R. Car, Phys.Rev. Lett. 83, 324 (1999).

13. K. Sugiura, K. Iwasaki, K. Umishita, S. Hino, H. Ogata, S. Miyajima, and T. Sakata, Chem. Lett. 841(1999).

14. U. Harten, A.M. Lahee, T. Peter, and Ch. Woll, Phys. Rev. Lett. 54, 2619 (1985).15. J.V. Barth, H. Brune, G. Etrl, and B.J. Behm, Phys. Rev. B 42, 9307 (1990).16. D.D. Chambliss, R.J. Wilson, and S. Chiang, Phys. Rev. Lett. 66, 1721 (1991).17. T. Yokoyama, S. Yokoyama, T. Kamikado, and S. Mashiko, J. Chem. Phys. 115, 3814 (2001).18. M.J.S. Dewar, E.G. Zoebisch, E.F. Healy, and J.J.P. Stewart, J. Am. Chem. Soc. 107, 3092 (1985).19. T. Yokoyama, S. Yokoyama, T. Kamikado, Y. Okuno, and S. Mashiko, Nature 413, 619 (2001).20. Y. Okuno, T. Yokoyama, S. Yokoyama, T. Kamikado, and S. Mashiko, J. Am. Chem. Soc. 124, 7218

(2002).21. M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb, J.R. Cheeseman, V.G. Za-

krzewski, 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).

Page 79: Nanotechnology and Nano-Interface Controlled Electronic Devices

This Page Intentionally Left Blank

Page 80: Nanotechnology and Nano-Interface Controlled Electronic Devices

Part B

NICE Devices

Page 81: Nanotechnology and Nano-Interface Controlled Electronic Devices

This Page Intentionally Left Blank

Page 82: Nanotechnology and Nano-Interface Controlled Electronic Devices

Nanotechnology and Nano-Interface Controlled Electronic DevicesEditors: M. Iwamoto, K. Kaneto and S. Mashiko© 2003 Elsevier Science B.V. All rights reserved

CHAPTER 6

Polymer optoelectronics – towardsnanometer 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 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 652. Excited states in polythiophenes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 663. Diffusion length of excited states . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 694. Stratified photodiodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 725. Excitation transfer in photodiodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 736. Models of charge generation in photodiodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 757. Nanodimension of electrodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 758. Nano-pattern application in photovoltaic devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80

1. Introduction

The unification of the natural sciences in a form now widely being called nanosciencesis one of the themes of the late 20th century scientific enterprise, and may well becomea dominant theme in the 21th century. This reunification of the natural sciences, dividedby processes of specialisation in the late 19th century, is central to the developmentof the field of conjugated polymers at the junction of physics and chemistry. Herechemistry delivers materials in the forms of conjugated polymers and molecules, usingwell-established methods of synthetic organic chemistry and polymer chemistry. Theobjects of synthesis are defined on the Å and nm length scale, and are synthesised insamples carrying millimoles of objects. The objects are quasizero- or one-dimensionalelectronic systems with a high degree of excitation and charge confinement, but also witha high mobility on the nm length scale. Physics offer one description of the electronicstructure of these systems, complementary to the quantum chemical description butall based in quantum mechanics; the language in which these objects are describedmay however differ between chemistry and physics. The description of the excitedstates, so crucial to the development of optoelectronic devices for emission of light by

Page 83: Nanotechnology and Nano-Interface Controlled Electronic Devices

66 O. Inganäs and F. Zhang

electronic injection of charge into molecules and polymers, or for creation of chargesfrom excited states, is an area of controversy, and where physical and chemical modelsmeet and collide. While these models may be fought over, using sophisticated modelsand spectroscopies, there is in the background a steady development of new molecules,materials and devices, which sometimes leads to results contradicting current dogma. Aparticular enigmatic example of these controversies is the issue of charge recombinationin organic layers, where a simple argument from symmetry shows that electrons andholes recombining in a solid should lead to singlet and triplet excited states in a ratioof 1 : 3. More recent experimental determinations do not agree with this rule [1], andthe ratio has been shown even to go beyond 1 : 1. As the ratio is now considered tobe a material and device parameter, rather than a fundamental property of recombiningcharges, one of the ultimate limiting factors of light emitting diodes is no longerpresent. Likewise, in the field of organic photodiodes, a central issue is the bindingenergy of the excited state, as it has a clear impact on the formation of mobile chargesin photodiodes. No clear consensus is yet found, and the range of binding energiesreported from experimental studies is broad. Internal quantum efficiencies of excitedstate dissociation into charges can be very high in donor–acceptor systems, and thekinetics of these processes is extremely fast [2]. In pure homopolymer systems, routestowards photogeneration of charges are present [3], but much less efficient than whatis found in donor–acceptor complexes where an electron acceptor receives an electronfrom the excited state on a conjugated polymer or molecule [4,5]. However, progress inorganic photodiodes has now generated photodiodes with external quantum efficiency atlow intensity monochromatic illumination of better than 60%, and solar cells with solarenergy efficiencies of 2.5% at AM 1.5 conditions [6–9]. A steady development of newmaterials, new devices and new patterning methods contributes towards incrementalprogress in this field. The length appropriate for analysis and characterisation in thisfield 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 forthe description of multilayer devices where a photoactive solid is confined betweenelectrodes separated by this thickness, and to 100–1000 nm, where electrodes andphotoactive 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 asthin films on flexible carriers of cm to dm dimensions, and which may one day be foundas an organic solar cell for energy conversion. Here conditions for large or giant areaelectronics are crucial, and it is of great importance to develop green chemistry and largearea deposition methods. This chapter focuses on some of our recent contributions tothe field of organic photodiodes, with particular emphasis on the nanometer dimensionsof materials, processes and devices.

2. Excited states in polythiophenes

We have chosen polythiophenes (PT) as our main group of conjugated polymers forphotovoltaics [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

Page 84: Nanotechnology and Nano-Interface Controlled Electronic Devices

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.

Page 85: Nanotechnology and Nano-Interface Controlled Electronic Devices

68 O. Inganäs and F. Zhang

of polymer modification by chemical substitution, and from the point of view of thehigher photostability of polythiophenes as compared to poly(paraphenylene vinylenes),for example. It is thus quite easy to move the optical absorption over much of thevisible spectrum, and the range of polythiophenes covers a sizable part of the solarspectrum. Another reason is the possibility of making high mobility polymers, as shownin recent years where poly(alkylthiophenes) have been induced to have high field effectmobility in polymer transistors, and even the possibility to field dope into the metallicand superconducting regime [11].

The nature of the excited state in PTs has been extensively studied through photo-luminescence (PL) quantum yield and photoluminescence and photoinduced absorptionkinetics [12,13]. The radiative lifetime of the excited state is typically 1–2 ns, andone source of decay is through internal conversion. That process is much enhancedin 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 areless significant for photovoltaics. More important are the low bandgap PTs, whereaggregates have been shown to lead to non-radiative recombination of excited states[13]. These non-radiative processes are not sufficiently fast to outcompete the processof photoinduced charge transfer in photodiodes. The possibility of intersystem crossingin PTs, induced by the spin–orbit coupling due to the presence of a sulphur atom in thecarbon conjugated chain, is of little importance in photovoltaics, as it appears that bothsinglets and triplets may be dissociated by the presence of acceptors.

The diffusion of the excited state is much influenced by the chemical structure ofthe polymer, as well as by the morphology of the polymer when aggregating into solidfilms. The typical behaviour is a drastic loss of PL quantum yield upon precipitationof PTs from good solvents. There are important alternatives; we have observed anenhanced PL yield of some soluble polythiophenes [14] when decreasing the solventquality on the route to the solid, where once more a general reduction of PL yieldis found. It may be that special aggregation forms may retain and even increase thequantum yield, but most forms of aggregation lead to an increase of the non-radiativedecay and thus quench luminescence. The spectral migration of excited states reveals adiffusion of these, which will couple the excited state to some source of non-radiativerecombination, which is never far away in a polymer film. Copolymers and oligomersincorporating chemically modified thiophene units have been shown to have a muchenhanced PL quantum yield [15].

The geometry of the polymer chain in good solvents has been studied by photophys-ical methods, where a sub-picosecond pulse of (polarized) monochromatic light is usedto excite polymer chains in solid films or in solutions; during the coming picosecondsthe intensity or polarisation state of the rapidly decaying luminescence is observed. Inthese studies is revealed how transport of the excited state along the single polymerchain in a dilute solution in a good solvent occurs [16]. This study reveals a tendency ofthe low bandgap PTs to give a stiff, close to planar, geometry as expected from the pointof theory. When now packing these chains into a polymer solid, a memory of the chainconformation in the solvent may persist, as argued in other studies.

Page 86: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 importantparameters controlling the design of photodiode materials. If the excited state is capableof finding a site for non-radiative recombination prior to finding a site for photoinducedcharge transfer, potential photocurrent is lost. Determination of the exciton diffusionlength is therefore an important item.

We have used a combination of photoluminescence measurements and opticalmodelling to deduce the diffusion length of the excited state which can be quenched bythe presence of C60 molecules [17]. In order to be able to do this experiment, we havechosen a polythiophene, which does not easily dissolve fullerene compounds, as to beable to obtain sharp bilayers of polythiophene/C60, where the fullerene is evaporatedonto a thin spin-coated polymer film, carried on a quartz substrate. These methods ofdeposition give reasonably flat films, and these films can be used to obtain the fulldielectric function of the materials. To do this we use spectroscopic ellipsometry, anon-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 spheremeasurements, and vary the polymer and C60 thickness. As the thickness of the polymerfilms increases, more excited states are created at a distance from the polymer/C60junction, and higher PL is obtained. With these methods, and with a detailed opticalmodel 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 be5 nm in the polymer studied. This is the distance in which an acceptor must be foundin order for an excited state to be able to generate charge. The diffusion length is amaterials property, and may also be influenced by morphology and packing of chains. Itis therefore an object of design and synthesis.

Fig. 2 shows the distribution of absorbed electromagnetic energy from an incidentmonochromatic 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 distributioncompared to that obtained from a simple-minded application of the Lambert–Beer lawas often used. The excitation or exciton distribution is actually not completely identicalto the distribution of excited states formed by absorption, as exciton diffusion may havean impact, in particular close to boundaries. Even more is this physics relevant in theemission of photons from excited states within the polymer layer, a distribution thatmust also be included in the complete analysis of the emission profile versus wavelengthin the far field.

We may also approach this problem of quenching of excited states in the presence ofacceptors by constructing objects where the excited state should always find an acceptorwithin 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 photogeneratedcharges, as they will hop between occupied and unoccupied states on the electron acceptormoiety and electron donor moiety, respectively. The term double cable refers to thisproperty, and by incorporation of these elements on each polymer chain means for excitedstate dissociation and for charge transport are found everywhere in the material.

Page 87: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 excitedat 2.34 eV, normalised to the incoming intensity. The axes thus show the fraction of the incoming intensityabsorbed 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 anexponential decay of the incoming radiation (e.g. I = I0 e−αz) is shown (thick solid line). Also shown is thesteady-state power density with an infinite sink at the polymer/C60 heterojunction and a diffusion length of5.3 nm (broken dashed line). The upper panel shows the calculated transmission of the polymer emissionfrom 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 nmC60 on top (dashed line). (From Ref. [17].)

A recent example of double cable polymers is found in a soluble polythiophene–C60copolymer, where copolymerisation of substituted thiophenes with a fraction of C60carrying substituted thiophenes leads to the polymer. Photoinduced charge transfer iseasily observed in these polymers, and is almost complete in the solid and partial in the

Page 88: Nanotechnology and Nano-Interface Controlled Electronic Devices

Polymer optoelectronics – towards nanometer dimensions 71

Fig. 3: PL spectra and their PL yield from pure POMeOPT film and copolymer POMeOPT with 10% C60film 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 oneC60 per 4 monomers of thiophene in the copolymer. The processing through polymersolutions 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

Page 89: Nanotechnology and Nano-Interface Controlled Electronic Devices

72 O. Inganäs and F. Zhang

C60 into a common solvent. This will lead to polymer films with enhanced photodiodeperformance [22], indicating that the density of C60 is not sufficient in the copolymerper se.

4. Stratified photodiodes

The sequel to photoinduced charge separation in photodiodes is the transport ofgenerated charges to the electrodes. This should preferably happen without the trappingor recombination of these charges, which are very plausible events in the materialsconcerned. Localisation of charges into traps is very likely due to the high degreeof disorder; recombination is likely when a stream of charges passes through fixedspace charges of the opposite polarity. It is therefore essential to organise the pathsof the electron and hole so that they can be transported with minimal trapping andrecombination.

Blends of donor and acceptor do not always deliver this. The size of domains inwhich donors and acceptors are aggregated is influenced both by kinetics and statics ofblending, 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 performanceof devices. The process of removing solvent from a blend may be fast – as in spin-coating – or slow, such as in dropcasting or doctor blading. The phase structure of theresulting film is probably very different under these different circumstances. We notethat the highest solar cell efficiencies are reported [9] for films prepared from particularblends of solvents, where the change of composition during evaporation of the differentsolvents at different rates is yet another mechanism that will influence the kinetics ofthin film formation and molecular organisation/morphology in the film. By influencingthe degree of phase separation by purely chemical means, as for example when usinga “immobilised solvent” as a side chain of a polythiophene, molecular miscibility ofpolymer/C60 is obtained, and the resulting phase structure does not show up above20–30 nm [23].

To gain some small means of control of the 10–100 nm organisation of materialsin the vertical dimension, we have developed stratified photodiodes [24,25]. These areassembled by using a soluble fullerene derivative to be spin-coated on top of a polymerfilm, previously deposited by spin-coating on a substrate. The structure of the stratifiedactive layer in photodiodes is shown in Fig. 4.

The second act of spin-coating could of course be expected to dissolve the firstlayer, unless solvents and conditions are carefully chosen so as not to dissolve the firstlayer when depositing the second. This has been possible for bilayers of high molecularweight PPV and soluble methanofullerene compounds, which form graded junctions[25]. The diffuse junctions of the polymer phase at the bottom, with a methanofullerenephase on top, is much more effective in generating photocurrent. This is very visiblein the photoluminescence of the polymer layer, which is almost completely quenchedin the presence of the methanofullerene, this indicating a very effective dissociation ofexcited states. The increase of photocurrent, due to the higher contact area betweendonor and acceptor molecules, now must be transferred through layers, which eventually

Page 90: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 contactsat the cathode side.

5. Excitation transfer in photodiodes

The collection of optical energy is in green plants very much helped by the constructionof the photosynthetic apparatus. Here antenna pigments in a primary step absorb theincoming photons. That energy is transferred to a reaction center chlorophyll in reactioncenters PSI and PSII, wherein the primary photochemistry occurs. The nature of theseantenna pigments varies, and they may be chlorophylls but they may also be distinctlydifferent, for instance in the form of xanthenes. The collection of energy from antennapigments to reaction center chlorophyll occurs through excitation transfer, also knownas Förster transfer. One advantage of this system is that more of the photon flow maybe absorbed, through the use of pigments of different absorption spectra. The sameadvantage is something we would like to utilise in photodiodes, to sensitize some partof the green and blue part of the solar spectrum. This requires the use of severalpigments, and we have implemented this in the form of polymer blends. In a study ofthree different red polythiophenes – absorbing in the green part of the spectrum – incombination with a polyparaphenylene vinylene absorbing in the blue-green, we havedemonstrated that this principle may be used to enhance the action spectrum of thephotodiodes [26].

These devices were built as bilayer photodiodes, where an evaporated C60 layer isthe acceptor located on top of a polymer blend layer. We compare the performanceof photodiodes incorporating blends with photodiodes where the polymer layer isa homopolymer. The high bandgap polymer shows an emission spectrum largelyoverlapping with the absorption spectra of the lower bandgap polythiophenes. Bystudying the photoluminescence spectrum of homopolymers and blend, we establishthat almost complete excitation transfer occurs in the polymer blends of 1 : 1 (byweight). This is very visible in the complete suppression of emission from the PPV highbandgap polymer, and in the considerable enhancement of PL from the polythiophenesused. This enhancement is very visible as the emission intensity from the three differentpolythiophenes is not high. Photoluminescence and excitation transfer in photodiodes is

Page 91: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 ratio1 : 1.

shown in Fig. 5. Therefore we conclude that geometries suitable for excitation transferare obtained in these blends obtained by spin-coating from a common solvent. Wecorroborate this interpretation by imaging the blend layer surface in scanning forcemicroscopy, with the tapping mode of interaction between tip and surface. We areunable to resolve any fine-structure in these blends, which may be due to structuresbeing smaller than the limits of resolution of the method, say 20–30 nm, or by acomplete phase separation into a pancake geometry, where one of the polymers wouldbe residing on top of the other. This does not appear to be a likely geometry, butis not excluded by our studies. The total absorption of the film is such as to makeimplausible that the degree of excitation transfer that is observed could be originatingfrom a double layer, as the individual layers would have to be too thin to generatethe optical absorption observed. It is therefore plausible that the dimensions of phaseseparation are small, and that most PPV chains are found close enough to a PT chain toallow excitation transfer.

In photodiodes incorporating polymer blends we find consistently higher photocur-rents and lower photovoltages. The photovoltages observed are similar to those thatwould be found in homopolythiophene/C60 layers, and we therefore conclude thatphotocurrent generation occurs through excitations found on the polythiophene chains,but to some degree delivered there by excitation transfer from the PPV. An enhancement

Page 92: Nanotechnology and Nano-Interface Controlled Electronic Devices

Polymer optoelectronics – towards nanometer dimensions 75

of photocurrent and an extension of the action spectrum is the result. We may in thiscase argue that the characteristic length of Förster transfer is smaller or comparable tothe phase separation dimensions. Later and more extensive studies of polymer blends inother polymer families have revealed a multitude of length scales in the phase separationbetween two polymers [27,28].

6. Models of charge generation in photodiodes

The same thin film optical phenomena found valuable in the analysis of photolumi-nescence quenching in polymer/C60 bilayers is also of importance in analysing theperformance of photodiodes constructed by sandwiching this bilayer in between twoelectrode layers, one metallic and highly reflective and one conducting but almosttransparent [29,30]. For all the materials in this photodiode, dielectric functions havebeen measured with the help of spectroscopic ellipsometry. With help of a detailedoptical model of all internal reflections and transmission at internal boundaries, we areable 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 differentthicknesses of C60. The distribution of optical absorption is thus known, and we cancouple this to the generation of photocurrent in the device by including this distributionas a source term in a diffusion equation appropriate for neutral excited states. With fixedthickness and optical parameters, due to the spectroscopic ellipsometrical data, we haveonly one free variable in fitting our real data of the external quantum efficiency of thedevices; that is the diffusion length of the excited state contribution to the formationof photocurrent. We implicitly assume many things, in particular that the photoinducedcharge transfer occurs only at the interface between polymer and C60. This assumptionis identical in PL quenching modelling experiments, and both types of experiment endup with an exciton diffusion length of 5 nm. Considering this short distance, we notethat we must position all polymer chains within a distance of less than 5 nm in order totransfer 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 bepatterned at a sub-micron scale; the vertical dimension of the device is almost alwaysless than a few hundreds of nm. In order to pattern at a nano scale, in additionto the traditional photolithography processing, soft lithography was developed and isnowadays used as a suitable method for patterning polymers [31,32]. Micro- or nano-patterning of polymers is not only necessary for electrical addressing and wiring ofthe circuits, but also has other functions, such as trapping more light to improve theperformance of polymer photodiodes [33] or increasing the efficiency, and controllinglight output from microstructured LEDs [34,35].

The conducting polymer poly(3,4-ethylenedioxythiophene) doped with poly(4-styr-enesulfonate) (PEDOT-PSS) is often used as a metal in polymer electronic devices,

Page 93: Nanotechnology and Nano-Interface Controlled Electronic Devices

76 O. Inganäs and F. Zhang

Fig. 6: Calculated distribution of the normalised modulus of the optical electric field |E|2 inside aphotovoltaic device: glass (1 mm)/ITO (120 nm)/PEDOT (11O nm)/PEOPT (40 nm)/C60/Al with a C60layer 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 enhancedperformance [36] and for improvement of the rectification ratio in polymer diodesPEDOT-PSS can be modified with glycerol or sorbitol to increase the conductivityby two orders of magnitude. This makes it possible to use this material as a flexibleelectrode for application in optoelectronic devices. The PEDOT-PSS has double func-tions in electronic devices: to enhance the electrode performance by adjusting the workfunction of an electrode [37] and for use as a flexible anode [38]. We here introduceour experimental results on nano-patterning of conducting polymer PEDOT-PSS, bothon glass and Si wafer, patterned by a soft lithography technique, MIMIC and liquidprinting, 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 Siwafers, by using capillary action in MIMIC structures. Two different elastomer submi-cron patterns were made by casting polydimethylsiloxane (PDMS) onto commerciallyavailable diffraction gratings. We put the elastomer stamp in conformal contact with apiece of cleaned glass or an Si wafer, and then applied a drop of PEDOT-PSS (BaytronP, Bayer AG, concentration 1.3%) aqueous solution in front of the capillary openings of

Page 94: Nanotechnology and Nano-Interface Controlled Electronic Devices

Polymer optoelectronics – towards nanometer dimensions 77

Fig 7: SFM image of polymer PEDOT-PSS nanowires and chemical structure of PEDOT-PSS. The period ofthe nanowires was 270 nm and the height ≈25 nm. The cross section picture shows the profiles of the wiresand their height.

the stamp. The sample was then left for several hours, for the solution to migrate into thecapillaries and subsequently dry out. After the solution had dried we carefully peeledoff the stamp leaving the polymer nanowires standing on the surface. In this way thegrating pattern was transferred from the stamp to the polymer layer. All the processeswere performed in ambient condition.

A scanning force microscope (SFM-Nanoscope III, Digital Instruments) was used intapping mode to image the polymer nanowires. Fig. 7 shows SFM images of polymerPEDOT-PSS nanowires (3600 lines per millimeter) with the same period as the stamp

Page 95: Nanotechnology and Nano-Interface Controlled Electronic Devices

78 O. Inganäs and F. Zhang

Fig. 8: SFM image of nanowires with a period 833 nm molded with a compressed stamp, showing a heightof 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 originalstamp (∼55 nm).

In order to make polymer nanowires of lesser height, we desired to reduce the size ofthe capillary channel. This we did by placing a weight (≈100 g), resulting in a pressureof ∼10 kPa on top of the stamp to compress the channel during molding. The pressurewe used to deform the stamp is similar to that previously analysed [39]. We also dilutedthe PEDOT-PSS solution with distilled water (1 : 1). Si wafer was used instead of glassto decrease the roughness of the substrate. The SFM images show that the height ofthe lines resulting in the MIMIC process could be decreased, depending on the pressureexerted on the top of the stamp. For example, the height of the nanowires for the 1200grating 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. Thenanowires in Fig. 8 are separate from each other, the distance between lines is ∼200 nmand the width of the wires is around 650 nm. The lengths of the nanowires can reach1 cm.

Page 96: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 morelight a thick film is needed, but due to the low mobility of charge carriers, this isassociated with increasing series resistance. In bilayer diodes built from the combinationof an electron donor and acceptor, the excited state dissociates at the interface betweenelectron donor layer and acceptor layer. The larger interface area thus causes moreefficient charge separation. These two aspects of device thickness motivate efforts tocombine donor and acceptor in micro and nano-patterned diodes, to control the geometryand thus influence photon propagation in the device as well as to control the interfacearea between donor and acceptor. We have used soft embossing to pattern active polymerlayer to enhance the external quantum efficiency (EQE) of photodiodes [33]. Here wecombine two soft lithography methods, liquid printing and soft embossing, to fabricatethe devices with two periodic patterns to increase the performance. Metallic polymeranode PEDOT-PSS was micro patterned in 600 lines/mm and polythiophene POMeOPTlayer was nano-patterned in 3600 lines/mm followed by vacuum deposition of C60as electron acceptor to keep a sharp interface. The bilayer diodes were fabricated asITO/PEDOT-PSS/POMeOPT/C60/Al. Our results show that the EQE of these diodeswith patterned anode and polymer was enhanced compared with non-patterned planardiodes (see Fig. 9). These results demonstrate the possibility of using sub-micro andnano-pattern by soft lithography to implant structure in polymer photodiodes to improvelight trapping in thin active layers, so that the performance of these devices may beenhanced.

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).

Page 97: Nanotechnology and Nano-Interface Controlled Electronic Devices

80 O. Inganäs and F. Zhang

In summary, we have demonstrated how processes and structures with nanometerextension are crucial to the operation of polymer photodiodes. Structuring of materialson the nanometer length scale is thus an important tool for obtaining higher detectionand conversion efficiencies in these devices, always relying on parallel progress inchemical synthesis, materials formulation and device assembly.

Acknowledgements

We wish to acknowledge all the coworkers and students contributing to the studies herereported, in particular Lucimara Stolz Roman, Lichun Chen, Mathias Theander, LeifPetterson, Yohannes Teketel and many more. Joint European projects with S. Sariciftci,C. Brabec, J. Hummelen, R. Janssen, M. Maggini, M.R. Andersson and M. Pratohave been instrumental in delivering materials. Fundings from the Göran Gustafssonfoundation and the Carl Tryggers foundation have been instrumental to achieve theseresults, as well as the Engineering Research Board (TFR).

References

1. Y. Cao, I.D. Parker, G. Yu, C. Zhang, and A.J. Heege, Nature 397, 414 (1999).2. C.J. Brabec, G. Zerza, G. Cerullo, S. De Silvestri, S. Luzzati, et al., Chem. Phys. Lett. 340, 232

(2001).3. A. Ruseckas, M. Theander, M.R. Andersson, M. Svensson, M. Prato, et al., Chem. Phys. Lett. 322,

136 (2000).4. N.S. Sariciftci, L. Smilowitz, A.J. Heeger, and F. Wudl, Science 258, 1474 (1992).5. G. Yu, J. Gao, J.C. Hummelen, F. Wudl, and A.J. Heeger, Science 270, 1789 (1995).6. C.J. Brabec, N.S. Sariciftci, and J.C. Hummelen, Adv. Func. Mater. 11, 15 (2001).7. C.J. Brabec and S.N. Sariciftci, Monatsh. Chem. 132, 421 (2001).8. O. Inganäs, L.S. Roman, F.L. Zhang, D.M. Johansson, M.R. Andersson, and J.C. Hummelen, Synth.

Metals 121, 1525 (2001).9. S.E. Shaheen, C.J. Brabec, N.S. Sariciftci, F. Padinger, T. Fromherz, and J.C. Hummelen, Appl. Phys.

Lett. 78, 841 (2001).10. M.R. Andersson, O. Thomas, W. Mammo, M. Svensson, M. Theander, and O. Inganäs, J. Mater.

Chem. 9, 1933 (1999).11. J.H. Schon, A. Dodabalapur, Z. Bao, C. Kloc, O. Schenker, and B. Batlogg, Nature 410, 189 (2001).12. M. Theander, O. Inganäs, W. Mammo, T. Olinga, M. Svensson, and M.R. Andersson, J. Phys. Chem.

B 103, 7771 (1999).13. A. Ruseckas, E.B. Namdas, T. Ganguly, M. Theander, M. Svensson, et al., J. Phys. Chem. B 105,

7624 (2001).14. M. Theander, M. Svensson, A. Ruseckas, D. Zigmantas, V. Sundstrom, et al., Chem. Phys. Lett. 337,

277 (2001).15. G. Gigli, O. Inganäs, M. Anni, M. De Vittorio, R. Cingolani, et al., Appl. Phys. Lett. 78, 1493 (2001).16. M.M.L. Grage, T. Pullerits, A. Ruseckas, M. Theander, O. Inganäs, and V. Sundstrom, Chem. Phys.

Lett. 339, 96 (2001).17. M. Theander, A. Yartsev, D. Zigmantas, V. Sundstrom, W. Mammo, et al., Phys. Rev. B 61, 12957

(2000).18. A. Cravino, G. Zerza, M. Maggini, S. Bucella, M. Svensson, et al., Chem. Commun. 24, 2487 (2000).19. A. Cravino, G. Zerza, H. Neugebauer, S. Bucella, M. Maggini, et al., Synth. Met. 121, 1555 (2001).

Page 98: Nanotechnology and Nano-Interface Controlled Electronic Devices

Polymer optoelectronics – towards nanometer dimensions 81

20. A. Cravino, G. Zerza, H. Neugebauer, M. Maggini, S. Bucella, et al., J. Phys. Chem. B 106, 70(2002).

21. G. Zerza, A. Cravino, H. Neugebauer, N.S. Sariciftci, R. Gomez, et al., J. Phys. Chem. A 105, 4172(2001).

22. F.L. Zhang, M. Svensson, M.R. Andersson, M. Maggini, S. Bucella, et al., Adv. Mater. 13, 1871(2001).

23. l.S. Roman, M.R. Andersson, T. Yohannes, and O. Inganäs, Adv. Mater. 9, 1164 (1997).24. D. Godovsky, L.C. Chen, L. Pettersson, O. Inganäs, M.R. Andersson, and J.C. Hummelen, Adv.

Mater. Optics Electr. 10, 47 (2000).25. L.C. Chen, D. Godovsky, O. Inganäs, J.C. Hummelen, R.A.J. Janssens, et al., Adv. Mater. 12, 1367

(2000).26. L.C. Chen, L.S. Roman, D.M. Johansson, M. Svensson, M.R. Andersson, et al., Adv. Mater. 12, 1110

(2000).27. A.C. Arias, J.D. MacKenzie, R. Stevenson, J.J.M. Halls, M. Inbasekaran, et al., Macromolecules 34,

6005 (2001).28. A.C. Arias, N. Corcoran, M. Banach, R.H Friend, J.D. MacKenzie, and W.T.S. Huck, Appl. Phys.

Lett. 80, 1695 (2002).29. L.A.A. Pettersson, L.S. Roman, and O. Inganäs, J. App. Phys. 86, 487 (1999).30. L.A.A. Pettersson, L.S. Roman, and O. Inganäs, Synth. Met. 102, 1107 (1999).31. Y.N. Xia and G.M. Whitesides, Angew. Chem. Int. Ed. 37, 551 (1998).32. E. Kim, Y.N. Xia, and G.M. Whitesides, Nature 376, 581 (1995).33. L.S. Roman, O. Inganäs, T. Granlund, T. Nyberg, M. Svensson, et al., Adv. Mater. 12, 189 (2000).34. J.A.E. Wasey, A. Safonov, I.D.W. Samuel, and W.L. Barnes, Phys. Rev. B 6420, art. no.-205201

(2001).35. 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).36. L.S. Roman, W. Mammo, L.A.A. Pettersson, M.R. Andersson, and O. Inganäs, Adv. Mater. 10, 774

(1998).37. T. Kugler, W.R. Salaneck, H. Rost, and A.B. Holmes, Chem. Phys. Lett. 310, 391 (1999).38. F. Zhang, M. Johansson, M.R. Andersson, J.C. Hummelen, and O. Inganäs, Adv. Mater. 14, 662

(2002).39. A. Bietsch and B. Michel, J. Appl. Phys. 88, 4310 (2000).

Page 99: Nanotechnology and Nano-Interface Controlled Electronic Devices

This Page Intentionally Left Blank

Page 100: Nanotechnology and Nano-Interface Controlled Electronic Devices

Nanotechnology and Nano-Interface Controlled Electronic DevicesEditors: M. Iwamoto, K. Kaneto and S. Mashiko© 2003 Elsevier Science B.V. All rights reserved

CHAPTER 7

Control of charge transfer and interfacestructures 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. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 832. Light harvesting efficiency based on dynamics of dye-sensitization . . . . . . . . . . . . . . . . . 853. Electron transport in nano-structured TiO2 electrodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86

3.1. Ambipolar electron diffusion mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 863.2. Evaluation method of electron diffusion in plain nano-structured TiO2 layers 883.3. Electron transport in nano-structured TiO2 layers . . . . . . . . . . . . . . . . . . . . . . . . . . 883.4. Effects of electrolytes on the electron transport . . . . . . . . . . . . . . . . . . . . . . . . . . . . 913.5. Effect of surface states in nano-structured TiO2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95

4. Charge transport in iodide/polyiodide electrolytes of DSCs . . . . . . . . . . . . . . . . . . . . . . . . 964.1. Mechanistic studies using quasi-solid-state electrolytes produced by low-

molecular-weight gelators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 964.2. Quasi-solid stated DSCs using imidazolium molten iodides as electrolytes . . 97

5. Importance of interface control in DSC fabrication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1005.1. Interface structures affecting efficiencies, ηei and ηhi . . . . . . . . . . . . . . . . . . . . . . . 1005.2. Important role of interfaces affecting efficiencies, ηec and ηhc . . . . . . . . . . . . . . 100

6. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102

1. Introduction

Almost 10 years have passed since dye-sensitized solar cells (DSCs) were innovatedby Grätzel’s group [1–3]. DSCs can be constructed by Ruthenium complex [forexample, N3 = cis-RuII (dcbpy)2(SCN−)2 (dcbpy = 2,2′-bipyridine-4,4′-dicarboxylicacid); Scheme 1] dye-anchored nano-structured TiO2 films and redox electrolyte madefrom iodine and iodide salts, and is currently under large, intense investigation, because

Page 101: Nanotechnology and Nano-Interface Controlled Electronic Devices

84 S. Yanagida et al.

Scheme 1.

a DSC is noted to have the following advantages over silicon solar cells; respectablephoto-conversion efficiency with high fill factor, low cost and earth-friendly materials(low material cost), economical production facilities like screen printing and ink jetprinting (low labor cost), high open-circuit photovoltage at low light intensity that isfavorable for indoor use, and less sensitiveness to angle of incidence of solar radiationwhich leads to an increase of total conversion of solar light per day. It is furtherinteresting 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 photocur-rent density (Jsc), open-circuit photovoltage (Voc) and fill factor (ff ) under one sunirradiation (Is) conditions (Eq. 1).

η = (Jsc × Voc × ff )/Is (1)

The respectable photo-conversion efficiency can be explained by efficiency of lightharvesting (ηlh) of the sensitizing Ru dye and efficiencies of transport of photo-formedelectrons 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 threefactors, i.e., efficiency of electron injection from the excited dye molecule, ηei, efficiencyof electron transport in nano-structured TiO2 phase, ηet, and efficiency of electroncollection 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 theefficiency of hole injection into the electrolyte as a hole-transport phase, ηhi, efficiencyof charge (hole) transport in the electrolyte phase, ηht, and efficiency of hole collectionat the counter electrode, ηhc (Eq. 4).

ηh = ηhi ×ηht ×ηhc (4)

Page 102: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 almost100% in the wide absorption band, which is comparable to amorphous silicon solarcells. 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 projectedone defined as roughness factor.

Current research topics to improve DSCs are electron transfer dynamics that de-termine 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 andcounter electrode will be discussed. On the basis of mechanistic points of view, we willpropose 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, toTiO2 was examined precisely by laser-induced ultra-fast transient spectroscopy assummarized in Fig. 1 [4–9]. These kinetics are largely affected by the composition ofthe electrolyte. For example, since decrease of surface pH makes the flat band potentialpositive, 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.

Page 103: Nanotechnology and Nano-Interface Controlled Electronic Devices

86 S. Yanagida et al.

presence of basic pyridine derivatives shifts it to the negative side [10]. Concentrationsof I− and I−3 formed by equilibrium of I− + I2 ↔ I−3 is also noted to affect chargetransfer kinetics. At the efficient DSC conditions, however, the two forward electron-transfer steps are much faster than the corresponding reverse electron-transfer steps(charge recombination), by a factor of 106 to 109. The high conversion efficiency wellexplains effective and vectorial electron transfer in a DSC with N3, suggesting highvalues 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-structuredTiO2 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 theDSC suggests that the nano-structured dye-coated TiO2 films should exhibit high ηet,i.e., anomalous electron transport properties. Interestingly, such electron transportingproperties of the nano-structured TiO2 films should appear only when the films are filledwith highly ionic electrolyte.

Electron transport in nano-structured TiO2 films in a liquid electrolyte has beendescribed in terms of carrier diffusion due to a lack of large electric field gradient in thefilm [11–26]. The diffusion coefficients of the dye-coated TiO2 and the plain TiO2 elec-trodes 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 thesteady state population of the trapped electrons affect the diffusion coefficients. For thenano-structured electrodes, the effect of traps is significant since the number of traps isextremely 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 effectsof electrolyte ion concentrations on measured diffusion coefficient were reported bySolbrand et al. [19]. The diffusion coefficient and total amount of charges decreasedas the electrolyte concentration decreased. Kopidakis et al. proposed an ambipolardiffusion mechanism to interpret the effect of electrolyte on the electron transportin a mesoporous TiO2–electrolyte system [23]. Photoinjected electrons in TiO2 weresurrounded by an electrolyte consisting of various kinds of ionic species. The ambipolardiffusion coefficient is expressed by

Damb = (n + p)

(n/Dp)+ (p/Dn)(5)

where n and p are the density of electrons and cations, and Dn and Dp are thediffusion coefficients of electrons and cations, respectively. The diffusion coefficient of

Page 104: Nanotechnology and Nano-Interface Controlled Electronic Devices

Control

ofcharge

transferand

interfacestructures

innano-structured

DSC

s87

Fig. 2: Schematic view of electron transport mechanism in nano-structured TiO2 electrode soaked in high ionic strength electrolyte.

Page 105: Nanotechnology and Nano-Interface Controlled Electronic Devices

88 S. Yanagida et al.

electrons depends not only on the density of electrons but also on the concentration ofions in electrolytes such as lithium perchlorate (LiClO4). The density of electrons isproportional 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 determinedby laser-induced transient current measurements, which is a kind of time-of-flighttechnique [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, sincethe absorption coefficient of TiO2 in the UV region is large (∼10−7 m−1). The chargeseparation in the TiO2 electrode after excitation is completed in the sub-nanosecondrange. The hole at the valence band of TiO2 is rapidly removed by supplying anelectron from quenchers in solvent, suppressing initial electron–hole pair recombinationinside the TiO2. When a laser is employed as the excitation light source, electrons aregenerated at the shallow region of the electrolyte side of the TiO2 electrode, and thentravel to the substrate side by a diffusion process because of thermal fluctuations in thesystem. Therefore, with increasing thickness of the electrode the carriers travel a longerdistance. The electron diffusion coefficient of the nano-structured TiO2 electrode canbe determined by analyzing the photocurrent transient response using Fick’s diffusionmodel. Neglecting a current due to electrostatic repulsion in the solution of the time-dependent 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 ofphotocurrent 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. Theelectrodes were immersed in an ethanolic or acetonitrile electrolyte solution composedof LiClO4, 0.7 M and using a platinum wire as a counter electrode. Short durationof excitation light was obtained using a 10 Hz Nd-YAG laser (The Quanta-Ray INDISeries Pulsed Nd : YAG Lasers, pulse width 7 ns, wavelength 355 nm) and time transientphotocurrent 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 ofthe 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 anelectrode was fixed as 0.093 cm−2 by an aperture. All the measurements were performedin air with at least 3-min intervals between measurements. Thickness of the films wasmeasured 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 thick-nesses are shown in Fig. 4 [28]. The positions of the peaks in the current rise wereshifted to longer times with increasing film thickness, indicating that the electrons have

Page 106: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 = 355nm), 7 ns, 0.98 mJ cm−2, 0.093 cm2.

Page 107: Nanotechnology and Nano-Interface Controlled Electronic Devices

90 S. Yanagida et al.

Fig. 5: The current peak vs. the square of each different nano-structured TiO2 film thickness sintered at450°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 currentwas slower for the samples with larger thickness. Therefore, the experimental observa-tions were qualitatively in agreement with those expected for the diffusive process ofelectrons in the films. The generated charge in our setup was evaluated as 1.2 μC for theelectrode with 5 μm in thickness by integrating the transient curve between the times 0and 100 ms and was almost constant for the films with <12 μm thickness, but decreasedto 0.5 μC for those 15 μm in thickness. This is probably due to the back electrontransfer 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 samplesaccording to the relationship expressed by Eq. 6 and obtained Fig. 5. A straight line wasobtained for all of the examined nano-structured TiO2 films, which indicates that thetransport of electrons in the TiO2 electrodes will follow the diffusion model.

The diffusion coefficients of electrons in nano-structured TiO2 (D) were determinedfor comparing the electron transport properties of the TiO2 films made with differentTiO2 nanoparticles [28] as follows. A1 was anatase TiO2 having spherical shape withan average size of 12 nm, which was prepared from titanium isopropoxide, containinga small amount of brookite crystal structure. A2 was anatase TiO2 having a highercrystallinity than A1. A3 was phase-pure anatase TiO2 having rod-like shape with anaverage size of 13 × 4 nm (supplied by Catalysis & Chemicals Industry (denoted asCCI, hereafter) HPW-25R). A4 was phase-pure anatase having cubic shape with an

Page 108: Nanotechnology and Nano-Interface Controlled Electronic Devices

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

Structure a Shape Size (nm) b D (×10−5 cm2 s−1)

A1 A/Amor Spheric 12 2.2A1TiCl4

c A/Amor Spheric 12 2.2A2 A Cubic 12 0.3A2550

d A Cubic 12 2.0A3 (CCI) A Rod-like 13/34 4.0A4 (CCI) A Cubic 11 4.1A5 (P25) A/R Spheric 21 4.0A5Large

e A/R Spheric 21 4.0R1 R Spheric 27 0.1R1TiCl4

c R Spheric 27 0.4R2 (CCI) R Rod-like 23/73 0.3

a A = anatase structure, Amor = amorphous phase, R = rutile structure.b Average size.c Treated with aq. TiCl4 solution.d Annealed at 550°C.e With 20 wt% of large TiO2 (Fluka).

average size of 11 nm (CCI, HPW-10R). A5 was nano-structured TiO2 (P25) havingmixed crystal structure (anatase : rutile = 8 : 2), spherical shape and an average size of21 nm (supplied by Nippon Aerosil). R1 was rutile TiO2 having spherical shape withaverage size 27 nm, which was prepared from titanium isopropoxide. R2 was rutilecrystal 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 arelisted in Table 1. In general comparison, the electron diffusion coefficient of the rutileelectrodes (DR1 and DR2) was almost one order smaller than that of the anataseelectrodes (DA1, DA3, DA4 and DA5). Similar diffusion coefficients were obtained forA3 and A4 which were prepared by the common synthesis method, but possesseddifferent particle sizes and shapes. The A2 having the anatase structure but highercrystallinity than A1 gave a smaller D value, but one which was highly increased byannealing at 550°C, being closed to that of DA1.

The detailed comparison of the diffusion coefficient depending on the morphology ofnano-structured TiO2 film well indicates the importance of surface structure of anataseTiO2 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–electrolytesystems 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 ofdye-sensitized solar cells [30–32].

Page 109: Nanotechnology and Nano-Interface Controlled Electronic Devices

92 S. Yanagida et al.

Fig. 6: Photocurrent transients for 7.2 μm-thick nano-structured TiO2 electrodes in acetonitrile with variousconcentrations of TBA+ClO−

4 at average electron density n = 2.1 × 1017 cm−3. From below 0.3, 0.7, 1.0,2.0, 5.0, 10, 20, 40, 80, 100, 200, 500 and 1000 mM of the cation concentration. Inset shows longer timescale 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 ofvarious densities of TBA+ at a photoelectron density of n = 2.1 × 1017 cm−3. Theshape of photocurrent transients changed in the series and the peak of the photocurrentincreased with an increase of TBA+ density at a comparable photogenerated electrondensity. Fig. 7 shows the change in diffusion coefficient (DAmb) as a function ofTBA+ 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 withthe measured DAmb at Dn = 2.1 × 10−4 cm2 s−1 and Dp = 2.8 × 10−6 cm2 s−1. TheDp obtained by the curve fitting is close to the limiting cation diffusion coefficientD∞

p = 2.32 × 10−6 cm2 s−1 derived from the Nernst equation using the limiting molarconductance Λ0 = 61.63 S cm2 mol−1 [33]. The diffusion coefficients (DAmb) increasewith 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

Page 110: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 iscalculated 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 cationand decreases in the order DMHI+ > TBA+ > Na+ > Li+ > Mg2+. Interestingly, thediffusion coefficients increased drastically in the presence of highly concentrated Li+ orDMHI+ without fitting well with the ambipolar diffusion mechanism. These behaviorswere in contrast with that in the case of TBA+.

We have reported previously that the measured D was well interpreted with theambipolar 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 thedifference 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 suggestthat the adsorption might form favorable trap states in the surface TiO2 and influencesthe electron transport.

The behavior of DAmb as a function of the cation density in the case of DMHI+ wassimilar to that in the case of Li+ rather than TBA+, although DMHI+ is a quaternaryammonium 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 monolayerof DMHI+ on the TiO2, suggesting multi-layered adsorption of DMHI+ on TiO2. Theproximity of multi-layered DMHI+ to all particles on the surface causes screening ofphotoinjected electrons [20,35], leading to their enhanced transport.

Page 111: Nanotechnology and Nano-Interface Controlled Electronic Devices

94 S. Yanagida et al.

Fig. 8: Diffusion coefficients of electrons in nano-structured TiO2 as a function of cation density of LiClO4

(a), NaClO4 (b), Mg(ClO4)2 (c) and DMHI+ClO−4 (d) when n = 2.1 × 1017 cm−3 (Li+), 1.6 × 1017 cm−3

(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.

Page 112: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 wasexpected to increase electron diffusion coefficients and to contribute to an increase ofηet of the TiO2 films when they are formed just below the conduction band. The oxygenvacancies should originate from the 3d-orbital of Ti3+ but that is unstable in the presenceof oxygen and water molecules. An alternative way to produce Ti3+ in the TiO2 films isfluorine doping, i.e., the chemical replacement of O2− with F−. A high physical stabilitycan be expected in the fluorine-doped TiO2 films (TiO2/F), because O2− sites come tobe 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(OiPr)4 followed by au-

toclaving in the presence of HF. The content of fluorine in the doped particles wasestimated as F/Ti = 0.0011 by applying the fundamental parameter method to theobserved F-Kα signal on X-ray fluorescent analysis (Rigaku, ZSX100e; RX35 analyz-ing crystal; F-PC detector). The films (TiO2/F) prepared in the similar procedure hadthe similar morphology to that of fluorine-free TiO2. For comparison, TiO2 films withexcess oxygen vacancies were prepared by calcination under N2 atmosphere, and theformation of oxygen vacancies was proved by absorption of UV-VIS spectroscopy atlonger wavelength.

Laser pulse induced photocurrent measurements were performed under comparableconditions (in ethanolic solution of LiClO4) for the TiO2/F film and fluorine-freeTiO2 films with and without introduction of oxygen vacancies. Fig. 9 shows that thephotocurrent transient is larger for the TiO2/F film. The diffusion coefficients of theTiO2/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 toopoor current to determine the diffusion coefficient. Fig. 10 shows I–V curves of thecells fabricated by using these films with ∼4 μm thickness. Introduction of oxygenvacancies 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.

Page 113: Nanotechnology and Nano-Interface Controlled Electronic Devices

96 S. Yanagida et al.

Fig. 10: Photocurrent density–voltage curve of dye-sensitized solar cells with 4 μm-thick nano-structuredTiO2 film under AM 1.5 irradiation: TiO2 (thin line), TiO2/F (bold line) and TiO2 with excess oxygenvacancies (dashed line).

the increase of photovoltage [38]. These results suggest that while the oxygen vacanciesmay produce deep trap states as back electron transfer site, the doped fluorine statesmay serve as shallow trap sites just below the conduction band that might serve effectiveelectron-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 bylow-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 cellat high illumination intensity up to AM 1.5 (one sun). This fact implies the specificcharge 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 gela-tion does not affect largely the conductivity of the electrolyte and that the conductivityincreased 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 I−3 and I−5 by addition of iodine. The self-diffusion of iodide species inthe gel electrolyte was about a quarter of that of I− in acetonitrile. Lindquist’s grouppreviously reported that the diffusion of I−2n+1 in nano-structured TiO2 space is one order

Page 114: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 elec-trolyte did not influence the charge transport process, which is in contrast to the lowerlimiting molar conductivity of polyiodide species than monoiodide [33]. In addition, therather high concentration (0.5 M) of polyiodide and iodide species in the electrolyteimplies 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 berationalized as due to electron hopping or iodine exchange (Grotthuss-type) mechanismcaused by the rather packed polyiodide species in the electrolytes. Fig. 11 depicts theelectronic transport in the iodide and polyiodide species in electrolyte. The effectiveelectron injection (see Fig. 1) from iodide species to oxidized dye may suggest chemicalinteraction of polyiodide species and thiocyanide groups of the ruthenium dye moleculeas 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 increasethe electron conductance of the redox electrolyte (Fig. 11) [40]. On the other hand,the presence of imidazolium cations accelerates the diffusion coefficient of electrons inthe mesoporous TiO2 phase (Dn) through their multi-layer adsorption [29]. In addition,

Page 115: Nanotechnology and Nano-Interface Controlled Electronic Devices

98 S. Yanagida et al.

Fig. 11: Effect of the concentration of Gelator 2 on conductivity for electrolytes containing 0.5 M of iodidesalts (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/iodideelectrolyte and electron transport mechanism in iodide/polyiodide electrolyte.

some imidazolium iodides are known as chemically inert room-temperature moltensalts, i.e., ionic liquids. Currently, many ionic liquids have attracted much attention aselectrolyte in electrochemistry because of their features such as high ionic conductivity,non-volatility, thermal stability and non-flammability [42–44]. In fact, such imidazoliumsalts were applied to DSCs as non-volatile electrolyte solvent [31,45]. However, the

Page 116: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 liquidelectrolyte containing organic solvent.

When we applied 1-alkyl-3-methylimidazolium iodides (Scheme 3) to the electrolyteof DSCs by combining iodine but without using any volatile solvents, the electrolytewith 1-hexyl-3-methylimidazolium iodide gave respectable photoenergy conversion ef-ficiency (η = 5.0%) when combined with 1/10 molarity of iodine. Further, solidificationof the molten imidazolium salts using gelator (Scheme 2, Gelator 1) showed compa-rable conversion efficiency (5.0%) under AM 1.5 irradiation (Fig. 13) [46]. Under dryheat test, the quasi-solid-state imidazolium DSCs showed a higher stability than thequasi-solid-state DSCs fabricated using organic solvent with the same gelator [47]. Itis interesting to note that the imidazolium DSC is non-flammable because of the highboiling 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 iodidecontaining 8.7 wt% of I2 under AM 1.5 irradiation without (dotted curve) and with (solid curve) 1.4 wt% ofGelator 1.

Page 117: Nanotechnology and Nano-Interface Controlled Electronic Devices

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-sensitizationmechanism. In the case of N3/TiO2 interface, the dye molecules are chemicallyadsorbed 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 importanceof a covalent bond between dye molecules and the hole transport layer was alsodisplayed when polypyrrole (PPy) as a hole transport layer and a Ru dye with pyrrolegroup, 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 electrontransfer to redox electrolytes by surface treatment with pyridine derivatives [53] orcholic acids [54,55]. Covering of solvent-exposed interface parts of OTE and TiO2

surfaces by insulating molecules such as poly(phenylene oxide-co-2-allylphenyleneoxide) 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 relatingto surface structures and morphology and the transparency especially at window OTEplay some decisive roles. On the other hand, Willig pointed out the importance ofthe junction at window OTE (SnO2/F) and nano-structured TiO2 as a driving force ofelectron transfer in DSCs [20]. Gregg et al. reported the difference in role of metaldeposition on window OTE in view of the work function [57]. These facts and findingsare closely related with interfacial control affecting ηec.

Fullerenes are good electron acceptors due to the small reorganization energy in

Page 118: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 transferreactions [58–60]. We successfully attempted to apply a C60 derivative to acceleration ofelectron capture at OTE in DSCs. Since the HOMO level of C60 is almost comparable tothe 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, 1,2-methanato[60]-fullerene-61,61-dicarboxylic acid [61,62] was adsorbed on an OTE filmor on OTE/TiO2 film from toluene solution. The resulting OTE/C60 film was coatedwith nano-structured TiO2 and sintered at 400°C for 30 min, and then N3 dye wasadsorbed on the electrodes. The resulting OTE/C60/TiO2/N3 and OTE/TiO2/C60/N3systems can be depicted as shown in the Scheme 5.

In the system of OTE/TiO2/C60/N3, the amount of the adsorbed dye moleculeswas decreased to about 40% of that of the OTE/TiO2/N3 system fabricated as areference system. On the other hand, the OTE/C60/TiO2/N3 system gave 80% dye-adsorption when compared to the OTE/TiO2/N3 system. While negligible responsein measurement of IPCE was observed in OTE/TiO2/C60/N3, the OTE/C60/TiO2/N3system gave a response comparable to that of the OTE/TiO2/N3 system in spite of thedecrease of dye adsorption (Fig. 14). This result would suggest that C60 has a potentialto 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 arequisite 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 aRu dye with pyrrole group, cis-RuII(dcbpy)2(pmp)2, carbon-based counter electrodeimproved 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 carbonelectrode will enhance the efficiency ηhc.

6. Conclusions

The respectable conversion efficiency of Grätzel solar cells is now rationalized by 7 highefficiencies, i.e., light harvesting efficiency, electron injection, transport and captureefficiencies, and hole injection, transport and capture efficiencies. Interfacial structures

Page 119: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 1.5irradiation: 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 windowOTE and counter electrodes have influence on the efficiency for each step. For keepinggood electron diffusion and charge transport both in TiO2 and in iodide/polyiodideelectrolyte, the presence of highly concentrated imidazolium species is favorable forDSC systems. Employment of a series of imidazolium iodides as ionic liquid electrolyteled to successful fabrication of DSCs with respectable efficiency and thermal stabilitywhen followed by solidification using a low-molecular-weight gelator.

Control of the interfaces to reduce the back electron transfer by chemically modifyingthe surfaces of TiO2 and OTE materials is also an important subject to improve DSCperformance. In the solidification of DSCs, we must maintain vectorial charge flow withkeeping 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 ofJapan.

References

1. B. O’Regan and M. Grätzel, Nature (London) 353, 737 (1991).2. 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).3. M. Grätzel, Nature (London) 414, 338 (2001).

Page 120: Nanotechnology and Nano-Interface Controlled Electronic Devices

Control of charge transfer and interface structures in nano-structured DSCs 103

4. Y. Tachibana, J.E. Moser, M. Grätzel, D.R. Klug, and J.R. Durrant, J. Phys. Chem. 100, 20056(1996).

5. T. Hannappel, B. Burfeindt, W. Storck, and F. Willig, J. Phys. Chem. B 101, 6799 (1997).6. J.S. Salafsky, W.H. Lubberhuizen, E. van Faassen, and R.E.I. Schropp, J. Phys. Chem. B 102, 766

(1998).7. S.A. Haque, Y. Tachibana, D.R. Klug, and J.R. Durrant, J. Phys. Chem. B 102, 1745 (1998).8. 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).9. 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).10. S.Y. Huang, G. Schlichthörl, A.J. Nozik, M. Grätzel, and A.J. Frank, J. Phys. Chem. B 101, 2576

(1997).11. S. Södergren, A. Hagfeldt, J. Olsson, and S.-E. Lindquist, J. Phys. Chem. 98, 5552 (1994).12. F. Cao, G. Oskam, G.J. Meyer, and P.C. Searson, J. Phys. Chem. 100, 17021 (1996).13. P.E. de Jongh and D. Vanmaekelbergh, Phys. Rev. Lett. 77, 3427 (1996).14. A. Solbrand, H. Lindström, H. Rensmo, A. Hagfeldt, S.-E. Lindquist, and S. Södergren, J. Phys.

Chem. B 101, 2514 (1997).15. G. Schlichthörl, S.Y. Huang, J. Sprague, and A.J. Frank, J. Phys. Chem. B 101, 8141 (1997).16. 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).17. G. Franco, J. Gehring, L.M. Peter, E.A. Ponomarev, and I. Uhlendorf, J. Phys. Chem. B 103, 692

(1999).18. G. Schlichthörl, N.G. Park, and A.J. Frank, J. Phys. Chem. B 103, 782 (1999).19. A. Solbrand, A. Henningsson, S. Södergren, H. Lindström, A. Hagfeldt, and S.-E. Lindquist, J. Phys.

Chem. B 103, 1078 (1999).20. K. Schwarzburg and F. Willig, J. Phys. Chem. B 103, 5743 (1999).21. J. Nelson, Phys. Rev. B 59, 15374 (1999).22. A.C. Fisher, L.M. Peter, E.A. Ponomarev, A.B. Walker, and K.G.U. Wijayantha, J. Phys. Chem. B

104, 949 (2000).23. N. Kopidakis, E.A. Schiff, N.-G. Park, J. van de Lagemaat, and A.J. Frank, J. Phys. Chem. B 104,

3930 (2000).24. J. van de Lagemaat and A.J. Frank, J. Phys. Chem. B 104, 4292 (2000).25. J. van de Lagemaat, N.-G. Park, and A. J. Frank, J. Phys. Chem. B 104, 2044 (2000).26. J. Nelson, S.A. Haque, D.R. Klug, and J.R. Durrant, Phys. Rev. B 63, 205321 (2001).27. S. Nakade, S. Kambe, T. Kitamura, Y. Wada, and S. Yanagida, J. Phys. Chem. B 105, 9150 (2001).28. S. Kambe, S. Nakade, Y. Wada, T. Kitamura, and S. Yanagida, J. Mater. Chem. 12, 723 (2002).29. S. Kambe, S. Nakade, T. Kitamura, Y. Wada, and S. Yanagida, J. Phys. Chem. B 106, 2967 (2002).30. C.J. Barbé, F. Arendse, P. Comte, M. Jirousek, F. Lenzmann, V. Shklover, and M. Grätzel, J. Am.

Ceram. Soc. 80, 3157 (1997).31. N. Papageorgiou, Y. Athanassov, M. Armand, P. Banhôte, H. Pettersson, A. Azam, and M. Grätzel, J.

Electrochem. Soc. 143, 3099 (1996).32. P. Banhôte, A.-P. Dias, M. Armand, N. Papageorgiou, K. Kalyanasundaram, and M. Grätzel, Inorg.

Chem. 35, 1168 (1996).33. J. Berthel, L. Iberl, J. Rossmaier, H.J. Gores, and B. Kaukal, J. Solut. Chem. 19, 321 (1990).34. K. Schwarzburg and F. Willig, Appl. Phys. Lett. 58, 2520 (1991).35. D. Cahen, G. Hodes, M. Grätzel, J.F. Guillemoles, and I. Riess, J. Phys. Chem. B 104, 2053 (2000).36. S.N. Subbarao, T.H. Yun, R. Kershaw, K. Dwight, and A. Wold, Mat. Res. Bull. 13, 1461 (1978).37. S.N. Subbarao, T.H. Yun, R. Kershaw, K. Dwight, and A. Wold, Inorg. Chem. 18, 488 (1979).38. S. Yamagida, T. Kitamura, and M. Kohmoto, Electrochem. 70, 399 (2002).39. W. Kubo, K. Murakoshi, T. Kitamura, Y. Wada, K. Hanabusa, H. Shirai, and S. Yanagida, Chem.

Lett., 1241 (1998).40. 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).

Page 121: Nanotechnology and Nano-Interface Controlled Electronic Devices

104 S. Yanagida et al.

41. Z. Kebede and S.-E. Lindquist, Solar Energy Mater. Solar Cells 51, 291 (1998).42. R. Hagiwara and Y. Ito, J. Fluorine Chem. 105, 221 (2000).43. T. Welton, Chem. Rev. 99, 2071 (1999).44. A.B. McEwen, H.L. Ngo, K. LeCompte, and J.L. Goldman, J. Electrochem. Soc. 146, 1687 (1999).45. H. Matsumoto, T. Matsuda, T. Tsuda, R. Hagiwara, Y. Ito, and Y. Miyazaki, Chem. Lett., 26 (2001).46. W. Kubo, T. Kitamura, K. Hanabusa, Y. Wada, and S. Yanagida, Chem. Commun., 374 (2002).47. Japanese Industrial Standard, Environmental and endurance test methods for amorphous solar cell

modules, JIS C 8938, (1995).48. K. Murakoshi, G. Kano, Y. Wada, S. Yanagida, H. Miyazaki, M. Matsumoto, and S. Murasawa, J.

Electroanal. Chem. 396, 27 (1995).49. K.S. Finnie, J.R. Bartlett, and J.L. Woolfrey, Langmuir 14, 2744 (1998).50. K. Murakoshi, R. Kogure, Y. Wada, and S. Yanagida, Chem. Lett., 471 (1997).51. K. Murakoshi, R. Kogure, Y. Wada, and S. Yanagida, Solar Energy Mater. Solar Cells 55, 113 (1998).52. T. Kitamura, M. Maitani, M. Matsuda, Y. Wada, and S. Yanagida, Chem. Lett., 1054 (2001).53. S.Y. Huang, G. Schlichthörl, A.J. Nozik, M. Grätzel, and A.J. Frank, J. Phys. Chem. B 101, 2576

(1997).54. A. Kay and M. Grätzel, J. Phys. Chem. 97, 6272 (1993).55. 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).56. B.A. Gregg, F. Pichot, S. Ferrere, and C.L. Fields, J. Phys. Chem. B 105, 1422 (2001).57. F. Pichot and B.A. Gregg, J. Phys. Chem. B 104, 6 (2000).58. H. Imahori, K. Hagiwara, M. Aoki, T. Akiyama, S. Taniguchi, T. Okada, M. Shirakawa, and Y.

Sakata, J. Am. Chem. Soc. 118, 11771 (1996).59. M. Knupfer, Surface Sci. Rep. 42, 1 (2001).60. N.S. Sariciftci, Prog. Quant. Electr. 19, 131 (1995).61. X. Campas and A. Hirsch, J. Chem. Soc. Perkin Trans. 1, 1595 (1997).62. I. Lamparth, G. Schick, and A. Hirsch, Liebigs Ann./Recueil, 253 (1997).

Page 122: Nanotechnology and Nano-Interface Controlled Electronic Devices

Nanotechnology and Nano-Interface Controlled Electronic DevicesEditors: M. Iwamoto, K. Kaneto and S. Mashiko© 2003 Elsevier Science B.V. All rights reserved

CHAPTER 8

Materials and devices for ultrafastmolecular 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. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1052. Materials and methods for ultrafast photoresponse measurements . . . . . . . . . . . . . . . . . . 1073. Absorption changes over wide ranges of wavelength and time by photoinduced

electrochromism of ion-pair charge-transfer complexes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1093.1. Photoinduced electrochromism in 4,4′-bipyridinium salts with various counter

ions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1093.2. Ultrafast dynamics of photoinduced electrochromism . . . . . . . . . . . . . . . . . . . . . . 1113.3. Charge resonance band in the near infrared region and its ultrafast dynamics 113

4. Parallel all-optical processing in guided wave geometry containing organic dyes . . . . 1164.1. Ultrafast spatial light modulation and parallel optical recording based on

photoinduced complex refractive index changes upon nanosecond laserexcitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116

4.2. Reversible reflectance control of fs white light in guided wave geometrycontaining photochromic compounds upon fs laser photoexcitation . . . . . . . . 124

5. Ultrafast nonlinear optical responses amplified by photoexcitation . . . . . . . . . . . . . . . . . . 1276. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 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. Ultrafastswitching materials which work in less than 1 ps are essential for teraHerz (THz) opticalcommunication. Several attempts have been made for this purpose, which include opticalswitching by tunneling bi-quantum well semiconductors or organic nonlinear opticalmaterials [1]. Magnetic “hard” disks and heat-mode optical disks such as phase change

Page 123: Nanotechnology and Nano-Interface Controlled Electronic Devices

106 T. Nagamura

Fig. 1: Why and how molecular photonics?

or magneto-optical memories have rapidly been increasing their recording density owingto the development of new pick-up heads or blue semiconductor lasers. But there is aphysical limit in such conventional memories recording data bit by bit in a serial way onthe surface or in the thin surface layer of recording materials.

Organic molecules have many useful optical and electronic functions that can beeasily controlled by the structures, substituents, or external fields. Specific interactionsor organization of molecules further can afford much higher functions than isolatedor randomly distributed molecules. Photons have many superior properties such aswavelength, polarization, phase, ultrashort pulse, or parallel processibility. Throughstrong interactions of molecules or molecular assemblies with photons, many superiorproperties of photons can be directly converted to changes in physical properties ofmaterials such as fluorescence, absorption, refractive index, conductivity, or opticalnonlinearity. These interactions will be utilized as molecular photonics with ultrahigh speed, ultra high density or high performance sensing as schematically shown inFig. 1. Excited state formation, photochromism, photoinduced electron transfer are someexamples among them. Photon-mode recording or switching based on these changes ofelectronic states can therefore achieve ultrafast multiple or three-dimensional recordingand parallel processing with ultimate spatial resolution at a molecular level. There willbe no doubt that molecular photonics based on interactions of molecules and photonshas many advantages as compared with electric or photoelectric switching, heat-modeor magnetic recording, switching based on liquid crystals or thermal phenomena.

We have been making efforts to develop new molecular photonics materials and de-vices by making various organized molecular systems and by optically controlling theirelectronic states. So far we have achieved photoinduced electrochromism, which is colorchanges due only to the photoinduced electron transfer and reverse reactions, molecularcontrol of the lifetime and the wavelength of colored species over extremely wideranges, amplified fluorescence quenching in LB films, photon-mode super-resolution

Page 124: Nanotechnology and Nano-Interface Controlled Electronic Devices

Materials and devices for ultrafast molecular photonics 107

to exceed the diffraction limit of light in optical memory based on transitory photo-bleaching of phthalocyanine derivatives, ultrafast all-optical two-dimensional controlof reflectance and parallel optical self-holding switch based on photoinduced complexrefractive index changes [2,3]. In the present chapter, some of our recent achievementswill 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−) saltsof polymeric 4,4′-bipyridinium (PV2+) were used together with polymer films cast orspin-coated from these solutions. The content of 4,4′-bipyridinium ions in a PV2+ poly-mer 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 absorptionchanges in the visible and near-infrared (NIR) region. Several derivatives of phthalocya-nines including water soluble copper-phthalocyanine (CuPcS) and zinc-phthalocyanine(ZnPcS) were used for photon-mode spatial light modulation. Phthalocyanines wereused in solutions or in poly(vinyl alcohol) (PVA) films. A photochromic spiropy-ran derivative, 1,3,3-trimethylindolino-6′-nitrobenzopyrylospiran (SP) was dispersed inpolystyrene 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 temperatureby the second harmonics (400 nm) of a femtosecond (fs) Ti : sapphire laser with aregenerative Ti : sapphire amplifier and a double path amplifier pumped with the secondharmonics (532 nm) of a Nd : YAG laser. The amplified Ti : sapphire laser deliveredpulses with a FWHM of 200–250 fs, 10 Hz repetition, a maximum power of 6 mJ/pulseat 800 nm. The fs probe white light was obtained by focusing the residual 800 nm lightinto a cell containing D2O/H2O (2 : 1) mixture after passing through a BBO crystal toobtain the second harmonics. The transient absorption and the dynamics were observedwith a Photonic Multi-channel Analyzer (PMA; Hamamatsu Photonics) system using adual photodiode array (Hamamatsu Photonics C6140) for the UV–visible and an InGaAsmultichannel detector (Hamamatsu Photonics C5890-256) for the NIR absorption usingan optical delay system. The intensities of the probe light with and without the pumppulses were averaged by 20 times. The block diagram of the fs transient absorptionmeasurement system is shown in Fig. 3.

For measuring basic properties of reflection type spatial light modulation or paralleloptical switching, the sample plate was index-matched with a BK7 prism, which wasset on a computer-controlled rotating stage. The writing beam was a ns OPO laserat 670 nm or the third harmonic (355 nm) of Nd : YAG laser; each with 8 ns pulsewidth, 0.03–2 mJ/pulse, and ca. 0.2 cm2 beam area. He–Ne laser (543.5 nm) througha half-wave plate, a polarizer, and a chopper was used as a reading beam. The timedependence of a reflected intensity at a given incident angle upon ns laser excitation atdifferent powers was detected with a photomultiplier and was recorded with a digitaloscilloscope terminated with 50 ohm. The spatial resolution was evaluated by using a

Page 125: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 lightas a reading light upon excitation by 400 nm fs pulses as a writing light, an Arton® filmwith spiropyran spin-coated on silver film was set with a prism on a rotating stage asschematically shown in Fig. 4.

Page 126: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 photoinducedelectrochromism of ion-pair charge-transfer complexes

3.1. Photoinduced electrochromism in 4,4′-bipyridinium salts with various counterions

Various photochromic systems employing polymeric thin films or Langmuir–Blodgett(LB) films have recently attracted much interest in view of their promising applicabilityto high-speed and high-density photon-mode optical memory [2–7]. The photochromismreported so far involves changes of chemical bonds such as heterolytic cleavage of apyran ring in spiropyrans, ring opening and closing in diarylethenes and fulgides,or trans–cis isomerization in azobenzenes [4–7]. Recently we have reported novelphotoinduced electrochromism as schematically shown in Fig. 5 [2,3,8–24]. It is thecolor 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 changesof chemical structure were involved in photochromism. TFPB− and iodide (I−) salts of4,4′-bipyridinium ions showed pale yellow and red colors, respectively, though each ionis colorless. These new absorption spectra above 350 nm in solutions were attributed tothe IPCT complexes with 4,4′-bipyridinium ion as an acceptor and TFPB− or I− as adonor. It was thus demonstrated that these ion pairs made electronic interactions at theground state partially transferring electronic charges from a donor to an acceptor. No

Page 127: Nanotechnology and Nano-Interface Controlled Electronic Devices

110T.N

agamuraFig. 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.

Page 128: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 toTFPB− salts.

From steady and laser photolysis results it has been shown that 4,4′-bipyridiniumradical cations, which escaped from the geminate reaction immediately after thephotoinduced electron transfer in less than 1 ps [22] upon IPCT excitation, becamemetastable owing to the bulk and chemical stability of TFPB−, to the restrictionof molecular motion by the microenvironment, and also probably to the very highexothermicity of the reverse reaction in the Marcus inverted region [26,27]. Highlysensitive detection of photoinduced electrochromism and transient absorption spectrain ultra-thin LB and polymer films have been achieved by the conventional and thewhite-light optical waveguide method [28–33]. Such photoinduced electrochromismmay 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, transientabsorption 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 absorptionspectra of PV2+(I−)2 in methanol solution. A marked increase in the absorbance ofthe 4,4′-bipyridinium radical cations took place with a rise time of about 0.3 ps uponexcitation. 4,4′-Bipyridinium radical cations were thus formed in a fs time scale by thephotoinduced electron transfer from a donor I− to an acceptor 4,4′-bipyridinium uponIPCT excitation [22]. The time profiles of transient absorption at 600 nm are shownin Fig. 7 for (a) PV2+(I−)2 in a film cast from DME and (b) PV2+(TFPB−)2 in DMEsolutions. Both of them showed a very rapid rise in about 0.3 ps, which was almostthe same as the time resolution of our fs Ti : sapphire laser measurement system witha regenerative amplifier. Similar extremely rapid formation of 4,4′-bipyridinium radicalcations was observed for PV2+(I−)2 salts in methanol and dimethylsulfoxide solutionsupon IPCT excitation, respectively. These results demonstrated that the charge separated4,4′-bipyridinium radical cations were formed directly upon IPCT excitation because of

Page 129: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 bandare transferred partially at the ground state and completely at the excited state. Sucha situation is very different from usual photochromism, which is caused by variouschanges of chemical bonds mainly via the excited singlet state. No transient absorptionwas observed for PV2+(I−)2 in DME solutions, which was most probably due to the

Page 130: Nanotechnology and Nano-Interface Controlled Electronic Devices

Materials and devices for ultrafast molecular photonics 113

decreased distance between ion pairs in such a less polar solvent and an appropriatethermodynamic driving force (−ΔG0) for reverse electron transfer reactions [26,27].

The decay behavior due to the reverse electron transfer was found to dependmarkedly on the microenvironment and the counter anion. The lifetime (τ ) and afraction of a major component for PV2+(I−)2 was 1.2 ps and 86% in films cast fromDME as compared with 4.0 ps and 73% in methanol solutions [22]. Photogenerated4,4′-bipyridinium radical cations disappeared completely during 10 Hz excitation,which corresponded well with the fact that no steady color changes were observed forPV2+(I−)2 in solutions and in cast films upon IPCT excitation. No decay was observedon the same time scale, as shown in Fig. 7(b) for PV2+(TFPB−)2. About a half decayedwith τ = 71 ps and the rest survived for an extremely long time corresponding to thereversible and persistent color changes observed by steady photolysis [20]. The lifetimesof photogenerated 4,4′-bipyridinium radical cations were thus controlled over a verybroad range from about 1 ps to almost infinity by the −ΔG0 value, the polarity ofsolvents and microenvironments in solid films. The present result of color change inabout 0.3 ps with IPCT complexes of PV2+ is the fastest response reported so far amongmaterials which show steady photochromism. It will help a great deal to develop noveloptical 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 todimer radical cation formation as a broad absorption with a peak at 950–2000 nm uponsteady photoexcitation of styrylpyridinium derivatives such as tetraphenylborate (TPB−)salts of 1-hexadecyl-4-(4-dicyanovinylstyryl)pyridinium (DCS+), 1-hexadecyl-4-(4-nitrostyryl)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]. Theabsorption spectra after irradiation (> 365 nm) for TPB− salts of DCS+ and NS+inDME are shown in Fig. 8 with respect to those before irradiation. In addition to theabsorption spectra in the visible region due to the radical formation, broad specificabsorption spectra were observed in the NIR region. The peak wavelength and shapeof the latter spectra depended on the substituents. They were assigned to a CR bandas schematically shown in Fig. 9 in a dimer radical cation which was formed betweena styrylpyridinium cation and a photogenerated styrylpyridinyl radical. We have alsoobserved the CR band with a peak at 1500–1700 nm as a charge resonance band forintramolecular 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 withpeaks at 950 and 1700 nm were observed only in NS+C3NS+ due to intramoleculardimer radical cations. The CR band energy is twice the stabilization energy of thedimer radical cations which is controlled by several factors such as the extent of overlapof two chromophores, their mutual distance, and/or the polarity of solvents. The CRbands at 950 and 1700 nm were assigned to fully and partially overlapped dimer radicalcations, respectively. Very recently we also reported fairly strong and unusual absorptionchanges in 800–2200 nm due to the intramolecular CR band of styrylpyridinyl radi-cals, formed by steady photolysis of newly synthesized meso-2,4-bis(4-(4′-nitrostyryl)

Page 131: Nanotechnology and Nano-Interface Controlled Electronic Devices

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, ΔH is thestabilization energy of dimer radical cations.

pyridinium)pentane ditetraphenylborates, during storage in the dark in solutions at roomtemperature, clearly indicating the change in conformation and overlapping of twochromophores 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 radicalformation and at the near-IR region due to the CR band was observed in less than 1

Page 132: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 inacetonitrile 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) timedependences 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 radicalcations were formed immediately after the photoinduced electron transfer reaction. TheCR band at 900 nm in NS+TPB− decayed single-exponentially (τ = 3.3 ps) [39]. Thetransient absorption at 580 nm showed double exponential decay with lifetimes of 3.3and 11.4 ps. Similar results were obtained for DCS+TPB−; the decay at 960 nm withτ = 3.2 ps, and that at 650 nm with lifetimes of 3.8 ps for a fast component and 17.4 psfor a slow one [39]. The difference in the decay behavior at the visible and the near-IRregion was explained as follows. While the NIR region absorption is attributed to the

Page 133: Nanotechnology and Nano-Interface Controlled Electronic Devices

116 T. Nagamura

dimer radical cation alone as the CR band, both the monomer radical and the dimerradical cation contribute to the absorption in the visible region as a HOMO–LUMOtransition as schematically shown in Fig. 9. The fast component in the visible regionand that in the NIR absorption gave almost the same lifetime. They were attributed tothe reverse electron transfer reactions from the dimer radical cation (SP+ · · ·SP•) to theTPB• radical. Then the slow component in the visible region was most probably dueto 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 inthe monomer radicals was explained by the classical Marcus theory as follows [26,27].The −ΔG0 values for the reverse electron transfer from NS• and DCS• to TPB• wereestimated as 1.69 and 1.61 eV from the redox potentials, respectively. The reductionpotential of the dimer radical cation should be less negative by 0.65 and 0.59 V than thatof the radical monomer radicals due to the stabilization energy. The −ΔG0 values forthe reverse electron transfer from the dimer radical cation to TPB• were thus estimatedto be 1.04 and 1.02 eV for (NS+ · · ·NS•) and (DCS+ · · ·DCS•), respectively [39].

The observed values of −ΔG0 in this study for the reverse electron transfer ofmonomer 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 cationto TPB• would become higher due to the smaller −ΔG0 in the inverted region. This isthe reason why the observed decay was faster for dimer radical cations. These resultsstrongly suggest the applicability of the present system to ultrafast optical switching inthe NIR region if an appropriate combination of a donor anion and an acceptor cation isused.

4. Parallel all-optical processing in guided wave geometry containing organic dyes

4.1. Ultrafast spatial light modulation and parallel optical recording based onphotoinduced complex refractive index changes upon nanosecond laser excitation

All-optical data processing has recently attracted much interest especially in the fieldsof 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 lightby another (writing) light, which plays an essential role in a projection TV, wavefrontcorrection and an optical correlator. However, no practical devices have been developedexcept some prototypes or SLMs based on liquid crystals (LCs). The response timeof LC-SLM is controlled by the electric field induced motion of LC; about a fewhundreds 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 10–20μm, because a photoconducting layer used to optically address LC limits it. Recentlymultiple-quantum-well (MQW) SLMs have been developed showing spatial resolutionsof 5–7 μm and a response time of 1 μs or faster [47–49]. MQW-SLMs will need furtherimprovements in properties as spatial resolution, contrast, or capability of handling largereadout intensities.

Page 134: Nanotechnology and Nano-Interface Controlled Electronic Devices

Materials and devices for ultrafast molecular photonics 117

Several types of new devices for optical parallel data processing have been proposedbased on surface plasmon resonance (SPR), guided wave mode and Fabry–Perot(FP) resonance [50–52]. However, all of these devices could not exceed the LC-SLM especially in the response time. Okamoto et al. [50] reported an all-opticalphotoaddressed SLM using a dye-doped polymer film in a surface plasmon resonance(SPR) configuration. They demonstrated the SLM based on photothermal changes in therefractive index of the methyl orange-doped poly(vinyl alcohol) (PVA) film using an Arlaser of 1–6 W cm−2 as a writing beam. The rise and fall times at 6 W cm−2 laser powerwere 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-dyepolymers [51]. They reported that their ATR-FP device enhanced optical modulationspeed and efficiency as compared with the conventional intensity modulation basedon photoinduced birefringence of Disperse Red 1 dye-doped poly(methylmethacrylate)[51]. The response time, 50–200 ms, was still relatively slow, though it was improvedas compared with the conventional modulation system [51]. Ho et al. proposed apolarization vectorial holographic recording based on birefringent polymeric materialscontaining a photochromic azo benzene dye. The response time was 80 μs with a 100mW power (80 μJ/cm2) of Ar laser [52]. Fichou et al. [53] proposed an incoherent-to-coherent optical converter based on photoinduced absorption of sexithiophene film. Noactual properties of such a device, including the response time, were reported.

We have also proposed a novel all-optical SLM based on complex refractive indexchanges upon photoexcitation of an organic dye-doped polymer thin film [54–60].Similar resonance shift in guided mode due to refractive index changes was firstreported by Sekkat et al. [61,62] in photoisomerization of azobenzene derivatives inpolymeric thin films by pumping at 546 nm. They studied photoisomerization andthermal relaxation by a shift of the reflectivity dips in the guided wave mode. Theobserved response time was 2–15 s, which was too slow to be used in spatial lightmodulation or optical memory [61,62]. The conformational changes (trans → cis)necessary in solid films and a smaller quantum yield of photoisomerization mightcontribute to such slow responses, though they did not show the value.

Our system is very unique as compared to the previously proposed “all-optical” lightmodulation systems as mentioned above. In principle fs response can be achieved inthis 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 thereal part of the refractive index due to transient absorption or its Kramers–Kronigtransformation, 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 complexrefractive index composed of real and imaginary parts, in general. If the imaginary partincreases due to transient absorption, for example, the reflectance increases from curvea to c as shown in Fig. 12. A change of the real part shifts the resonance from curve ato b as shown in Fig. 12 for the case of a decrease. The intensity of the probe beam canbe two-dimensionally controlled by the pump (writing) beam through photoexcitationof a dye. The main advantages using the guided mode are (1) its high sensitivity tosmall changes in refractive index and thickness, and (2) its sensitivity to both p- ands-polarized light. So far we have achieved repeated light modulation using a pulsed ns

Page 135: Nanotechnology and Nano-Interface Controlled Electronic Devices

118 T. Nagamura

Fig. 12: Schematic representation of the all-optical parallel processing in guided mode geometry and thecalculated reflectance for a polymer film (1600 nm) on a silver layer (50 nm). The complex refractive indexof 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 wascontrolled by the triplet lifetime of phthalocyanines, 30 ns for CuPcS and 0.55 ms forZnPcS. We are making efforts to achieve much faster responses using a ps or fs laserand appropriate materials.

We have also demonstrated self-held ultrafast parallel optical switching based onthe same geometry and using photochromic compounds instead of phthalocyanines[55–58,60]. Absorption spectra of spiropyran derivative (SP) dispersed in polystyrenewith a weight ratio of 1 : 10 are shown in Fig. 13 before and after UV irradiation for5 s. Strong absorption in the visible region due to photomerocyanine (PM) can beheld for a long time and be reverted to that of SP by visible irradiation. Spectra ofextinction-coefficient and refractive-index changes (Δk and Δn) of polystyrene thin filmcontaining SP upon UV excitation for 5 s are shown in Fig. 14. The former is based onthe observed difference absorption spectra before and after UV irradiation as shown inFig. 13. The latter is calculated from the extinction coefficient Δk by Kramers–Kronigtransformation. Refractive-index changes with different signs can be seen near strongabsorption changes. The extinction-coefficient and/or refractive-index changes over awide wavelength range approximately from 400 to 800 nm can be utilized to operatea wide range all-optical switch. The guided mode structures are very important to

Page 136: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 apolystyrene 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 SPand PM. The former is based on the observed difference absorption spectra shown in Fig. 13 before andafter UV irradiation. The latter was calculated from (a) by Kramers–Kronig transformation.

utilize ultrafast changes of molecular electronic state upon photoexcitation for practicalphotonics devices.

The incident-angle dependences of measured reflectance of a probe beam at 543 nmare 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 thecomposite thin film before (b) and after (c, d) excitation by a pulsed Nd–YAG laser

Page 137: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 laserexcitation (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 reflectanceand slightly shifted the dip to lower incident angles from curve b to c or d as shownin 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 ofthe 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 uponexcitation as shown in Fig. 15. The reflectance increase and the shift were found tobe due to the increase of extinction coefficient and the decrease of refractive indexat 543.5 nm of the polystyrene thin film containing SP as shown in Fig. 13 by theformation of the PM form. The changes of refractive index and extinction coefficientwere estimated to be −0.015 and +0.024 at 2.5 mJ/pulse from comparison between themeasured and the calculated dependences. Transmittances of the probe beam before andafter excitation at 2.5 mJ/pulse were calculated as comparison to the present reflectancechanges by using the same value for the film thickness and the extinction coefficientchanges. 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 tobe much better than that of transmission changes, 1.15.

The reflectance at the incident angle of 50.76° increased by 5–10 times very rapidlywith a rise time of about 20 ns upon pulsed laser excitation at 355 nm, depending onits power. This rise time corresponded to the response of a photomultiplier. Much betterswitching is expected if a picosecond laser and an experimental setup with much better

Page 138: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 polystyrenefilm 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 highvalue without applied power. The switching OFF was also demonstrated to be very fastwith a response time similar to that of switching ON, although its accurate estimationwas difficult due to a smaller S/N ratio. These results indicate that this fast reflectancedecrease was caused by the reverse photochromic reaction from PM to SP, and notby the thermal reaction. The observed smaller reflectance change as compared withswitching ON was due to a lower quantum yield of ring closure of PM. Photochromicdyes which have a higher quantum yield for reverse photochromic reaction will switchfrom ON to OFF with a larger dynamic range and fast response.

Wavelength dependences of the refractive index and the extinction coefficient changesevaluated from reflectance changes upon pulsed laser excitation in the polystyrenefilm containing SP corresponded well with the spectra of extinction-coefficient andrefractive-index changes estimated from steady photolysis as shown in Fig. 14. Theseresults confirmed the mechanism responsible for the reflectance changes in guided wavegeometry and also demonstrated the wide range of operation wavelength of the presentall-optical device.

In addition to very fast photoresponses, it is essential to write and read a two-dimensional image pattern for optical parallel data processing. Fig. 16 shows micro-scopic photographs (×500) of the image written through a USAF Test Target as amask by a single shot of ns 355 nm laser (1 mJ/pulse) on a spiropyran-doped PS film

Page 139: Nanotechnology and Nano-Interface Controlled Electronic Devices

122 T. Nagamura

deposited on a silver film. At least 128 line pairs/mm was clearly seen, which corre-sponds to a spatial resolution of 3.9 μm. Hickel et al. reported that the lateral resolutionof optical waveguide microscopy based on the resonance shift by using guided wavesas illumination light source is better than 10 μm [64]. The spatial resolution in theseimaging devices based on guided mode geometry is limited by the propagation length ofthe guided waves, which is reduced by coupling to surface plasmon states in the case ofp-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 wereproposed [58]. Optical parallel AND and OR devices are composed of two presentswitches and one switch only, respectively. They are operated by two parallel data astwo input signals, Input 1 and Input 2. An optical parallel NOT device will also becomposed by a polymer film containing a photochromic dye which will be colored onlyduring irradiation. Then, the excitation light as input signal causes an intensity decreaseof a probe beam as output signal. A combination of the present all-optical switch andthe photon-mode spatial light modulator will also contribute a great deal to constructan ultrafast parallel processing optical correlator. Several spatial optical logic devicescan be composed simply and easily from the present all-optical switch which employsreflectance increase or decrease by irradiation. An example of possible all-opticalparallel 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 presentdevice.

Page 140: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 of50.0°. (B) Relationship between the dip wavelength and incident angle for a spin-coated 360 nm thick filmof spiropyran/Arton® (1 : 2) observed with fs white light.

Page 141: Nanotechnology and Nano-Interface Controlled Electronic Devices

124 T. Nagamura

4.2. Reversible reflectance control of fs white light in guided wave geometrycontaining photochromic compounds upon fs laser photoexcitation

As mentioned in the previous section, our all-optical SLM or optical switching devicebased on complex refractive index changes upon photoexcitation of an organic dye-doped polymer thin film can make in principle fs response, because we use resonancecondition changes of the guided optical waves (guided mode) in the ATR geometrybased 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-probemeasurement system as shown in Fig. 4. Profiles of detected fs white light are shown inFig. 18(A) at an incident angle θ = 50.0° without and with a 360 nm thick Arton® filmcontaining SP. The dip found around 588 nm in Fig. 18(A) is due to the guided wavemode. 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, asshown in Fig. 19. Changes of reflected light intensity for SP/Arton® (1 : 2) film (360nm) at θ = 53.0, 50.0, 42.6, and 39.6° upon fs laser excitation at 400 nm and He–Nelaser at 543.5 nm are shown in Figs. 20–23, respectively. All reflectance changes werehighly reversible, corresponding to photochromism between SP and PM forms. The“direction” of changes of reflectance clearly depended on the incident angle or the dipwavelength. By photochromism from SP to PM, the reflectance increased with shiftingthe 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°, thedip wavelength shifted to the longer side with almost no changes of reflectance. UponCW 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 with20–160 shots.

Page 142: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 of53.0° 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 clearlyindicate that fs white light can be used as a probe light in guided mode geometryand that the fs pump-probe method will be used in such geometry. Efforts are being

Page 143: Nanotechnology and Nano-Interface Controlled Electronic Devices

126 T. Nagamura

Fig. 21: Changes of reflected light intensity for SP/Arton® (1 : 2) film (360 nm) at an incident angle of50.0° 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 ultrafastphotoresponsive materials.

Page 144: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 of42.6° 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, wavelengthconversion, electro-optical or pure optical control of the refractive index, and all-optical

Page 145: Nanotechnology and Nano-Interface Controlled Electronic Devices

128 T. Nagamura

Fig. 23: Changes of reflected light intensity for SP/Arton® (1 : 2) film (360 nm) at an incident angle of39.6° 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 mainproblems especially in organic compounds is the small nonlinear optical coefficient. Wehave also been making efforts to modulate or enhance the second- and the third-ordernonlinear optical properties by changing the electronic state or the extent of electronic

Page 146: Nanotechnology and Nano-Interface Controlled Electronic Devices

Materials and devices for ultrafast molecular photonics 129

distribution upon photoexcitation [65–73]. We also observed considerable enhancementof 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 thelifetime of the colored state by photoinduced electrochromism, the complex refractiveindex, and the nonlinear optical responses by the interactions with photons. Applicationof such materials to novel optical devices based on photoinduced complex refractiveindex changes was proposed and was successfully demonstrated in photon-mode record-ing and reflection control. These results will contribute a great deal to realize parallelall-optical ultrafast data processing devices. Considering that our vision is initiated bya simple photoisomerization of 11-cis retinal and is processed in a parallel way toachieve extreme high functions, molecular photonics based on elegant combination ofmolecules, photons and appropriate devices is expected to be the very promising way ofultrafast 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, andMr. I. Yoshida for their great contributions. Partial supports by the Grant-in-Aids forScientific 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. T. Kamiya, F. Saito, O. Wada, and H. Yajima (Eds.), Femtosecond Technology, Springer Series inPhotonics 2, Series Eds. T. Kamiya, B. Monemar, and H. Venghaus (Springer-Verlag, Berlin, 1999).

2. T. Nagamura, Ch. VI-4 in Ref. [1], (1999), p. 376.3. 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. 387.4. H. Dürr and H. Bouas-Laurent, Photochromism Molecules and Systems (Elsevier, Amsterdam, 1990).5. M. Irie, Chem. Rev. 100 (5), 1685 (2000) and references cited therein.6. Y. Yokoyama, Chem. Rev. 100 (5), 1717 (2000) and references cited therein.7. G. Berkovic, V. Krongauz, and V. Weiss, Chem. Rev. 100 (5), 1741 (2000) and references cited

therein.8. T. Nagamura and K. Sakai, J. Chem. Soc. Faraday Trans. 1 84, 3529 (1988).9. T. Nagamura and S. Muta, J. Photopolym. Sci. Technol. 4, 55 (1991).

10. T. Nagamura and K. Sakai, J. Chem. Soc. Chem. Commun., 810 (1986) .

Page 147: Nanotechnology and Nano-Interface Controlled Electronic Devices

130 T. Nagamura

11. T. Nagamura and K. Sakai, Ber. Bunsenges. Phys. Chem. 93, 1432 (1989).12. T. Nagamura, K. Sakai, and T. Ogawa, J. Chem. Soc. Chem. Commun., 1035 (1988).13. T. Nagamura and K. Sakai, Thin Solid Films 179, 375 (1989).14. T. Nagamura and Y. Isoda, J. Chem. Soc. Chem. Commun., 72 (1991).15. T. Nagamura, Polym. Int. 27, 125 (1992).16. T. Nagamura, S. Muta, and K. Sakai, J. Photopolym. Sci. Technol. 5, 561 (1992).17. T. Nagamura, Mol. Cryst. Liq. Cryst. 224, 75 (1993).18. T. Nagamura, H. Sakaguchi, S. Muta, and T. Ito, Appl. Phys. Lett. 63, 2762 (1993).19. T. Nagamura, H. Sakaguchi, T. Ito, and S. Muta, Mol. Cryst. Liq. Cryst. 247, 39 (1994).20. T. Nagamura, H. Sakaguichi, and S. Muta, Proc. SPIE 2514, 241 (1995).21. T. Nagamura, Pure Appl. Chem. 68 (7), 1449 (1996).22. H. Inoue, H. Sakaguchi, and T. Nagamura, Appl. Phys. Lett. 73, 10 (1998).23. T. Nagamura and K. Sakai, Chem. Phys. Lett. 141, 553 (1987).24. T. Nagamura and K. Sakai, Ber. Bunsenges. Phys. Chem. 92, 707 (1988).25. H. Nishida, N. Takada, M. Yoshimura, T. Sonoda, and H. Kobayashi, Bull. Chem. Soc. Jpn. 57, 2600

(1984).26. R.A. Marcus, J. Chem. Phys. 24, 966 (1956).27. J.R. Miller, L.T. Calcaterra, and G.L. Closs, J. Am. Chem. Soc. 106, 3047 (1984).28. T. Nagamura, H. Sakaguchi, K. Suzuki, C. Mochizuki, and K. Sasaki, J. Photopolym. Sci. Technol. 6,

133 (1993).29. T. Nagamura, H. Sakaguchi, K. Sasaki, C. Mochizuki, and K. Suzuki, Thin Solid Films 243, 660

(1994).30. T. Nagamura, D. Kuroyanagi, K. Sasaki, and H. Sakaguchi, Proc. SPIE 2547, 320 (1995).31. K. Sasaki and T. Nagamura, J. Photopolym. Sci. Technol. 9, 129 (1996).32. K. Sasaki and T. Nagamura, Mol. Cryst. Liq. Cryst. 294, 145 (1997).33. H. Kawai, K. Nakano, and T. Nagamura, Chem. Lett., 1300 (2001).34. T. Nagamura, A. Tanaka, H. Kawai, and H. Sakaguchi, J. Chem. Soc. Chem. Commun., 599 (1993).35. T. Nagamura, H. Kawai, T. Ichihara, and H. Sakaguchi, Synth. Metals 71, 2069 (1995).36. H. Kawai and T. Nagamura, Mol. Cryst. Liq. Cryst. 267, 235 (1995).37. T. Nagamura, T. Ichihara, and H. Kawai, J. Phys. Chem. 100, 9370 (1996).38. T. Nagamura, S. Kashihara, and H. Kawai, Chem. Phys. Lett. 294, 167 (1998).39. H. Kawai and T. Nagamura, J. Chem. Soc. Faraday Trans. 94, 3581 (1998).40. W.-S. Xia, H. Kawai, and T. Nagamura, J. Photochem. Photobiol. A: Chem. 136 (1–2), 35 (2000).41. H. Kawai and T. Nagamura, Mol. Cryst. Liq. Cryst. 344, 209 (2000).42. S.H. Park, H. Kawai, and T. Nagamura, J. Photopolym. Sci. Technol. 13 (2), 197 (2000).43. S.H. Park and T. Nagamura, Mol. Cryst. Liq. Cryst. 370, 249 (2001).44. S.H. Park and T. Nagamura, J. Photopolym. Sci. Technol. 14 (2), 227 (2001).45. S.H. Park and T. Nagamura, J. Chem. Soc. Chem. Commun., 2344 (2001).46. S.H. Park, H. Kawai, and T. Nagamura, J. Chem. Soc. Perkin Trans. 2 (3), 508 (2002).47. A. Partori, A.M. Glass, T.H. Chiu, and D.T.H. Liu, Opt. Lett. 18, 906 (1993).48. P. Tayebati, E. Canoglu, C. Hantzis, and R.N. Sacks, Appl. Phys. Lett. 71, 1610 (1997).49. 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).50. T. Okamoto, T. Kamiyama, and I. Yamaguchi, Opt. Lett. 18, 1570 (1993).51. A. Yacoubian and T.M. Aye, Appl. Opt. 32, 3073 (1993).52. Z.Z. Ho, G. Savant, J. Hirsh, and T. Jannson, Proc. SPIE 1773, 433 (1992).53. D. Fichou, J.-M. Nunzi, F. Charra, and N. Pfeffer, Adv. Mater. 6, 64 (1994).54. T. Nagamura and T. Hamada, Appl. Phys. Lett. 69, 1191 (1996).55. K. Sasaki and T. Nagamura, Appl. Phys. Lett. 71, 434 (1997).56. K. Sasaki and T. Nagamura, J. Appl. Phys. 83, 2894 (1998).57. T. Nagamura and K. Sasaki, Proc. SPIE 3466, 212 (1998).58. T. Nagamura and K. Sasaki, Mol. Cryst. Liq. Cryst. 344, 199 (2000). .59. T. Nagamura, T. Adachi, I. Yoshida, H. Inoue, H. Heckmann, and M. Hanack, Mol. Cryst. Liq. Cryst.

370, 97 (2001).

Page 148: Nanotechnology and Nano-Interface Controlled Electronic Devices

Materials and devices for ultrafast molecular photonics 131

60. T. Nagamura, K. Sasaki, F. Iizuka, T. Adachi, and I. Yoshida, Opt. Commun. 205, 107 (2002).61. Z. Sekkat and M. Dumont, Appl. Phys. B 53, 121 (1991).62. Z. Sekkat, D. Morichere, M. Dumont, R. Loucif-Saibi, and J.A. Delaire, J. Appl. Phys. 71, 1543

(1992).63. T. Ito, M. Hiramatsu, M. Hosoda, and Y. Tsuchiya, Rev. Sci. Instrum. 62, 1415 (1991).64. W. Hickel and W. Knoll, Appl. Phys. Lett. 57, 1286 (1990).65. H. Sakaguchi, T. Nagamura, and T. Matsuo, Jpn. J. Appl. Phys. 30, L377 (1991).66. T. Nagamura, H. Sakaguchi, and T. Matsuo, Thin Solid Films 210/211, 160 (1992).67. H. Sakaguchi, L.A. Gomez-Jahn, M. Prichard, T.L. Penner, D.G. Whitten, and T. Nagamura, J. Phys.

Chem. 97, 1474 (1993).68. H. Sakaguchi and T. Nagamura, Nonlinear Optics 15, 73 (1996).69. 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, 209.70. A. Harada and T. Nagamura, Mol. Cryst. Liq. Cryst. 316, 79 (1998).71. A. Harada and T. Nagamura, Nonlinear Optics 22, 169 (1999).72. H. Sakaguchi and T. Nagamura, Nonlinear Optics 22, 413 (1999).73. T. Nagamura, T. Umeda, and H. Sakaguchi, Mol. Cryst. Liq. Cryst. 370, 119 (2001).

Page 149: Nanotechnology and Nano-Interface Controlled Electronic Devices

This Page Intentionally Left Blank

Page 150: Nanotechnology and Nano-Interface Controlled Electronic Devices

Nanotechnology and Nano-Interface Controlled Electronic DevicesEditors: 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. Conducting organic materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1332. Conduction in organic LED: experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1343. Conduction in organic LED: modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1354. Band model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1365. Hopping and tunneling models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1376. Carrier injection model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1397. Space Charge Limited Current (SCLC) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140

7.1. Theoretical introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1407.2. Experimental verification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142

8. Simulation of carrier transport by directly calculated hopping model . . . . . . . . . . . . . . . 1448.1. Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1448.2. Model in detail . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1458.3. Carrier behaviors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1498.4. Transient response characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1518.5. Summary of the simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153

9. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154

1. Conducting organic materials

Organic materials have been regarded as insulating before the appearance of conductivepolymers. The concept of “organic semiconductor” was revealed from the studies ofπ-conjugated polymers, that polyacethylene was discovered, and that the doping methodwas developed [1,2]. In addition, research fields such as organic functional materialsand organic electronics are growing through the application of photosensitive materials(photoconduction materials) to electrophotography. However, it is very unstable state foran essentially neutral organic molecule to ionize by negatively or positively dischargingas 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

Page 151: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 H2O.

2. Conduction in organic LED: experiment

The current–voltage (I–V ) characteristics in OLEDs show a non-linear behavior. Forexample, 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-oxideand 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. Alq3is 8-hydroxyquinoline aluminum (Alq3) and a most famous emitting material. CuPcis 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 bilayerOLED, 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 steeplyand 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.

Page 152: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 thefollowing 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, andE is an electric field. However, if the current in a material is caused by many kindsof charged carriers (i.e. electron, hole, anions, cations), the current density must bedescribed as

J =k∑

i=1

qi niμi E , (2)

where qi is the charge of the i th carrier species, ni is the carrier density of the i th carrierspecies, μi is the mobility of the i th 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 layerand 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, theirconduction mechanism is very complicated.

Let us discuss the simple bilayer OLED with TPD as a hole transport layer and Alq3as an emitting layer. TPD is well-known to be a hole transport material and then weconsider only hole conduction in the TPD layer. In addition, the electron injection fromAlq3 into TPD is strongly blocked because of the high barrier height between TPD andAlq3. On the other hand, Alq3 is a weak electron transport material because its electron

Page 153: Nanotechnology and Nano-Interface Controlled Electronic Devices

136 T. Mori and T. Mizutani

mobility is 100 times larger than its hole mobility [7]. Strictly speaking, the current ofthe TPD/Alq3 device must be written as

J = JTPD = JAlq3 = Jh,TPD + Je,TPD = Jh,Alq3 + Je,Alq3

= epTμhT ET + enTμeT ET = epAμhA EA + enAμeA EA, (3)

where p and n are hole and electron densities, respectively. Subscripts T , A, e, and hmean TPD, Alq3, electron and hole, respectively. In a steady state, JTPD = JAlq3 (the lawof continuity of current). Even if TPD thickness agrees with Alq3 thickness, the dividedvoltage of TPD layer is different from that of Alq3 because the former conductivity islower than the latter. When the above experimental results are considered, Je,TPD can beneglected. 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 densityinjected from TPD. Consequently, Eq. 3 becomes

J = epTμhT ET = epAμhA EA + enAμeA EA. (4)

The next problem is that carrier density and mobility in organic materials dependon electric field. That is, carrier density and mobility cannot be regarded as constantparameters. In addition, electric field is obtained as a function of position as well aseach layer. As ET or EA is each average electric field in the TPD or Alq3 layer, thisexpression is ambiguous and it is right that the electric field should be described asE(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 ofthe device. In addition, as high-performance OLEDs have a complicated multi-layerstructure, one can understand that it is not easy to describe an analytical solution as theconduction model of OLEDs.

In OLEDs, the fact that organic materials have low carrier mobilities is thought tolead to that the conduction mechanism in OLEDs is due to the space charge limitedcurrent (SCLC) model. In PLEDs, the SCLC model is comparably easy to be acceptedbecause their layer structures are simpler than those of low-molecular LEDs.

The conductive mechanism in OLEDs is categorized by two models: One is thatthe current in OLEDs is strongly controlled by injected carrier density since organicmaterials have low carrier concentration. The other is that it is strongly controlled bycarrier 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 orelectron injection. However, the value of a physical parameter (a dielectric constant,the barrier height of carrier injection, etc.) determined from the approximate I–V curveis very different from that estimated from a direct measurement [8,9]. Although someinterpretations for the conflict are suggested, they are not thought to be consistent withthe physical phenomena. The carrier transport in OLEDs cannot be explained only bythe unipolar carrier injection model.

4. Band model

In general, the conduction behavior in OLEDs is often explained using an energydiagram on the basis of the band model. For example, the OLEDs in Figs. 2 and 3 can

Page 154: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 conductionband and valence band will be matched with LUMO (lowest unoccupied molecularorbital) and HOMO (highest occupied molecular orbital) levels, respectively. However,since the interaction between organic molecules is a van der Waals force, which ismuch weaker than covalent bond and metallic bond, the band width of the energyband becomes narrow even if an energy band may be formed in organic materials. Thenarrow band means low mobility for carrier transport in the band model. In nature, thecarrier mobility in the band model is more than several hundreds cm2/V s. The largestmobility in organic materials is at most 1 cm2/V s, the carrier mobility in a pentacenecrystal [10]. We must think that it is not appropriate to apply the band model to organicmaterials.

5. Hopping and tunneling models [11]

Let us remember that the carrier transport in organic materials is caused by alternateionization between ionized molecules and neutral molecules, as shown in Fig. 1. That is,

M+(−) +M −→ M+M+(−).

Page 155: Nanotechnology and Nano-Interface Controlled Electronic Devices

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. hoppingand tunneling, processes.

Now we regard the potential diagram of neighboring molecules in Fig. 5. U is thebarrier height of the potential. a is the distance between two neighboring molecules.The hopping probability of thermally activated charge is given by

P = ν exp

(− U

kT

), (5)

where ν is trial frequency factor and k is Boltzmann constant. The mobility underelectric field is given by the Einstein relation

μ = eD

kT, (6)

where D is the diffusion coefficient. In addition, using D = Pa2, we obtain thefollowing relation,

μ = ea2ν

kTexp

(− U

kT

). (7)

When temperature increases, the preexponential factor decreases inversely proportionalto temperature but the exponential term increases steeply. Consequently, the thermallyactivated hopping process has a positive temperature dependence.

On the other hand, the charge transfer between molecules may be caused by atunneling process. The tunneling probability, PT depends on the number of carrierscolliding with the potential barrier, N and the tunneling factor, T . PT can be describedas the product of N and T , i.e. PT = N T . T can be written as

T = T0 exp

(−2w

√2m(U − E)

h

), (8)

where T0 is a constant, w is the barrier width, m is the electron mass, U is the barrierheight 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 notaffected by temperature, it strongly depends on the distance between one molecule andthe counter as well as the electric field. Usually the effective distance for tunnelingtransfer is said to be < 1 nm.

Page 156: Nanotechnology and Nano-Interface Controlled Electronic Devices

Carrier transport behavior in OLED 139

6. Carrier injection model

Some researchers propose that the current in OLEDs can be simulated by the Schottkyinjection model [8,9]. The ground for their proposal is the temperature dependenceof the current as shown in Fig. 6. If the current in OLEDs could be simulated bythe tunneling injection model, it would not have shown a remarkable temperaturedependence. However, let us remember Eq. 1. Since organic materials do not haveintrinsic carrier density because they are essentially insulators, carrier density in bulkis due to carrier injection from the electrodes. Therefore, if current depended on onlycarrier density, we would consider the Schottky injection for the conduction in OLEDssince the current in OLEDs has temperature dependence. But we need to remember thatthe carrier transportation in organic materials is not caused by band conduction, but bysuch a discontinuous process as hopping conduction. The hopping conduction modelhas temperature dependence.

The current due to Schottky injection is described as

J = AT 2 exp

(βE1/2 −φ

kT

), (10)

where A is Richardson–Dushman’s constant, φ is the barrier height of carrier injection,and β is defined as

β =√

e3

4πε. (11)

This current depends on the squared electric field as well as temperature. In order tojudge whether Schottky current can be applied to the conduction current of a materialor not a Schottky-plot, ln J : E1/2, is often used. When the dielectric constant estimatedfrom the gradient of the graph agrees with the experimental value, it is possible thatthe 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 theconduction mechanism should be treated carefully.

Fig. 6: The temperature dependence of the current density–electric field characteristics of ITO/TPD[50nm]/Alq3[70nm]/Mg.

Page 157: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 aninjection-controlled conduction but a bulk-controlled one. Some researchers believe thatthe current density, J , due to SCLC equation is

J = 9

8

εμV 2

d3, (12)

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 Poissonequation and the boundary condition, E(0) = 0, V (0) = 0. In this section, we discuss theproblem of the SCLC model.

The SCLC model ought essentially to be applied to unipolar conduction. Whenvoltage is applied to an insulator (organic material) interposed by two electrodes and thecharged carriers injected from an electrode are not neutralized by the counter chargedcarriers injected from the counter electrode, the injected charged carriers form a spacecharge around the electrode. This space charge modifies the electric field between theelectrodes in the case of low mobility. The homo space charge accumulated in front ofan electrode reduces the electric field on the electrode. Therefore, the carrier injectionafter forming the space charge strongly depends on the modified electric field due tospace charge.

The necessary conditions that a conduction current becomes a SCLC are thefollowing:

1. the current due to injected carriers has the same or higher value as the Ohmiccurrent;

Fig. 7: Typical SCLC characteristics.

Page 158: Nanotechnology and Nano-Interface Controlled Electronic Devices

Carrier transport behavior in OLED 141

2. the dielectric relaxation time of the material is longer than the carrier drift timebetween the electrodes in the material.

A typical SCLC example is the anode current–voltage characteristics of a two-electrode vacuum tube. In general, it is not easy for organic materials to satisfy thenecessary condition 1.

Let us find Eq. 12 from Poisson’s equation. We regard a one-dimensional systemcontaining the sample with two electrodes. The interface between cathode and sample isx = 0. Now we imagine that electrons are injected into the sample. Poisson’s equation is

d2ϕ

dx2= −en

ε, (13)

where ϕ is the potential in the sample, n is injected electron density, and ε is thedielectric 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,

dE

dx= J

εμE. (15)

Integrating Eq. 15 with respect to x after separating variables,

E2(x) = 2J

εμx +C , (16)

Using E(0) = 0, C = 0, therefore:

E(x) = ±√

2J

εμx1/2. (17)

However, as the positive solution is not appropriate for this case,

E(x) = −√

2J

εμx1/2. (18)

V (x) is given by integrating E(x) with respect to x .

V (x) = −∫ x

0E(x)dx =

∫ x

0

√2J

εμx1/2 dx

=√

8J

9εμx3/2 +C ′ (19)

We can use V (0) = 0, C ′ = 0. When the sample thickness is d and the applied voltage isV , the following equation can be given (Fig. 8),

J = 9

8

εμV 2

d3. (12)

Page 159: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 someproblems for applying the SCLC model to organic LEDs. The system of the organicLED may not satisfy the above necessary conditions.

Fig. 9 shows the experimental and calculated J–V characteristics of the OLEDsas shown in Figs. 2 and 3. The lines with marker are the experimental curves andthe lines without marker are calculated curves. Estimated values are used for theparameters: ε = εrε0, the dielectric constant of organic material, εr is about 3 and ε0 isthe permittivity of vacuum, μ is 10−3–10−6 cm2/V s. The thickness, d, is the sum of theCuPc and NPD layers in the dash-dotted line. d is the total thickness of organic layersfor the other lines. The carrier mobility in both the dash-dotted line and the solid line is10−3 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 thecalculated curves. In the high current region, the value of the former approaches thecalculated 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 transferdue to photoexcited carriers with high energy (> 3 eV), it is possible to overestimatethe intrinsic mobility which will be excited by thermal activation (∼0.026 eV). Carriermobility of organic materials needs to be discussed in detail.

Fig. 10 shows the conductivity, dielectric relaxation time and drift time–voltagecharacteristics of ITO/TPD[50nm]/Alq3[50nm]/AlLi. Each parameter is calculated bythe following. The apparent conductivity, σ is calculated by σ = J/E = Jd/V . The

Page 160: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 calculatedcurves. TPD thickness is used as d in the dash-dotted line. d is the total thickness of organic layers in theother lines. The carrier mobility in both the dash-dotted line and solid line is 10−3 cm2/V s. That of thedotted line, short-dashed one, and long-dashed one is 10−4, 10−5, and 10−6 cm2/V s, respectively.

10-15

10-13

10-11

10-9

10-7

σ [

S/c

m]

12840Voltage [V]

10-7

10-5

10-3

10-1

101

τ [s

]

p

TPD/Alq3

Fig. 10: The apparent conductivity, dielectric relaxation time and drift time as function of voltage inITO/TPD/Alq3/AlLi: closed circles mean apparent conductivity, open circles mean dielectric relaxationtime, and solid line is calculated drift time.

Page 161: Nanotechnology and Nano-Interface Controlled Electronic Devices

144 T. Mori and T. Mizutani

dielectric relaxation time, τ , is estimated from τ = ε/σ . We used 3ε0 as the dielectricconstant of organic materials: ε0 is the permittivity of vacuum. The drift time, td iscalculated by td = d/v = d/μE . We used a constant mobility, 10−6 cm2/V s, as averagemobility. The hole mobility in TPD was estimated to be ∼ 10−3 cm2/V s [19] and thehole 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, wethink that the carrier mobility obtained by the time-of-flight method is overestimatedas the carrier mobility of organic material. In addition, the exact drift time in organicLEDs can be obtained by summing the hole drift time in TPD and the electron drift timein 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/cmbelow ∼ 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 notsatisfied in OLEDs. We have to consider the conduction mechanism in OLEDs on thebasis 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 twoelectrodes is simple, the light-emitting mechanisms of the device are quite complicated.These mechanisms may be roughly divided into three processes: the carrier injectionprocess from each electrode, the carrier transport process, and the emission process viaexcitons generated by electron–hole recombination. For example, many researchers triedto explain the carrier injection mechanism of OLED from the viewpoint of experimentalcurrent–voltage characteristics. However, such external information is insufficient toexplain the injection mechanism. Clarification of each process will ease improvement ofcurrent performance of the device.

When we improve on the device performance, it is important to discuss the balancebetween electron and hole injections. Rate of electron–hole recombination, electric field,and space-charge distributions in the OLED are also important. However, it is impossibleto obtain and evaluate these parameters experimentally because these parameters are“internal” OLED parameters. In the present work, we assumed a simple model andattempted to calculate carrier behavior in OLED in order to clarify light-emittingmechanisms.

Many groups have attempted to simulate I–V characteristics of devices [27–35].Calculations were carried out using a “continuous model” in which conduction currentdensity 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–Boltzmannstatistics as accompanied with Schottky and tunneling injection [28].

Page 162: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 conductionmechanism 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 ofsemiconductors [30]. They used Fowler–Nordheim and SCLC currents.

Malliaras et al. used numerical methods to calculate the current and the efficiencyof a single-layer organic LED, taking into account field-dependent mobilities, diffusion,and thermionic injection [31]. Staudigel et al. quantitatively simulated the conductionand 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 dependenceof the Pool–Frenkel form [33]. And they treated the conduction of PLEDs as a bipolarmechanism. They tried to explain the change of conduction in a single-layer PLEDcaused by the difference of cathode metal using their model. They claimed that theirmodel successfully describes the I–V characteristics of a single-layer PLED. Croneet al. applied their conduction model to single-layer OLEDs [34]. They calculated thespatial 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 alwaysan equilibrium carrier density [36]. In general, space charge limited current (SCLC) isused to explain conduction of organic thin films such as OLEDs. In this model, theinjection 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 proposedby Iwamoto and Hino [37]. Although carrier density is not limited in continuousmodels, the carrier number accepted by a molecule is limited in our model. Because thecarrier transport between organic molecules is regarded as an intermolecular oxidation–reduction, our model approximates carrier behavior more accurately than conventionalcontinuous models. In continuous models, the carrier number accepted by a molecule isnot limited.

We assumed a bilayer OLED of ITO/TPD/Alq3/Al. Thickness of each organic layeris 50 nm. Since an Alq3 molecule is represented by a sphere of 0.8 nm diameter, weapproximate that these molecules are arranged with average distance of 1.73 nm in anelectric field. The number of sites is 30. Molecular stacking is not considered. Maximumcarrier 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 hoppingsite, (II) a site can be occupied by an electron or a hole at most, (III) carriers move onlyto adjacent sites, and (IV) the hopping rate depends on not only to carrier density, butalso to the rate of unoccupied adjacent sites. Conduction currents from the kth site to the

Page 163: Nanotechnology and Nano-Interface Controlled Electronic Devices

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

Jp(k,k+1) = ν ′pqpk

[N − pk+1 − rk+1

N

]exp

(qaF(k,k+1)

2kBT

)

−ν ′pqpk+1

[N − pk − rk

N

]exp

(−qaF(k,k+1)

2kBT

)(k = 1,2, . . . ,m) (20)

ν ′p = ν exp

(−U ′p

kBT

), (21)

and those for electrons (Jn(k,k+1)) are

Jn(k,k+1) = ν ′nqnk

[N −nk+1 − rk+1

N

]exp

(qaF(k,k+1)

2kBT

)

−ν ′nqnk+1

[N −nk − rk

N

]exp

(−qaF(k,k+1)

2kBT

)(k = 1,2, . . . ,m −1) (22)

ν ′n = ν exp

(−U ′n

kBT

), (23)

Page 164: Nanotechnology and Nano-Interface Controlled Electronic Devices

Carrier transport behavior in OLED 147

where N represents the maximum site density for a molecular layer (= 1018 m−2);pk , nk , 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 U ′n (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, andq are temperature, the Boltzmann constant, and elementary charge, respectively. Thehopping distance a is assumed to be 1.73 nm, which is the average distance betweenthe two centers of adjacent molecules. Up and Un were calculated using the equationfor conventional hopping transport from experimental carrier mobility, μp and μn [7].Electron mobility is about 100 times higher than hole mobility in Alq3; hole mobilityin the TPD bulk is about five orders of magnitude higher than that in the Alq3 bulk.We use Up (0.27 eV) as U ′

p and Un (0.15 eV) as U ′n , respectively, in Alq3. At the

TPD/Alq3 interface, U ′p and U ′

n are Up +φbp and Un +φbn , where φbp (0.26 eV) andφbn (0.83 eV) are barrier heights for the hole and the electron, respectively. Since φbn isso high that electrons are almost blocked at the TPD/Alq3 interface, electron behaviorcan be ignored in the TPD bulk. We use both Schottky emission and Fowler–Nordheimemission for electron injection from the cathode. The electron current density passingbetween the Alq3 and the cathode interface, Jn(m,m+1), is assumed as

Jn(m,m+1) =[

N −nm − rm

N

][An T 2 exp

(−φn

kBT

)exp

(q

kBT

√q F(m,m+1)

4πεrε0

)

+ρq F2

(m,m+1)

8πhφnexp

(−8π

√2m∗φ3

n

3qh F(m,m+1)

)]−ν ′

nqnm exp

(−qaF(m,m+1)

2kBT

), (24)

where φn (eV) is the barrier height for electron injection from the cathode to an Alq3molecule and is estimated to be 0.67 eV, and An , ε0, and εr are initial parameters basedon the Richardson–Dushman constant for electrons, vacuum permittivity, and dielectricconstant of Alq3 bulk, respectively. Hole injection from an anode is assumed to be dueto 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 chargedensity is negligible in the TPD bulk except for the site adjacent to the Alq3. Thus, theinterface is assumed to be a hole reservoir as shown in Eq. 25. As holes are accumulatedin the TPD site closest to the TPD/Alq3 interface, we can regard this site as a reservoirfor holes. Hole density is represented as pres, that is, p0 = pres. Therefore, the holeconduction current passing through the TPD/Alq3 interface is obtained by substitutingpres into Eq. 20:

Jp(0,1)) =[

N − pres

N

]ApT 2 exp

(−φp

kBT

)exp

(q

kBT

√q F(0,1)

4πεrε0

)

−ν ′pqp1

[N − pres

N

]exp

(−qaF(0,1)

2kBT

). (25)

The barrier height, φbp, for hole injection from the TPD molecule to Alq3 isestimated to be 0.26 eV. Current density flowing in an external circuit consists of the

Page 165: Nanotechnology and Nano-Interface Controlled Electronic Devices

148 T. Mori and T. Mizutani

hole conduction component Eq. 26 and the electron conduction one Eq. 27, both ofwhich are derived from the continuity equation under DC field, Jp is hole currentdensity; Jp(0,1), Jp(m,m+1), wJp(0,1) and the sum of Jp(k,k+1) are hole current densitiesflowing in the TPD/Alq3 interface, the Alq3/Al interface, the TPD bulk, and the Alq3bulk, respectively.

Jp = 1

2{Jp(0,1) + Jp(m,m+1)}+wJp(0,1) +

m−1∑k=1

{Jp(k,k+1)}. (26)

In the equation above, Jn is electron current density; Jn(0,1), Jn(m,m+1), and the sum ofJn(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.

Jn = 1

2{Jn(0,1) + Jn(m,m+1)}+

m−1∑k=1

{Jn(k,k+1)}. (27)

Here, w is the number of sites in TPD. Time variation of hole density is shown in Eq. 28and that of electron density is shown in Eq. 29.

dpk

dt= 1

q{−Jp(k,k+1) + Jp(k−1,k)}− Rnk pk , (28)

dnk

dt= −1

q{−Jn(k,k+1) + Jn(k−1,k)}− Rnk pk , (29)

where R is the electron–hole recombination coefficient for Alq3 molecules. The fieldsare expressed as Eqs. 30–32, which are derived from the Poisson equation.

F(k,k+1) = −qa

εrε0d

[k∑

s=1

(s − 1

2

)(ps −ns)

]

+ qa

εrε0d

[m∑

s=k+1

(m − s + 1

2

)(ps −ns)

]− Va

d, (30)

F(0,1) = −qa

εrε0d

[m∑

s=k+1

(m − s + 1

2

)(ps −ns)

]− Va

d, (31)

F(m,m+1) = −2qa

εrε0d

[m∑

s=k+1

(m − s + 1

2

)(ps −ns)

]− Va

d. (32)

In these equations, d and Va are thickness and applied voltage of the device. When L isthe length of exciton diffusion and τ is the fluorescence lifetime in Alq3, the diffusioncoefficient, D, is shown by

D = L2

τ. (33)

Time variation of exciton density is shown by

dnk

dt= Rnk pk + D(N − pk −nk − rk)

d2rk

dk2+ Drk

d2

dk2(N − pk −nk − rk)− rk

τ. (34)

Page 166: Nanotechnology and Nano-Interface Controlled Electronic Devices

Carrier transport behavior in OLED 149

Electroluminescence (EL) intensity is assumed to be proportional to the sum ofRnk pk in the Alq3 layer (k: from 1 to m),

EL ∝ 1

τ

m∑k=1

Rnk pk . (35)

8.3. Carrier behaviors

In this simulation, the carrier (electron and hole) distribution and field distribution aswell as current density and EL intensity are calculated when a DC step voltage isapplied.

Distributions of hole density, electron density, and exciton generation density areshown in Figs. 12, 13, amd 14, respectively. In these calculations, a recombination rateR = 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. Holedensity decreases with distance from the TPD/Alq3 interface. Holes are accumulatedwithin 10 nm distance from the interface (Fig. 12) because of the low hole mobilityin the Alq3 layer. In the emission layer (Alq3), electrons injected from a cathodemove to the TPD/Alq3 interface. Electrons are comparatively uniformly distributed inAlq3 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, asshown in Fig. 13. Distribution of hole density differs from that of electron densitybecause the electron mobility is 100 times faster than the hole mobility in the Alq3layer. Fig. 14 shows distribution of generated exciton density after 30, 100, and 250ns. 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 recombinationof electrons and holes, resulting in generating excitons near the TPD/Alq3 interface.

Fig. 12: Distribution of hole density in Alq3 layer.

Page 167: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 interfacewith time. Since electron density is lower than hole density, all electrons are consideredto recombine with holes before reaching the TPD/Alq3 interface.

Fig. 15 shows the field distribution in both organic layers at an average field ofFa = 140 MV/m. Field distortion in TPD bulk is little observed at Fa = 140 MV/mwhere an OLED shows strong luminance of over 600 cd/m2. Our one-dimensionaldiscontinuous calculation model suggests that conduction in OLEDs cannot be explainedby a typical SCLC conduction model since field distortion is not observed near bothcathodes in organic layers.

Page 168: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 intensityat Fa = 140 MV/m. The hole current density, Jp, at 100 ns decreases until it reaches90% 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 nearthe interface (Fig. 15). Also, the amount of injected electrons decreases. At 30 ns, theelectron current density is 90% of that at 100 ns and saturated. Thus, electron currentdensity appears to be saturated after 100 ns. It has a turning point at 30 ns; afterwhich 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.

Page 169: Nanotechnology and Nano-Interface Controlled Electronic Devices

152 T. Mori and T. Mizutani

Fig. 17: Time dependence of EL intensity at an average field of 140 MV/m.

100

101

102

103

104

Curr

ent densi

ty [A

m-2]

1401201008060

F [MVm-1]

experiment calculation

Fig. 18: Comparison of calculated and experimental field dependence of current density.

equalize due to exciton generation near the TPD/Alq3 interface. When the appliedelectric field is small, (Fa = 100 MV/m), the delay time of EL (solid line) and 90% ofEL 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. Thecalculated current density normalized by the current density at Fa = 100 MV/m are usedto calculate those at other Fa. The calculated curves (solid line) agree to the experimentalones (dashed line), as shown in Fig. 18. In our previous work, we considered onlySchottky emission as electron injection mechanism. EL intensity (Fig. 19) did not agreewith experimental values at low electric field, although the calculated density agreedwith experimental data. Considering both Fowler–Nordheim emission and Schottkyemission into the electron injection mechanism, Fowler–Nordheim emission is dominantin high fields, as shown in Fig. 20.

Page 170: Nanotechnology and Nano-Interface Controlled Electronic Devices

Carrier transport behavior in OLED 153

1019

1020

1021

1022

EL Inte

nsi

ty [arb

.units

]

1401201008060

F [MVm-1]

100

101

102

103

104

Lu

min

an

ce [

cdm

-2]

experiment

calculation

Fig. 19: Comparison of calculated and experimental field dependence of EL intensity.

10-1

100

101

102

103

Jn(m

+w

,m+

w+

1)

[Am

-2]

1601401201008060

F [MVm-1]

total ( i + ii ) i. Fowler Nordheim ii. Schottky

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 emittingmolecule corresponds to a hopping site simulating actual charge transfer between adja-cent molecules. Time dependence of carrier, exciton and EL intensity, and distributionsof field and carrier density are calculated.

Hole and electron densities decrease near the TPD/Alq3 interface. As a result, thedensity of exciton generation achieves its maximum within 10 nm from the TPD/Alq3interface. Field distribution due to the space charge effect is not apparent in the TPD

Page 171: Nanotechnology and Nano-Interface Controlled Electronic Devices

154 T. Mori and T. Mizutani

bulk. These results suggest that the conduction mechanism in bilayer OLEDs cannotbe explained by a typical SCLC conduction model. This model accommodates Fowler–Nordheim emission as an electron injection mechanism. As a result, behavior of currentdensity and EL intensity agree with measured current density and luminance. Fromabove results, a simple bilayer and discontinuous model is effective for investigatingOLED carrier behavior.

9. Conclusion

We showed that it is difficult to directly apply the SCLC model to OLEDs. If aresearcher believes that the conduction mechanism of a material can be explained bya particular mechanism, that is, an equation, he can analyze the conduction current bythe equation. And he may obtain the various information on the conduction mechanismof a material. However, if he does not verify his calculated parameters by means ofother experimental results, his analysis will be almost nonsense in the special case withadditional assumptions. Of course, our simulation is still incomplete and also needs toreflect the experimental results. Since OLEDs have multi-layer structure and a bipolarconduction mechanism, we have to treat the complicated conduction discretely.

References

1. H. Shirakawa, T. Ito, and S. Ikeda, Polym. J. 2, 231 (1971).2. M. Hirooka and T. Doi, Synthetic Metals 17 372 (1987).3. C.W. Tang and S.A. Vanslyke, Appl. Phys. Lett. 51 913 (1987).4. T. Mori, K. Imaizumi, K. Yamashita, T. Mizutani, and H. Miyazaki, Synthetic Metals 111–112 79

(2000).5. T. Mori, K. Obata, and T. Mizutani, J. Phys. D 32 1198 (1999).6. T. Kato, T. Mori, and T. Mizutani, Thin Solid Films 393 109 (2001).7. 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).8. E.g., J. Kalinowski, M. Cocchi, V. Fattori, and P.D. Marco, J. Phys. D 34 2274 (2001).9. M. Matsuura, T. Akai, M. Saito, and T. Kimura, J. Appl. Phys. 79 264 (1996).

10. S. Naka, H. Okada, and H. Onnagawa, Synthetic Metal 91 129 (1997).11. S.F. Nelson, Y.Y. Lin, D.J. Gundlach, and T.N. Jackson, Appl. Phys. Lett. 72 1854 (1998).12. 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].13. T. Mori, T. Ogawa, D.C. Cho, and T. Mizutani, The 12th Int. Conf. on Solid Films and Surface,

Marseille, France, 2002.14. P.W.M. Blom, M.J.M. de Jong, and J.J.M. Vleggaar, Appl. Phys. Lett. 68 3308 (1996).15. S. Karg, M. Meier, and W. Riess, J. Appl. Phys. 82 1951 (1997).16. A.J. Cambell, D.D.C. Bradley, and D.G. Lidzey, J. Appl. Phys. 82 6326 (1997).17. J.C. deMello, N. Tessler, S.C. Graham, and R.H. Friend, Phys. Rev. B 57 12951 (1998).18. U. Wolf, S. Barth, and H. Bässler, Appl. Phys. Lett. 75 2035 (1999).19. T. Mori, E. Sugimura, and T. Mizutani, J. Phys. D: Appl. Phys. 26 452 (1993).20. K. Imaizumi, K. Kaneko, T. Mori, and T. Mizutani, 1999 Autum Meeting, Materials Research

Society, Boston, USA (1999).

Page 172: Nanotechnology and Nano-Interface Controlled Electronic Devices

Carrier transport behavior in OLED 155

21. K. Imaizumi, K. Kaneko, T. Mori, and T. Mizutani, Int. Symp. on Oranic Molecular Electronics Proc.Nagoya, Japan, 2000.

22. K. Imaizumi, K. Kaneko, and T. Mori, T. Mizutani, The 10th Int. Work. on Inorganic & Organic.Electroluminescence, Hamamatsu, Japan, 2000.

23. K. Imaizumi, K. Kaneko, T. Mori, and T. Mizutani, Trans. IEE Jpn. 121-A 332 (2001) [in Japanese].24. K. Imaizumi, K. Kaneko, T. Mori, and T. Mizutani, Trans. IEE Jpn. 121-A 666 (2001) [in Japanese].25. K. Imaizumi, K. Kaneko, T. Mori, and T. Mizutani, Jpn. J. Appl. Phys. 41 366 (2002).26. T. Ogawa, D.-C. Cho, K. Kaneko, T. Mori, and T. Mizutani, IEICE Trans. Electron. E85-C, 1239

(2002).27. D.V. Khramtchenkov, H. Bässler, and V.I. Arkhipov, J. Appl. Phys. 79 9283 (1996).28. P.S. Davids, I.H. Campbell, and D.L. Smith, J. Appl. Phys. 82 6319 (1997).29. B.K. Crone, P.S. Davids, I.H. Campbell, and D.L. Smith, J. Appl. Phys. 84 833 (1998).30. Y. Kawabe, M.M. Morrell, G.E. Jabbour, S.E. Shaheen, B. Kippelen, and N. Peyghambarian, J. Appl.

Phys. 84 5306 (1998).31. G.G. Malliaras and J. Scott, J. Appl. Phys. 85 7426 (1999).32. J. Staudigel, M. Stöel, F. Steuber, and J. Simmerer, J. Appl. Phys. 86 3895 (1999).33. 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).34. B.K. Crone, P.S. Davids, I.H. Campbell, and D.L. Smith, J. Appl. Phys. 87 1974 (2000).35. E. Tutis, M.N. Bussac, B. Masenelli, M. Carrad, and L. Zuppiroli, J. Appl. Phys. 89 430 (2001).36. T. Tsutsui, C.P. Lin, and S. Saito, Mol. Cryst. Liq. Cryst. 256 63 (1994).37. M. Iwamoto and T. Hino, Trans. IEE Jpn. A 100 291 (1980) [in Japanese].

Page 173: Nanotechnology and Nano-Interface Controlled Electronic Devices

This Page Intentionally Left Blank

Page 174: Nanotechnology and Nano-Interface Controlled Electronic Devices

Nanotechnology and Nano-Interface Controlled Electronic DevicesEditors: M. Iwamoto, K. Kaneto and S. Mashiko© 2003 Elsevier Science B.V. All rights reserved

CHAPTER 10

Electrical characterization of organicsemiconductor films by in situ field-effect

measurements

Kazuhiro Kudo

Department of Electronics and Mechanical Engineering, Faculty of Engineering, ChibaUniversity, 1-33 Yayoi-cho, Inage-ku, Chiba 263-8522, Japan

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1572. Experimental details of in situ field-effect measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . 1583. Field-effect characteristics of organic films . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160

3.1. Merocyanine evaporated films . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1603.2. Phthalocyanine evaporated films . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1623.3. Hole transporting materials for EL devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1683.4. Perylene evaporated films . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173

4. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179

1. Introduction

Organic semiconductors have recently received increasing interest because of theirpotential applications in low-cost and large-area devices such as organic light emittingdiodes (LEDs), organic field-effect transistors (FETs), and optoelectronic integratedcircuits (OEIC). High carrier mobilities comparable to amorphous silicon have beenobtained for several organic transistor materials. In particular, field-effect mobilitiesin the range from 1 to 5 cm2/V s and on/off current ratios larger than 108 werereported in single crystal organic FETs [1–3]. Although many kinds of p-type organicsemiconductors have been reported, there are few examples of n-type behavior. Forpractical device applications, both p-type and n-type semiconducting materials withhigh stability against air are highly desirable. From these points of view, it is importantto investigate the intrinsic electrical properties of organic semiconductor films beforeand after exposing in atmospheric gasses, especially oxygen gas. In situ field-effectmeasurement is a promising method for the evaluation of conduction type (p or

Page 175: Nanotechnology and Nano-Interface Controlled Electronic Devices

158 K. Kudo

n), carrier mobility (μ), electrical conductivity (σ ), and carrier concentration (N ) ofevaporated films [4,5].

Phthalocyanine (Pc) and merocyanine (MC) derivatives have potential for applicationto 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 extendfrom the ultraviolet to the infrared region. In particular, the coordinated metals in Pcare important factors for the photoelectrical properties. It is necessary to investigatethe relationship between their chemical structure and intrinsic electrical properties,especially without the influence of atmospheric gasses and impurities. The in situ field-effect measurement [4,5] is a promising method to evaluate the electrical parameters oforganic thin films.

In the present work, we have carried out the in situ field-effect measurement of severalkinds of organic semiconductors expected for optoelectronic devices, and estimated theintrinsic electrical parameters, such as carrier mobility (μ), carrier concentration (N )and electrical conductivity (σ ), and excluded the influence of atmospheric gasses andimpurities with in situ field-effect measurements. The organic films were fabricated bya standard vacuum evaporation technique and the FET characteristics were investigatedbefore 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 structureare shown in Figs. 1 and 2. The highly doped Si substrate which works as a gateelectrode was covered with thermally grown SiO2 with a thickness of approximately200 nm. The interdigital source and drain electrodes were formed on the substrate usingstandard vacuum evaporation and photolithographic techniques. The metal materials ofthe source and drain electrodes were chosen to make an ohmic contact to the organic

Fig. 1: Schematic of the in situ field-effect measurement.

Page 176: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 lengthand width were 0.1 and 56 mm, respectively. Substrates were thoroughly cleaned withorganic solvent in an ultrasonic bath and prebaked at 373 K over 30 min in thevacuum chamber. Subsequently, organic semiconductor materials were evaporated asthe 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-effectmeasurements 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 sampleat 373 K for 1 h in vacuum (10−5 Torr). All of the electrical measurements were carriedout 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 standardTFT is given by [10]

IDS = −(W/L)COX[(VG − Vth)VDS − (1/2)V 2DS], (1)

where W is the channel width, L is the channel length, μ is the carrier mobility, COX isthe capacitance of the SiO2 layer, and Vth is the threshold voltage. When IDS does notincrease 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 electricalconductivity (σ ) was obtained from the slope of the IDS–VDS plot at VG = 0. σ is alsoexpressed by

σ = q Nμ, (3)

where q is the elementary electric charge and N is the carrier concentration. N wasobtained from Eq. 3.

Page 177: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 andtheir 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, carrierconcentration, and electrical conductivity by field-effect measurements and discussedthe transport mechanism of MC film from the temperature dependence of the FETcharacteristics.

Fig. 3 shows the molecular structures of MC derivatives examined here. Typicalcharacteristics 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 resultindicates that negative gate voltages form a hole accumulation layer and MC films showp-type semiconducting properties without influences of impurities and atmosphericgasses. From this procedure, the following electrical parameters of MC(a) film wereestimated: the field-effect mobility was 1.4 × 10−6 cm2/V s, the electrical conductivitywas 2.6×10−10 S/cm, and the carrier concentration was 1.3×1015 cm−3. The estimatedvalue of μ depends on the thickness of the dye films. Fig. 5 shows μ variation as afunction 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 theeffects of interface traps between SiO2 and organic film, and the carrier conduction ofthe discontinuous parts in thin films. Both effects appear strongly in thinner films andtend to be inconspicuous in thicker films. From this result, we have chosen as standardthickness 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.

Page 178: Nanotechnology and Nano-Interface Controlled Electronic Devices

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) 1.5×10−5 1.0(b) 5.0×10−6 0.3(c) 1.0×10−7 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 theilluminated monochromatic photon density and electrons produced as a short-circuit

Page 179: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 areclosely related to the field-effect mobilities.

The temperature dependence of electrical parameters obtained by in situ field-effectmeasurement is shown in Fig. 6. From these Arrhenius plots, the activation energies ofconductivity and field-effect mobility in MC film were estimated to be 0.44 eV, and theactivation energy of carrier concentration was very low (0.04 eV). These results indicatethat the generation of thermally activated carriers from the shallow acceptor level isvery 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 thermallyactivated, 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 aretrapped 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 previoussection represents the number of hopping carriers.

3.2. Phthalocyanine evaporated films

Phthalocyanine (Pc) films have been expected as gas sensors [6,7] and the coordinatedmetals in Pc are important factors for their electrical properties. It is necessary toinvestigate the relationship between their chemical structure and intrinsic electricalproperties, especially without the influence of atmospheric gasses and impurities. AsPc derivatives, copper-phthalocyanine (CuPc), lead-phthalocyanine (PbPc), metal-freephthalocyanine (H2Pc), and fluoro-phthalocyanine (F16CuPc) were examined here, andn-type behavior of F16CuPc was reported [13,14]. Fig. 7 shows the molecular structuresof 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 afterthe deposition at Tsub of 373 K are shown in Fig. 8(a). IDS increases with negative

Page 180: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 thatnegative gate voltages enlarge the conduction channel due to the formation of a holeaccumulation layer, thus, CuPc films show p-type semiconducting properties. We havealso investigated the effect of oxygen gas and annealing on the electrical properties ofthe films. Fig. 8(b) and (c) shows FET characteristics of the CuPc sample after theoxygen gas exposure and after the thermal treatment in vacuum, respectively. Althoughthe IDS increases after the oxygen gas exposure, the IDS decreases by the thermaltreatment at 373 K in vacuum for 1 h. It was also confirmed that H2Pc and PbPc showedp-type semiconducting properties in the absence of atmospheric gasses. However, the

Page 181: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 PbPcfilm is significant but that on H2Pc films is small.

On the other hand, typical FET characteristics of F16CuPc films are shown in Fig. 9.Electrical properties of F16CuPc films are rather stable against the oxygen gas exposure.The experimental results in Table 2 and Fig. 9 indicate that the influence of oxygen gasexposure is small and the fluorine atoms in F16CuPc molecules seem to prevent oxygenadsorption.

The carrier mobility, μ, conductivity, σ , and carrier concentration, N obtained by thein situ field-effect measurements are shown in Table 2. Fig. 10 shows the variations ofN 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 exposureto oxygen gas. Particularly in PbPc films, N increased when the sample was exposedto oxygen gas and decreased when the sample was annealed in vacuum. The effectof oxygen on N is marked in PbPc films compared with those in H2Pc and F16CuPcfilms. The effect of oxygen on CuPc film is between those of the PbPc and H2Pc films.Thus, oxygen gas acts as an acceptor impurity and increases the net charge carriers. It isconsidered that electrons transfer from phthalocyanine molecules to oxygen molecules

Page 182: Nanotechnology and Nano-Interface Controlled Electronic Devices

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).

Page 183: Nanotechnology and Nano-Interface Controlled Electronic Devices

166 K. Kudo

Fig. 9: Typical FET characteristics of F16CuPc films, (a) as-grown, (b) oxygen exposure (5 h), (c) annealingat 373 K in vacuum (1 h).

Page 184: Nanotechnology and Nano-Interface Controlled Electronic Devices

Electrical characterization of organic semiconductor films 167

Table 2

Electrical parameters of phthalocyanine derivatives obtained by in situ field-effect measurements

Material Tsub (K) As-grown O2 gas exposure Annealing in vacuum

Hole mobility (cm2/V s)PbPc 373 3.8×10−6 5.7×10−5 3.2×10−5

CuPc 373 2.4×10−6 1.2×10−5 1.1×10−5

H2Pc 373 2.0×10−6 2.1×10−6 3.6×10−6

F16CuPc 373 1.6×10−3 1.0×10−3 1.8×10−3

Conductivity (S/cm)PbPc 373 1.3×10−9 7.7×10−7 6.6×10−8

CuPc 373 2.1×10−10 4.6×10−10 1.3×10−9

H2Pc 373 1.1×10−10 1.4×10−10 7.5×10−11

F16CuPc 373 5.6×10−5 2.1×10−5 4.4×10−5

Carrier concentration (cm−3)PbPc 373 2.1×1015 8.4×1016 1.3×1016

CuPc 373 5.4×1014 2.4×1015 8.2×1014

H2Pc 373 3.4×1014 4.1×1014 1.3×1014

F16CuPc 373 2.2×1017 1.2×1017 1.5×1017

and oxygen molecules are directly related to the composition of the central metal ofphthalocyanine molecules. This result is closely related to reports that the interactionbetween H2Pc and oxygen is weak [15] or that the adsorption site of H2Pc is differentfrom that of metal phthalocyanines [16].

F16CuPc has the largest mobility and H2Pc the smallest, indicating that the carriermobility is dependent on the coordinated metal and fluorine atoms. These seem to beintrinsic properties of the materials measured, since the effect of oxygen was excludedduring measurements. It should be noted that the deposition at high temperature gavehigh μ for all Pc films.

Evaluation of the thermal activation energy (Ea) is important to discussion ofthe carrier transport mechanism in organic semiconducting films. In particular, thetemperature dependence of σ , μ, and N is necessary, because Ea estimated by σ

contains both μ and N , as expressed in Eq. 3. Fig. 11 shows Ea(σ ), Ea(μ), and Ea(N )estimated by Arrhenius plots between 295 and 380 K for CuPc film. Plots of logσ ,logμ, log N against 1/T follow a straight line. The carrier mobility of CuPc film has anactivation energy of about 0.26 eV. Ea(μ) obtained by the field-effect measurement canbe interpreted in terms of a barrier which prevents the carrier transport at the interfacebetween the micrograins or between the organic film and the electrodes. The effect ofoxygen on Ea(σ ) and Ea(N ) depends on the chemical structure of the Pc molecules.Although Ea of F16CuPc and H2Pc remain almost the same values, those of PbPc andCuPc change to lower values after the oxygen exposure. In particular, Ea(σ ) of PbPcfilm varied from 0.53 to 0.23 eV by introducing oxygen gas. After thermal treatmentin vacuum, Ea(σ ) recovers to almost the same value of 0.54 eV. These results indicatethat oxygen molecules interact with the PbPc molecules and a shallow acceptor level isformed near the valence band edge or HOMO (highest occupied molecular orbital) level

Page 185: Nanotechnology and Nano-Interface Controlled Electronic Devices

168 K. Kudo

Fig. 10: The variations of N and μ of as-grown sample, after oxygen gas exposure and after annealing invacuum.

of the PbPc. Thus, oxygen gas acts as an acceptor impurity and increases the net chargecarriers.

3.3. Hole transporting materials for EL devices

Organic electroluminescent (EL) devices have already been put to practical use fordisplay panels. However, basic electrical parameters of organic materials used in ELdevices have been hardly reported. Thin-film transistors (TFTs) using hole trans-porting materials of organic EL devices were fabricated and the electrical param-eters were estimated by in situ field-effect measurements. Fig. 12 shows molec-ular structures of TPD (N ,N ′-diphenyl-N ,N ′-di(3-methylphenyl)-1,1′-biphenyl-4,4′-

Page 186: Nanotechnology and Nano-Interface Controlled Electronic Devices

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), andm-MTDATA (4,4′,4′′-tris-(3-methylphenyl-phenylamino)triphenylamine). The growthtemperatures (Tsub) of these materials were selected below the glass transition temper-ature (Tg). The glass transition temperatures of TPD, a-NPD, and m-MTDATA are 63,96, and 75°C, respectively.

Page 187: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 depositionare shown in Figs. 13(a), (b), and (c), respectively. For all the materials, IDS increaseswith increasing negative VG. These results indicate that a hole accumulation layer isformed by negative gate voltages and evaporation films of hole transport materialshave p-type semiconducting properties. From these results, the estimated field-effectcarrier mobilities of TPD, a-NPD, and m-MTDATA were 1.6 × 10−6, 9.3 × 10−8, and6.3 × 10−7 cm2/V s, respectively (Table 3). These values of μ are smaller than thoseobtained by the time-of-flight method [17]. The difference in the μ values using the FET

Page 188: Nanotechnology and Nano-Interface Controlled Electronic Devices

Electrical characterization of organic semiconductor films 171

Fig. 13: Typical FET characteristics of TPD, a-NPD and m-MTDATA just after deposition.

Page 189: Nanotechnology and Nano-Interface Controlled Electronic Devices

172 K. Kudo

Table 3

Electrical parameters of hole transporting materials obtained by in situ field-effect measurements

Material Tsub (K) As-grown O2 gas exposure Annealing in vacuum

Hole mobility (cm2/V s)TPD 333 1.6×10−6 1.3×10−6 2.3×10−6

α-NPD 353 9.3×10−8 2.3×10−7 3.2×10−7

m-MTDATA 343 6.3×10−7 6.0×10−7 6.2×10−7

Conductivity (S/cm)TPD 333 1.3×10−11 8.1×10−12 7.6×10−12

α-NPD 353 1.1×10−12 9.4×10−13 6.5×10−12

m-MTDATA 343 1.7×10−10 2.4×10−10 1.3×10−10

Carrier concentration (cm−3)TPD 333 2.2×1013 2.6×1013 1.1×1013

α-NPD 353 7.6×1013 3.6×1013 1.3×1014

m-MTDATA 343 1.7×1015 2.5×1015 1.3×1015

and time-of-flight method is probably due to the difference in the electric field, currentdirection though the organic films, etc.

Fig. 14 shows the variations of N and μ of as-grown sample, after oxygen gasexposure and after annealing in vacuum. As shown in Fig. 14(a) and (b), these holetransporting materials show stable against the oxygen exposure. Fig. 15 shows theelectrical properties of TPD, a-NPD, and m-MTDATA films as a function of substratetemperature (Tsub) during the deposition. The conductivity of m-MTDATA is higher thanthat of the other materials. The conductivities of a-NPD and m-MTDATA have weakdependence on Tsub below Tg. However, the conductivity of TPD has a peak at around55°C. The conductivities of all the materials show a drastic change at Tsub over Tg. Thecarrier mobilities of these materials show clearer Tsub dependence. The carrier mobilityof a-NPD and m-MTDATA changes to a low value under Tg, the carrier mobility ofm-MTDATA is one order larger than that of a-NPD. On the other hand, the carriermobility of TPD has a sharp peak at Tsub around 55°C.

Similar phenomena were observed by the thermal treatment in air after the filmdeposition. The electrical properties of all the materials show a drastic change afterthe thermal treatment over Tg. Fig. 16 shows AFM (atomic force microscope) imagesof TPD films after thermal treatment below and above Tg. Though the surface of theTPD 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 thatof an island-like film (Fig. 16(b)). The steepness of the change in electrical propertiesat a higher temperature is mainly due to the discontinuity of the conductive channelcaused by the formation of island-like grains after the thermal treatment (Fig. 17). Theseresults indicate that the degradation of organic EL devices at higher temperatures isclosely related to the formation of island-like grains in hole transport layers at highertemperature over Tg.

Page 190: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 invacuum.

3.4. Perylene evaporated films

Three kinds of perylene derivatives (PTCDI, PTCDA, BPPC) were used and n-typecharacteristics 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 conductionof all perylene derivatives increases with positive VG and the FETs operate in anenhancement mode. From the results of field-effect measurements, as deposited films of

Page 191: Nanotechnology and Nano-Interface Controlled Electronic Devices

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.

Page 192: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 absenceof atmospheric gasses. The introducing of oxygen gas to the n-type films works todecrease the conductivity and the degree of change depends on the molecular species.FET characteristics of PTCDI film obtained just after the deposition were completely

Page 193: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 valuesas those of as-grown samples.

The hole mobility, μ, conductivity, σ , and carrier concentration, N , obtained bythe in situ field-effect measurements are shown in Table 4. The effect of oxygen gasexposure on perylene films is in a marked contrast to that of p-type materials. Thevariations 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 gasacts as an acceptor impurity and produces good results for p-type materials, but oxygen

Page 194: Nanotechnology and Nano-Interface Controlled Electronic Devices

Electrical characterization of organic semiconductor films 177

Fig. 19: Typical FET characteristics of PTCDI films.

Page 195: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 beexplained as follows: oxygen gas acts as an acceptor impurity and carrier compensationoccurs in n-type films. The variation of the electrical properties, however, depended onthe molecular structure and the growth condition of the films.

4. Conclusions

The basic electric parameters of several kinds of organic thin films were evaluatedby in situ field-effect measurement. The effects of thermal treatment and introducingoxygen gas on the electrical properties were also investigated. A marked change inthe electrical parameters corresponding to the adsorption and desorption of oxygenmolecules was observed and the variation of the electrical properties strongly depended

Page 196: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 2.9×10−5 < 10−7 3.3×10−6

PTCDA 373 4.4×10−6 2.4×10−7 1.4×10−6

BPPC 373 2.4×10−7 < 10−7 1.1×10−8

Conductivity (S/cm)PTCDI 373 2.1×10−7 < 10−12 6.6×10−10

PTCDA 373 6.1×10−7 8.2×10−10 7.4×10−9

BPPC 373 3.5×10−11 < 10−12 1.4×10−12

Carrier concentration (cm−3)PTCDI 373 4.5×1016 < 1014 1.2×1015

PTCDA 373 8.7×1017 2.1×1016 3.4×1016

BPPC 373 9.0×1014 < 1014 8.2×1014

on the molecular structure and the growth conditions of the films. These resultsdemonstrate that the influence of atmospheric gases is significant for organic deviceapplications and the in situ field-effect measurement is a powerful method to investigatethe fundamental properties of organic materials.

References

1. C.D. Dimitrakopoulos and P.R.L. Malenfant, Adv. Mater. 14, 99 (2002).2. H. Klauk, D.J. Grundlach, J.A. Nichols, C.D. Sheraw, M. Bonse, and T.N. Jackson, Solid State Tech.

43, 63 (2000).3. C.R. Kagan, D.B. Mitzi, and C.D. Dimitrakopoulos, Science 286, 945 (1999).4. K. Kudo, M. Yamashina, and T. Moriizumi, Jpn. J. Appl. Phys. 23, 130 (1984).5. K. Kudo, T. Sumimoto, K. Hiraga, S. Kuniyoshi, and K. Tanaka, Jpn. J. Appl. Phys. 36, 6994 (1997).6. C. Hamann, A. Mrwa, M. Muller, W. Gopel, and M. Rager, Sens. Actuat. B 4, 73 (1991).7. A. Wilson, J.D. Wright, and A.V. Chadwick, Sens. Actuat. B 4, 499 (1991).8. A.K. Ghosh, D.L. Morel, T. Feng, R.F. Shaw, and C.A. Rowe Jr., J. Appl. Phys. 45, 230 (1974).9. K. Kudo, T. Shinohara, T. Moriizumi, K. Iriyama, and M. Sugi, Jpn. J. Appl. Phys., Suppl. 20-2, 135

(1981).10. S.M. Sze, Physics of Semiconductor Devices (Wiley, New York, 1969) p. 425.11. K. Kaneto, K. Yamanaka, K. Rikitake, T. Akiyama, and W. Takashima, Jpn. J. Appl. Phys. 35, 1802

(1996).12. H. Bassler, G. Schonherr, M. Abkowitz, and D.M. Pai, Phys. Rev. B 26, 3105 (1982).13. Z. Bao, A.J. Lovinger, and A.Dodabalapur, Adv. Mater. 9, 42 (1997).14. Z. Bao, A.J. Lovinger, and J. Brown, J. Am. Chem. Soc. 120, 207 (1998).15. J. Simon and J.J. Andre, Molecular Semiconductors (Springer-Verlag, Berlin, 1985) p. 116.16. J.P. Contour, P. Lenfant, and A.K. Vijh, J. Catal. 29, 8 (1973).17. S. Naka, H. Okada, H. Onnagawa, Y. Yamaguchi, and T. Tsutsui, Synth. Met. 111–112, 331 (2000).18. G. Horowitz, F. Kouki, P. Spearman, D. Fichou, C. Nogues, X. Pan, and F. Garnier, Adv. Mater. 8,

242 (1996).19. A. Brown, D.M. de Leeuw, E.J. Lous, and E.E. Havinga, Synth.Met. 66, 257 (1994).20. T. Suga, M. Iizuka, S. Kuniyoshi, K. Kudo, and K. Tanaka, Synth. Met. 102, 1050 (1999).

Page 197: Nanotechnology and Nano-Interface Controlled Electronic Devices

This Page Intentionally Left Blank

Page 198: Nanotechnology and Nano-Interface Controlled Electronic Devices

Part C

Smart Soft Materials

Page 199: Nanotechnology and Nano-Interface Controlled Electronic Devices

This Page Intentionally Left Blank

Page 200: Nanotechnology and Nano-Interface Controlled Electronic Devices

Nanotechnology and Nano-Interface Controlled Electronic DevicesEditors: M. Iwamoto, K. Kaneto and S. Mashiko© 2003 Published by Elsevier Science B.V.

CHAPTER 11

Introducing ruber into theLangmuir–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, Germanyb Max-Planck-Institut für Kolloid- & Grenzflächenforschung, Berlin, Germany

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1832. Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1843. Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189

1. Introduction

The Langmuir–Blodgett (LB) technique offers the opportunity to generate suspendedmembranes by assembling a monolayer at the air–water interface and transferring it tocover a hole in a solid substrate. However, the preparation of suspended membranes viaLB-transfer is generally more difficult than LB-transfer of thin organic coatings ontocontinuous smooth surfaces: Because it is not supported by an underlying substrate,the suspended membrane itself must be tough enough to withstand mechanical stressduring fabrication and final use. Monolayers that are made from low molecular weightcompounds or from liquid polymers easily rupture during transfer across a hole.Suspended membranes have been fabricated using glassy polymers, often stabilised bycross-linking [1,2]. These membranes are usually rigid. For certain applications, likemembranes in micro mechanical valves and pumps, it might be advantageous to haveelastomeric thin membranes available and to take advantage of the comparatively largereversible deformation of these materials. Here, we show that tough and mechanicallystable freely suspended membranes – spanning millimetre sized holes in solid substrates– can be obtained from cross-linked monolayers of low Tg polymers with ionic headgroups.

Page 201: Nanotechnology and Nano-Interface Controlled Electronic Devices

184 H. Xu et al.

2. Theory

When applied to a water surface, liquid polymers like perfluoropolyethers [3], polyiso-prene [4], polybutadiene [5] or polydimethylsiloxane [6] with ionic head groups easilyform smooth and continuous monolayers. By variation of the polymer chain length andsurface concentration, the thickness of these monolayers can easily be tuned in the rangeof 10 to 100 nm thickness [7]. If these monolayers are transferred to substrates withopenings (e.g. an electron microscopy grid) as schematically depicted in Fig. 1 theyinitially cover the openings as approximately 50 nm thin bilayers. However, these mem-branes rupture within minutes after transfer. As an example two images of a membranemade via LB-transfer of polyisobutene with a single head group and a chain length of300 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 asoapy membrane made out of a water core that is coated from both sides with a liquidlayer of amphiphiles. Such membranes usually are metastable and rupture, especiallyif 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 shortlyafter 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 frommonolayers of poly-4-n-butylstyrene with trimethylammoniumbromide head groups.Polybutylstyrene has a glass transition temperature of 25°C. Hence, at elevated tem-peratures polybutylstyrenes with ionic head groups, applied to a water surface, behaveessentially like polyisoprenes at room temperature and form smooth and continuous

monolayer

solidsupport

Water

transfer to grid

Fig. 1: Scheme of the formation of a freely suspended membrane via Langmuir–Blodgett transfer of ananchored polymer monolayer to substrates with openings.

Page 202: Nanotechnology and Nano-Interface Controlled Electronic Devices

Introducing ruber into the Langmuir–Blodgett technique 185

Fig. 2: A freely suspended membrane generated by Langmuir–Blodgett transfer of a monolayer ofpolyisobutene with a single ionic head group ruptures within 30 min. (Light microscopy image with top andbottom illumination.)

Fig. 3: Light microscopy image of an approximately 20 nm thick freely suspended membrane ofpolybutylstyrene-N+ prepared by spreading at 40°C, cooling to 10°C and transfer to an electron microscopygrid.

monolayers. Upon cooling, the polybutylstyrene monolayers vitrify to room temperatureand thus can be transferred to yield solid freely suspended membranes [8] (Fig. 3).

Freely suspended membranes stabilised by chemical cross-linking have been obtainedby irradiation of monolayers of polyisoprenes with ionic head groups and anthraceneside 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-linkingpoints. Since the polyisoprene chains have a low glass transition temperature, thiscross-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 resultingfreely suspended membranes are long-term stable (at least several months).

Page 203: Nanotechnology and Nano-Interface Controlled Electronic Devices

186 H. Xu et al.

Fig. 4: Scheme of the preparation of elastomeric membranes via cross-linking of the side chains ofanthracene tagged polyisoprene with ionic head groups.

Fig. 5: Top and side view of an elastomeric cross-linked polyisoprene membrane reversibly deformed by asmall overpressure from below.

The elastomeric properties of these freely suspended membranes can be shown byapplying a small pressure from one side. When the pressure is applied, the membranebulges. When the pressure is released, the membrane flattens itself reversibly (seeFig. 5). In this procedure the monolayer is cross-linked and converted into a solidlayer on the water surface; after cross-linking it can sustain neither shear flow norextensional flow. When a monolayer is transferred to a substrate, which is smaller thanthe Langmuir trough, it has to undergo two-dimensional flow. Otherwise it will developstress or will wrinkle, especially if several substrates are coated consecutively. Thus,

Page 204: Nanotechnology and Nano-Interface Controlled Electronic Devices

Introducing ruber into the Langmuir–Blodgett technique 187

Fig. 6: Scheme of physical cross-linking of a suspended membrane of polymer chains with multiple headgroups. Upon drying of the water core, which is initially present in the transferred membrane, the headgroups aggregate and form physical cross linking sites.

Fig. 7: Top view of an elastomeric polyisobutene membrane cross-linked via aggregation of multiple ionichead groups. (a) Chemical formula of the polyisobutene. (b) Membrane spanning a 300 μm hole in a brassplate. (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 themonolayer had to be discarded.

The two-dimensional-flow problem can be avoided by transferring liquid monolayersand cross-linking them shortly after the transfer. This can be accomplished quite easilyby 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 thelinear polymers depicted in Fig. 2, the monolayers of the three-arm-star polymers canbe transferred to cover holes in solid substrates. In the case of the star polymers withseveral ionic head groups, however, the ionic groups form inverted micelles when themembrane dries. These inverted micelles efficiently cross-link the polymer and thus giverise 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 cross-linked membranes are elastomeric. In Fig. 7(b) to (d) a continuously increasing pressure

Page 205: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 monolayerobtained after removal of the colloids.

of approximately 10–100 Pa is applied to a freely suspended membrane. The higherthe applied pressure, the more does the membrane bulge upward. Upon release of thepressure this deformation is completely reversible and can be repeated multiple times.

Cospreading polymers with ionic anchor groups and hydrophobised silica colloidson a water surface, followed by transfer to solid substrates of electron microscopygrids gives rise to mixed monolayers. In these monolayers, domains of silica particlesare embedded in a continuous matrix of polymeric monolayers (see Figs. 8 and 9).o Fig. 10. Inserted

s. 8 and 9. Pleaseck.

Exposure of these membranes to hydrofluoric acid vapour removes the silica particles.This gives rise to porous monolayers and porous membranes of controlled porosity witha uniform pore size distribution [11]. These membranes are promising for applicationslike ultrafiltration, bio-encapsulation and as masks and moulds for generating newnanoscopic and mesoscopic structures and surface patterns.

3. Experimental

Linear polyisoprene with a sulfonate head group and anthracene side groups hasbeen synthesised via living anionic polymerisation followed by platinum catalysedhydrosilylation of the sulfonate terminated parent polyisoprene as published in [4]and [9]; polybutylstyrene with ammonium head group has been synthesised via living

Page 206: Nanotechnology and Nano-Interface Controlled Electronic Devices

Introducing ruber into the Langmuir–Blodgett technique 189

anionic polymerisation as described in [8]; polyisobutenes with sulfonate head groupshave been synthesised via living cationic polymerisation as described in [12–14].Monolayers on a water surface were prepared using a 20 cm × 46 cm rectangularLangmuir trough made of polytetrafluoroethylene, equipped with one compressionbarrier and a floating barrier for the detection of the surface pressure via the Langmuirmethod (Lauda FW2, Germany). The polymers were usually spread from chloroformsolutions 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 (1/50by weight) (polyisobutenes). UV-illumination was made through the thermostatted,transparent lid of the trough using an array of four 30 cm long fluorescence lampsmounted 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 atλmax = 355 nm. 30 min before and during illumination, the air space above the air–waterinterface was flushed with nitrogen (5 l/min). Silica colloids coated with polyisobuteneamphiphiles (mean radius = 70 nm, polydispersity = 11%, suspended in cyclohexane)were obtained from Utrecht Colloid Synthesis Facility, Van ’t Hoff Laboratory forPhysical and Colloid Chemistry, Utrecht University, The Netherlands.

Acknowledgements

This work was partially conducted in the Max-Planck-Institute of Colloids and Interfacesand 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) forintroducing us into living anionic polymerisation and M. Grasmüller and O. Nuyken(TU-Munich) for introducing us to living cationic polymerisation. We thank Dr. Carlosvan Kats, and Dr. Judith Wijnhoven, A. van Blaadern, A. Phillipse (Utrecht ColloidSynthesis Facility) for providing the silica colloids and for helpful discussions.

References

1. M. Seufert, C. Fakirov, and G. Wegner, Adv. Mater. 7, 52 (1995).2. M. Kunitake, T. Nishi, H. Yamamoto, K. Nasu, O. Manabr, and N. Nakashima, Langmuir 10, 3207

(1994).3. W.A. Goedel, C. Xu, and C.W. Frank, Langmuir 9, 1184 (1993).4. R. Heger and W.A. Goedel, Macromolecules 29, 8912 (1996).5. P. Christie, M.C. Petty, and G.G. Roberts, Thin Solid Films 134, 75 (1985).6. T.J. Lenk, D.H.T. Lee, and J.T. Koberstein, Langmuir 10, 1857 (1994).7. H. Baltes, M. Schwendler, C.A. Helm, R. Heger, and W.A. Goedel, Macromolecules 30, 6633 (1997).8. W.A. Goedel, C. Peyratout, L. Ouali, and V. Schädler, Adv. Mater. 11, 213 (1999).9. W.A. Goedel and R. Heger, Langmuir 14, 3470 (1998).

10. F. Mallwitz and W.A. Goedel, Angew. Chemie Int. Ed. 40, 2557 (2001); Angew. Chemie 113, 2716(2001).

11. Hui Xu and W.A. Goedel, Langmuir 18, 2363 (2002).12. R. Santos, J.P. Kennedy, and M. Walters, Polymer Bull. 11, 261 (1984).

Page 207: Nanotechnology and Nano-Interface Controlled Electronic Devices

190 H. Xu et al.

13. 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).

14. R.F. Storey and Y. Lee, J. Polym. Sci. Polym. Chem. 29, 317 (1991).

Page 208: Nanotechnology and Nano-Interface Controlled Electronic Devices

Nanotechnology and Nano-Interface Controlled Electronic DevicesEditors: M. Iwamoto, K. Kaneto and S. Mashiko© 2003 Elsevier Science B.V. All rights reserved

CHAPTER 12

Design of functional interface betweenliving 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. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1922. Engineering of semiconductor surfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193

2.1. Grafting of alkylsiloxane monolayers on ITO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1932.2. Electrochemical passivation of bulk GaAs electrodes with thiol monolayers . 1942.3. Passivation of semiconductor nano-structures close to surface . . . . . . . . . . . . . . 1972.4. Outlooks – engineering of GaAs with functional 4,4′-mercaptobiphenyls . . . 199

3. “Soft cushions” at interface – biocompatible polymer films as physical models ofextracellular matrices (ECMs) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2013.1. Hydration and wetting of polysaccharide films. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2023.2. Chemical control of wetting interaction – grafting of synthetic polymer

brushes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2053.3. Outlooks – chemical “switching” of cell/surface interactions . . . . . . . . . . . . . . . 207

4. Physical models of cell surface glycocalix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2094.1. Thermodynamics and hydration of glycolipid monolayers . . . . . . . . . . . . . . . . . . 2094.2. Morphology of glycolipids – structural basis of carbohydrate complexes . . . . 2114.3. Viscoelasticity of oligosaccharides on cell membranes – rheology at the

interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2145. Native and model cell membranes on semiconductors – potential candidates for

sensorics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2155.1. Uptake of antibiotic peptides into model membranes on semiconductor

electrodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2185.2. Novel charge sensor based on bio-membrane/semiconductor hybrids . . . . . . . 2215.3. Orientation-selective immobilization of native cell membranes on ultra-thin

polymer films . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228

Page 209: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 aspectsas well as from technological point of view [1] (Fig. 1).

Scientifically, glycolipids, membrane receptors, or proteins of the extracellular matriximmobilized onto semiconductor devices can provide physical models of cell- andtissue surfaces, which allows the investigation of the basic principles of their complexfunctions in nature. With the aid of various surface sensitive techniques, (a) quantitativedetection of protein–protein recognition processes at membranes and (b) fine-tuning ofthe adhesion forces between cells and surfaces by the interplay of specific (lock-and-key) forces and universal interfacial forces are possible.

Technological applications should profit from the possibilities to immobilize variousmembrane proteins under non-denaturing conditions to obtain fast screening of theuptake of drugs/toxins and highly sensitive detection of “local” binding and adsorptionto the surfaces. Especially, model cell membranes reconstituted on bulk and nano-structured semiconductors (e.g. ITO, GaAs-, and Si-based) have a large potentialtowards intelligent sensors based on electro-optical transducers. For example, isolatedsingle quantum dots could be utilized as point-like luminescence sources, while thetransport of the charged species across the bio-membranes can be detected by in-planegate transistors with narrow (width ∼ 100 nm) channels.

To date, such strategies have been impeded upon (1) the electrochemical decom-position 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 proteinsand cell death (apoptosis). Thus, to achieve this goal, fabrication of biocompatible andfunctional inter-layers to bridge “wet and soft” biological matters and “dry and hard”solid surfaces is required.

G GS

DΨshν

Fig. 1: Schematic representation of a bio-membrane/semiconductor hybrid, based on stratified molecularconstructs. Micro- or nano-structured semiconductors can be used as “local” sensors to detect the membrane-mediated processes, e.g. optical sensing of protein functions with near-surface quantum dots, detection ofchanges in surface potentials due to ligand–receptor coupling, monitoring selective material transport acrossthe membrane using in-plane FET and 2 dimensional electron gas (2DEG), etc.

Page 210: Nanotechnology and Nano-Interface Controlled Electronic Devices

Living systems and semiconductor nano-structures 193

This review deals with the step-wise construction of hierarchical molecular assem-blies, starting from surfaces of semiconductor nano-structures up to cell membranes.Required functionalization steps include: (A) engineering of the semiconductor surfaceswith organic molecules; (B) deposition of biocompatible polymer films and controltheir interfacial characteristics; (C) physical modeling of cell surface glycocalix; (D)reconstitution and characterization of native- and model cell membranes. To optimizethe interfacial interaction between soft and stratified layers, systematic combination ofdifferent surface-sensitive methods is necessary. Several examples will be introduced inthe following sections.

2. Engineering of semiconductor surfaces

Self-assembled monolayers (SAMs) are suitable both to render the surface hydrophobicfor the later modification and to form very thin insulating layers that prevent leakcurrents, unspecific adsorption, and the surface decomposition in aqueous electrolytes[2]. To date, the most intensively studied systems along this strategy are SAMsof alkanethiols on gold electrodes, which is mainly due to the chemical stabilityof the interfaces. The investigations of blocking efficiencies of the SAMs againstheterogeneous electron transfer and ion penetration enable to determine the surfacecoverage quantitatively [3–7].

Another class of SAMs widely used is based on alkylsiloxane monolayers on differentoxide surfaces (e.g. SiO2, Al2O3, SnO2, TiO2, etc.) [8,9]. A variety of characterizationtechniques have been applied to characterize the monolayer structures, such as contactangle measurements [10,11], X-ray reflectivity or ellipsometry [12]. In addition, thesemolecules can provide self-assembled monolayers with different functional groups[13].

To date, there have been only a few reports concerning the electrical proper-ties of SAMs on semiconductors. One example is the electrical characterization ofalkyltrichlorosilane monolayers on Si/SiO2, discussed in terms of the suppression ofcharge carrier tunneling [14,15]. Nevertheless, systematic studies under physiologicalconditions (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 polar-izable properties [16,17], maintaining high sensitivity without insulating oxide layers.Furthermore, ITO is transparent to visible light, which enables multiple parametermeasurements by using optical and electrical techniques.

Previously, we reported the deposition of octadecyl trichlorosilane (OTS) via sily-lation with surface oxides [18]. The local defect area ratio, i.e. surface coverage, wasquantitatively evaluated using impedance spectroscopy by measuring changes in thecharge transfer resistance in the presence of redox couples (Fe2+/3+), yielding a defectarea ratio of about 0.9%. However, the electrical properties of OTS monolayers were notobserved in buffered electrolytes. Indeed, OTS is highly reactive and often makes poly-

Page 211: Nanotechnology and Nano-Interface Controlled Electronic Devices

194 M. Tanaka

15

10

5

0

-5

-10

Cur

rent

[μA

]

0.40.20.0-0.2-0.4

Bias potential [V]

Fig. 2: Cyclic voltammogram of ITO electrodes at scan rate 25 mV/s. Solid line: Bare ITO in 10 mM Hepesbuffer (pH = 7.5); broken line: bare ITO in 10 mM Hepes buffer containing 1 mM K3/4Fe(CN)6 (pH = 7.5);dotted line: ITO coated with octyltrimethoxysilane (OTMS) monolayer in redox buffer (pH = 7.5). Thevoltage was applied versus a Ag/AgCl reference electrode.

meric siloxane structures, and decomposition of ITO by a sub-product of the couplingreaction (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. Hydrophobicityand homogeneity of the surfaces are discussed by measuring contact angles betweenthe ITO electrodes and water droplets before and after the monolayer deposition. Elec-trochemical properties of the SAMs (e.g. resistance, capacitance, defect area ratio, ionpenetration) were measured quantitatively by using cyclic voltammetry and impedancespectroscopy (Figs. 2 and 3). The voltammograms proved the polarizable properties ofITO electrodes in buffered electrolyte over a large potential range. In addition, cyclicvoltammetry in the presence of redox couples demonstrated the significant passivationeffect of the SAMs against surface electrochemistry, in spite of the intrinsic roughnessof the ITO electrodes (r.m.s. ∼ 2.5 nm). The impedance spectrum of the SAMs exhibitedan obvious difference compared to that of the pure electrode in the frequency regionbetween 1 kHz and 1 Hz. The measured spectra were analyzed in terms of equivalentcircuit models (Fig. 4) that the monolayers behave as a diffusion barrier for ions in theelectrolyte. The effects of alkyl chain length on the electrochemical properties were alsostudied using a silane with a longer chain (octadecyltrimethoxysilane, ODTMS). Theimpedance measurements in the presence of redox couples yielded a defect area ratio ofas 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 highelectron mobility in nano-structures such as a two-dimensional electron gas (2DEG)

Page 212: Nanotechnology and Nano-Interface Controlled Electronic Devices

Living systems and semiconductor nano-structures 195

104

2

4

6

8105

2

4

6810

6

2

4

abso

lute

impe

danc

e [O

hm]

10-1

100

101

102

103

104

frequency [Hz]

-1.6

-1.4

-1.2

-1.0

-0.8

-0.6

-0.4

-0.2

0.0

phase [rad]

Fig. 3: Absolute impedance and phase shift of the ITO electrodes before (�) and after (�) the deposition ofthe OTMS monolayer in 10 mM Hepes buffer (pH = 7.5). The symbols represent the measured data, whilethe lines are corresponding to the fits using the equivalent circuit models from Fig. 4. The bare ITO surfacecould be analyzed by the equivalent circuit I (solid lines). Fitting of the spectrum after the silanization withthe ideal equivalent circuit II shows clear deviations from the measured data (solid lines). However, byintroducing a Randles circuit for the monolayer (equivalent circuit III in Fig. 4) the fit could be improved(dotted lines).

Equivalent circuits

Hepes buffers Redox buffers

Ro CIF

W

Ro CIF

CS

RPT ZW

α

Ro CIF

RS

CS / CPE

W

Ro

CIF

CSZW

RS

Rct

W

Ro

CIF

Rct 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 con-stant phase element; RPT phase-transfer resistance between SAM and electrolyte; ZW Warburg impedance;Rct charge transfer resistance.

Page 213: Nanotechnology and Nano-Interface Controlled Electronic Devices

196 M. Tanaka

4

6

8104

2

4

6

8105

2

4

6

8106

abso

lute

impe

danc

e [O

hm]

10-1

100

101

102

103

104

frequency [Hz]

-1.2

-1.0

-0.8

-0.6

-0.4

-0.2

0.0

phase [rad]

Fig. 5: Absolute impedance and phase shift of the ITO electrodes before (�) and after (�) the depositionof the ODTMS monolayer with 1 mM K3/4Fe(CN)6 in 10 mM Hepes buffer (pH = 7.5). The measuredimpedance data were analyzed by the equivalent circuits IV and V. The deposition the ODTMS monolayerled to an increase in the charge transfer resistance from Rct0 = 110 � cm2 to Rct = 50 k� cm2, yielding thelocal defect area ratio of 0.2%.

[20] and a quantum well (QW) [21]. However, the application of GaAs to livingsystems is still difficult because of the complex electrochemical processes at theGaAs/electrolyte interface. Indeed, most of the electrochemical studies of GaAs havebeen performed in acidic or basic solutions [22–24]. There have been several reportson the functionalization of GaAs surfaces with various types of sulfides and mercaptosin contact with air or with metals, and the passivation effects were mostly discussed interms of photoluminescence (PL) from the bulk GaAs [25,26] or electrical propertiessuch as the Schottky barrier height [27–29]. Nevertheless, systematic studies on thefunctionalization 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 procedureswere attempted to optimize the surface pre-treatment. The chemical composition ofthe surface was evaluated by X-ray photoelectron spectroscopy (XPS), demonstratingthat the photochemical etching procedure (named “etch P” in this study) can generatea 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 ODTmonolayer. The monolayer showed high stability in air, as monitored by the constantellipsometric thickness (18∼20 Å). Cyclic voltammetry showed that both oxidation andreduction current at the interface were significantly suppressed by monolayer deposition

Page 214: Nanotechnology and Nano-Interface Controlled Electronic Devices

Living systems and semiconductor nano-structures 197

As 3d

Ga 3d

Inte

nsity

[a.

u.]

45 40 35 25 20 15

Binding Energy [eV]

As-oxide

Untreated

Etch A

Etch B

Etch C

Etch P

Fig. 6: Ga 3d and As 3d core-level spectra by XPS (survey scan) for the untreated surface (+), and forthe surfaces prepared by four different etching procedures; etch A (�), etch B (�), etch C (◦), and etchP (�). The symbols represent the measured data points, while the solid lines correspond to fitting. Thearsenide-oxide peak (EB ∼ 45 eV, indicated by an arrow) had disappeared after all the etching procedures.

(Fig. 7). The electrochemical passivation of the monolayer-coated surface was verifiedby impedance spectroscopy under current minimum potential (Uj=0 = −360 mV,determined by cyclic voltammetry) for more than 20 h (Figs. 8 and 9). Thickness of themonolayer can be estimated by the change in interface capacitance (18∼22 Å), whichis in good agreement with the ellipsometric thickness measured in air. The stabilityof the interface was further monitored under different bias potentials. Electrochemicalpassivation of the GaAs surface has been achieved for the first time under physiologicalconditions, which allows the potential application of GaAs electrodes to biologicalsystems.

2.3. Passivation of semiconductor nano-structures close to surface

Towards novel hybridization of (bio-)organic molecular assemblies and semiconductornano-structures, the surface of low-dimensional semiconductor was also chemicallymodified. As the first step, photoluminescence properties of the surface-near indium

Page 215: Nanotechnology and Nano-Interface Controlled Electronic Devices

198 M. Tanaka

100

50

0

-50C

urre

nt [μ

A]

-1.0 -0.5 0.0 0.5Potential [V]

with ODT

without ODT

Fig. 7: Cyclic voltammograms of n-GaAs before (dotted line) and after (solid line) the deposition of ODTmonolayer in a PBS buffer (pH = 7.5). For both cases, the surface was prepared by Etch P. The currentobserved before the deposition was significantly suppressed.

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 transferringalmost the same functionalization protocols, ODT monolayers were deposited onthe surface of surface-near InAs QDs [33]. The monolayer deposition resulted ina significant enhancement in the photoluminescence from the QDs, which can beattributed to the effective suppression of the surface state densities by arsenide–sulfidecoupling (Fig. 12). It is notable that the shown luminescence signals from chemicallymodified 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

10-3

Frequency [Hz]

10-1 105101 103

Abs

olut

e Im

peda

nce

[Οημ

]

105

107

103

RsC

Rp

Fig. 8: Impedance spectra of n-GaAs directly after Etch P (�), the same sample after 22 hours (�), andn-GaAs functionalized with ODT (•) are shown. The symbols represent the measured data, while the solidlines correspond to fits according to the equivalent circuit model composed of a serial resistance, Rs, acapacitance, C , and a parallel resistance, Rp. As the surface of freshly etched GaAs was not stable, thedata of this surface were measured in a smaller frequency range from 50 mHz to 10 kHz. After thefunctionalization, the spectrum was totally stable for more than 24 hours in the frequency range from 1 mHzto 100 kHz.

Page 216: Nanotechnology and Nano-Interface Controlled Electronic Devices

Living systems and semiconductor nano-structures 199

(a)

5Par

alle

l Res

ista

nce

[Ohm

cm

2 ]

105

106

5

5

Time [hours]24201612840

(b)

Cap

acita

nce

[μF

cm

-2]

1.0

1.5

2.0

0.8

Time [hours]24201612840

Fig. 9: Changes in (a) parallel resistance, Rp, and (b) interface capacitance, C , versus time before (◦) andafter (•) the monolayer deposition. Before the functionalization, Rp0 = 63 k� cm2 continuously increasedto 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 luminescencesignal was more significant when the QDs approached closer to the surface. In fact, theenhancement 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. Sincethe hydrophobic surface of the monolayer can be functionalized with polymer films andmodel cell membranes, this strategy is promising for the design of local detectors in thevery 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 aderivative of rigid 4′-substituted 4-mercaptobiphenyls, which provides various surfacefunctions by flexible substitution. Furthermore, Gölzhäuser et al. demonstrated that thisaromatic thiol monolayer could be cross-linked with low-energy electrons, which canbe applied as negative resists for nanolithography. Recently, we demonstrated the highly

Page 217: Nanotechnology and Nano-Interface Controlled Electronic Devices

200 M. Tanaka

Grow

th Direction

GaAs

AlGaAs

InAs

GaAs

AlAs/GaAs

GaAs

5 nm

5 - 25 nm

30 nm

(a)

CB

VB

hvex

GaAs

GaAs

Surface

AlGaAs E

z

hv

(b)

Fig. 10: Schematic illustrations of (a) the layered nano-structure and (b) the band diagram of the InAsquantum dots (QDs) in the vicinity of the surface. The carriers are created in a GaAs layer between QDsand AlAs/GaAs superlattice structures. The QDs were grown at T = 803 K in Stranski–Krastanov growthmode, resulting in ∼ 2×1010 dots per cm2.

Inte

nsity

[a.

u.]

1.21.11.0

Energy [eV]

1.3 1.2 1.01.1

Wavelength [μm]

QD/I30

QD/I20

QD/I10

T = 4.2 K150

0

50

100

Inte

nsity

[a.

u.]

1.21.11.00.9

1.3 1.2 1.1

Wavelength [μm]

Energy [eV]

QD/I30

QD/I20

QD/I10

T = 300 K75

0

50

25

(a) (b)

Fig. 11: Photoluminescence spectra of QD/I10, 20, and 30 at T = 4.2 K (a) and 300 K (b), excited bya 50 W cm−2 Ar+ laser (513 nm). The continuous decrease in the intensity due to the non-radiativerecombination via surface states and the red shift of the peak due to the local strain near the dots wereobserved. The photoluminescence signal could be detected when the distance to the surface was as close asd = 10 nm.

stable surface coating of GaAs [100] surface with a mercaptobiphenyl monolayer [34].Homogeneity and hydrophobicity of the surfaces were characterized by atomic forcemicroscopy (AFM) and contact angles to a water droplet. Electrochemical propertiesof the monolayer such as resistance and capacitance were quantitatively measured bycyclic voltammetry and impedance spectroscopy. The current–voltage scans showedthat the monolayer deposition led to a remarkable suppression of electrochemistry atGaAs/electrolyte interface. The electrochemical stability of the monolayer-coated GaAswas carefully verified by impedance spectroscopy in the wide frequency region between

Page 218: Nanotechnology and Nano-Interface Controlled Electronic Devices

Living systems and semiconductor nano-structures 201

Inte

nsity

[a.

u.]

1.00.9

1.3 1.21.4

Wavelength [μm]

Energy [eV]

as-grown

monolayer-coated

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 spectralred shift of 15 meV after the monolayer coating, which can be attributed to the reduction in distance to thesurface 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 ofextracellular matrices (ECMs)

Cell surface glycocalix and extracellular matrices (ECMs) in nature maintain high localdisjoining pressures and generate hydrated “cushions” between cells and tissues by thecombination of weak, generic forces at interfaces [35,36]. One of the possible strategiesto mimic such interlayers is the deposition (e.g. grafting, casting, spin-coating and layer-by-layer transfer) of thin (d < 100 nm) hydrated polymer films on solid surfaces. Thisstrategy can be applied for proliferation and stress-free immobilization of cells, enzymesand receptors under non-denatured conditions, as well as for creation of model ECMs tocontrol cell–cell and cell–tissue interactions by the chemical nature and morphology ofpolymers.

Nature utilizes such soft interlayer at cell–cell and cell–tissue interfaces, and controlstheir 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

Page 219: Nanotechnology and Nano-Interface Controlled Electronic Devices

202 M. Tanaka

Inte

nsity

[a.

u.]

1.11.00.9

1.3 1.2

Wavelength [μm]

Energy [eV]

right after coatingafter 30 days

Fig. 13: Photoluminescence spectrum of QD/I10 right after the coating (broken line) and that of the samesample stored under ambient conditions (in air, at room temperature) for one month (solid line). No changesin intensity or in the peak position were observed.

when the interfacial interaction becomes too attractive [37]. Indeed, recent studieshave demonstrated that the phase separation induced by strong adhesion can evenbe interpreted as the first-order wetting/de-wetting transition [38]. Thus, the precisecontrol of “wetting affinity”, i.e. interfacial forces at soft interface under water, is veryimportant to fabricate thermodynamically and mechanically stable biological molecularcomposites [1].

To date, a sufficient biocompatibility that prevents from nonspecific adsorption ofproteins and cells could be achieved by grafted films of dextran [39] and poly(ethyleneglycol) (PEG) derivatives [40], which have been used in numerous fields. When the filmsof functionalized PEG derivatives are covalently end-grafted, the protein/cell resistanceof 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 surfaceelastic 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 cellwalls of plant cells. In the 1990s, Wegner and his colleagues have established chemical

Page 220: Nanotechnology and Nano-Interface Controlled Electronic Devices

Living systems and semiconductor nano-structures 203

OO

OH

SiO

OSi n

Fig. 14: Chemical structure of trimethylsilylcellulose (TMSC), which has the degree of substitutionD.S. = 2.0. Hydrophobic trimethylsilyl side chains can be cleaved by treatment with HCl vapor, resulting inregenerated 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 LBfilms [52–54]. We also demonstrated the biocompatibility of the regenerated celluloseby deposition of native and model cell membranes [18] (ref. Section 3). Several typesof cellulose derivatives with rigid ’rod-like’ backbones and ’hairy’ hydrophobic sidechains have been applied. Since the thickness of the monolayer is 4∼9 Å, the totalfilm thickness can be controlled in nm accuracy simply by changing the number ofdeposited monolayers. Furthermore, intra- and inter-layer structures can be stabilized bycross-linking of the side chains or by hydrogen bonding.

Here, appropriate numbers of trimethylsilylcellulose (TMSC) layers (10 and 20 lay-ers, with thicknesses of about 10 and 20 nm, respectively) were transferred successivelyonto silicon wafers, that were hydrophobized with ODTMS monolayers (ref. Section1.1). 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 (20layers), respectively (Fig. 14).

To investigate static and dynamic swelling behaviors of such ultra-thin polysaccharidefilms under controlled thermodynamic conditions, the film thickness was measuredquantitatively by ellipsometry coupled to a climate chamber (Fig. 15a, b). Ellipsometryis a non-invasive, powerful technique with a high thickness resolution (±0.1 Å) to studysoft, hydrated polymer films. The static swelling behavior was studied by measuring theequilibrium film thickness as a function of the relative humidity to obtain force–distancecurves, i.e. disjoining pressure plotted as a function of film thickness. The dynamicswelling behavior was studied by monitoring the change of the film thickness as afunction of time after an ’osmotic shock’ (i.e. abrupt change of the relative humidity) toobtain the kinetics of the hydration.

Interestingly, both the maximum swelling ratio ρmax = dmax/d0 (∼ 1.7) and thenormalized decay length λ∗ = λh/d0 (∼ 0.12) were independent from the number oflayers, i.e. initial thickness of the films (Fig. 16a, b). This could be attributed to thelayer-by-layer structure of LB films that is maintained even after the regeneration. Thedynamic swelling experiments suggested that the characteristic time constant of thewater uptake of the cellulose film could be much smaller than that of the atmosphereexchange (τ ∼ 30 s). Further improvement to trace faster kinetics of hygroscopicpolymers is required.

As physical models of heterogeneous (i.e. multi-functional) cellular surfaces pos-

Page 221: Nanotechnology and Nano-Interface Controlled Electronic Devices

204 M. Tanaka

air in

air out

polarizer

analyzer

detector

He-Ne-laser

humiditychamber

film lift

sample

polarizer

analyzer

CCD camera

He-Ne-laser(HBO lamp)

sample

microscopeobjective

tube lens

air in

air out

(a) (b)

Fig. 15: Schematic views of (a) a point-like ellipsometer and (b) an imaging ellipsometer coupled to aclimate chamber. The static swelling behavior can be obtained by measuring the equilibrium film thicknessas a function of the relative humidity, while dynamic swelling behavior was monitored as changes in thefilm 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 Δ andΨ can be obtained for each pixel on the image.

18

16

14

12

10

8

6

laye

r th

ickn

ess

[nm

]

100806040200

relative humiditiy X [%]

10 layers cellulose 20 layers cellulose

6

8

106

2

4

6

810

7

2

4

6

810

8

2

6

disj

oini

ng p

ress

ure

[Pa]

1.81.61.41.21.0

swelling ratio ρ

10 layers cellulose 20 layers cellulose

Hydrationregime}Repulsionregime}π ~ ρ-n

π = π0 exp (-ρ/λ0)

(a) (b)

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 20monolayers).

sessing self-assembled patterns of functional domains (e.g. rafts, clustered ligands), itis interesting to generate micro-patterns of polysaccharide films with different sizesand distributions by UV photolithography (Fig. 17). When coupled to a cooled CCDcamera, two ellipsometric parameters Δ and Ψ can be obtained locally by applyingthe rotating analyzer principle for each pixel on the image. Different from scanningprobe microscopy such as atomic force microscopy (AFM), this technique allows thenon-invasive study on structure and kinetics of swelling behavior of structured, softinterfaces. Mathe et al. reported “local” swelling of physisorbed films of dextran and

Page 222: Nanotechnology and Nano-Interface Controlled Electronic Devices

Living systems and semiconductor nano-structures 205

Fig. 17: Micro-patterned polysaccharide films as multi-functional cell and tissue surfaces. Native and modelcell 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 thepresented (a) Δ and (b) Ψ maps, one can gain ellipsometric parameters within a lateral resolution of 1 μmand a vertical resolution of 1 nm.

hyaluronic acid by imaging ellipsometry, where not only the “swelling” in the verticaldirection 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 (Ψand Δ 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, synthesizedby living cationic polymerization, provide the possibility to tailor macromolecules bythe choice of different side substitutions and terminal groups [56,57]. Since the lengthof the alkyl chain in the 2-position determines the hydrophilicity of each monomer

Page 223: Nanotechnology and Nano-Interface Controlled Electronic Devices

206 M. Tanaka

R1N

R2 O

NH

Si

OCH3

OCH3H3CO

n

R1- = CH3-R2- = CH3-, C2H5-

n = 15, 30Fig. 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, biocompatibilityof poly(2-oxazoline) derivatives was investigated, where lipid vesicles with poly(2-oxazoline) lipopolymers showed remarkably larger blood circulation times than ordinaryphospholipid 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 withtrimethoxy coupling groups (n = 15 and 30, PDI = 1.09–1.24) onto silicon wafers, andstudied 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 thePEOX30 film, ρmax(PEOX30) = 1.6–1.8, was obviously smaller than the correspondingvalues obtained for the films of PMOX15 and PEOX15 (Fig. 20). At high disjoining

5.5

5.0

4.5

4.0

3.5

3.0

2.5

2.0

poly

mer

thic

knes

s [n

m]

100806040200

relative humiditiy [%]

1.52

1.50

1.48

1.46

1.44

1.42

1.40

thickness [nm] refractive index

Fig 20: Equilibrium thickness (left ordinate) and refractive indices derived from the Garnett equation (rightordinate) of a grafted PMOX15 film plotted versus relative humidity of the atmosphere. Significant changesin the film thickness were observed under relative humidity conditions > 90%.

Page 224: Nanotechnology and Nano-Interface Controlled Electronic Devices

Living systems and semiconductor nano-structures 207

Table 1

Physical parameters of the grafted poly(2-oxazoline) films measured by ellipsometry. PMOX15, poly(2-methyloxazoline) with n = 15; PEOX15, poly(ethyloxazoline) with n = 15; and PEOX30, poly(2-ethyl-oxazoline) with n = 30. Initial dry thickness d0 was determined at a relative humidity of ∼ 4%, by assumingthe refractive index n0 = 1.52. ρmax is the maximum swelling ratio, n1 and n2 stand for exponents for thepower-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 lowerdisjoining pressure regime, P < 1×107 Pa.

Polymer d0 [Å] ρmax n1 n2 λ∗

PMOX15 18.2–21.7 2.7–3.4 18–24 6.6–8.5 0.49–0.56PEOX15 12.4–20.8 2.8–3.5 12–34 6.5–8.4 0.48–0.63PMOX30 24.4–39.7 1.6–1.8 32–33 12 0.16–0.18

pressure conditions (P > 1×107 Pa), we observed two regimes where strong repulsiveforces obey a power-law, P(ρ) ∼ ρ−n : the first regime with large exponents (n1 = 34–12) can be almost explained by a theta function, while the exponents in the secondregime 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 normalizedswelling ratio ρ. The decay constant λ∗ also exhibited a clear dependence on the chainlength: 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 thechain length, but is not affected by the side chains. The characteristic time constantfor PEOX30 films (τPEOX30 = 200 s) is larger than those for the shorter PEOX15 andPMOX15 (τ = 140–150 s) (Fig. 21).

Since poly(2-oxazoline) used in this study can be terminally functionalized withalkyl 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) allowsa precise control of interfacial properties of the resulting brushes. Recently, we couldaccomplish 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 interac-tion based on weak cationic polymer brushes. The monolayer of diblock copolymers(poly(2-(dimethylamino)ethyl methacrylate-block-methyl methacrylate), synthesized bythe group of S.P. Armes, Univ. Sussex), whose hydrophilic part can be charged/de-charged by pH titration under physiological conditions (pH = 5∼9) [61], is depositedfrom the air/water interface onto the hydrophobized silicon wafers by the Langmuir–

Page 225: Nanotechnology and Nano-Interface Controlled Electronic Devices

208 M. Tanaka

thic

knes

s [a

rb. u

nits

]

time[s]

160012008004000

humid dry

Fig. 21: Dynamic swelling curves of a PEOX15 film under several cycles of ‘osmotic shocks’. The filmthickness (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 berecovered 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 fluidityof the membranes can be controlled by grafting density, side chain functionalities, and mixing behavior withmatrix lipids.

Schaefer method. Preliminarily, we measured the pressure–area isotherms of the diblockmonolayers at the air/water interface to confirm the stability of the films and their ther-modynamic properties. Despite the “glassy” nature of MMA backbones, the monolayershowed no compression/relaxation hysteresis or material loss until it collapsed at 50mN/m, when the barrier was driven at 20 μm s−1. The application of reflectivity tech-niques, such as neutron reflection in D2O, will enable us to observe the conformationalchanges of weak polyelectrolyte brushes and interaction with water soluble proteins(lectins, fibronectin, etc.) under pH titration.

Page 226: Nanotechnology and Nano-Interface Controlled Electronic Devices

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, glycolipidsin animal cells are localized in the outer leaflet of the lipid bilayer and stabilizethe 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 distinctconformation and to interact with counter-part lectins and cell adhesion receptors[35,62]. Examples are blood group- and tumor-associated antigens such as sialyl-Lewis-X and sialyl-Lewis-A that interact specifically with selectins [63–65]. Interestingly,besides carbohydrate–protein interactions, it has been demonstrated that cell surfacecarbohydrates 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 ofglycoscience. However, the physical basis of glycocalix functions has not yet beenunderstood experimentally. As physical models of glycocalix, we have investigated theinterfacial 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 lactoseoligomers (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 areexpected to be rather simple glycocalix models. In this study, Langmuir pressure–areaisotherms were first measured at several temperatures to estimate thermodynamic andstructural parameters of the monolayers at the air/water interface. The molar transitionentropy and the molar latent heat were calculated by applying the Clausius–Clapeyronequation. It has been demonstrated that the phase behavior of the glycolipid monolayersis comparable to that of ordinary phospholipids [68], in spite of the lower degree of co-operativity between the larger head groups (Fig. 24a–c). These results suggest entropiceffects of the head groups on the interaction between the neighboring molecules. At the

O

OH

HOOH

O

OH

O

HOOH

O

OH

OC16H33

OC16H33N

N = 1 - 3

Fig. 23: Structures of the glycolipids studied, Lac N (N = 1, 2, and 3).

Page 227: Nanotechnology and Nano-Interface Controlled Electronic Devices

210 M. Tanaka

35

30

25

20

15

10

5

0

120100806040

303 K 298 K 293 K 288 K

π [m

N/m

]

Area [ 2]

35

30

25

20

15

10

5

0

120100806040

303 K 298 K 293 K 288 K

π [m

N/m

]

Area [ 2]

35

30

25

20

15

10

5

0

120100806040

303 K 298 K 293 K 288 K

π [m

N/m

]

Area [ 2]

a

b c

Å

Å ÅFig. 24: Langmuir isotherms of the monolayers of: (a) Lac 1, (b) Lac 2, and (c) Lac 3 at various subphasetemperatures. The phase coexistence line was fitted by a polynomial of 4th order (broken line). At hightemperature and pressure conditions, the coexistence line can be approximated as parabola, whose vertexcoincides with the critical point. The points of intersection of the linearly extrapolated lines in (a) weretaken 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 thesaccharide 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 themonolayer was transferred. The obtained swelling curve of Lac 1 was similar to thoseof 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 behaviorsand (ii) “polymer-like” swelling behaviors, might explain the distinct conformationof oligosaccharides on cellular surfaces, which can be recognized by proteins andcarbohydrates selectively.

Page 228: Nanotechnology and Nano-Interface Controlled Electronic Devices

Living systems and semiconductor nano-structures 211

20 40 60 80

0

4000

8000

12000 (a)C

p [c

al/m

ol ˚

C]

T [ ˚C]

0.60.50.40.30.20.1

20 ˚C

40 ˚C

60 ˚C

80 ˚C

2.01.81.61.41.21.00.8

20 ˚C

80 ˚C

3.76 Å

4.45 Å

4.57 Å

7.50 Å

Log

(I)

[ar

b. u

nits

]

SAXS WAXS

q [Å-1]

(b)

Fig. 25: (a) Differential heat capacity scan of the Lac 1 dispersion (1 mg/mL) recorded at a heating rate of20°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, polymorphismof Lac N lipids (N = 1,2,3) was studied in lamella phase [73]. By a combinationof differential scanning calorimetry (DSC) and small- and wide-angle X-ray scatteringexperiments (SAXS and WAXS), the effect of hydrophilic/hydrophobic balance ontheir thermotropic phase behaviors was systematically studied. The dispersion of Lac 1exhibited the crystalline–fluid phase transition (Fig. 25a, b). Both the thermotropic phase

Page 229: Nanotechnology and Nano-Interface Controlled Electronic Devices

212 M. Tanaka

(a)

20 40 60 80

0

2000

4000

6000

Cp

[cal

/mol

˚C

]

T [˚C]

0.60.50.40.30.20.1

20 ˚C

40 ˚C

60 ˚C

80 ˚C

Log

(I)

[ar

b. u

nits

]

SAXS WAXS

20 ˚C

80 ˚C

4.45 Å

2.01.81.61.41.21.00.8

4.17 Å

q [Å-1]

(b)

Fig. 26: (a) Heat capacity trace of the Lac 2 dispersion (1 mg/mL), where the phase transition enthalpy wasalso clearly reduced to �H = 9.2 kcal/mol. (b) SAXS diffraction patterns of the lamellar dispersion of Lac2 at T = 20, 40, 60, and 80°C (left). The lamellar spacing showed a transition between 40°C (dSAXS = 87Å) 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) ofthe Lac 1 dispersion are higher in comparison to those of other synthetic lipids withdihexadecyl chains, Tm = 40∼50°C and �H = 8∼9 kcal/mol. Here, the alkyl chainsare strongly correlated by van der Waals interaction, which even enabled them toform crystalline-like tight packing with almost no tilting. The additional enthalpiccontribution can be due to the hydrogen bonding between the Lac 1 head groups that arefree 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 chainsis shifted to the hydrophilic side. This shift in the balance reduces the cooperativitybetween the alkyl chains, resulting in a decrease in the transition temperature and thephase transition enthalpy (Fig. 26a, b). The strongly crystallized alkyl chain packing was

Page 230: Nanotechnology and Nano-Interface Controlled Electronic Devices

Living systems and semiconductor nano-structures 213

20 40 60 80

(a)

0

2000

4000

-2000

Cp

[cal

/mol

˚C

]

T [˚C]

WAXS

2.01.81.61.41.21.00.8

80 ˚C

20 ˚C

4.19 Å4.46 Å

7.61 Å

q [Å-1]

SAXS

Log

(I)

[ar

b. u

nits

]

80 ˚C

20 ˚C

0 0.1 0.2 0.3 0.4 0.5 0.6

60 ˚C

40 ˚C

(b)

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 lamellarspacing showed no transition at all measurement conditions, dSAXS = 108 Å. The WAXS peaks suggestedthat 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. Differentfrom 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 sharpscattering peaks due to the slightly tilted alkyl chains and one sharp correlation peak dueto the head groups. The small-angle scattering also demonstrated the highly ordered 3Dlamellar 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 hexasaccharidehead groups forced the alkyl chain to take the distorted, crystalline-like packing, whichis obviously different from the ideal hexagonal lattice. Since the attractive interactionbetween the head groups is so strong, the hydration does not take place any longer. Togain more insight in the structural characteristics of this phase, further conformationalanalysis by spectroscopic techniques such as FTIR and the systematic link to themorphology in two dimensions, i.e. morphology of monolayers, is further progressed.

Page 231: Nanotechnology and Nano-Interface Controlled Electronic Devices

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) operatingon cellular surfaces is the measurement of the rheological (viscoelastic) properties ofglycocalix. Several studies have been conducted on the morphology and elasticity ofbacteria surfaces using atomic force microscopy (AFM) [74–76]. However, extremelytoxic fixatives (e.g. glutaraldehyde, osmium tetroxide) or strong mechanical stressesoften used in the traditional fixation procedures can damage the cell irreversibly.Although there are some studies on the elasticity of the isolated sheath [77] orsacculi [78], there has been no quantitative study on rheology of artificial glycocalixmodels, such as the monolayers of synthetic glycolipids at the air/water interface.The rheological study of insoluble monolayers (Langmuir monolayers) at the air/waterinterface can provide a quantitative measure of hydrogen bonding between neighboringmolecules, physical entanglement, and cross-linking under dynamic conditions. Incomparison to other conventional devices [79], a new class of rheometers, referred to asinterfacial stress rheometer (ISR) [80], developed by G. Fuller’s group, allows for highlysensitive and real-time measurements of viscoelastic parameters at different frequenciesunder controlled thermodynamic conditions (surface pressure, temperature) (Fig. 28).

Using this device, the rheology of Lac N monolayers at the air/water interfacewas investigated [81]. The viscous and elastic surface moduli of the monolayerswere measured as a function of the length of the linear oligosaccharide head groupsquantitatively. The Lac 1 monolayer was highly viscoelastic, which can be attributed tostrong chain–chain correlations. The introduction of another lactose unit further reducedthe chain–chain correlation, and the head groups are more hydrated (Fig. 29a). Thisresults in a fluid nature of the Lac 2 monolayer. In contrast, a clear rheological transitionof Lac 3 monolayers from a viscous to an elastic film could be observed for the firsttime at a surface pressure of 6∼8 mN/m, suggesting the formation of a cross-linkedphysical gel (Fig. 29b). On the other hand, if one considers the short (the stretchedlength 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 oligosaccharidehead groups. Interestingly, the surface pressure at which the rheological transition ofthe Lac 3 monolayer takes place (π = 6∼8 mN/m) corresponds to the end point ofthe coexistence of the liquid expanded and the liquid condensed phase. Above thistransition, the hydrating water is excluded and hydrogen bonding “bridges” the Lac 3head groups during lateral compression to higher pressures (Fig. 30). This is in contrastto the monolayers of Lac 1 and Lac 2, where the phase transition to the liquid condensedphase results in a significant increase in the film viscosity. Thus, it can be concludedthat the rheological transition of the Lac 3 monolayer is not caused by the correlationbetween the condensed alkyl chains, but by the strong coupling between the linearhexasaccharide (Lac 3) head groups, which might mimic a function of glycocalix tointroduce elasticity to the plasma membrane. Indeed, the obtained results are in goodagreement 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

Page 232: Nanotechnology and Nano-Interface Controlled Electronic Devices

Living systems and semiconductor nano-structures 215

Glass Wall

L ~ 30 mm

W ~ 20 mm

BMagnetic Rod

Glass Wall

Detector (Photodiode)

Helmholtz Coils

BarrierPressure Sensor

Light Source

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 enableus to investigate the effects of carbohydrate–carbohydrate and carbohydrate–proteininteraction on the membrane rheology, proposing a realistic physical model of cellsurface glycocalix (Fig. 31).

5. Native and model cell membranes on semiconductors – potential candidates forsensorics

In nature, the plasma membrane defines the periphery of each cell and separates itscytoplasmic contents and extracellular spaces. The fundamental components of plasmamembranes are lipid bilayers, which support the hierarchical structures of variousmembrane proteins.

Artificial lipid monolayers and bilayers on solid supports are treated as plasmamembrane models [83,84], which can be deposited onto solid surfaces by one ofthe following methods: (i) monolayer transfer, (ii) vesicle fusion, (iii) single bilayerspreading, or (iv) solvent exchange method (Fig. 32). The first method includestransfer of the lipid monolayers from air/water interface by the Langmuir–Blodgett

Page 233: Nanotechnology and Nano-Interface Controlled Electronic Devices

216 M. Tanaka

1.4

1.2

1.0

0.8

0.6

0.4

0.2

0.0

434241403938

G' G''

Area per Molecule [Å2]

G',

G''

[mN

/m]

30

20

10

0

120100806040Area per Molecule [Å2]

Late

ral P

ress

ure

[mN

/m]

A

C

B

A

B

C

C

B

(a) 1.4

1.2

1.0

0.8

0.6

0.4

0.2

0.056504240

G' G''

Area per Molecule [Å2]

G',

G''

[mN

/m]

Area per Molecule [Å2]

Late

ral P

ress

ure

[mN

/m]

30

20

10

0

120100806040

AB

C

A

C

B

C

5452 5848464438

(b)

gelation

Fig. 29: (a) Dynamic moduli of a “fluid” Lac 2 monolayer, measured at T = 20°C, ω = 1 rad/s. Themonolayer remained viscous fluid (G ′ < G ′′) through the measurements. (b) Rheological transition of a Lac3 monolayer (at T = 20°C, ω = 1 rad/s). At a surface area of about 0.5 nm2 (corresponding to a surfacepressure of 6 to 8 mN/m), the monolayer became elastic (G ′ > G ′′).

0.16

0.14

0.12

0.10

0.08

0.06

0.04

0.80.70.60.50.40.30.2

ω [rad/s]

Lac2

Lac3

G''

[mN

/m]

π = 20 mN/m

Fig. 30: The loss modulus G ′′ of Lac 2 (�) and that of Lac 3 (�) at π = 25 mN/m, plotted as a function ofoscillation frequency.

and Langmuir–Schaefer methods. This is a laborious procedure, but enables one todeposit asymmetric bilayers [85]. In the second method, lipid vesicles are spreadonto the substrate from vesicle suspensions [86]. By adjusting the adhesion forcesto the substrate (e.g. electrostatic attraction, wetting conditions), vesicles rupture onthe surface and form adherent bilayer patches, which fuse with each other to formcontinuous bilayers. The third method, single bilayer spreading, can be achieved bypasting a lipid reservoir onto substrates, followed by the swelling under water. A singlebilayer is spontaneously pulled over the surface by adhesion forces [87]. The fourthmethod results in the formation of lipid membranes simply by exchange of solvent fromalcohols (e.g. ethanol, isopropanol) to aqueous buffers [88]. According to the latter threemethods, membranes are formed according to the self-assembling process to heal theirlocal defects and pores.

Page 234: Nanotechnology and Nano-Interface Controlled Electronic Devices

Living systems and semiconductor nano-structures 217

O(CH2)15CH3

O(CH2)15CH3O

O

O NHAc

OH

OH

OO

OH

OH

OH

O

OH

Me

O

OH

OHO

OH

O

HOOH

OH

OAcNH O

HO

CO2ÐOH

OHHO

O

HO

O(CH2)15CH3

O(CH2)15CH3O

O

O NHAc

OH

OH

OO

OH

OH

HO

OH

O

OH

O

OH

OHO

OH

O

HOOH

OH

O

Me

HO

(a)

(b)

Fig. 31: Other functional glycolipids with (a) sialyl Lewis-X and (b) Lewis-X head groups with lactosespacers. Synthetic modification of head groups and hydrophilic/hydrophobic balance even enables to formfunctional domains (artificial rafts) at the interface.

Step 1 Step 2

Langmuir-Blodgett(vertical lifting)

Langmuir-Schääaefer(holizontal dipping)

(i) (ii)

Step 1 Step 2

adsorption, rupture spreading, fusion

Step 2

swellng, spreading

Step 1

dry lipid reservoir

(iii) (iv)

Step 1alcohol

Step 2aqueous buffer

Fig. 32: Deposition of model cell membranes (lipid mono- and bilayers) on solid supports: (i) monolayertransfer, (ii) vesicle fusion, (iii) single bilayer spreading, or (iv) solvent exchange method.

To introduce biological functions to the supported lipid membranes, it is necessaryto incorporate membrane proteins in non-denatured state, maintaining their native ori-entation and function (Fig. 33). One of the less-laborious methods is the direct insertionof proteins (peptides) from the solution into the pre-formed membranes. This is suit-able especially for the incorporation of small antibiotic peptides [89], but it requires

Page 235: Nanotechnology and Nano-Interface Controlled Electronic Devices

218 M. Tanaka

(a) (b)

Fig. 33: Introduction of bio-functionalities to model membranes: (a) incorporation of transmembraneproteins/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 theminto artificial lipid vesicles (proteoliposomes) and to deposit them onto substrates [93–98]. To “dock” proteins without hydrophobic cores onto a membrane surface, severalnatural and synthetic anchors can be used: biotin tethers for avidin/streptavidin [99],glycan–phosphatidyl inositol (GPI) linkages [100], and nitrotriacetic (NTA) anchorsagainst recombinant proteins with histidine tags [101,102]. Although it is challengingand practically difficult, the direct deposition of native cell membranes on solid surfaces[103] includes numerous advantages to maintain the original orientations and popula-tions of the transmembrane proteins. In the following sub-sections, several experimentalapproaches 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 werefunctionally modified with alkylsiloxane monolayers (ref. Section 1.1) and LB filmsof regenerated cellulose (d = 5 nm, ref. Section 2.1). ITO is transparent and allowsfor in situ investigations both with optical microscopy and electrical techniques. Sincethe regenerated cellulose films are hydrated and behave like an electrolyte, this systemcan be characterized as an electrolyte/membrane/electrolyte/semiconductor structure(EMES) (Fig. 34). Lipid membranes have been deposited by vesicle fusion, and thehomogeneity and fluidity of the resulting membranes were confirmed by fluorescencemicroscopy and fluorescence recovery after photobleaching (FRAP). Reflection inter-ference contrast microscopy (RICM) can be used to monitor the spreading of a lipidbilayer on a polymer cushion, exhibiting the self-healing of bilayers due to lateraldiffusion (Fig. 35). In fact, deposition of the lipid bilayer resulted in a significantchange in interface resistance to Rm = 0.44 M� cm2, which is slightly smaller than thatof solvent-free black lipid membranes. Noteworthy, the electrical properties achievedhere remained stable for more than a week, suggesting thermodynamic and mechanicalstability of the membranes. Compared to the work of Gritsch et al. [17], where the samelipid bilayer was directly deposited onto ITO electrodes, the membrane resistance is

Page 236: Nanotechnology and Nano-Interface Controlled Electronic Devices

Living systems and semiconductor nano-structures 219

Electrolyte

Membrane

Electrolyte

Semiconductor

EMES StructureFig. 34: Polymer-supported lipid bilayer deposited onto an ITO electrode. As the hydrated cellulosefilms 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 byreflection 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 canbe attributed to the reduction of surface roughness by cellulose “cushions” as well asto the wetting affinity between cellulose and lipid bilayers. As a preliminary check ofthe membrane quality for the protein incorporation, an antibiotic peptide, gramicidin D,was incorporated into membranes by incubating small vesicles with gramicidin D for6∼10 hours. Although function and ion selectivity of the incorporated channels werereasonable from their native functions, the amount of the incorporated channels was

Page 237: Nanotechnology and Nano-Interface Controlled Electronic Devices

220 M. Tanaka

reference counter

flow cell

Fig. 36: Schematic drawing of the system studied. A lipid bilayer was deposited on Si/SiO2 electrodescoupled to a flow cell by vesicle fusion, and a trifluoroethanol solution of antibiotic peptide gramicidinD was injected. The formation of a continuous membrane, insertion kinetics of peptides, and theirfunctionalities can be monitored by impedance spectroscopy.

certainly small. By assuming a single ion channel conductance of 0.7 pS, the surfaceconcentration 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 chemicalstability, flexibility for the manufacturing of a variety of device structures [105], therehave been not so many applications of Si/SiO2 in biological systems, except for livingcells on transistors [106,107]. It is still difficult to assign the electrical propertiesof membranes separately from the background signals [108,109], especially whenthe capacitance of the oxide layer is competitive or even less than that of the lipidmembranes, 0.7 μF cm−2 [110]. Fromherz et al. had overcome this problem by touchinga giant vesicle to the open gate of a field-effect transistor [111]. The lipid bilayer keptits continuous shape without rupturing, and adhered onto the cationic poly-lysine filmdue to electrostatic attraction.

We took a different strategy to deposit lipid membranes onto highly doped p-type Si/SiO2 electrodes. First, important physical parameters of the substrates suchas the doping ratio and thickness of the oxide layer were carefully optimized by acombined study using ellipsometry and impedance spectroscopy. A lipid bilayer wasthen deposited by fusion of small unilamellar vesicles on the electrode (Fig. 36).Self-assembling of the bilayer patches and the continuous growth of a homogeneousmembrane could be monitored by measuring electrical impedance over a wide frequencyrange (from 100 kHz to 1 mHz). Despite of a relatively larger active electrode area (0.50cm2), the resistance of the membrane can be compared to that of the solvent-free, blacklipid membrane, 1 M� cm2. To increase the incorporation amount and to acceleratethe uptake, gramicidin D was incorporated into this membrane from a trifluoroethanolsolution. By measurement of membrane resistance as a function of time, we couldmonitor the kinetics of peptide uptake (Fig. 37). Furthermore, the membrane withgramicidin showed certainly high conductivity and reasonable selectivity to monovalentcations (Na+, K+) against anions (Cl−) (Fig. 38). Such strategy can be applied in fastscreening of the uptake of toxins and other antibiotic peptides as well as in the in situinvestigation of pore formation in bio-membranes.

Page 238: Nanotechnology and Nano-Interface Controlled Electronic Devices

Living systems and semiconductor nano-structures 221

1000

900

800

700

600

500

400

300

200

100

0

Mem

bran

e R

esis

tanc

e [ k

Ωcm

2]

806040200Time after Injection [min ]

injection

Fig. 37: Insertion of gramicidin D, monitored as a change in membrane resistance as a function of time. Theinsertion 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 ofhighly insulating polymer/lipid composite films on indium-tin-oxide (ITO) semiconduc-tor electrodes [112] (Fig. 39). The lipid monolayers were deposited on LB multilayers ofcellulose derivatives (isopentylcellulosecinnamate, IPCC) [113] by continuous exchangeof solvent. Prior to membrane deposition, the intra- and inter-membrane structures werestabilized by cross-linking cinnamoyl side chains under UV illumination. The resultingpolymer/lipid composite system showed an electric resistance of 2.5 M� cm2 and alateral diffusion constant for the lipids of 0.1 μm2/s, which were obtained by impedancespectroscopy and FRAP, respectively. Such highly insulating and fluid composite films onsemiconductors can be utilized as a membrane charge sensor to detect changes in surfacecharge by treating this electrolyte/(organic) insulator/semiconductor (EIS) system as ananalogue of the metal/oxide/semiconductor (MOS) system. For this purpose, we incor-porated 10 mol% of lipids with chelator headgroups in order to switch the membranecharge reversibly. In the presence of Ni2+-loaded and EDTA-loaded (EDTA: ethylene-diaminetetraacetic acid) buffers, significant changes in the impedance spectra could beobserved (Fig. 40). Noticeably, the “switching” of electrical properties between “un-charged (Ni2+-loaded)” and “charged (unloaded)” states was reversible and reproduciblefor more than 2 weeks, showing no degradation. Semi-quantitative Mott–Schottky anal-ysis (due to a high electron density (1021 cm−3) of our ITO, it cannot be quantitativelytreated as a “standard” semiconductor, but a “degenerated” semiconductor) demonstratedthat the difference in surface charge density of �Q ∼ 10−6 C/cm2 changed the flatband potential of the EIS system by nearly 50%. From our measurement accuracy to

Page 239: Nanotechnology and Nano-Interface Controlled Electronic Devices

222 M. Tanaka

103

104

105

106

107

Abs

olut

e Im

peda

nce

|Z| [

Ω]

10-2

10-1

100

101

102

103

104

Frequency [Hz]

bare electrode membrane with gramicidin, Cl -

with gramicidin, Na +

with gramicidin, K +

Fig. 38: Impedance spectra of the supported lipid bilayer before (◦) and after incorporation of gramicidin Dat Ubias = 0 mV. The ion selectivities of the incorporated channels were reversible and reproducible for morethan a week in the presence of Na+, K+, and Cl− ions.

Uncharged

O

N

OO

R

Ni

OO

O

Ni-loaded

Charged

R

N+

O

O

O-

O

O

O

H

+++

Unloaded

Ni2+

EDTA

Δq = 1eE

I

S

--

-

- - - - - - - - --- -

Fig. 39: Principle of membrane charge sensor based on biological electrolyte/(organic) insulator/semi-conductor (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 aschanges in capacitance of space charge region or flat-band potential of semiconductors. Here, reversiblecomplexation of nickel ions changes the molecular net charge of the NTA-lipid by 1 e−.

Page 240: Nanotechnology and Nano-Interface Controlled Electronic Devices

Living systems and semiconductor nano-structures 223

103

104

105

106

107

Abs

olut

e im

peda

nce

|Z| i

n [O

hm]

10-2

10-1

100

101

102

103

104

Frequency [Hz]

Uncharged (Ni-NTA)

Charged (with EDTA)

Cs

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 ofnickel ions (Ni-loaded), the lipid monolayer is baring 1 e−/NTA. After treatment with EDTA, the NTAhead groups are unloaded and the monolayer is charged with 2 e−/NTA. A semi-quantitative Mott–Schottkyplot under potential sweeps revealed that changes in surface charge density of �Q ∼ 10−6 C/cm2 changedthe 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 pernm2. The obtained result suggests that the sensitivity limit of this strategy is promising todetect selective coupling of proteins on a membrane surface. In fact, the chelator complexof 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 lipidscan be used as a recyclable platform for the selective detection of charged protein bindingtowards non-invasive applications in pharmaceutical screening tests.

5.3. Orientation-selective immobilization of native cell membranes on ultra-thinpolymer films

In comparison with various methods to reconstitute transmembrane proteins in artificialmodel membranes, the direct deposition of “native” cell membranes includes several

Page 241: Nanotechnology and Nano-Interface Controlled Electronic Devices

224 M. Tanaka

Ni2+

EDTA ---(I) Charged

---

-----

(II) Uncharged

(III) Protein BoundEDTA charged proteins

his-tag

---

-----

---

-----

+

Fig. 41: Potential application of biological EIS systems for sensing single protein binding. As the chelatorcomplex of NTA can serve as docking sites for the recombinant proteins with his-tags, such systems can beused as a recyclable platform for selectively detecting binding of charged proteins.

monoclocal AB

polyclocal AB with FITC

glycocalix

sialic acid residue

lectin with FITC

neuraminidase lectin

inside label

outside label

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 couplingof a first monoclonal mouse antibody and a second fluorescence labeled polyclonal goat anti-mouse IgGantibody (inside label, top). On the other hand, the extracellular glycocalix of erythrocyte membranes canbe detected by fluorescence labelled peanut agglutinin after cleaving sialic acid residues by neuraminidase(outside label, bottom).

Page 242: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 theincubation, the cytoplasmic domains of Band 3 were identified with inside label. However, the membranepatches did not fuse with each other even after prolonged incubation time. The patches that might correspondto the ruptured ghosts (∼ 145 μm2) are highlighted by white circles. Noteworthy, outside labelling resultedin 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 proteinsare the same as those in native cells, as well as the asymmetric orientation oftransmembrane proteins to the membrane surface is strictly controlled by nature.

In our recent study [114], right-side-out (RSO) human erythrocyte ghost membraneswere 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 cellulosederivatives. The resulting membrane orientation was identified with selective fluores-cence markers (Fig. 42). The extracellular glycocalix of erythrocyte membranes can bedetected by fluorescence labeled peanut agglutinin (outside label), while the cytoplasmicdomain of transmembrane Band 3 protein can be recognized with a specific coupling ofa first monoclonal mouse antibody and a second fluorescence labeled polyclonal goatanti-mouse IgG antibody (inside label). When RSO erythrocyte ghosts were incubatedwith planar glass cover slides, no adsorption or rupture of erythrocytes could be ob-served. To increase the interfacial attraction between cells and the surface, two types ofhydrated polymer films were deposited on the glass cover slides. On poly-lysine films,patches of the ruptured membranes could be observed, exposing their cytoplasmic sideto 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 prolongedincubation time. This might be attributed to the strong attraction between negativelycharged glycocalix and positively charged poly-lysine surfaces, which “pin” the adheredmembrane patches [103,115]. Actually, such a stable, strong coupling (de-wetting) hadbeen often reported for the strongly adsorbed polyelectrolytes on oppositely chargedsurfaces [116,117]. It is also well worth noting that the outside label did not yieldany fluorescence signals, suggesting all of the immobilized membrane patches inverted

Page 243: Nanotechnology and Nano-Interface Controlled Electronic Devices

226 M. Tanaka

spreading

rupture

adhesion

RSO Ghost

polymer support

Fig. 44: Schematic illustration of the orientation selective immobilization of erythrocyte membranes on theplanar surface: (1) approach of vesicles, (2) adhesion to the surface, (3) rupture of vesicles, and (4) lateralspreading and membrane fusion.

their orientation (Fig. 44). On the other hand, RSO ghosts were likely to coat the sur-face of cellulose films more continuously (Fig. 45). The immune-fluorescence labelingdemonstrated that immobilized erythrocyte membrane selectively inverted their nativeorientation, where the cytoplasmic side of the membrane is exposed to the bulk elec-trolyte. Tentatively, we interpreted this larger surface coverage on the cellulose film interms of the “weak attraction” between the cell surface glycocalix and the polysaccha-rides [38] (ref. Section 2). Such ultrathin (thickness 5–10 nm), biological polysaccharidefilms have a large potential to immobilize native cell membranes without denaturingtheir structure, membrane orientation, and functions. Further challenges along this linewill be the electrical detection of membrane proteins on semiconductor devices as

Page 244: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 withlectin. Compared to the immobilization on poly-lysine films, the surface coverage was larger and morehomogeneous.

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. electricfield, chemical potential gradient) to fabricate “protein arrays” on semiconductor devicesurfaces (Fig. 46).

Acknowledgements

I am sincerely indebted to E. Sackmann for his warm encouragement and fruitfulcomments while carrying out these studies. I am also grateful to my collaborating

Page 245: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 toG. Abstreiter’s group (Walter–Schottky Institute, Tech. Univ. Munich, fabrication andcharacterization of semiconductor nano-structures) and T. Bolom and S. Veprek (Tech.Univ. Munich, XPS). The works presented in Section 2 were carried out under thecollaborations with G. Wegner (Max-Planck Institute for Polymer Research, cellulosechemistry) 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. Gegeand R.R. Schmidt (Univ. Konstanz), and X-ray diffraction studies were carried out withthe aid of R. Zantl. I am also thankful to K. Lim and G.G. Fuller (Stanford Univ.) forISR 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, andNational Science Foundation (NSF) through the Center on Polymer Interfaces andMacromolecular Assemblies (CPIMA). Last but not least, I am grateful to JapanSociety for Promotion of Science (JSPS) and Alexander von Humboldt Foundation forthe postdoctoral fellowship and DFG for the Habilitation fellowship (Emmy NoetherProgram).

References

1. E. Sackmann and M. Tanaka, Trends Biotechnol. 18, 58 (2000).2. A. Ulman, An Introduction to Ultrathin Organic Films: From Langmuir–Blodgett to Self-Assembly

(Academic Press, 1991).3. M.D. Porter, T.B. Bright, D.L. Allara, and C.E.D. Chidsey, J. Am. Chem. Soc. 109, 3559 (1987).4. C.E.D. Chidsey and D.N. Loiacono, Langmuir 6, 682 (1990).5. C. Miller and M. Grätzel, J. Phys. Chem. 95, 5225 (1991).6. E. Sabatani, J. Cohen-Boulakia, M. Bruening, and I. Rubinstein, Langmuir 9, 2974 (1993).7. H.O. Finklea, D.A. Snider, J. Fedyk, E. Sabatani, Y. Gafni, and I. Rubinstein, Langmuir 9, 3660

(1993).8. R. Maoz and J. Sagiv, J. Coll. Inter. Sci. 100, 465 (1984).9. E.P. Pluedemann, Silane Coupling Agents (Plenum Press, 1990).

10. S.R. Wasserman, Y.T. Tao, and G.M. Whitesides, Langmuir 5, 1074 (1989).11. A.N. Parikh and D.L. Allara, J. Phys. Chem., 98, 7577 (1994).12. 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).13. N. Balachander and C.N. Sukenik, Langmuir 6, 1621 (1990).14. P. Fountain, D. Goguenheim, D. Deresmes, D. Vuillaume, M. Garet, and F. Rondelez, Appl. Phys.

Lett. 62, 2256 (1993).15. C. Boulas, J. V. Davidovits, F. Rondelez, and D. Vuillaume, Phys. Rev. Letters 76, 4797 (1996).16. O.M.-R. Chyan and K. Rajeshwar, J. Electrochem. Soc. 132, 2109 (1985).17. S. Gritsch, P. Nollert, F. Jähnig, and E. Sackmann, Langmuir 14, 3118 (1998).18. H. Hillebrandt, G. Wiegand, M. Tanaka, and E. Sackmann, Langmuir 15, 8451 (1999).19. H. Hillebrandt and M. Tanaka, J. Phys. Chem. B 105, 4270 (2001).20. P. Baumgartner, C. Engel, G. Abstreiter, G. Böhm, and G. Weimann, Appl. Phys. Lett. 66, 751

(1995).

Page 246: Nanotechnology and Nano-Interface Controlled Electronic Devices

Living systems and semiconductor nano-structures 229

21. P. Baumgartner, C. Engel, G. Abstreiter, G. Böhm, and G. Weimann, Appl. Phys. Lett. 64, 592(1994).

22. S. Menezes and B. Miller, J. Electrochem. Soc. 130, 517 (1983).23. I. Uhlendorf, R. Reinecke-Koch, and R. Memming, J. Phys. Chem. 100, 4930 (1995).24. Z. Hens and W.P. Gomes, J. Phys. Chem. B 103, 130 (1999).25. S.R. Lunt, G.N. Ryba, P.G. Santangelo, and N.S. Lewis, J. Appl. Phys. 70, 7449 (1991).26. K. Asai, T. Miyashita, K. Ishigure, and S. Fukatsu, Surf. Sci. 306, 37 (1993).27. O.S. Nakagawa, S. Ashok, C.W. Sheen, J. Märtensson, and D.L. Allara, Jpn. J. Appl. Phys. 30, 3759

(1991).28. C.W. Sheen, J.X. Shi, J. Martenson, and D.L. Allara, J. Am. Chem. Soc. 114, 1514 (1992).29. E.A. Miller and G.L. Richmond, J. Phys. Chem. B 101, 2669 (1997).30. K. Adlkofer, M. Tanaka, H. Hillebrandt, G. Wiegand, E. Sackmann, T. Bolom, R. Deutschmann, and

G. Abstreiter, Appl. Phys. Lett. 76, 3313 (2000).31. K. Adlkofer and M. Tanaka, Langmuir 17, 4267 (2001).32. 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).33. 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).34. K. Adlkofer, W. Eck, M. Grunze, and M. Tananka, J. Phys. Chem. B, in print (2003).35. H.J. Gabius and S. Gabius, Glycoscience (Chapmann & Hall, 1997).36. W.D. Comper, Extracellular Matrix (Harwood Academic Publishers, 1996).37. G. Elender and E. Sackmann, J. Phys. II France 4, 455 (1994).38. R. Bruinsma, A. Behrisch, and E. Sackmann, Phys. Rev. E. 61, 4253 (2000).39. S. Löfas and B. Johnson, J. Chem. Soc., Chem. Commun. 21, 1526 (1990).40. J.M. Harris, Poly(ethyleneglycol) Chemistry (Plenum Press, 1992).41. H.J. Taunto, C. Toprakcioglu, L.J. Fetters, and J. Klein, Nature 332, 712 (1988).42. S.I. Jeon and J.D. Andrade, J. Colloid Interface Sci. 142, 159 (1991).43. S.I. Jeon, J.H. Lee, J.D. Andrade, and P.D. de Gennes, J. Colloid Interface Sci. 142, 149 (1991).44. A. Halperin, Langmuir 15, 2525 (1999).45. Z. Yang, J.A. Galloway, and H. Yu, Langmuir 15, 8405 (1999).46. 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).47. S.J. Sofia, V. Premnath, and E.W. Merrill, Macromolecules 31, 5059 (1998).48. B.J. Herren, S.G. Shafer, J. van Alstine, J.M. Harris, and R.T. Snyder, J. Colloid Interface Sci. 115,

46 (1987).49. E. Kiss and C.G. Gölander, Coll. Surf. 49, 335 (1990).50. N.L. Burns, J.M. van Alstine, and J.M. Harris, Langmuir 11, 2768 (1995).51. F. Rehfeldt and M. Tanaka, Langmuir, in print (2003).52. W. Wegner, Thin Solid Films 216, 105 (1992).53. M. Schaub, G. Wenz, G. Wegner, A. Stein, and D. Klemm, Adv. Mater. 5, 919 (1993).54. V. Buchholz, G. Wegner, S. Stemme, and L. Ödberg, Adv. Mater. 8, 399 (1996).55. G. Mathe, A. Albersdörfer, K. R. Neumaier, and E. Sackmann, Langmuir 15, 8726 (1999).56. S. Kobayashi, S. Iijima, T. Igarashi, and T. Saegusa, Macromolecules 20, 1729 (1987).57. T. Lehmann and J. Rühe, Macromol. Symp. 142, 1 (1999).58. T.G. Bassiri, A. Levy, and M. Litt, J. Polym. Sci. B 7, 463 (1969).59. M.C. Woodle, C.M. Engbers, and S. Zalipsky, Bioconjugate Chem. 5, 493 (1994).60. F. Rehfeldt, L. Pagnoni, R. Jordan, and M. Tanaka, Langmuir (2002), in print.61. A.S. Lee, A.P. Gast, V. Butun, and S.P. Armes, Macromolecules 32, 4302 (1999).62. W. Curatolo, Biochim. Biophys. Acta 906, 137 (1987).63. P.R. Crocker and T. Feizi, Curr. Opin. Struct. Biol. 6, 679 (1996).64. A. Varki, Proc. Natl. Acad. Sci. USA 91, 7390 (1994).65. J. Vogel, G. Bendas, U. Bakowsky, G. Hummel, R.R. Schmidt, U. Kettmann, and U. Rothe, Biochim.

Biophys. Acta 1372, 205 (1998).

Page 247: Nanotechnology and Nano-Interface Controlled Electronic Devices

230 M. Tanaka

66. S. Hakomori, J. Biol. Chem. 265, 18713 (1990).67. M.F. Schneider, G. Mathe, M. Tanaka, C. Gege, and R.R. Schmidt, J. Phys. Chem. B 105, 5178

(2001).68. H. Möhwald, in Structure and Dynamics of Membranes, edited by R. Lipowsky and E. Sackmann

(Elsevier, 1995).69. S. Alexander, J. Phys. II France 38, 983 (1977).70. P.G. De Gennes, J. Phys. 37, 1445 (1976).71. S.T. Milner, T.A. Witten, and M.E. Cates, Macromolecules 21, 2610 (1988).72. S.T. Milner, T.A. Witten, and M.E. Cates, Europhys. Lett. 5, 413 (1988).73. M.S. Schneider, R. Zantl, C. Gege, R.R. Schmidt, and M. Tanaka, Biophys. J. (2003), in print.74. A. Razatos, Y.-L. Ong, M.M. Sharma, and G. Georgiou, Proc. Natl. Acad. Sci. USA 95, 11059

(1998).75. Y.-L. Ong, A. Razatos, G. Georgiou, and M.M. Sharma, Langmuir 15, 2719 (1999).76. M. Arnordi, M. Fritz, E. Bäuerlein, M. Radmacher, E. Sackmann, and A. Boulbitch, Phys. Rev. E

62, 1034 (2000).77. W. Xu, P.J. Mulhern, B.L. Blackford, M.H. Jericho, M. Firtel, and T.J. Beveridge, J. Bacteriol. 178,

3106 (1996).78. X. Yao, M. Jericho, D. Pink, and T. Beveridge, J. Bacteriol. 181, 6865 (1999).79. B. Warburton, Curr. Opinion Colloid Int. Sci. 1, 481 (1996).80. C.F. Brooks, G.G. Fuller, C.W. Franck, and C.R. Robertson, Langmuir 15, 2450 (1999).81. M.F. Schneider, K. Lim, G.G. Fuller, and M. Tanaka, Phys. Chem. Chem. Phys. (2002), in print.82. C. Gege, G. Bendas, J. Vogel, U. Rothe, and R.R. Schmidt, Chem. Eur. J. 6, 111 (2000).83. L.K. Tamm and H.M. McConnell, Biophys. J. 47, 105 (1985).84. E. Sackmann, Science 271, 43 (1996).85. R. Merkel, E. Sackmann, and E. Evans, J. Phys. France II 50, (1989).86. E. Kalb, S. Frey, and L.K. Tamm, Biochim. Biophys. Acta 1103, 307 (1992).87. J. Nissen, S. Gritsch, G. Wiegand, and J. Rädler, Eur. Phys. J. B 10, 335 (1999).88. C. Miller, P. Cuendet, and M. Grätzel, J. Electroanal. Chem. 278, 175 (1990).89. D.W. Urry, D.F. Mayers, and J. Haider, Biochemistry 11, 487 (1972).90. H. Tournois, J.A. Killian, D.W. Urry, O.R. Bokking, J. de Gier, and B. de Kruijff, Biochim. Biophys.

Acta 905, 222 (1987).91. J.A. Killian, K.U. Prasad, D. Hains, and D.W. Urry, Biochemistry 27, 4848 (1988).92. M. Bouchard and M. Auger, Biophys. J. 65, 2484 (1993).93. A.A. Brian and H.M. McConnell, Proc. Natl. Acad. Sci. USA 81, 6159 (1984).94. P. Chan, M.B. Lawrence, M.L. Dustin, L.M. Ferguson, D.E. Golan, and T.A. Springer, J. Cell Biol.

10, 245 (1991).95. P.B. Contino, C.A. Hasselbacher, J.B.A. Ross, and Y. Nemerson, Biophys. J. 67, 1113 (1994).96. R. Naumann, A. Jonczyk, R. Kopp, J. van Esch, H. Ringsdorf, W. Knoll, and P. Gräber, Angew,

Chem. Int. Ed. Engl. 34, 2056 (1995).97. J. Salafsky, J.T. Groves, and S.G. Boxer, Biochemistry 35, 14773 (1996).98. G. Puu, E. Artursson, I. Gustafson, M. Lundström, and J. Jass, Biosens. Bioelectron. 15, 31 (2000).99. D. Savage, G. Mattson, S. Desai, G. Nielander, S. Morgensen, and E. Conklin, Avidin–Biotin

Chemistry: A Handbook (Pierce Chemical Company, 1992).100. G.A.M. Cross, Annu. Rev. Cell Biol. 6, 1 (1990).101. A.L. Beauchamp, J. Israeli, and H. Saulnier, Can. J. Chem. 47, 1269 (1969).102. L. Schmitt, C. Dietrich, and R. Tampé, J. Am. Chem. Soc. 116, 8485 (1994).103. B.S. Jacobson and D. Branton, Science 195, 302 (1976).104. O. Purrucker, H. Hillebrandt, K. Adlkofer, and M. Tanaka, Electrochim. Acta 47, 791 (2001).105. S.M. Sze, Semiconductor Devices: Physics and Technology, (John Wiley & Sons, 1985).106. P. Fromherz, A. Offenhäuser, T. Vetter, and J. Weiss, Science 252, 1290 (1991).107. 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).108. M. Stenberg, H. Arwin, and A. Nilson, J. Colloid Interface Sci. 72, 255 (1979).

Page 248: Nanotechnology and Nano-Interface Controlled Electronic Devices

Living systems and semiconductor nano-structures 231

109. A. Sinner and A. Offenhaeusser, Thin Solid Films 327, 758 (1998).110. J.P. Dilger, S.G. McLaughlin, T.J. McIntosh, and S.A. Simon, Science 206, 1196 (1979).111. P. Fromherz, V. Kiessling, K. Kottig, and G. Zeck, Appl. Phys. A 69, 571 (1999).112. H. Hillebrandt, M. Tanaka, and E. Sackmann, J. Phys. Chem. B 106, 477 (2002).113. M. Seufert, F. Christo, and G. Wegner, Adv. Mater. 7, 52 (1995).114. M. Tanaka, S. Kaufmann, J. Nissen, and M. Hochrein, Phys. Chem. Chem. Phys. 3, 4091 (2001).115. N.E. Thomas, W.T. Coakley, and C. Winters, Coll. Surf. B 6, 139 (1996).116. T. Afshar-Rad, A.I. Baily, P.F. Luckham, W. MacNaughtan, and D. Chapman, Biochim. Biophys.

Acta 915, 101 (1987).117. G. Decher, H.D. Hong, and J. Schmitt, Thin Solid Films 210–211, 831 (1992).

Page 249: Nanotechnology and Nano-Interface Controlled Electronic Devices

This Page Intentionally Left Blank

Page 250: Nanotechnology and Nano-Interface Controlled Electronic Devices

Nanotechnology and Nano-Interface Controlled Electronic DevicesEditors: M. Iwamoto, K. Kaneto and S. Mashiko© 2003 Elsevier Science B.V. All rights reserved

CHAPTER 13

Structural color forming system composedof 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] National Institute of Advanced Industrial Science and Technology,

Hirate-cho 1-1, Kita-ku, Nagoya 462-8510, Japanc Fujinomiya Research Laboratories, Fuji Photo Film co., LTD.,

Fujinomiya Shizuoka 418-8666, Japan

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2332. Self-organized two-dimensional patterning by α-helical block-copolypeptides . . . . . . 2363. Two-dimensional structural color formation and regulation using polypeptide LB

films . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252

1. Introduction

Fig. 1 shows an example of an intelligent function of a living system. Living cellssense their environment as a stimulus and respond to environmental changes for themaintenance of life. For example, a physical or chemical stimulus is accepted by thesignaling cell and the induced signal is transmitted to the nerve system with a nerveimpulse, 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 atbiological interfaces such as the biological membranes.

30 years ago, Singer and Nicolson [1] predicted that biological membranes are anassembly of molecular machines, each responsible for an elementary function. This ideahas been reasonably accepted now. And it is recognized that the biological membranesare a self-organized system composed of amphiphilic molecules such as lipids andmembrane proteins, and the specific location and/or molecular orientation of theseelements are essential to lead the intelligent functions at the biological interfaces.

Page 251: Nanotechnology and Nano-Interface Controlled Electronic Devices

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, andmore recently, the formation of molecular membrane systems composed of α-helicalpolypeptides is also investigated.

For example, Kimura et al. [2] reported on a α-helix regularly standing system as aself-assembled monolayer (SAM) on Au substrate (Fig. 2). Higashi et al. [3] showed thepreparation of a α-helix rich PLGA monolayer and its enantioselectivety.

For the last few years, we [4] have tried to construct novel artificial membrane orinterfacial systems using polypeptides and their derivatives as a very simple modelpolymer of membrane proteins. And at the same time we proposed two approaches formimicking the intelligent functions of biological membrane systems (Fig. 3). One isfunctional modeling such as stimulus–response coupling, information transfer, and so

Page 252: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 membranecomponents such as membrane proteins and the surface and/or interface structure ofliving systems. Based on this concept, we will show here two types of polypeptidemembrane systems (Fig. 4). One is the self-organized two-dimensional patterning in anamphiphilic helix monolayer. And the other is the polypeptide Langmuir–Blodgett (LB)

Page 253: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 oftwo-dimensional structure. And for the latter, structural color formation and regulationwill be shown.

2. Self-organized two-dimensional patterning by α-helical block-copolypeptides

A monolayer was firstly prepared by a diblock-copolypeptide composed of hydrophilicand 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-helixstructure was confirmed by circular dichroism measurement of the TFE solution ofthe polymer. So this polymer, as schematically shown in Fig. 5, is composed of ahydrophobic larger-diameter helix and a smaller-diameter and longer hydrophilic helixrod.

The monolayer was prepared by spreading a DMF/benzene mixed solution of thepolymer on a water surface at pH 5 (Fig. 6). And the surface pressure was monitored bycompressing the monolayer by the Wilhelmy method.

Fig. 7 shows the surface pressure–area isotherm obtained at pH = 5. The limitingarea was estimated to be ca. 6 nm2/molecule. The calculated limiting area when thehelix is normal to the air/water interface is shown in Table 1. We expected the normalorientation of the helix after compressing the membrane, however, the big differencebetween the calculated and observed limiting area indicated that it is impossible toobtain a standing helix monolayer using this system.

Page 254: Nanotechnology and Nano-Interface Controlled Electronic Devices

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-benzyloxy-carbonyl L-lysine)25–poly[(γ -methyl L-glutamate)42/(L-glutamic acid)18]).

Fig. 6: Surface pressure measurements.

Page 255: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 aqueoussolution at pH 5.

Table 1

Limiting area estimated from π–A isotherm.

Observed (nm2/molecule) Calculated (nm2/molecule)ApH=5 A⊥

PLLZ25–P(MLG42/LGA18) 6.26 2.39

So the monolayer was transferred onto a mica surface at 25 mN/m to get an LB filmwith a single layer. And the morphology of the LB film was observed by atomic forcemicroscopy (AFM). Fig. 8 shows the AFM image of the LB film on μm scale. Fromthe depth of the cavity that was made by scratching with a cantilever, the thickness ofthe membrane was estimated to be ca. 1.4 nm. This value is almost consistent with thediameter 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 thickand thin layers was observed. And the difference of the thickness was ca. 0.3 nm. Thisvalue is almost consistent with the difference between the radii of the hydrophilic andhydrophobic helix rods. This may mean, therefore, that the thick layer is a molecular

Page 256: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 hydrophilicsmaller helix. The interval of the stripes was estimated to be ca. 24 nm. This valueis almost the twice the length of this polymer. These results suggest that this diblock-copolypeptide aggregates by head to head and tail to tail, resulting in the formation of ananophase-separated structure as is shown in Fig. 10.

Page 257: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 filmon 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 domaina branching pattern was also made. Fig. 11(b) shows a schematic illustration of thespeculated monolayer structure. We think that the hydrophobic interaction and hydrogenbonding between helix rods promote the phase-separated regular pattern, however, thelarge 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

Page 258: Nanotechnology and Nano-Interface Controlled Electronic Devices

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) andits schematic illustration (b).

of hydrophobic, hydrophilic and hydrophobic helix segments (Fig. 12). We expectedthat this dumbbell shape will block the development of the branching pattern. Thehydrophobic segment is poly(L-leucine) and the hydrophilic segment is poly(L-glutamicacid). The stable α-helix structure was confirmed by FT-IR measurements of the LBfilm on Au surface. In this case, the monolayer on aqueous solution at pH = 4.0 wastransferred on an Au surface for the FT-IR measurements. So this polymer in theacidic condition, as schematically shown in Fig. 12, is composed of a hydrophobiclarger-diameter helix and a smaller-diameter hydrophilic helix.

And then, the monolayer on aqueous solution at pH = 4.0 was transferred onto amica surface to get the LB film with the single layer for AFM measurements. Fig. 13(a)

Page 259: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 patterncomposed 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-separatedstructure of the monolayer is schematically shown in Fig. 13(b). As is expected, thetriblock-copolypeptide made a more sophisticated stripe pattern with very few branchingstructures.

We think that this is a good example showing a relation between the molecular shapeand character of the polypeptides and their self-organized two-dimensional structure. Itis expected, therefore, that the size of this pattern can be regulated by the degree ofpolymerization of each helix and the functions might be controllable by the side chainstructure of the α-helix units.

Thus, we showed the possibility of nanometer-size control of two-dimensionalstructure for the formation of novel functional interfaces.

3. Two-dimensional structural color formation and regulation using polypeptideLB 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 displaysof portable telephones, computer games, and displays of sound systems in automobiles.Furthermore, it is expected that the novel display systems should answer the problemsof 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 colorof butterflies, beetles, and tropical fish and shells. The structural color or interferencecolor, in this case, is based on a nanometer-scale layered structure of the body surface,without any pigments.

Page 260: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 itsschematic illustration (b).

When light with a certain incident angle (α) is reflected at the surface layer, it canbe emphasized or weakened depending on the wavelength (λ) of the light (Fig. 14). Theλ-value of emphasized or weakened light is given by these two equations [6].

λ = 2h

m

√n2 − sin2 α (emphasized) (1)

λ = 4h

2m −1

√n2 − sin2 α (enfeebled) (2)

where h is thickness of the corresponding layer, n is refractive index of the layer, andm is natural number (m = 1,2, . . .). For α-helix LB films, in our experimental range ofh,m ≤ 2, these two equations can be written as these equations, where d is the diameter

Page 261: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 theparticular color from the surface layer according this optical principle.

Structural color materials are also investigated and applied as a color coatingtechnique for ceramic surfaces, such as for example the structural color formation ofoxidized titanium [7]. The temperature, in this case, can control the thickness of theoxidized 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, wehave tried to construct structural color LB films of stimulus–response type as a noveldisplay system [10]. We are using, in this case also, α-helical polypeptide, because thepolypeptide can produce a well-defined layered structure on the substrate by the LBmethod, as we showed above.

The preparation and characterization of polypeptide LB films were investigated byWegner’s group [11–14] using polypeptides with long alkyl side chains (Fig. 15). Theycall it “hairy-rod”. And they showed the supra-molecular structure and optical propertiesof 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, evenat room temperature. This will have a negative influence on the optical properties, sowe selected poly(n-hexyl L-glutamate) (PHeLG, Fig. 16), which has not so long alkylchains, as the LB film component. In this case, the stable α-helix conformation of thispolymer was also confirmed by CD measurements.

The monolayer was prepared by spreading a DMF/benzene mixed solution of thispolymer on a water surface. And the surface pressure was monitored by compressing themonolayer by the Wilhelmy method at 20°C. Fig. 17 shows the surface pressure–areaisotherm 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 isparallel to the air/water interface. So the monolayer at 15 mN/m was transferred onto asilicon substrate. The substrate is hydrophobized before use by coating a silane-couplingagent with octadecyl moiety and heat-fixed at 110°C. Fig. 18 shows the relation between

Page 262: Nanotechnology and Nano-Interface Controlled Electronic Devices

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/watersurface.

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 thesubstrate, the second was up mode, and then this repeated alternately. Each depositionratio is ca. 1, indicating the monolayer was transferred onto the substrate keeping thestructure 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 arefour major peaks: a side chain C=O peak, two amide I bands associated with the α-helixstructure and the β structure, respectively, and amide II of the α-helix structure. Theα-helix amide I band is much larger than that of the β-structure, indicating PHeLG keptthe 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) showsan AFM image of the 20 layers LB film with the boundary between the top of the 20layers and the surface of the substrate itself. The difference in height between them is

Page 263: Nanotechnology and Nano-Interface Controlled Electronic Devices

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.

Page 264: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 substratesurface. (b) Section of the line in (a).

ca. 32 nm (Fig. 20(b)). This value is corresponding to 20 layers. Therefore, the thicknessof one layer is estimated to be ca. 1.6 nm. This value is almost consistent with thediameter (1.5 nm) of PHeLG obtained from the limiting area of the π–A isotherm ofthis 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 filmsshowed various colors depending on the number of layers; brown at 40 layers, dark blueat 80 layers, yellow at 120 layers, and red at 160 layers.

As a next step, the color formation behavior could be quantitatively analyzed by thereflective VIS spectra of these LB films. Fig. 22 shows reflective VIS spectra of the LBfilms with the incident light angle at 10°. The 80 layers film shows a peak at 418 nm

Page 265: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 at456 nm corresponding to brown or dark orange which is the complementary color toblue. The 120 layers film shows a peak at 619 nm corresponds to yellow. As a result,

Page 266: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 numberof layers. For example, in the case of 40 layers there is one emphasized peak and oneenfeebled peak. And for 80 layers, there are two emphasized peaks and two minimumpeaks. The solid lines were obtained by Eq. 1, i.e., λ of the emphasized light, and thedashed 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 measuredpeaks. Furthermore, Fig. 24 shows observed and calculated peak positions as a functionof incident light angle. The open circles and squares are maximum peaks and the filledcircles and squares are minimum peaks of these layers, respectively. And the lines Aand B were obtained using Eqs. 1 and 2 under the condition A and under the conditionB. The condition B yields the same calculated line B. The observed peak positionsare almost consistent with the calculated values for all the incident light angles. Theseresults suggest that we can get a structural color forming system by polypeptide-basedLB films.

Thus, we obtained a structural color forming system based on the structural modelingapproach in Fig. 3. So, we tried to implement stimulus–response coupling functions inour prepared structural color forming system.

First trial is photo-regulation of the structural color. For that, azobenzene containingpolypeptides were prepared. It is expected that the photoisomerization of the azobenzenemoiety 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 -methylL-glutamate) containing 28% azobenzene side-chains showed photo-induced changes inthe reflective spectrum of the LB film, ca. 10 nm peak shift to lower wavelength. It wasalso 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 isan example of a chemo-responsive structural color forming system. According to the

Page 267: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 thelayer was estimated by the experimental value of the amount of sorption of dioxane tothis LB film, ca. 5% when the relative vapor pressure of dioxane is 0.2 [15]. And therefractive index of the solvent is almost the same as that of the LB matrix. As a result, Itwas shown that the color should change from the yellow to the red region. We directlyobserved the color change of the LB film in the sorption chamber. The vapor pressurewas controlled by the temperature. As a result, we could observe the expected colorchange, yellow to pink, thus confirming the solvent-induced color change based on theswelling of the LB film.

On the other hand, methanol shifted λmax to lower wavelength. Opposite to the shiftfrom dioxane. In this case, methanol is a poor solvent, so there is no significant swellingof the layer, so we think that the lower refractive index of the methanol may induce theshift of λmax to lower values.

Page 268: Nanotechnology and Nano-Interface Controlled Electronic Devices

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°.

Page 269: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 changelike litmus paper. We are now trying to construct such stimulus–response LB films usinglight, temperature and chemicals as a stimulus (Fig. 27). This type of novel display andnovel sensing system will be more sophisticated in 1–2 years, based on the fundamentalresults today.

Acknowledgements

This study has been supported by the New Energy and Industrial Technology Develop-ment Organization (NEDO), Japan, for the project on Technology for Material StructureControl.

References

1. S.J. Singer and G.L. Nicolson, Science, 175, 720 (1972).2. Y. Miura, S. Kimura, Y. Imanishi, and J. Umemura, Langmuir 14, 6935 (1998).3. N. Higashi, T. Koga, Y. Fujii, and M. Niwa, Langmuir 17, 4061 (2001).4. T. Kinoshita, Prog. Polym. Sci. 20, 527 (1995); J. Photochem. Photobiol. B 42, 12 (1998).5. H. Yokoi, T. Kinoshita, Y. Tsujita, and H. Yoshimizu, Chem. Lett., 1210 (2000).6. E. Hecht and A. Zajac, Optics (Addison-Wesley, Reading, MA, 1974).7. S. Sakka, K. Kamiya, and T. Yoko, ACS Symposium Series 360 (American Chemical Society,

Washington, DC, 1988) pp. 345–353.8. K. Naitoh, Y. Ishii, and K. Tsujii, J. Phys. Chem. 95, 7908 (1991).9. M. Hayakawa, T. Onda, T. Tanaka, and K. Tsujii, Langmuir 13, 3595 (1991).

10. T. Kinoshita, S. Hayashi, and Y. Yokogawa, J. Photochem. Photobiol. A 145, 101 (2001).11. G. Duda, A.J. Schouten, T. Arndt, G. Lieser, G.F. Schmidt, C. Bubeck, and G. Wegner, Thin Solid

Films 159, 221 (1988).12. 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).13. A. Mathy, K. Mathauer, G. Wegner, and C. Bubeck, Thin Solid Films 215, 98 (1992).14. M. Büchel, Z. Sekkat, S. Paul, B. Weichart, H. Menzel, and W. Knoll, Langmuir 11, 4460 (1995).15. T. Kinoshita, N. Okazaki, A. Takizawa, and Y. Tsujita, Polymer 20, 791 (1979).

Page 270: Nanotechnology and Nano-Interface Controlled Electronic Devices

Nanotechnology and Nano-Interface Controlled Electronic DevicesEditors: M. Iwamoto, K. Kaneto and S. Mashiko© 2003 Elsevier Science B.V. All rights reserved

CHAPTER 14

Generation of a strong dipole layer and itsfunction 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 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2532. Primary amphiphilicity and orientation on water . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2543. Molecular orientation of helical peptide SAM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2564. In situ FTIR-RAS measurement on subphase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2585. Helical peptide monolayer spread at limited area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2636. Helical peptide monolayer spread on a mixture of water and methanol . . . . . . . . . . . . . . 2647. Multilayer formation on gold . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2648. Surface potential of helical peptide SAM. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2659. Photocurrent generation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26510. Future aspects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267

1. Introduction

α-Helix is one of the major secondary structures found in proteins. The molecular shapeof the helix is cylindrical, and the two ends, the N and C terminals, have an oppositekind of partial charge to each other. The N terminal is charged positively because theamide protons point to the N terminal. On the other hand, the C terminal is chargednegatively 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 haspointed out that the large dipole moment of the helix should contribute to the proteinfunctions such as binding a substrate to the active center and electron transfer in thephotosynthetic 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, andthe π–A isotherm was measured. It was found that helix axes in the monolayer lay

Page 271: Nanotechnology and Nano-Interface Controlled Electronic Devices

254 S. Kimura et al.

flat on the water surface. Upon compression, change from monolayer to bilayer wasobserved [4]. These phenomena are commonly observed with other helical peptides, andit is a general understanding that it is a hard task to obtain helical peptide layers with avertical orientation. There are several factors which influence the orientation of helicalpeptides 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 hydrophobicityin the molecule, is presumably the most reasonable strategy for obtaining a monolayerwith a vertical orientation. With respect to this point, naturally occurring biologically ac-tive peptides tell us many things. There are many naturally occurring peptides which arecomposed both of hydrophilic and hydrophobic amino acid residues in regular spatialarrangements. The arrangement is classified into two types: primary and secondary am-phiphilicity. The former type has a block property where hydrophilic and hydrophobicresidues form each cluster and are arranged in series. On the other hand, the latter typeshows hydrophilic and hydrophobic surfaces in the molecule when it takes a secondarystructure such as α-helix. It has been pointed out that amphiphilicity of peptides iscrucial in the biological activity, for example, by taking an active conformation forbinding to a receptor. An interesting and useful idea, which correlates the amphiphilicityof peptides with the biological activity, has been proposed [5]. The idea is called themembrane compartment concept, and describes in detail the way of induction of activeconformation on the basis of the amphiphilicity and other properties of the peptideespecially when the peptide binds to a cell membrane [6]. According to the concept, aninterface provides a gradient environment to induce specific orientation of the peptide.For example, a helical peptide with primary amphiphilicity will be incorporated intoa membrane surface as if the peptide feels a torque force due to the tendency of thehydrophilic part to be exposed to aqueous compartment and the hydrophobic part to beburied in the hydrophobic core of the cell membrane [7]. Under certain conditions, thehelical peptide therefore takes a vertical orientation against the surface.

An air/water interface is also regarded as gradient environment in terms of am-phiphilicity where the hydrophilic part of a primary amphiphilic peptide tends to bedistributed to the water subphase and the hydrophobic part to air. On the basis ofthis 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 ofsarcosine (hydrophilic residue) with ammonium. The hydrophobic part was composedof 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 thepeptides because of favorable intermolecular interaction for obtaining regular molecularassemblies. All the peptides took α-helical conformation in solution as found by CDmeasurement. These peptides were spread on water, and the orientation of the helicalpeptides 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 aliquid state to a solid state [10]. Observation of the peptide monolayer by fluorescence

Page 272: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 apeptide monolayer on water.

Page 273: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 liquidcondensed 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 orientationof 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 molecularsectional area along the helix axis. Some peptides, AdmA16OH and HA16FB, showeda slow slope with a relatively low collapse pressure in the π–A curve, suggesting thatthe 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 ofthe peptide by connecting crown ether at the C terminal (Fig. 3). 18-Crown-6-ether iswell known to form complex selectively with K+ ions. It was expected that the crownether part would form a complex on water subphase containing KCl and would becomehydrophilic to promote vertical orientation of the hydrophobic peptide part. However,the π–A curves of the peptide–crown ether conjugates indicated parallel orientation tothe surface even after complexation with K+.

These results strongly suggest that there should be other factors which influencethe orientation of helical peptides on water subphase. After many attempts, the authorshave succeeded in obtaining a helical peptide monolayer with a vertical orientationby taking other factors into consideration. Before stating that, however, self-assembledmonolayers (SAMs) of helical peptides on gold are explained in the following section toclarify 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 formationof Au–S bonds. The assembling property of the alkane part is very important for obtain-

Page 274: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 peptidemonolayer on water.

ing a well-packed monolayer free from defects. Helical peptides having lipoic acid atthe N terminal, Lipo-(Ala-Aib)8-OBzl (Lipo and OBzl represent lipoic acid and benzylester, respectively; LipoA16B) and Lipo-(Lys(Z)-Aib)8-OMe (Z and OMe representbenzyloxycarbonyl and methyl ester, respectively; LipoKZ16M), were synthesized andSAMs were prepared on gold (Fig. 4) [11]. The orientation of the helical peptides ongold was determined by FTIR-reflection absorption spectroscopy (RAS). The tilt anglesof 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 SAMon gold.

Page 275: Nanotechnology and Nano-Interface Controlled Electronic Devices

258 S. Kimura et al.

36° and 55°, respectively. The latter value means that the distribution of helix axes israndom. The difference in helix orientation between the two peptides should be ascribedto the difference in the size of the side chain between LipoA16B and LipoKZ16M. Thelarge side chain of LipoKZ16M should be less effective in promoting self-assemblingof helices with a vertical orientation. Indeed, the monolayer of Boc-(Lys(Z)-Aib)8-OMeon water stayed in a liquid condensed state upon compression, while the monolayerof Boc-(Ala-Aib)8-OMe formed a two-dimensional crystal. The highly self-assemblingproperty of the peptide itself is therefore an important factor for obtaining a SAM witha vertical orientation.

The self-assembling property of the peptide is affected by the solvent system inthe preparation of the SAM. For example, LipoA16B SAM, which was prepared froman ethanol solution of the peptide, showed a tilt angle of 36°, but that increasedto 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-assemblingproperty of the peptide probably by favorable solvation to the peptide. Therefore, thegood self-assembling property appearing in ethanol should contribute to the vertical ori-entation 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 theSAMs 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 normaldecreases in the order of Lipo-(Ala-Aib)8-OBzl > Lipo-(Ala-Aib)10-OBzl > Lipo-(Ala-Aib)12-OBzl. The smallest tilt angle was obtained with the 24mer peptide. The higherself-assembling property of the longer peptide should be due to the larger Van der Waalsinteraction to promote a tight packing.

In the cases of helical peptide SAMs, the orientation of the helix on gold is thereforedetermined by several factors such as the type of component amino acid, solvent forpreparation, and chain length. On the other hand, the helical peptides, even those thattook 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 thereaction energy to increase the peptide density as high as possible.

What is then the factor which promotes horizontal orientation of hydrophobic helicalpeptides on water? The orientation of helical peptides on water is generally evaluatedfrom the π–A isotherm comparing the molecular area with the sectional area of themolecule. This method is rather indirect to get information on molecular orientation. Toovercome the ambiguity of the evaluation method, we have used in situ FTIR-RAS todetermine the molecular orientation because that method is powerful to obtain directlythe tilt angle from the surface normal. The principle has been already clarified andreported by several researchers. The application of this method to the helical peptidemonolayer 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 insitu FTIR-RAS. The tilt angle of the helix axis from the surface normal is, therefore,

Page 276: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 spectrawhich are obtained by using polarized incident light (s or p). The theoretical ratio ofthe band intensities is simulated for the peptide layer with a defined tilt angle, andis compared with the experimental value, giving the tilt angle of the helix axis onsubphase.

For in situ FTIR-RAS measurements, a trough with a small surface area with amonolayer/grazing angle accessory attached is used, and a wire-grid polarizer is set justbefore the subphase surface to obtain the polarized incident light. First, the backgroundspectrum of the subphase itself is collected, and then the peptide monolayer is preparedon the subphase. After equilibrium, the sample spectrum is collected and transformed tothe absorbance using the background spectrum.

Spectral simulation is carried out by using computer programs principally accordingto the mathematical formalism developed by Ohta and Ishida [12]. This formalism isbased on the Abelès matrix method [13] describing stratified layers of homogeneousfilms, and includes appropriate modifications to interpret absorbing and anisotropicproperties of the layers [14].

An optical model of a three-phase (air/monolayer/water) system is considered. Theoptical property of the j th medium is described by the anisotropic complex refractiveindex,

n jc = njc + ikjc (1)

where i2 = −1, nj and kj are the refractive index and the extinction coefficient of thej th medium, respectively, and c represents x , y, z coordinates. Although air ( j = 0) is anon-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 indicesare, thus, expressed as Eq. 2:

n0x = n0y = n0z = n0 = n0, n1x = n1y �= n1z , n2x = n2y = n2z = n2 (2)

In the formalism, the relation between the amplitudes of the electric fields of the incidentwave E+

0 , reflected wave E−0 , and transmitted wave after the monolayer E+

2 is expressedas Eq. 3:(

E+0

E−0

)= C0,1 · C1,2

t0,1 · t1,2

(E+

2

E−2

)(3)

with

C0,1 =[

exp(−iδ0) r0,1 exp(−iδ0)r0,1 exp(iδ0) exp(iδ0)

]

C1,2 =[

exp(−i δ1) r1,2 exp(−i δ1)r1,2 exp(i δ1) exp(i δ1)

](4)

where r j−1, j and tj−1, j are the Fresnel reflection and transmission coefficients, respec-tively, between the ( j −1)th and j th medium. These coefficients for s- and p-polarized

Page 277: Nanotechnology and Nano-Interface Controlled Electronic Devices

260 S. Kimura et al.

light are given by Eqs. 5 and 6, respectively:

for s-polarized light:

r j−1, j = n j−1 y cos θj−1 − n j y cos θj

n j−1 y cos θj−1 + n j y cos θjtj−1, j = 2n j−1 y cos θj−1

n j−1 y cos θj−1 + n j y cos θj(5)

for p-polarized light:

r j−1, j = n j−1 x cos θj − n j x cos θj−1

n j−1 x cos θj + n j x cos θj−1tj−1, j = 2n j−1 x cos θj−1

n j−1 x cos θj + n j x cos θj−1(6)

The complex refractive angles θj in different media are related to the incidence angle θ0

by Snell’s law:

for s-polarized light: n0 sinθ0 = n j y sin θj

for p-polarized light: n0 sinθ0 = n j z sin θj (7)

The term δj represents the phase thickness of the j th medium, and is expressed as Eq. 8:

for s-polarized light: δj = 2πνdj n j y cos θj

for p-polarized light: δj = 2πνdj n j x cos θj (8)

where ν and dj represent the wave number of the incident light in the vacuum and thethickness of the j th medium, respectively. δ0 is equal to zero because air is the initialsemi-infinite medium. E−

2 is equal to zero in Eq. 3 because there is no reflection in thewater subphase. By taking the matrix product as Eq. 9:

C01 · C02 =[

a bc d

](9)

the reflection coefficient (r ) and reflectance (R) of the system are given by Eq. 10:

r = E−0

E+0

= c

a= r0,1 exp(−i δ1)+ r1,2 exp(i δ1)

exp(−i δ1)+ r0,1r1,2 exp(i δ1)R = |r |2 (10)

On the other hand, the reflection coefficient (r0) and reflectance (R0) of the watersubphase are easily obtained from the Fresnel formula:

r0 = n0 cosθ0 − n2 cos θ2

n0 cosθ0 + n2 cos θ2R0 = |r0|2 (11)

The absorbance of the system is, therefore, given as Eq. 12:

A = − log10R

R0(12)

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 literaturevalues [15]. The anisotropic extinction coefficient of the peptide monolayer along the caxis k1c(ν) (c = x , y, z) was calculated by using a uniaxial symmetrical model, in whichhelical 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 andII bands are expressed by two components, parallel (w) and perpendicular (u) to the

Page 278: Nanotechnology and Nano-Interface Controlled Electronic Devices

Generation of a strong dipole layer and its function 261

helix axis [16]. The wave numbers, maximum extinction coefficients, and peak widths athalf-height of the transition dipole moments of the amide I and II bands were taken fromliterature values which have been reported with those of poly(γ -benzyl-L-glutamate) ina film [17,18]. The maximum values of k1c(ν) of helical peptide with a tilt angle φ fromthe 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 )+ 1

2 f sin2 α]

(13)

k1zmaxTDM = kmaxTDM[

13 (1− f )+ f cos2 α

](14)

with the orientational distribution order parameter:

f = 12 (3cos2 φ −1) (15)

where α represents the tilt angle of each transition dipole moment component from thehelix axis, and this value is 0° for the transition parallel to the helix axis and 90° forthe transition perpendicular to the helix axis. Each extinction coefficient over the wholewave number is expressed as an antisymmetrical linear combination of two Lorentzianfunctions [19]:

k1cTDM(ν) = k1cmaxTDM(fwhh/2)2

(ν −ν0)2 + (fwhh/2)2− k1cmaxTDM(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 ofthe absorption band. The refractive indices are obtained from the Kramers–Kronigtransformation of Eq. 16:

n1cTDM(ν) = n∞ − k1cmaxTDM(ν −ν0)(fwhh/2)

(ν −ν0)2 + (fwhh/2)2+ k1cmaxTDM(ν +ν0)(fwhh/2)

(ν +ν0)2 + (fwhh/2)2(17)

where n∞ is the constant refractive index in the near-infrared region and was set to be1.50 for the simulations. The sum of the extinction coefficients and refractive indicesfor the each transition dipole moment gives the complex refractive index of the peptidemonolayer along the c axis:

k1c(ν) = k1cIw(ν)+ k1cIu(ν)+ k1cIIw(ν)+ k1cIIu(ν)

n1c(ν) = n1cIw(ν)+n1cIu(ν)+n1cIIw(ν)+n1cIIu(ν)

n1c(ν) = n1c(ν)+ ik1c(ν) (18)

The simulated RAS spectra of a helical peptide monolayer for s- and p-polarizedlight with varying the tilt angle of the helix axis from the surface normal are shown inFigs. 5 and 6, respectively. These simulations were performed with taking the incidenceangle being 65° because a larger angle makes the reflected light from the water surfacestronger and 65° is the maximum value possible for our experimental apparatus. Underthis condition, the reflectance of s-polarized light is about 10 times larger than that ofp-polarized light. In order to increase the sensitivity, measurements using s-polarizedlight were, therefore, chosen for the determination of the molecular orientation. Therelationship between the intensity ratios of the amide I and II bands and the tilt angles

Page 279: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 helicalhexadecapeptide monolayer with a thickness of 3 nm in the case of s-polarized light being used for themeasurement. 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 exceptp-polarized light is used for the measurement.

Page 280: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 surfacenormal and the intensity ratio of the amide I and II bands in in situ FTIR-RAS measurements on watersubphase 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 inthe simulations is set to be 3.0 nm, which does not influence the ratio of the bandintensities.

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 gasphase, they took a parallel orientation even after compression. The conventionalmethod for preparation of monolayers was not successful for obtaining helical peptidemonolayers with a vertical orientation even though hydrophobic peptides with acylindrical shape were used. However, the peptides took a vertical orientation when thepeptide was deposited in the liquid phase instead of the gas phase in the preparationof the monolayer from the beginning. For example, the intensity ratio of the amideI and II bands in an in situ FTIR-RAS spectrum was 1.47 under the condition thatBoc-(Leu-Aib)8-OBzl was spread at the molecular area of 1.5 nm2. According to therelationship between the intensity ratio and the tilt angle, Boc-(Leu-Aib)8-OBzl wasfound to stand up on water with a tilt angle of 28°. The helical peptides may be forced totake a vertical orientation and to be packed as densely as possible during the evaporationprocess of the spreading solvent [20].

Page 281: Nanotechnology and Nano-Interface Controlled Electronic Devices

264 S. Kimura et al.

Among the three peptides, Boc-(Leu-Aib)8-OBzl stood up more vertically on thesubphase than the other longer peptides. For example, the tilt angle of Boc-(Leu-Aib)16-OBzl monolayer was 39°, which is larger than the 28° of Boc-(Leu-Aib)8-OBzlmonolayer. In the preparation process, a large number of rod-shaped molecules werespread over a limited area. Under this condition, strong intermolecular interaction of thelonger peptides may work locally among the peptides, which may hinder an orderedalignment 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 thatthere must be an important factor which contributes to the parallel orientation. Moleculeswith dipole moment deposited on gold or water should experience an electrostaticattractive force as if there were an image dipole at the same distance on the other side ofthe interface. The attractive force should act to bring the parallel orientation, because theinteraction energy with the image dipole becomes larger with decreasing the distance ofthe dipole from the interface. However, this interaction energy will become smaller withlowering the dielectric constant of the substrate or subphase, because the image dipole isproportional to (εs − εm)/(εs + εm) (εs and εm represent dielectric constants of substrateor subphase and monolayer, respectively). Therefore, using subphase having a lowerdielectric constant may promote the formation of vertically oriented monolayer. On thebasis of this idea, helical peptides were spread on a mixture of water and methanol (1/1v/v).

The tilt angles of helical peptide monolayers spread on a mixture of water andmethanol were 23° for Boc-(Leu-Aib)8-OBzl and 33° for Boc-(Leu-Aib)16-OBzl, whichindeed 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 replacementof the subphase and influence the orientation, the reduction of the dipole–image-dipoleinteraction is considered to contribute significantly to the vertical orientation of thehelical peptides on the subphase. This interpretation is also applied successfully to thecase of multilayer formation on gold.

7. Multilayer formation on gold

The helical peptide monolayers on subphase were transferred on gold by evaporationof the subphase which was directly put on gold [20]. With repeating the evaporation ofthe subphase deposited with a peptide monolayer, multilayers of helical peptides wereobtained on gold as shown by the increase of the amide I band intensity by FTIR-RASmeasurement. The intensity ratio of the amide I and II bands measured on gold wasused to determine the tilt angle of the helix from the surface normal [21]. The effect ofthe subphase solution on the tilt angle was significant. The average tilt angle of 10-layerBoc-(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

Page 282: Nanotechnology and Nano-Interface Controlled Electronic Devices

Generation of a strong dipole layer and its function 265

image dipole should be diminished by using a mixture of water and methanol to inducethe vertical orientation.

The tilt angle was dependent on the number of helical peptide layers on gold. Thetilt 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 verticalorientation of the peptide monolayer on the subphase was affected probably because ofthe 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 transferringpeptide layer with the image dipole in gold to promote the vertical orientation. Thedriving force to align the peptides horizontally therefore becomes insignificant for thethick multilayer.

8. Surface potential of helical peptide SAM

The helical peptide SAMs with a vertical orientation are theoretically indicated togenerate a surface potential due to the large dipole moment. The surface potentials ofLipoA16B and BA16Lipo SAMs in vacuum were measured by the Kelvin probe methodto be about −120 mV and 400 mV, respectively [22]. LipoA16B was connected to goldvia the N terminal and exposed the C terminal to the outside to generate the negativesurface potential. On the other hand, the positive surface potential was generated byexposing the positively charged N terminal to the surface. Therefore, the helix peptidelayer with a parallel and vertical orientation indeed generates a surface potential due tothe dipole moment. However, the observed values were lower than the calculated valuesbased purely on the dipole moment. The difference is successfully explained by twoways depending on the direction of the dipole moment in the SAM. In the case of thepeptide SAMs immobilized via the N terminal, the effects of the mutual depolarizationand the ionic property of Au+–S− linkage should work on the surface potential to reducethe absolute values. Whereas, the peptide SAMs immobilized via the C terminal maypromote electron transfer from gold to the peptide layer due to the large positive surfacepotential. The generation of the surface potential based on the dipole moment should becompromised by the electron transfer. The effect of the electron transfer should becomelarger with increase of the dipole moment. Indeed, similar positive surface potentialswere observed independently of the peptide layer thickness as long as the dipoles arealigned to point to the vacuum side.

9. Photocurrent generation

Helical peptides may mediate electrons through hydrogen bonds with combination of thethrough space mechanism. Such electron transfer should be accelerated in the presenceof 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 thanthat against the dipole moment [23]. A helical peptide carrying an N -ethylcarbazolegroup at the C terminal and a disulfide group at the N terminal was synthesized, and a

Page 283: Nanotechnology and Nano-Interface Controlled Electronic Devices

266 S. Kimura et al.

Fig. 8: Schematic illustration of photocurrent generation by the helical peptide SAM. Anodic current in thepresence of an election donor in the aqueous phase was observed with the peptide SAM where the dipolemoment is directed to the gold.

peptide SAM was prepared on gold. Interestingly, an anodic photocurrent was observedin the presence of an electron donor (ethylenediamine tetraacetate or triethanolamine) inthe aqueous phase (Fig. 8), but a cathodic photocurrent was not clearly detected uponphotooxidation of the N -ethylcarbazole by an electron acceptor (methylviologen) in theaqueous 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 aqueousphase, but an anodic photocurrent was small in the presence of triethanolamine in theaqueous phase. The direction of the electron transfer coincides with that of the dipolemoment, indicating that the dipole moment is crucial for determining the direction ofthe photocurrent.

10. Future aspects

A surface potential of a few hundred mV was generated by helical peptide SAMshaving 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 isaccelerated or hindered depending on the direction of the electron transfer for or againstthe electric field in the case of photocurrent generation using the helical peptide SAM inaqueous solution. On the basis of these results, further study is progressing on electric

Page 284: Nanotechnology and Nano-Interface Controlled Electronic Devices

Generation of a strong dipole layer and its function 267

properties of a single helical peptide immobilized on gold. The helical peptide SAMsare now subjected to scanning probe microscopies such as STM, STS, and KFM. Theeffect of the dipole moment of the helical peptide on the electron transfer through asingle peptide molecule is now under investigation. The helical peptide may be a goodcandidate for a molecular device which shows a diode property.

References

1. A. Wada, Adv. Biophys. 9, 1 (1976).2. W.G. Hol, Prog. Biophys. Mol. Biol. 45, 149 (1985).3. S. Kimura, In Handbook of Surfaces and Interfaces of Materials, edited by H.S. Nalwa, (Academic

Press, San Diego, 2001) pp. 207–231.4. F. Takeda, M. Matsumoto, T. Takenaka, and Y. Fujiyoshi, J. Colloid Interface Sci. 84, 220 (1981).5. R. Schwyzer, Biopolymers (Peptide Science) 37, 5 (1995).6. D.F. Sargent and R. Schwyzer, Proc. Natl. Acad. Sci. USA 83, 5774 (1986).7. R. Schwyzer, Biochem. 25, 6335 (1986).8. K. Fujita, S. Kimura, Y. Imanishi, E. Rump, and H. Ringsdorf, Langmuir 10, 2731 (1994).9. K. Fujita, S. Kimura, Y. Imanishi, E. Okamura, and J. Umemura, Langmuir 11, 1675 (1995).

10. K. Fujita, S. Kimura, Y. Imanishi, E. Rump, and H. Ringsdorf, Langmuir 11, 253 (1995).11. Y. Miura, S. Kimura, Y. Imanishi, and J. Umemura, Langmuir 14, 6935 (1998).12. K. Ohta and H. Ishida, Appl. Optics 29, 1952 (1990).13. F. Abeles, Ann. Phys. 3, 504 (1948).14. W.N. Hansen, J. Opt. Soc. Am. 58, 380 (1968).15. J.E. Bertie and M.K. Ahmed, J. Chem. Phys. 93, 2210 (1989).16. T. Miyazawa, J. Chem. Phys. 35, 693 (1961).17. T. Buffeteau, E. Le Calvez, B. Casteno, B. Desbat, D. Blaudez, and J. Dufourcq, J. Phys. Chem. B

104, 4537 (2000).18. T. Buffeteau, E. Le Calvez, B. Cesbat, I. Pelletier, and M. Pezolet, J. Phys. Chem. B 105, 1464

(2001).19. K. Ohta and H. Ishida, Appl. Spectrosc. 42, 952 (1988).20. K. Kitagawa, T. Morita, J. Umemura, and S. Kimura, Polymer 43, 3533 (2002).21. Y. Miura, S. Kimura, Y. Imanishi, and J. Umemura, Langmuir 15, 1155 (1999).22. Y. Miura, S. Kimura, S. Kobayashi, M. Iwamoto, Y. Imanishi, and J. Umemura, Chem. Phys. Lett.

315, 1 (1999).23. E. Galoppini and M.A. Fox, J. Am. Chem. Soc. 118, 2299 (1996).24. T. Morita, S. Kimura, S. Kobayashi, and Y. Imanishi, J. Am. Chem. Soc. 122, 2850 (2000).

Page 285: Nanotechnology and Nano-Interface Controlled Electronic Devices

This Page Intentionally Left Blank

Page 286: Nanotechnology and Nano-Interface Controlled Electronic Devices

Part D

Interfacial Dynamic Technology

Page 287: Nanotechnology and Nano-Interface Controlled Electronic Devices

This Page Intentionally Left Blank

Page 288: Nanotechnology and Nano-Interface Controlled Electronic Devices

Nanotechnology and Nano-Interface Controlled Electronic DevicesEditors: 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. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2712. Optical guided waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 274

2.1. Optical waveguide modes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2742.2. The field distributions of optical guided modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2772.3. Coupling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282

3. Liquid crystal waveguide geometries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2863.1. Fully guided geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2883.2. Fully leaky geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2913.3. Half-leaky guided mode geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2933.4. Improved fully leaky guided mode geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301

4. Dynamic guided mode technique . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3065. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 308Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 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 inflat panel displays having low power consumption and compact dimensions has occurredonly during the last 30 years. The worldwide demand for flat panel displays is hugeand continues to drive further scientific investigations in liquid crystal (LC) science andtechnology. This has resulted in developments in materials synthesis giving rise to novelmaterials and new discoveries in the fundamental science of liquid crystal phases. Inaddition there has been substantial new device structure development strongly pushedby 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

Page 289: Nanotechnology and Nano-Interface Controlled Electronic Devices

272 F. Yang and J.R. Sambles

this response, coupled with the optical anisotropy, which has led to applications inflat panel display technology, including alpha-numeric displays, high resolution TV,data projection systems and monitors for desk and laptop computers, etc. They alsochange their behaviour quite markedly with temperature and they are readily aligned bymolecular scale surface structures. Consequently LC materials are also being used assensors for temperature, stress and flow, etc. For many of these applications, especiallyfor displays, the optical behaviour of liquid crystals is vitally important and the opticalresponse 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 asrefractive index, dielectric permittivity, viscosity and elasticity coefficients, etc., changewith temperature, or how the apparent optical response of a thin LC layer changesunder application of a field. From both a fundamental and a device perspective wemay wish to investigate how the response of LC layers vary with surface treatment, thespecifics 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 changewith applied field is essential for developing a full understanding of device functionfrom which further devices and applications may arise. For example for fast devicesdetails of the electroclinic coefficients may be required, or viscosities may be needed.This may demand determining both the static and dynamic director response. In orderto characterise fully the optical properties of a given liquid crystal, in a given cellgeometry, with defined thickness and boundary treatments it is obvious that some formof appropriate structural study needs to be undertaken.

In many liquid crystal research laboratories the most common method used to explorethe optical structure within a LC cell is that of polarised microscopy [3]. This provides aquick and simple procedure for exploring the approximate director profile allowing thestudy of cell uniformity, defect structures, phase transitions, the influence of aligninglayers 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 thecell as a whole. Thus it cannot readily be used to resolve details of the director profilethrough the cell thickness or its variation with time during switching.

A second procedure, often used for the study of liquid crystals structures, even withthin cells (< 10 μm), is X-ray scattering [4]. In bulk materials the X-ray scatteringgives the symmetry of the phase and in thin layers it is primarily used to explore onlythe density wave layering found in SA, SC, S∗

C and other more ordered phases, forexample in the study of ferroelectric LC materials [4]. This method may then give veryuseful information on the elastic deformation of the density wave, the layers, but it saysalmost nothing directly as regards the optical properties of the LC cell, which are vitallyimportant for display applications.

At the same time as new display technologies using liquid crystalline materials werebeing developed in the early 1970s, new optical techniques were being introduced inthe broad area of light guiding. This is particularly obvious in the area of opticalfibre communications. Accompanying this development of optical fibre technology lightguiding in thin films and layered structures has also received considerable attention bothfor integrated optics and other applications. During these developments a new optical

Page 290: Nanotechnology and Nano-Interface Controlled Electronic Devices

Guided mode studies of liquid crystal layers 273

probing technique, the guided mode technique [5,6], for investigating the optical indexprofiles in guiding structures has materialised. This technique utilises the fact that aseries of discrete (quantised waveguide momentum) modes may be excited in thin filmsor 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 spectrumit is possible to characterise in some detail the optical properties of thin films or fibreguides. At the heart of LC displays are liquid crystals in the form of a thin layer, witha thickness of the order of a few microns with a director structure which varies littlelaterally, but may vary substantially through the cell. This layer is sandwiched betweentwo 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 crystaldisplay structures are not designed as optical waveguides they often are just that, albeitrather ‘leaky’ ones. So in parallel with the development of new liquid crystal deviceswe have seen the evolution of a range of optical waveguide characterization techniqueswhich allow the probing of the details of the director structure in these devices. It isthese new techniques which are the primary focus of this chapter and which will bediscussed later in detail.

In the early 70s, both guided wave optics (often narrowly labelled integratedoptics) and liquid crystal studies were in their infancy and, based upon the materialsyntheses and the requirements of display devices, much work centred simply on opticalmicroscopy of the nematic mesophase of liquid crystals. Theoretical modelling of thenematic phase was developed quite early and the director profile in waveguides madeof such nematics is quite simple, even in the case of finite surface tilt and twist inthe cell. Because of this, even though the optical guided mode technique has greatpotential 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 inaddition to polarised microscopy only the monitoring of the transmitted intensity fornormal incidence light together with modelling the optical response of liquid crystalcells based upon 2 × 2 Jones’ matrix are used as standard procedures in most liquidcrystal research laboratories. Often liquid crystal ‘waveguides’ are used as devices, asdisplays, in modulators [8,9], switches [10] or deflectors [11] but with limited studiesof details of the optical tensor distribution in cells except for a few measurements ofrefractive indices and surface pre-tilt angle.

However, several new areas of liquid crystal research have led to the guided modetechnique becoming more and more useful and important. Firstly, in the early 80s theprediction [12] and experimental observation [13] of the surface stabilised ferroelectric,S∗

C, liquid crystal (SSFLC) state introduced a challenge to unravel and understand thecomplex optical tensor profile in such LC cells. In view of the fact that thin cellscontaining 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 stillnot a convenient theoretical model, then the experimental unravelling of the directorstructures in such cells is vital. Secondly, even for some simpler phases, such asnematic or SA phases, the director structure through the whole cell is not necessarilystraightforward. For example there may exist coexisting nematic and SA phases ina twisted liquid crystal cell [14] in which the director profile through the LC cell

Page 291: Nanotechnology and Nano-Interface Controlled Electronic Devices

274 F. Yang and J.R. Sambles

is stepped in several sections, which can only be found in detail by the guidedmode techniques. Any other method, such as polarised microscopy or monitoring thetransmission intensity from normal incidence, cannot distinguish this stepped structure.Also from theoretical modelling and experimental demonstration the method based onmonitoring the transmission intensity from normal incidence and 2 × 2 Jones’ matrixwhich treats the LC cell as a simple slab may sometimes introduce serious errors inthe measured parameters [15]. By contrast the guided mode technique together withtheoretical modelling based upon the 4 × 4 Berreman’s matrix method, treating the LCcell as a real multi-layer optical system will give the correct results. Finally, in dynamicstudies of liquid crystal films the director profile and its change with time and field arevery complex. The serious profile degeneracy problems associated with the usual opticalprocedures may only be solved by use of the recently developed dynamic LC guidedmode technique [16,17].

In this chapter we first introduce briefly the background of the optical guidedmode technique, including the guided mode spectrum, the optical field distribution fordifferent order modes and various optical coupling methods to couple the radiationinto the waveguides. Then we illustrate four different types of liquid crystal waveguidegeometry, including the fully guided mode geometry, the fully leaky mode geometry, thehalf-leaky guided mode geometry and the improved fully leaky guided mode geometry.Finally we briefly discuss the dynamic LC guided mode technique. Various experimentalresults obtained recently using different types of LC optical guided mode techniques arepresented 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 ge-ometry comprising a planar slab of perfectly transparent isotropic dielectric surroundedby semi-infinite blocks of perfectly transparent isotropic dielectric. The geometry isinvariant in the plane normal to the page, i.e. the y–z plane as shown in Fig. 1. Thethree materials in the waveguide geometry have been labelled as the cladding layer, ofindex nc, the guiding layer, of index ng and the substrate layer of index ns. For full-

Fig. 1: The geometry of a planar optical waveguide, with cladding area, guiding layer and substrate area.

Page 292: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 ng > ns and ng > nc is required. For the ensuing discussionwe further suppose ns > nc, unless explicitly stated otherwise, so ng > ns > nc.

A ray optics picture is first used here to describe the optical modes [5] propagating inthe 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 theboundary of the geometry) to the interface, propagating inside the guiding layer asshown in Fig. 2. According to Snell’s Law (basically conservation of momentum in they–z plane) the critical angles at the top and bottom interfaces of the geometry are givenby

βc = sin−1

(nc

ng

)(1)

and

βs = sin−1

(ns

ng

)(2)

where βs > βc. In Fig. 2 there are three different zigzag pictures for different ranges ofthe internal angle β.

When βs < β < π/2 the light is primarily contained inside the guiding layer bytotal internal reflection at both the top and bottom boundaries. According to the rayoptics picture the light propagates along the zigzag path shown in Fig. 2a. This casecorresponds to a fully guided mode. Even if the ray is totally inside the guiding layer, theoptical radiation field is not entirely constrained to the slab since there are evanescentlydecaying fields present in both the substrate and cladding layers. However, if there areno imaginary parts to the refractive indices of the two layers the optical energy flowof the modes will be strictly along the z-direction and no radiation propagates away ineither of the semi-infinite blocks.

Secondly when βc < β < βs, as shown in Fig. 2b, only one interface, the claddinginterface, acts as a totally reflecting surface, part of the radiation energy of the modesescapes 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. Thistype of waveguide may be labelled as a half-leaky waveguide – it only leaks radiationinto one half space of the geometry.

Page 293: Nanotechnology and Nano-Interface Controlled Electronic Devices

276 F. Yang and J.R. Sambles

Finally for β < βc the radiation energy of the modes will leak into both the substrateand cladding half spaces across both interfaces as shown in Fig. 2c. Thus this is labelleda fully leaky waveguide and the confinement of radiation within the guided layer is quiteweak.

In all three cases described above, but primarily in the first case of fully guiding, sincethe light is propagating either entirely or partially in the guiding layer, then exploringmodes excited in the geometry will give information on the optical properties of theguiding layer. Different order modes, which propagate with different momenta alongthe z-direction in the waveguide geometry, will have different numbers of optical fieldmaxima across (in the x-direction) the guiding layer. Both the in-plane momentum andthe optical field distribution of different order modes are dependent upon the thicknessand refractive index profile of the guiding layer. Thus each mode will have a differentsensitivity 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 adescription of a mode with oblique-up and oblique-down beams, from the wave-natureof the light a sustainable mode propagating in the guiding layer is only one that does notdestructively interfere with itself. Thus there will only be a few angles of propagationfor the zigzag which give correct constructive interference, i.e. there can only be a finiteset 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 forconstructive interference, that is

kx = mπ

d(3)

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 correctlythe phase shifts (−2Φgs) and (−2Φgc) at the substrate and cladding interfaces a moregeneral equation can be given as

2kx d −2Φgs −2Φgc = 2mπ (4)

Of course kx is simply given through the relationship

kx = k0ng cosβ (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

kz =[

k20n2

g −(

mπ +Φgs(β)+Φgc(β)

d

)2]1/2

(6)

where 2Φgs and 2Φgc are also functions of k0 itself. This equation still gives a discreteset of guided modes but it is clearly not that trivial to solve. We will take this equationa stage further by replacing Φgs(β) and Φgc(β) by the appropriate functional forms

Page 294: Nanotechnology and Nano-Interface Controlled Electronic Devices

Guided mode studies of liquid crystal layers 277

found from Fresnel equations. According to the Fresnel equations it is clear that bothΦgs(β) and Φgc(β) will be polarisation sensitive and it is generally the situation thatthere will be two families of discrete momentum modes, one set transverse electric (TEor s-polarised) and one set transverse magnetic (TM or p-polarised).

For the situation of both half-leaky and fully leaky waveguides these standing wavemode 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 itis 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 thegeometry of the waveguide, there will still be some selected in-plane momenta at whichthere 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 spatialregions of a waveguide wave optics should be used and optical field profiles in aslab waveguide need to be explored. From Maxwell’s electromagnetic theory the twoimportant equations in isotropic (non-magnetic) lossless dielectrics are

∇ × E(r , t) = −μ0∂ H(r, t)

∂ t(7)

∇ × H(r, t) = ε0n2 ∂ E(r, t)

∂ t(8)

where ε0 and μ0 are the dielectric permittivity and magnetic permeability of free spacerespectively and n is the refractive index of the dielectric.

When a plane wave propagates along the z-direction (Fig. 1) with the propagationconstant γ (= kz) then the electromagnetic field may be expressed as

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 Ez = Hz = 0, ∂/∂ t ≡ iω;∂/∂z ≡ −iγ ; ∂/∂y = 0, gives us two independent solutions of the form

∂2 Ey

∂x2+ [k2

0 n2 −γ 2]Ey = 0 (11)

∂2 Hy

∂x2+ [k2

0 n2 −γ 2]Hy = 0 (12)

These two equations are the solutions for transverse electric and transverse magneticoptical fields, respectively.

To obtain the optical field distribution in the waveguide appropriate boundaryconditions have to be imposed to give Ey and Hy as functions of x . The boundaryconditions are conservation of tangential E and H and also conservation of normalD and B. For the waveguide geometry mentioned above the fields are not zero inthe 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

Page 295: Nanotechnology and Nano-Interface Controlled Electronic Devices

278 F. Yang and J.R. Sambles

non-infinite potential well.) For the TE case Ey and ∂ Ey/∂x are continuous across theboundary while for the TM case Hy and ∂ Hy/∂x are continuous across the boundary.The phase shifts at the interfaces are introduced by these continuity conditions leadingto 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

Φgs = tan−1

(δs

kx

)(13)

Φgc = tan−1

(δc

kx

)(14)

and for the TM case

Φgs = tan−1

[(ng

ns

)2δs

kx

](15)

Φgc = tan−1

[(ng

nc

)2δc

kx

](16)

where

δs = [γ 2 − k20n2

s

]1/2(17)

and

δc = [γ 2 − k20 n2

c

]1/2(18)

are the exponential decay coefficients for the evanescent fields in the substrate andcladding media, respectively. From Eqs. 17 and 18 it is clear that as γ continues todiminish (β continuously becoming smaller) δs and δc will become imaginary and onemoves over to the ‘leaky’ situation. When γ < k0ns the situation becomes a continuousspectrum in γ rather than the discrete values found for the trapped modes. The variousfield profiles of the electromagnetic modes for different ranges of the propagationconstant (in-plane momentum) are shown in Fig. 3 for the TE case. As expected it isclear 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 differentTE modes propagating in a planar waveguide with different propagation constants. (a) substrate-claddingradiation mode, γ < k0nc, (b) substrate radiation mode, k0nc < γ < k0ns, (c) guided mode with k0ns < γ <

k0ng.

Page 296: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 = 1and 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 theradiation mode range the optical fields of course propagate out into the cladding andsubstrate media. For the guided mode situation the field profiles produced from Eqs. 11and 12 for TE modes of order 0, 1 and 2 with the equivalent ray optics model are shownin Fig. 4.

From Eq. 6 it is apparent that the fundamental mode, with m = 0, has the largestpropagation constant γ (= kz), close to the limit value of a plane wave in the guidinglayer, ngk0. 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 thesituation is much the same. The next order mode, with m = 1, has a smaller propagationconstant than that of the mode m = 0 and will generally have zero optical field near tothe centre of the guiding layer as shown in Fig. 4. Thus it is clear that different orderoptical guided modes will have different field distributions through the guiding layer. Asshown in Fig. 4 the zero order mode will be much more sensitive to the centre of theguiding layer than would be the first order mode. Therefore it would be not difficult tosee that this guided mode technique applied to the study of a liquid crystal layer havinga complex director distribution through a cell may allow discrimination of details ofthe optical tensor varying through the layer. This information would be unobtainable byintegral (polarised microscopy, etc.) techniques.

According to the mode equation Eq. 4 it is also abundantly clear that the thickerthe guiding layer the more modes it will support. In addition more modes can alsobe excited for shorter probing wavelengths of the incident light unless there is verystrong optical dispersion of the guide indices with wavelength. Further details of opticalwaveguide 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 toisotropic, lossless and uniform materials, however, in an optical waveguide having

Page 297: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 willhold. Of the three assumptions that of low loss is generally not too worrying. If thelosses are not too large, as in liquid crystals in the optical range, they only changethe mathematics somewhat and make otherwise infinitely sharp modes available forexcitation and detection. However, the non-uniformity is of fundamental interest andit is the anisotropy which most significantly affects the discussion thus far. These twopoints are briefly discussed in the following.

In general the eigenmodes of a waveguide geometry having an anisotropic guidinglayer will not be pure TE or TM polarisations except for some special, high-symmetrystructures. Thus the complex eigenmode solutions can not be readily be simply analysedusing Maxwell equations, even though some methods based upon electromagnetic fieldtheory have been developed [19] to try to solve this problem. Here we will explore a fewsituations which allow us to illustrate the behaviour of an anisotropic waveguide.

Let us restrict the discussion to just uniaxiality in the guiding layer itself. Considera uniformly aligned uniaxial layer of liquid crystal as shown in Fig. 5 surrounded bysemi-infinite cladding and substrate media with isotropic refractive indices nc and ns,respectively. The liquid crystal is specified by indices parallel and perpendicular to thedirector, the optical axis N , ne and no, respectively, and we assume ne > no, i.e. theliquid crystal has positive anisotropy. For the general situation the director of the liquidcrystal layer is tilted by θ from the x-axis and twisted by ϕ from the xoy plane as shownin 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 theplane 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 issimple to see that the TE guided modes depend only on no, while the TM eigenmodes

Page 298: Nanotechnology and Nano-Interface Controlled Electronic Devices

Guided mode studies of liquid crystal layers 281

depend on both ne and no, with the lower order modes (with high β) depending mostlyon no. Thus for this simple case the effective index for the TM eigenmodes changeswith mode order. If now the optic axis is tilted so that it is still in the x–z-planenothing fundamentally changes. The TE eigenmodes will still sense no while the TMeigenmodes sense a different combination of ne and no. Secondly, if the optical axis liesinstead along the x-axis, then the lowest order TM mode will sense ne with the higherorder modes becoming more sensitive to no. Of course all order TE modes will stillsense no. In practice this means that when the optic axis is along the z-axis the TE andTM modes will have the same upper momentum limit, while for any other tilt of theoptic axis in the x–z-plane the limit of the TM modes will, for ne > no, move abovethe TE limit. Thirdly, if the optic axis lies along the y-axis, which is the simplest casefor the optical axis lying out of the incidence plane xoz, now the TE modes are givenby ne while the TM modes are given by no, a very simple situation. For these threespecial cases mentioned above the eigenmodes propagating in an anisotropic waveguidegeometry are pure TE or pure TM modes, even though their propagation constants mayvary.

However, as soon as the optic axis is rotated out of the y-axis to some arbitraryangle in the x–y plane, or to some arbitrary angle in the y–z plane, or both, i.e. theoptical axis is at a general position in the frame as shown in Fig. 5, the eigenmodesare no longer pure TE and TM. Thus an experimental investigation of such a systemusing radiation of a given linear polarisation, either TE or TM, will lead to polarisationconversion, the output radiation having some of the orthogonal polarisation componentpresent. The signals from the polarisation conversion are very useful for investigatingthe director structure of liquid crystal waveguides since they are so clearly sensitive toboth tilt and twist of the optical axis (the director) out of the plane of incidence of theexciting radiation.

Some analytical explanations can also be provided for the situation of polarisationconversion in the uniaxial guiding layer [20,21]. The geometry of Fig. 5 gives

cosψ = cosθ cosβ − sinθ sinβ sinϕ (19)

where ψ is the angle between the optical axis (the director) and the wave-front normalof 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 theplane perpendicular to the wave front AO and the index ellipsoid of the uniaxial layer.From Fig. 5 this gives

n′e = none√

n2o sin2 ψ +n2

e cos2 ψ

. (20)

Of course the extraordinary eigenmode index lies within the range ne ≥ n′e ≥ no and

will correspond to a pure TE or a pure TM eigenmode for the three special situationsmentioned 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 incidentradiation, two eigenmodes are excited in the uniaxial layer, one with the E field alongthe short semi-axis OB in the ellipse BOF, and normal to the plane AON, and a secondwith the E field along the major semi-axis OF in the ellipse BOF, and normal to OB.

Page 299: Nanotechnology and Nano-Interface Controlled Electronic Devices

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, conversionsignal when either pure s or pure p radiation enters the uniaxial layer. From Fig. 5 weobtain

cosΩ = sinβ cosθ + sinθ cosβ sinϕ√1− (cosθ cosβ − sinθ sinβ sinϕ)2

(21)

Obviously, only when Ω equals 0 or π/2 is there no polarisation conversion. Thiscorresponds to one of three special cases; (1) ϕ = π/2, i.e. the optic axis lying in theincidence 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 situationsmentioned above. Of course the amplitudes and the phases of different order guidedmodes, propagating in the uniaxial guiding layer are more complex than describedabove because of the interference between the reflections at the two interfaces. However,whether or not there is creation of polarisation conversion is still essentially correctlydescribed 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. Inaddition in real liquid crystal cells significant variations of the director twist and tiltthrough the layer will generally exist and these very important situations have to beconsidered. Thus for most investigation of liquid crystal waveguides simple analyticexpressions tend not to be utilised, instead full multilayer optics theory [22–25] is usedto model reflectivities, transmissivities and optical field profiles. This then allows theincorporation of the full optical tensor with a spatially varying (through the cell) directorprofile, allowing the prediction of optical response functions which may be used tocompare with data.

Up to now the optical waveguide has only been considered as isolated fromthe outside environment by infinite dielectric slabs. We now need to describe theexperimental procedures used to couple incident radiation into the guiding layer to allowa 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 claddingor substrate area unless the light is introduced either from the ends of the guide (end-coupling) or by some secondary mechanism such as fluorescence in the guiding layer.However, end-coupling or fluorescence excitation inside the guiding layer is highlyimpractical with many liquid crystal waveguides, thus some light-coupling mechanism,which will inevitably perturb the waveguide, has to be introduced to allow experimentalexploration of the waveguide.

For conventional procedures used in laboratories there are, broadly speaking, twomechanisms, prism or grating coupling, for coupling external radiation to the guidinglayer. Both give the possibilities of enhancing the momentum of the incident radiationalong the propagating direction of the guided waves, essential if the external radiation isto be coupled into a truly guided system. For the first, prism-coupling, the momentum ofthe incident radiation is enhanced by the refractive index of the prism. The momentum

Page 300: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 thegrating momentum.

In the prism-coupling procedure a high-index prism is generally used to excite theguided waves by providing the necessary phase matching between the evanescent fieldsof the incident radiation and a guided wave [26]. A geometry for this type of coupling isshown in Fig. 6. A high-index prism, ideally with np > ng, is put in close proximity toan air-clad waveguide and radiation is made incident at an incident angle β so that itsmomentum along the interface matches that of the guided mode, that is

kz = npk0 sinβ = γ ′ (22)

The propagation constant, γ ′, of the waveguide mode excited via evanescent couplingacross the air-gap thickness d is modified from its original value by the proximity ofthe coupling prism. Of course as the air-gap tends to infinity γ ′ tends to γ while thecoupling tends to zero. In most experiments the incident angle β is varied and thecoupling to the waveguide is monitored in some way, then the observed features whichcorrespond to resonant mode coupling can yield information on the mode structure. Asshown in Fig. 6 the true external incidence angle, β ′, against the entrance face of theprism is related to the internal angle β by the prism angle σ and the refractive index np

of the prism through

β ′ = sin−1

[np

n0sin(

90° −β − σ

2

)](23)

where σ is the apex angle of the symmetric prism and n0 ≈ 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 guidinglayer the angle σ satisfies

cos(σ

2

)= ng

np(24)

Thus for a typical guiding layer with an index of the order 1.55 and a prism of index1.80 a symmetric prism with apex angle of the order 60° may be used to couple theradiation into the waveguide. This is quite easy to fabricate.

Page 301: Nanotechnology and Nano-Interface Controlled Electronic Devices

284 F. Yang and J.R. Sambles

By the same evanescent mechanism which allows the radiation to be coupled intothe waveguide the prism can also couple the radiation out of the waveguide. Twoindependent 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 fewmicrons 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 andcoupling out in standard experiments.

Of course, for practical liquid crystal waveguides, apart from free-standing films, anair coupling gap is not practicable since a constraining wall is needed to contain theliquid 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 eithercase deposited by evaporation on to the base of the prism. For more convenience asimilar thin layer may be deposited on to a surface of a high index glass plate whichis then index matched with a high index matching fluid to the coupling prism. In thisfashion a planar liquid crystal cell can first be fabricated by a normal commercial-likeprocedure 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 nextsection.

The second important technique for coupling the incident radiation into the wave-guide is grating-coupling [6]. In this arrangement a grating, which may be an amplitudemodulation (surface grating) or a phase modulation (index grating), is used to giveextra 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 agrating surface modulated waveguide is shown in Fig. 7, in which the grating profileis located at the interface between the waveguide layer and the cladding, although thewaveguide–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 gratinggeometry having pitch (or wavelength), Λ and height g as shown in Fig. 7, then theincident optical field on the interface of the grating may be written

Ey = A{i [(k0nc sinβ)z − (k0nc cosβ)x −ωt]} (25)

where A is an amplitude coefficient and, for a sinusoidal grating, we have

x = g

2

(2π z

Λ

)(26)

Fig. 7: Geometry of a grating-coupled waveguide structure. The grating wavevector 2π/Λ adds momentumto the incident radiation allowing it to couple to the guided mode.

Page 302: Nanotechnology and Nano-Interface Controlled Electronic Devices

Guided mode studies of liquid crystal layers 285

It is clear that if the boundary between the cladding and guiding layer is a flat surfacewith x not a function of z, i.e. g = 0, the propagation constant in the z direction willsimply be k0nc sinθ which will not be big enough, according to the definition of the crit-ical angle for true waveguiding, to couple to guided modes. However, if g is finite thenat the grating boundary x is a function of z and for a sinusoidal shape of the grating then

Ey = A exp

{i

[(k0nc sinθ)z −

(g

2k0nc cosθ sin

2π z

Λ

)−ωt

]}(27)

and it is obvious that the propagation constant can no longer be readily extracted sincethere is no simple term just multiplying z in the equation. However, if we expand theabove equation in terms of Bessel functions as follows

Ey = A∞∑

m=−∞Jm

(−g

2k0nc cosθ

)exp

{i

[(k0nc sinθ +m

Λ

)z −ωt

]}(28)

then the problem can be overcome and, as expected, an infinite series of propagationconstants has been introduced in the grating coupling

kzm = k0nc sinθ +m2π

Λ(29)

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 thata single incident plane wave will perhaps produce several diffracted plane waves, as wellas local evanescent diffracted fields. The coupling mechanism of the grating-coupling isthat if we can find an integer m such that kzm is equal to the propagation constant of awaveguide mode then the mode matching condition will be satisfied and some incidentradiation may be coupled into this guided mode. Of course the radiation propagatingin the waveguide can also equally be coupled out of the guide, just as in the caseof evanescent prism coupling. As mentioned above for a planar structure the strengthof 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 thecoupling strength and with a non-planar boundary involved the optics is much moredifficult to model and to compare with experimentally recorded data.

In practice these two coupling methods mentioned above can also be combined togive prism–grating coupling systems, or, in addition, holographic couplers [27].

For completeness two other simple coupling methods should also be briefly intro-duced. One is the end-coupling method [28], in which radiation with a field profilesimilar in form to the field profile of a guided mode is fed into a waveguide through itsend-face. The mode propagation direction in the waveguide is normal to the end-face asshown in Fig. 8. By using a focusing lens the incident radiation coupling is localisedat the end of the waveguide. A high quality, defect free end-face, normally producedby polishing or cleaving, of the waveguide is required in this technique. Thus it will beextremely difficult to arrange a geometry to end-couple incident radiation into a liquidcrystal waveguide, even though such a procedure may easily be applied to an opticalfibre 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

Page 303: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 incidentbeam 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 beingradiative to being totally internally reflected in the guiding layer.

picture the incident radiation is coupled by total internal reflection with a zigzag pictureinto a thin waveguide terminated by a taper. Due to the slowly changing propagationconstant in the wedge section of the waveguide end the wavefront is not exactly a planewave. Because fabricating a cell to give appropriately graded faces to the liquid crystallayer is not particularly easy, this method is also mainly used to couple the radiation intoa optical fibre or some solid waveguide in integrated optics.

As mentioned above it is very clear that for the optical probing of liquid crystalwaveguides prism or grating coupling are by far the most appropriate techniques. Ofcourse prism-coupling is a simple and convenient method for investigating a fabricatedliquid crystal cell, however, grating coupling in some senses is also quite convenientsince the grating may be fabricated within the cell during construction and thus no extraoptical elements are needed in the study of the cell. In addition aligning liquid crystalsin the cell may be achieved by grating and there can be found several studies of theusing of grating/liquid crystal geometries [30–33]. However, the cost and the complexprocedure of grating fabrication, the extra complexity of the overall optical response ofthe cell when used as a device may be drawbacks of the grating-coupling technique. Inaddition the geometry of a liquid crystal waveguide with a grating structure is muchmore difficult to model optically by comparison to the rather simple prism-coupledplanar geometry. Since only a limited amount of quantitative work has yet to materialiseusing grating-coupling to detail the director profile in a liquid crystal cell we will largelyconfine 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 modespectra are sensitive to the parameters of the guiding layer including the profiles of

Page 304: Nanotechnology and Nano-Interface Controlled Electronic Devices

Guided mode studies of liquid crystal layers 287

the optical indices (tensor) through the layer and the thickness of the layer. This arisesparticularly from the fact that each different order mode will be sensitive to differentparts of the guiding layer through different optical field distributions for each the guidedmode. Of course, this sensitivity is only in one dimension, through the thickness of theguiding layer, but since a mono-domain of liquid crystal in a cell should be invariant inthe plane of the guiding layer, then this is exactly the sensitivity required for studyingliquid crystal thin films. In addition, comparing with other optical methods the guidedmode is the only technique which is going to yield the required spatial selectivitythrough the thickness of the guiding layer. Over recent years four somewhat differentguided mode geometries have been exploited by using prism-coupling to liquid crystalfilms [34].

For all of the four geometries which are described in more detail below, theessential experimental procedure comprises that of monitoring the angle dependentreflectivity and/or transmissivity of a plane parallel, monochromic linearly polarisedoptical beam, incident through a coupling-in prism at the glass/liquid crystal layerboundary. According to the optical waveguide models mentioned above, at certainangles of incidence the momentum of the incident radiation along the surface will matchthat of one of the guiding modes in the layered planar waveguide structure. Then ifthe geometry is appropriate there will be a reduction in the polarisation-conservingreflectivity at these angles. Of course some related change in the polarisation-conservingtransmissivity at these angles will also occur, if another coupling-out prism is used at thebottom of the waveguide structure. Hence by simply monitoring the reflectivity and/ortransmissivity as a function of angle of incidence we will find all the mode momentum,the momentum spectra, of the waveguide. These momentum spectra are generallyenough to determine the optical properties of the guiding layer for isotropic and losslessmaterials as may be used in normal integrated optics. However, for a waveguidestructure incorporating liquid crystals the situation is more complex. The momentumspectra are only a minimum set of information and although useful, will not readily givethe full director profile of the liquid crystal through the cell. If the director is twisted outof the plane of incidence or tilted from the surface the modes in the cell are neither pureTE or TM and the reflected and transmitted radiation will generally have a polarisationconverted component. This will not only create polarisation-conversion reflectivity andtransmissivity signals but it will also lead to more complex polarisation-conservingreflectivity and transmissivity spectra. More generally, instead of just discussing themode momentum, one monitors accurately the angle dependent reflectivity (and/ortransmissivity) over a wide range of angles and then fits these data to predictionsfrom a multilayer optical model of the geometry. In particular, using angle dependentpolarisation conversion reflectivity and/or transmissivity, the sensitivity to the directortwist/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 dependson the particular geometry studied, of which, as mentioned above, there are essentiallyfour.

Page 305: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 indexpyramid 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 refractiveindex cladding, a high index guiding layer and low index substrate. Although inprincipal any low index materials can be used as cladding and/or substrate, in generalmetal-clad waveguides have been the focus of attention in this geometry. There aretwo main reasons for using metal-clad liquid crystal waveguides in the fully guidedgeometry. One is that the metal layers may be used as the electrodes as well as providinga low refractive index (nreal < 1) and the other is that surface plasmons may be excitedat the interface between the metal and the alignment film/liquid crystals. With theirexponentially decaying optical fields these surface plasmon excitations can be used toexplore the director profile near the aligning surface of the liquid crystals. A typicalsample geometry for a metal-clad liquid crystal waveguide is illustrated in Fig. 10. Thisis a nearly symmetric metal-clad dielectric waveguide. In this geometry the thin, about30–50 nm, metal layer coated directly on to a high index pyramid, acts as both a mirrorfor trapping guided modes and as a tunnel barrier for radiation coupling. The thicknessof this top metal coating is critical and depends on the kind of metal and the wavelengthof the radiation. Too thick a metal layer will result in weak coupling to the guidedmodes, while too thin a layer will allow the guided modes to become more ‘leaky’ andthus 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 qualitymirror, if the transmission data are not needed. Of course, two aligning films, e.g. SiOX ,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 theliquid crystal layer the refractive index of the pyramid should be chosen to have a highervalue than any other layer in the geometry. Both high index glass and some high indexanisotropic crystal materials, e.g. sapphire, may be used as coupling prisms, althoughcare has to be taken in production to ensure the optic axis of any anisotropic crystal isorthogonal to the plane of incidence.

In this metal-clad waveguide geometry, provided the thickness of liquid crystallayer is greater than the cutoff thickness [6], which is nearly always the case for

Page 306: Nanotechnology and Nano-Interface Controlled Electronic Devices

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, abroader surface plasmon–polariton (SPP) resonance may also be excited [35] at themetal/alignment layer/liquid crystal interfaces using TM polarised incident radiation.While each guided mode is sensitive to different parts of the director profile throughthe cell, the TM polarised surface plasmon resonance will be sensitive to the director atthe 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 dictatedby the transverse optical index of the system and so allow further details of the directorprofile to be extracted. Hence it is very clear that only using this relatively simplereflectivity-monitoring technique the director profile inside a cell may be unravelled insome detail.

A significant amount of work has been done using this novel technique. For thesimple nematic phase, the fully guided technique has been first used to explore thedirector reorientation at the surface [37] allowing an estimation of the surface anchoringenergy [38]. The voltage response of a 90° twisted nematic (TN) cell has also beenexplored by the metal-clad waveguide technique [39], while, by the use of pulsedvoltages very small field-induced changes in the optical permittivity of a nematic havealso been investigated in some detail [40,41]. This has allowed the study of macroscopicdirector effects as well as other induced order-parameter effects. Other examples includethe study of bulk director reorganisation in a nematic having finite surface tilt [42], anda very sensitive measurement of the electro-optic pretransitional effects in the isotropicphase using a differential variant [43] of the direct waveguide technique.

Comparing with studying the rather simple nematic phase the guided mode techniqueis an even more powerful and important tool for experimentally unravelling the directorprofile of the more complex chiral smectic C (S∗

C) liquid crystal inside cells. Byobserving the guided mode spectra with the metal-clad fully guided geometry thedirector alignment in the ‘chevron’ structure of a surface stabilised ferroelectric liquidcrystal (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 alignedlayer of ferroelectric liquid crystal (FLC) SCE3 the angle dependent reflectivities havebeen 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 afunction of angle of incidence are shown fitted by multilayer optics modelling theory inFig. 11 [44]. A series of sharp resonance dips in the reflectivity are due to the excitationof fully guided modes in the guiding ferroelectric liquid crystal layer, while the broadresonant dip at about 66° is due to the SPP on the silver/SiOX/FLC interface. In the caseof 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 reflectivitycurve. 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 ofthe original nematic alignment direction as shown in Eq. 21. For the orthogonal plane ofincidence, Fig. 11b, some sharp guided modes appear in the region of the broader SPPresonance dip. This indicates that the twist of the primary director of the FLC is not toofar from the surface alignment direction since these s-like guided modes here (from the

Page 307: Nanotechnology and Nano-Interface Controlled Electronic Devices

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) perpendicularto the plane of incidence. The solid lines indicate the fit of the data (crosses) to theory. The cell walls, firstcoated 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 islargely dictated by the SiOX and the situation of the director near the surface. By usingmulti-layer optics, with the liquid crystal divided into a large number of sub-layers, tocarefully fit such angle dependent data in detail, the director profile in the SSFLC cell isfound to be largely a uniformly twisted slab, with a twist angle of 13° (from the originalnematic alignment direction) with rather limited tilts of less than 2°. Two thin regions oforder 100 nm are near both surfaces over which the director twists out from the originalalignment direction. This distribution of the optical director across the cell is in goodaccord 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 opticalstudies, such as the guided mode technique here, the layer information can only deducedfrom the director profile determined if the information about the cone angle of the FLCmaterial is also available.

Many fundamental research studies about SSFLC cells have been performed withthe metal-clad fully guided geometry, including studies of various SSFLC cells under

Page 308: Nanotechnology and Nano-Interface Controlled Electronic Devices

Guided mode studies of liquid crystal layers 291

applied AC and DC electric fields [45–47], half-splayed states [48], including adetermination of the low level of optical biaxiality [49]. Further studies have alsoexplored the homeotropically aligned state [50] finding for the smectic A phase that thesmectic layers are not necessarily parallel to the cell walls. In view of the high definitionof the optical reflectivity response, that is the resonant modes are relatively narrowin angle, the experiments mentioned above have been used to quantify in some detailthe director distribution through liquid crystal cells. Thus the metal-clad liquid crystalwaveguide technique is a powerful tool for studying the behaviour of liquid crystalinside cells, specially for the more complex ferroelectric phases. Unfortunately there areseveral limitations that inhibit the usefulness of the metal-clad fully guided techniquefor 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 thestrongly rubbed polymer alignment layers, which are found in most commercial cellson the transparent conducting electrodes (indium tin oxide – ITO) coatings. Even forthose situations where quite strong gold films have been used with polymer alignmentthe alignment created may be different to that with ITO coatings. Thirdly, the thin metaltunnel layer gives very different optical response for the two orthogonal polarisationdirections of the incident radiation, TE and TM. A thin metal layer, such as 40–50 nmof silver as typically used in the visible part of the radiation spectrum, tends to reflectstrongly TE radiation. This results in rather weak coupling to TE-like modes in theguiding layer and more especially gives a weak polarisation conversion signals. Since itis just these conversion signals that are particularly sensitive to the director twist and/ortilt, then its weakness limits the detailed determination of the director profile throughthe cells by this guided technique. A way needs to be found to explore more realisticdevice-like structures.

3.2. Fully leaky geometry

It is obvious that if the metallic layers are removed from the previous geometry than allthree limitations will be avoided. However this will inevitably mean losing the surfaceplasmon resonance, and also all of the guided modes will now become leaky. Now thesample geometry becomes that of a liquid crystal layer sandwiched between a pyramidand a glass substrate plate, both with a higher index than the primary index of the liquidcrystal and both coated with transparent ITO as a conducting film on top of which arethin transparent rubbed polymer aligning layers as shown in Fig. 12. The use of highindex pyramid means that there are still critical edges available to help in quantifyingthe refractive index tensor of the liquid crystal.

For this fully leaky geometry, when the incident angle β is smaller than the angle fortotal reflection between the pyramid and the liquid crystal layer, the incident light beamenters the liquid crystal layer and, at the substrate surface, it is partly reflected back intothe liquid crystal again with some of the radiation being refracted into the substrate. It isclear that true guided modes cannot be excited and supported in this geometry. Howeverwhen the electromagnetic wave reflected at the two liquid crystal/glass interfacessatisfies a constructive interference condition then a partial localisation of the radiation

Page 309: Nanotechnology and Nano-Interface Controlled Electronic Devices

292 F. Yang and J.R. Sambles

Fig. 12: Geometry for a fully leaky liquid crystal waveguide. The pyramid and substrate both have refractiveindices greater than the liquid crystal layer.

inside the liquid crystal layer will occur. Of course, these are fully leaky guidedmodes, with propagating radiation leaking out both into the substrate (transmission)and cladding area (reflection). Thus, if the angle dependent reflectivity is once againrecorded, then the sharp features obtained from the metal-clad waveguide are nowcompletely absent and a series of much broader resonance will be obtained. However, ina similar fashion to the true guided modes, the electromagnetic waves interfering withinthe multi-layer system, giving these poorly defined resonances, will also have their owndifferent field distribution across the liquid crystal layer. Hence a study of these fullyleaky modes should also provide details of the director profile through the cell. Thistype of study is an extension of conoscopy, measuring much higher internal angles ofincidence by using a matching fluid and coupling pyramid and also providing muchmore quantitative information concerning the director profile in a given cell.

This fully leaky geometry has also been used to explore the director alignment inSSFLC cells [51,52]. A typical set of reflectivity data, compared with a multi-layeroptics 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 caserubbed polyimide was used to provide the alignment layers. (From [51].)

Page 310: Nanotechnology and Nano-Interface Controlled Electronic Devices

Guided mode studies of liquid crystal layers 293

features recorded for the metal-clad fully guided geometry are completely absent andthe sensitivity of data fitting to the specifics of the director profile is much reduced. Thislimits the precision with which the technique may be used to determine the directortwist/tilt structure through the cell. However unlike the metal-clad waveguides there islittle constraint on the strength of the p to s conversion signal which may then be usedto 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 couplein radiation it may be applied to commercial-like cells. In addition it also gives good p tos 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 seriousprofile degeneracy problem arising from fitting model data to the wide modes obtainedexperimentally can be overcame. However before moving on to an improved fully leakyguided mode geometry we will first introduce another very powerful technique – thehalf-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-leakyguided mode geometry which avoids drawbacks from both of the above geometriesand approximates quite well to a real cell geometry, while at same time giving sharperresonant features and strong polarisation conversion signals.

The chosen geometry for the half-leaky guided mode (HLGM) technique is that ofa high index glass prism (which may for convenience of cell fabrication be replacedby a prism, matching fluid and a glass plate), ITO coating, rubbed polymer alignmentlayer, an aligned liquid crystal layer and a low index glass substrate with first ITOcoating and then alignment layer on it. It is the asymmetry of the glass index whichprovides the essential new function of this geometry, since now there are a range ofangles of incidence in the upper prism for which light is totally reflected at the liquidcrystal/low index glass substrate interface. Thus this interface acts as a perfect mirror(dielectric/dielectric interface beyond critical angle) over a certain angle range providinga half-guided or half-leaky guide system. For this new geometry the high index glassshould have an index, nc, which is greater than the highest index available in the liquidcrystal while the low index glass should have an index, ns, lower than the lowest indexavailable in the liquid crystal. The new half-leaky guided mode geometry is shown inFig. 14. For the convenience of discussion we suppose that the liquid crystal is positiveuniaxial with the extraordinary refractive index, ne, greater than the ordinary index,no. This means that the ideal condition of the new geometry is nc > ne and ns < no.According to the guided mode theory it is obvious that there can be no true guided wavesin the liquid crystal guiding layer with these conditions satisfied. However, analyticaltreatment, as well as numerical modelling, of this situation shows [54] that there isa special wavevector range in which there are strong polarisation conversion signalsin the reflectivity spectrum. This in-plane wavevector range is between k0ns to k0n′where n′ is the maximum effective index of the liquid crystal probed by the radiation,no < n′ < ne. When the angle of incidence (in-plane momentum) is in this window

Page 311: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 indexsubstrate.

the optical field will be evanescent in the substrate area since nck0 sinβ is greater thannsk0 (so it is beyond the prism/substrate critical angle) while it propagates in the liquidcrystal layer because nck0 sinβ is less than k0n′ (so it is below the prism/liquid crystalpseudo-critical angle). In this incident angle window the radiation reflected from theprism/liquid crystal boundary will interfere with that mirror-reflected from the liquidcrystal/substrate boundary producing sharp interference features, ‘resonances’, in theangle dependent reflectivity response. From the illustration mentioned above it is clearthat this special wavevector window may be quite narrow between βs = sin−1(ns/nc) andβ ′ = sin−1(n′/nc), where βe ≥ β ′ ≥ βo with βo = sin−1(no/nc) and βe = sin−1(ne/nc).Then by scanning the incident radiation over this narrow angle window the sharphalf-leaky guided mode reflectivity spectra will be experimentally recorded.

Comparing with the fully guided and fully leaky geometries the major advantages ofthis half-leaky guided mode geometry are obvious. Firstly, since the high index glassplate has been replaced by a low index one from the fully leaky geometry then over thelimited angle range the optical energy leaking into the substrate is zero. Thus the modesrecorded in this range are quite sharp, and are hence more sensitive to the details of thedirector profile than the fully leaky geometry. Secondly, and possibly more importantly,because there are no metallic coatings needed the test cell can be constructed followingthe commercial cell fabrication process with ITO and rubbed polyimide alignment.Thirdly and also associated with the absence of metallic coatings, the polarisationconversion signal may now be quite strong, no longer limited by the reflectivity of themetal layer for TE radiation. This is very important for determining the director profilethrough 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 angledependent TM to TE conversion reflectivity. Fitting these data provides even more detailon the twist/tilt director profile across the cell.

The advantages of the HLGM technique have been analysed in detail [54]. Fromnumerical modelling it has been found that the technique has very good sensitivity to allchanges in the director twist and tilt, even less than 1°. It may be used to monitor opticalbiaxiality as low as 0.0002, and it also gives details of the director configuration quitenear the bounding surfaces of the cell.

Page 312: Nanotechnology and Nano-Interface Controlled Electronic Devices

Guided mode studies of liquid crystal layers 295

From the experimental point of view a practical point worthy of note is that althoughthe geometry has to have a high index prism there is no essential need to use very highindex glass since for normal liquid crystals the refractive indices are not very high inthe visible. We may use an index of 1.73 (632.8 nm, He–Ne) for which matching fluidsare available, thus the prism used in the geometry such as shown in Fig. 14 may thenbe replaced by a high index upper glass plate, n = 1.730, matching fluid (e.g. CH2I2)layer, and a high index, n = 1.730, prism. The advantage of this arrangement is notonly for simpler cell fabrication but it also allows rotation of the cell under the prismthereby guaranteeing, provided there is not simple homeotropic liquid crystal alignment,that some strong TE to TM conversion signal may be obtained by a suitable choiceof rotation of the cell. So by using two glass plates with the requisite high and lowindices having similar thermal expansion coefficients, so that heating the cell does notcause mechanical stress problems, and following commercial procedures a HLGM testcell can be fabricated and the detailed characterisation of the director profile in such astructure undertaken.

Since the HLGM technique was developed in 1993 [55] a significant amount ofwork has been undertaken using this powerful tool. The first study, using this technique,was to explore the detailed optical tensor configuration in a homogeneously alignedSSFLC (Merck-BDH SCE3) [55]. In the test cell the aligning surface layers were siliconoxide deposited by evaporation at 60° to create in-plane homogeneous alignment inthe nematic phase. The experimental recorded data has shown as high as 60% p to sconversion reflectivity signals in the incident angle window as discussed above. Usingthe prediction from Fresnel multilayer optics theory to fit the angular dependent p to sconversion reflectivities yields, in unprecedented detail, the optical tensor configurationthrough the cell. From the fitting results the director profile of the SSFLC across thecell is that of a slightly bent ‘chevron’ director structure with a small tilt angle of order1.5°, with near-surface region of order 0.3 μm in thickness. Additionally a permittivitybiaxiality of the FLC material as small as 0.0035 can be found from the fits. In additioninformation on a small amount of optic tensor axes dispersion is also provided by usingtwo 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 togetherwith theoretical fits are shown in Figs. 15a and 16a, the corresponding director tilt andtwist profiles being given in Figs. 15b and 16b, respectively. From Figs. 15b and 16bit is clear that there are finite tilt surface angles in the cell, which is contrary perhapsto earlier expectations, since Cognard [56] indicates that for 60° obliquely evaporatedSiOX and a nematic liquid crystal, there will be little surface tilt. It looks like theevidence for tilt with the S∗

C phase is overwhelming. From Figs. 15b and 16b the optictensor axes dispersion with the wavelength of the radiation is also very clear, sincethere 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 beconfirmed by a particular example. Consider only the right hand side of Fig. 16a, whichhas been expanded and shown in Fig. 17a. In Fig. 17a a particular peak, indicated by anarrow, 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

Page 313: Nanotechnology and Nano-Interface Controlled Electronic Devices

296 F. Yang and J.R. Sambles

Fig. 15: (a) The p to s conversion reflectivity from liquid crystal SCE3 using the HLGM technique witha wavelength of 632.8 nm. The solid line indicates the fit of theory to the experimental data (shown ascrosses). (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 forFig. 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 amaximum tilt of 2°. It is very clear that the amplitude of this chosen half-leaky guidedmode is very sensitive, with resolution much less than 0.5°, to the director tilt profileacross the cell. This also strongly supports the director tensor axes dispersion shown inFigs. 15b and 16b. Comparing with a metal-clad waveguide, in which even though themodes may have been sharper, because of the weakness of the p to s conversion signalby 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 techniqueis well suited to the study of the electroclinic effect. This effect has been investigatedfor 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]

Page 314: Nanotechnology and Nano-Interface Controlled Electronic Devices

Guided mode studies of liquid crystal layers 297

Fig. 17: (a) Angular dependent reflectivity, for a wavelength of 514.5 nm, obtained by theoretical modellingusing the twist profile in Fig. 16(b) and the tilt angle profiles shown in (b). (b) The tilt angle profiles usedto generate the theoretical curves shown in (a). The reflectivities and corresponding profiles are indicated bythe different dashed lines. (From [55].)

transitions have been studied. The extra director tilts induced by the electroclinic effecthave been comparing with mean field theory and the electroclinic coefficients have alsobeen quantified. From the experimental results strong surface anchoring constraints arealso 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 suchcell has been studied. Although the arrangement of the S∗

C director should be particularlysimple in this geometry the optical results show that, somewhat surprisingly, the smecticlayers are not parallel to the cell walls but tilted by as much as 4°. When the cell isrotated so that the director lies in a plane perpendicular to the plane of incidence a finitep to s conversion signal appears to indicate this layer tilt and this phenomenon can beextended 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 coneangle 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 theabove experimental results [60] and, in addition, detailed information on the opticaltensor components of both the S∗

C and SA phases has also been given [62]. In suchexperiments 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 itslayers flat in the plane. For a homeotropically aligned FLC with no surface treatment theextra tilt induced in the SA phase by an in-plane field can be very accurately measuredby the p to s conversion signal. Thus it is possible to quantify the electroclinic effect inthis almost unconstrained environment, a simple linear relationship between the inducedtilt and the DC field being found even under very weak fields [61]. The predictions from

Page 315: Nanotechnology and Nano-Interface Controlled Electronic Devices

298 F. Yang and J.R. Sambles

a second order Landau mean-field theory, which includes the coupling between the tiltangle 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 tofit with multiplayer optics modelling theory the more complex reflectivity data obtainedas the unwinding DC field is removed. This then will give the pitch of the helical S∗

Cphase at zero voltage applied, and the ratio of the spontaneous polarisation PS to thetwist elastic constant B3 will be given by fitting the distorted helix for finite fields. Inadditional 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 HLGMmethod [64]. Under the application of an AC voltage the director configuration changesover a time-scale of the order of seconds. Fitting model results to the recorded angledependent 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 ofthe optical permittivity, implying a suppression of fluctuations. However the expectedHelfrich-like deformation is not recorded, with detailed analysis showing not an increaseof 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 thedirector structure is quite complex due to external constraints then the HLGM techniquemay provide a vital method for determining the profile. Two recent examples are givenas follows.

The first concerns the observation of coexisting nematic and smectic-A phases ina twisted liquid-crystal cell [65]. While the properties of twisted nematic cells, whichare the important elements of the most widespread electro-optic liquid crystal displaydevices, are well understood, there is no information about the structure of such cellsbelow the nematic–smectic transition temperature. Since the smectic layers cannotsustain twist the question is what does the system do on cooling? Does it become fullof defects or can some other defect-free configuration exist? This phase transformationis also interesting from the fundamental point of view because it represents a nontrivialexample 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 cellthickness. So a powerful technique for exploring this is required to accurately determinethe director configuration through a thin cell in some detail. As mentioned above theHLGM geometry provides the required technique.

The sample comprises a twisted homogeneously planar-aligned cell with an angleof 87° between the two rubbing directions of the top and bottom polyimide coatedsubstrates. The liquid crystal material in this study is the ferroelectric liquid crystalSCE13. After filling the liquid crystal in the isotropic phase (∼110°C) the temperatureis 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 transitiondata are taken at intervals of about 1° down to the S∗

C phase. From fitting the angle

Page 316: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 usingthe 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-Amaterial that grows with decreasing temperature. The insert shows a schematic of the geometry of theexperiment. (From [65].)

dependent reflectivity data recorded several points have been obtained. Firstly, asexpected, 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 phasetransition point by the sudden appearance, between 67.0 and 67.8°C, of an extra opticalmode feature in the HLGM reflectivity spectrum. Thirdly, when the temperature is lowerthan the N∗ to SA phase transition point there is still a good monodomain in the cellwhich 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 rangeof the SA phase the director profile has a very different form to the linearly twistednematic. For a thinner cell with thickness of 1.60μm an untwisted homogeneous regionforms 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 twistgradient 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 cellthickness more separated SA areas appear in the cell. For the 2.0 μm cell there are tworegions of uniform azimuthal angle (SA phase) separated by a twisted nematic with twotwisted nematic boundary regions as shown in Fig. 19a, while for the 2.4 μm cell thereare three SA regions separated by two twisted layers with two twisted boundary layersas shown in Fig. 19b. The uniqueness of the director profile through these test cells canonly be achieved by the HLGM technique, since it is very sensitive to the variation ofthe director twist/tilt distributions across the cells. So the director profiles discovered inthis work gives a picture of coexisting nematic and smectic-A phases in a twisted liquidcrystal cell. The experimental results also show that comparing with the bulk phasesequence of the FLC material the SA phase exists over a very small temperature range inthese twisted thin cells. This suppression in phase transition is caused by the high twistgradient in the liquid crystal and has to be expected since the SA phase cannot tolerate

Page 317: Nanotechnology and Nano-Interface Controlled Electronic Devices

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] togive a complete explanation.

The second example concerns establishing the direction of the ‘easy’ axis at a twistednematic liquid crystal wall determined by the half-leaky guided mode technique [66].For the measurement of surface torsional or azimuthal anchoring energy, which is veryimportant both from the fundamental science perspective as well as for the productionof display devices, a geometrical technique using a thin twisted nematic (TN) cell hasbeen widely used. Since no external field is needed to distort the director profile inthe cell both the mathematics and the experimental procedures are quite simple. In thisTN geometry the twist-off of the director at the surface of the cell is brought aboutby equilibrium between the two surface torsional anchorings mediated through the bulktwist elastic constant, K22. From continuum theory the key numbers required to quantifythe torsional anchoring strength are the elastic constant K22, the gradient of the twistangle, dφ/dz, and the deviation of the twist angle from the easy axis, φ −φe, at the wallof the TN cell. K22 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. Thenhalf of the difference between the two angles φt and φ0

t , which is the angle between thetwo easy axes on the two interfaces of the cell, is taken as the deviation of the twistangle at both boundaries, if the alignments are assumed to be identical. Both φt anddLC can be accurately determined in the geometrical technique using for example theHLGM technique. However, when a thin TN cell has been assembled the twist anglesof the director on the two boundaries always deviate from the easy axes, hence accuratedetermination of φ0

t is not simple. It is in fact much easier to measure φt rather than φ0t .

It is apparent that after rubbing the polyimide, assembling the cell and mounting it on asample holder an accurate and direct determination of the easy axes directions of a TNcell is very important to allow characterisation of the surface torsional anchoring forceby the geometrical technique.

From numerical modelling using continuum theory for a thin TN cell it is apparent

Page 318: Nanotechnology and Nano-Interface Controlled Electronic Devices

Guided mode studies of liquid crystal layers 301

that under high voltage (a typical value for a practical cell about 4–5 V) the gradientdφ/dz near both surfaces is zero. This is because in the middle part of the cell the di-rector, driven by the high electric field, is almost homeotropically aligned. This removesthe influence of the two boundaries from each other through the twist elastic constantK22, i.e. now the direction of the director at the wall will coincide with the direction ofthe easy axis. Thus all that is required is some optical procedure for determining thisdirector 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 atthe surface: (1) For an interface between an isotropic and a uniaxial anisotropic mediumif the optic axis of the uniaxial medium is in the incidence plane there will be nopolarisation conversion reflectivity signal. By contrast a very small angle of optic axistwist out of the incidence plane creates a small but detectable polarisation conversionsignal in the reflected beam. (2) If the director of the top surface is close to the incidenceplane of the radiation beam then in the HLGM geometry there is a pseudo-critical angledependent on the high index of the isotropic medium and the low ordinary index of theuniaxial medium. Also in an incident angle range beyond this pseudo-critical angle, fora s-polarised incident beam the optical field in the anisotropic medium exponentiallydecays away from the interface. (3) For a practical TN cell at high voltages the directorlies in the direction of the easy axis for a distance of about 0.5 μm which is much greaterthen the decay distance mentioned above. Thus there will be no p to s (or s to p) conver-sion within this depth if the director lies in the incidence plane. According to the abovethree factors an experimental procedure is designed to accurately measure the easy axisat the cell surface [66]. Firstly matching fluid is placed between a high index prism andthe top glass plate of the HLGM cell to allow the cell to be freely twisted against theincidence plane with the rubbing direction set near to the incidence plane. Then withthe 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 fordifferent cell twist angles. The minimum point of the polarisation conversion reflectivitysignal accurately gives the director direction, the easy axis direction, at the top surfaceof the cell. The experimental results of the polarisation conversion reflectivity againstthe easy axis twist angle from the incidence plane are shown in Fig. 20 [66]. Using thisprocedure 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 avery powerful procedure for exploring the optical tensor configuration of a liquid crystalin a thin cell. However, this powerful technique still has a limitation, i.e. the two glassplates from which the test cell is comprised are very different from each other andfrom 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 modetechnique, 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 beinvestigated by a guided mode technique to unravel the director profile through the

Page 319: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 cellat 4.0 V. (a) Six RPS curves which correspond to the director in the incident plane and twisted away by1.0°, 2.0°, 4.0°, 5.0° and 6.0° from bottom to top, respectively. (b) Six RPS curves which correspond tothe director in the incident plane and twisted by the same angles in the opposite direction as (a). (c) Theintensity of a selected p to s conversion mode against the twist-off angle of the director from the incidentplane of the radiation. (From [66].)

cell, then only the low index fully leaky guided mode (FLGM) technique may bechosen. As mentioned before, because all the guided modes will now be leaky andwill give correspondingly broad features in the reflectivity data, this may severely limitthe precision with which the director twist/tilt structure through the liquid crystal cellmay be determined. In addition, the use of low index glass means that no longer willthere be any critical angle available to help determine the refractive indices of the liquidcrystal. However, recently some improvements [21] have been introduced to the fullyleaky geometry, in which two refinements to the original technique have been made tomake 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. Thepolarisation-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

Page 320: Nanotechnology and Nano-Interface Controlled Electronic Devices

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. Inaddition, this allows the acquisition of data for a set of different azimuthal angle settingsthat, by fitting of all the data sets, further removes ambiguity in determining the directorprofile.

The new FLGM geometry is shown in Fig. 21. As mentioned above this is asymmetrical structure in which two identical low index prisms with matching fluidcouple the radiation in and out from a symmetrical commercial-like liquid crystal cellcomprised of two low index glass plates with suitable ITO coatings and alignment layerson their inner surfaces. Of course, comparing with the other three geometries an extradetector and an extra polarizer are needed in the experimental arrangement for detectingsuitable transmission signals.

Since the improved FLGM technique was introduced in 1999 [21] a substantialbody of work has been done using this technique to investigate the commercial-likeliquid crystal cells. To demonstrate the potential use of this new technique for thedetermination of the director profile in a liquid-crystal layer, a conventional surface-stabilised ferroelectric liquid crystal cell has been studied [21]. The glass of the cellis ordinary glass with an index close to 1.52, as are the two coupling prisms. Thesample used in this study contains the ferroelectric liquid crystal SCE8∗. The alignmentis homogeneous with a slight pretilt achieved by the use of rubbed polyimide in aparallel arrangement. The glass plates, which are 1 mm thick and coated with ITO onthe inner face, have an index of 1.517 at 632.8 nm. All measurements were conductedon a monodomain at room temperature, 23.7° C. Experimental data, including thepolarisation-conserving signals (RPP, RSS, TPP and TSS) and the polarisation-conversionsignals (RSP, RPS, TSP and TPS), are recorded with the cell configured so that the surface-alignment axis (rubbing direction) is close to the plane of incidence. An example of thepolarisation-conversion reflectivity signals, RSP and RPS, are shown in Fig. 22. FromFig. 22 we note that these RSP and RPS signals are quite weak, being less than 1%. Thisindicates that the director of the ferroelectric liquid-crystal layer is only twisted a small

Page 321: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 theliquid-crystal being represented as 150 sub-layers. These sub-layers are part of anoverall 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 linearvariation 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 realand imaginary permittivities and ten sets of linked layer parameters. The final directorprofiles, obtained by fitting the recorded data with the predictions from the multi-layeroptics, are shown in Fig. 23a and b, in which there is a clear director-tilt chevron closeto the centre of the cell with thin boundary layers with the liquid crystal being in theC2U state [68]. The smectic layer tilt, δ, may also be extracted by using a value for thecone 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 upperportion it is 17.30°. There is also a discernable, but small variation of this tilt near theupper bounding surface.

Thus it is clear that the new FLGM technique, using both reflectivity and transmis-sivity signals, is capable of giving details of the director profile in a commercial-like cellto almost the same level of precision as had been previously obtained by use of eithermetal-clad guiding structure or cells with different index glass plates in the half-leakygeometry. Unlike the other two procedures, there are no very sharp features in the angle-dependent signals owing to the weak (leaky) nature of the guiding; furthermore, thereare no true critical angles to yield the refractive indices of the liquid crystal because the

Page 322: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 coneangle 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 angleof study to yield all the requisite information. This means that more detail can still beextracted regarding the spatial distribution of the director profile through the cell by thenew FLGM technique.

After this first demonstration of the improved fully leaky technique some commer-cially like standard liquid-crystal cells have also been investigated. This includes thequantification of the azimuthal anchoring energy [69] and the surface- and bulk-orderparameters [70] of a homogeneously aligned nematic liquid crystal under an in-planeelectric field, and the determination of the polar anchoring energies of both homoge-neously [71] and a homeotropically [72] aligned nematic liquid crystals. Some mixedalignment liquid crystal cells, which have a zero-order grating alignment on the su-perstrate and rubbed polyimide alignment on the substrate, have also been studied by

Page 323: Nanotechnology and Nano-Interface Controlled Electronic Devices

306 F. Yang and J.R. Sambles

this new technique to investigate the influence of the groove depth of the grating onthe alignment [73] and the torsional anchoring energy [74]. Because the new techniqueallows the use of standard liquid crystal cells then combined X-ray scattering and fullyleaky guided mode studies have been undertaken to explore the smectic layer and theoptic tensor configuration in a ferroelectric liquid crystal cell [75]. All of this work,using the guided mode technique to explore conventional commercial-like cells, pavesthe way for detailed exploration of the behaviour of such cells under various conditionsand also allows dynamic studies.

4. Dynamic guided mode technique

As mentioned in the above sections the success of the guided mode techniques indetermining the director distribution through a thin liquid crystal layer can be attributedto the fact that each optical mode has a different field intensity profile across the liquidcrystal 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 transmissivitydata with the predictions from multi-layer optic theory. From both a fundamentalscientific perspective and also for device development the switching process of a liquidcrystal layer under an external voltage is very important. Thus establishing a clearpicture of the transient director profile through a cell as it varies on time scales oforder ms, the dynamic director profile, is very important. It is also obvious that theguided mode technique would be an attractive tool for this dynamic director profiledetermination. Some early studies successfully resolved the dynamic director profileacross 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, thesestudies were based upon the standard angle-scan collimated beam procedure. Theexperiment is a fairly simple optical arrangement in which a plane polarised collimatedlaser beam is incident through a coupling prism onto the liquid crystal cell. By rotatingthe liquid crystal cell and prism arrangement around an axis perpendicular to theincident wave vector, the angular dependent reflectivity and transmissivity features arerecorded. Of course, this is a relatively slow data acquisition procedure, every switchingprocess has to be repeat again for every rotating step, i.e. every incident angle, takinganywhere from several minutes to two hours to perform a single angle scan [77]. Theseponderous studies of dynamic processes using a slow angle scan with time dependentdata taken at each angle demand highly repeatable voltage cycling of the cells, completethermal stability and total lateral invariance within a given cell. Thus an improved guidemode technique has been developed for fast dynamic studies.

The new approach involves the use of a convergent beam and has several advantagesover the standard collimated beam angle scan procedure [78,79]. This convergentbeam guided mode (CBGM) technique uses a highly focused beam spot that allowssimultaneously the excitation of many guided modes and produces reflectivity andtransmissivity data over a wide incident angle range. There are several advantages to thisprocedure. Firstly the CBGM technique removes the need to physically rotate the liquidcrystal cell geometry and consequently the focused beam remains completely stationary

Page 324: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 thetechnique with either half-leaky [80] or fully leaky [81] geometries the static directorprofiles for nematic liquid crystal cells have been determined. Secondly, possibly evenmore importantly, data from the converged beam technique may be captured with asuitable charge coupled device (CCD) array in a time equal to the line transfer rate ofthe 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 thecollimated beam procedure, allowing for real time studies of liquid crystal dynamics bythe guided mode technique.

Since the fully leaky geometry allows the study of standard commercial-like liquidcrystal cells the first COGM study [16] of liquid crystal dynamics was undertaken withthe improved fully leaky guided mode geometry. In this study the cell consisted of twoordinary glass plates (n = 1.52) coated with ITO upon which is a rubbed polyimidelayer. The rubbing directions on the top and bottom plates are antiparallel, thus inducinga uniform, nearly planar alignment of the director through the cell. The cell is filled withthe nematic liquid crystal ZLI-2293 (Merck). For directly coupling the convergent beamin and out from the sample geometry two low index hemispheres are optically matchedonto the cell by use of a suitable low volatility silicon based oil as shown in Fig. 24. Theuse of the matching fluid not only provides a continuous optical medium for couplinglight into and out of the waveguide modes but also enables the cell to be easily rotatedto 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 laserbeam (632.8 nm, 75 mW) is expanded, polarised, and focused through the hemisphereonto the liquid crystal layer. The reflected and transmitted beams are then captured witha linear CCD array (DALSA SPARK). The diffuser arrangement situated in the beamexpander serves two purposes. First, it breaks up the spatial and temporal coherence ofthe laser, and second, it provides an intensity profile across the expanded beam such thatapproximately the same intensity is available to excite each guided mode. The dynamicstudy in this work uses an integration time of 0.3 ms and therefore it is desirable forthe rotating diffuser to complete at least one whole revolution in this time. A turbine

Page 325: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 diffuserrotation rate of 120,000 rpm (period ∼0.5 ms).

Initially, the total of eight data sets of the reflectivity and transmissivity were takenwith no voltage and 1.5 V rms (10 kHz) applied to the cell to characterise the staticoptical parameters of the liquid crystal cell. Once the optical parameters of the LC celland the static director profile at 0 and 1.5 V had been ascertained by fitting the datarecorded with multi-layer optics theory, the dynamics of the director profile as it relaxesfrom the 1.5 V to the 0 V configuration was studied. The switching dynamics of the LCcell were recorded by synchronising the CCD array to capture data when the voltageacross the LC cell was changed. After removal of the voltage both RPP and RSS signalswere captured as a function of time and fitted together. The fitted experimental data forthe RPP and RSS signals and the corresponding director profiles used to obtain each leastsquares fit are shown in Fig. 26. From the fitting results the tilt of the director in themiddle of the cell against time can be extracted and then the exponentially decayingtime constant of the relaxation process is revealed to be τ = 51.4±0.1 ms. Then, fromthe relation between the time constant, τ , the thickness of the cell, d, the splay elasticconstant, K11, and the effective rotational viscosity, γ ∗, the γ ∗ value is deduced. Usinga value for K11 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 experimentalarrangement as mentioned above the ‘back-flow’ phenomenon of the relaxation processin 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 FLGMgeometry is a very powerful technique for investigating dynamic processes within liquidcrystal 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

Page 326: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 varioustimes. 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 fundamentalunderstanding, material parameter determination and device design procedures aredependent upon the knowledge of the director profile and its changes under variousconditions. Because of the nature of optical guided modes the liquid crystal waveguidetechniques provide an extremely powerful method for studying this tensor profile indetail allowing exploration of both static and dynamic processes.

In this chapter first the principles of optical guided waves in general have beendescribed with some discussion of optical field profiles, coupling procedures, etc., thishas been followed by a fuller discussion of several types of waveguiding techniquesused to investigate liquid crystals in aligned cells. These techniques include that of themetal-clad fully guiding geometry; the fully leaky waveguide with no metal layers butwhich gives poor resolution; the half-leaky waveguide technique which again has nometal layers but which yield sharper optical features giving much finer detail of thedirector profile; the improved fully leaky guided mode technique which can be directlyused to investigate a commercial-like standard liquid crystal cell and still gives enoughinformation for characterising the director profile through such cells, and finally the

Page 327: Nanotechnology and Nano-Interface Controlled Electronic Devices

310 F. Yang and J.R. Sambles

convergent beam guided mode technique with the improved fully leaky geometry todirectly and rapidly study the dynamic process in a standard liquid crystal cell. A rangeof experimental results using these techniques have been presented to illustrate theirusefulness with a range of cells structures and liquid crystal phases.

All these guided mode techniques are very powerful tools for exploring the staticor, very recently, the dynamic director profiles and determining the optical or physicalproperties of thin liquid crystal films. Which technique is chosen is dependent onthe sample and the information required. However it seems likely that the advantagesof being able to use off-the-shelf cells will see progressively more use of the fullyleaky technique. It is very clear that by using a wide range of incident angles andmultilayer optics theory to fit the obtained data in these guided mode techniques someof the uncertainties and erroneous conclusions associated with integrated techniques(e.g. simple crossed polariser microscopy), which often ignore the surface layers, areavoided. However, it should be added that, dependent on the sample geometry and themeasurements taken, the guided mode technique may require elaborate data analysisbased on the Berreman transmission/reflection matrix and a complex fitting procedureto obtain the required results. It is essential that sufficient consideration is given to thesample geometry suitable for obtaining the information required and as many of theunknown parameters are pre-determined by other procedures before any one particularoptical guided mode technique is chosen.

In this review chapter we have chosen not to discuss to any length the other veryinteresting and useful technique, grating-coupling of radiation into guided modes. Thisis primarily because this needs special cell fabrication and also the theoretical analysisof the optics is much more complicated. By contrast the prism-coupled proceduresdiscussed use simple planar multilayer optics theory, leading to accurate detailedcomparisons of predicted responses with those observed experimentally. This leads tosubstantial confidence in the director profiles thus deduced, which is fundamentally whythese 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. F. Reinitzer, Monatsh. Chem. 9, 421 (1888).2. O.Z. Lehmann, Phys. Chem. 4, 462 (1889).3. M.A. Handschy and N.A. Clark, Appl. Phys. Lett. 41, 39 (1982).4. 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).5. P.K. Tien, Appl.Opt. 10, 2395 (1971).6. D. Marcuse (editor), Intergrated Optics, (IEEE Press, New York, 1973).7. H.A. Weakliem, D.J. Channin, and A. Bloom, Appl. Opt. 14, 560 (1975).8. D.J. Channin, Appl. Phys. Lett. 22, 365 (1973).

Page 328: Nanotechnology and Nano-Interface Controlled Electronic Devices

Guided mode studies of liquid crystal layers 311

9. J.P. Sheridan, J.M. Schnur, and T.C. Giallorenzi, Appl. Phys. Lett. 22, 561 (1973).10. J.P. Scheridan, OSA Topical Meeting on Integrated Optics, New Orleans, (1974) TUA S-1.11. Chenming Hu, J.R. Winnery, and N.M. Amer, IEEE J. Quan. Elect. QE10, 218 (1974).12. R.B. Meyer, Mol. Cryst. Liq. Cryst. 40, 33 (1977).13. N.A. Clark and S.T. Lagerwall, Appl. Phys. Lett. 36, 899 (1980).14. L.Z. Ruan, M.A. Osipov, and J.R. Sambles, Phys. Rev. Lett. 86, 4548 (2001).15. Fuzi Yang and J.R. Sambles, Jpn. J. Appl. Phys., 37, 3998 (1998).16. N.J. Smith and J.R. Sambles, Appl. Phys. Lett. 77, 2632 (2000).17. N.J. Smith, M.J. Tillin, and J.R. Sambles, Phys. Rev. Lett., (2002), accepted.18. 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. 5.19. M.S. Kharusi, J. Opt. Soc. Am. 64, 27 (1974).20. Fuzi Yang and J.R. Sambles, J. Opt. Soc. Am. B 11, 605 (1994).21. Fuzi Yang and J.R. Sambles, J. Opt. Soc. Am. B 16, 488 (1999).22. R.M.A. Azzam and N.M. Bashara, Ellipsometry and Polarised Light, (North Holland, Amsterdam,

1979).23. M. Born and E. Wolf, Principle of Optics, (Pergamon, Oxford, 1964).24. J. Lekner and M.C. Dorf, J. Opt. Soc. Am. A 4, 2092 (1987).25. D.Y.K. Ko and J.R. Sambles, J. Opt. Soc. Am. A 5, 1863 (1988).26. J.H. Harris, R. Shubert, and J.N. Pelky, J. Opt. Soc. Am. 60, 1007 (1970).27. E.A. Ash, E. Seaford, O. Soares, and K.S. Pennington, Appl. Phys. Lett. 24, 207 (1974).28. R. Shubert and J.H. Harris, IEEE Trans. Microwave Theory Tech. MTT-16, 1048 (1968).29. N.H. Hartshorne and A. Stuart, Crystals and the Polarising Microscope, 2nd ed. (Arnold, London,

1950).30. A. Sugimura and T. Kawamura, Jpn. J. Appl. Phys. 23, 137 (1984).31. Y. Kawata, K. Takatoh, M. Hasegawa, and M. Sakamoto, Liq. Cryst. 16, 1027 (1994).32. G.P. Bryan-Brown, J.R. Sambles, and K.R. Wolford, J. Appl. Phys. 73, 3603 (1993).33. E.L. Wood and J.R. Sambles, J. Mod. Opt. 40, 493 (1993).34. 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.

35. E. Kretschmann, Z. Phys. 241, 313 (1973).36. G.J. Sprokel, R. Santo, and J.D. Swalen, Mol. Cryst. Liq. Cryst. 68, 29 (1981).37. K.R. Welford, J.R. Sambles, and M.G. Clark, Liq. Cryst. 2, 91 (1987).38. K.R. Welford and J.R. Sambles, Appl. Phys. Lett. 50, 871 (1987).39. Lizhen Ruan, S.J. Eolston, and J.R. Sambles, Liq. Cryst. 10, 369 (1991).40. Lizhen Ruan, G.W. Bradberry, and J.R. Sambles, Liq. Cryst. 11, 655 (1992).41. Lizhen Ruan, G.W. Bradberry, and J.R. Sambles, Liq. Cryst. 12, 799 (1992).42. Lizhen Ruan, T.W. Preist, Fuzi Yang, and J.R. Sambles, Liq. Cryst. 13, 541 (1993).43. E.L. Wood, J.R. Sambles, and P.S. Cann, Liq. Cryst. 16, 983 (1994).44. S.J. Elston, J.R. Sambles, and M.G. Clark, J. Mod. Opt. 36, 1019 (1989).45. S.J. Elston and J.R. Sambles, Appl. Phys. Lett. 55, 1621 (1989).46. S.J. Elston, J.R. Sambles, and M.G. Clark, J. Appl. Phys. 68, 1242 (1990).47. S. Ito, F. Kremer, E. Aust, and W. Knoll, J. Appl. Phys. 75, 1962 (1994).48. S.J. Elston and J.R. Sambles, Jpn. J. Appl. Phys. 29, 41 (1990).49. S.J. Elston and J.R. Sambles, Mol. Cryst. Liq. Cryst. 220, 99 (1992).50. M.N. Gong, J.R. Sambles, and Fuzi Yang, Liq. Cryst. 13, 637 (1993).51. C.R. Lavers and J.R. Sambles, Ferroelectrics 113, 339 (1991).52. C.R. Lavers and J.R. Sambles, Jpn. J. Appl. Phys. 30, 729 (1991).53. C.R. Lavers and J.R. Sambles, Liq. Cryst. 8, 577 (1990).54. Fuzi Yang and J.R. Sambles, J. Opt. Soc. Am. B 10, 858 (1993).55. Fuzi Yang and J.R. Sambles, Liq. Cryst. Opt. 13, 1 (1993).56. J. Cognard, Mol. Cryst. Liq. Cryst. 78, Supplement (1981).

Page 329: Nanotechnology and Nano-Interface Controlled Electronic Devices

312 F. Yang and J.R. Sambles

57. Lizhen Ruan, J.R. Sambles, E.L. Wood, and J. Seaver, Liq. Cryst. 18, 401 (1995).58. Lizhen Ruan, J.R. Sambles, and J. Seaver, Liq. Cryst. 19, 133 (1995).59. Fuzi Yang and J.R. Sambles, Mol. Cryst. Liq. Cryst. 250, 143 (1994).60. Fuzi Yang, G.W. Bradberry, and J.R. Sambles, Phys. Rev. E 50, 2834 (1994).61. Fuzi Yang, J.R. Sambles, and G.W. Bradberry, Liq. Cryst. 18, 407 (1995).62. Fuzi Yang, J.R. Sambles, and G.W. Bradberry, J. Appl. Phys. 78, 2187 (1995).63. Fuzi Yang, G.W. Bradberry, and J.R. Sambles, Phys. Rev. 53, 674 (1996).64. L. Ruan, J.R. Sambles, and J.M. Towler, Liq. Cryst. 18, 81 (1995).65. L.Z. Ruan, M.A. Osipov, and J.R. Sambles, Phys. Rev. Lett. 86, 4548 (2001).66. F.Z. Yang, H.F. Cheng, H.J. Gao, and J.R. Sambles, J. Appl. Phys. 88, 4553 (2000).67. F.Z. Yang, H.F. Cheng, H.J. Gao, and J.R. Sambles, J. Opt. Soc. Am. B 18, 994 (2001).68. N. Itoh, M. Koden, S. Miyoshi, and T. Wada, Liq. Cryst. 15, 669 (1993).69. B.T. Hallam, Fuzi Yang, and J.R. Sambles, Liq. Cryst. 26, 657 (1999).70. B.T. Hallam, C.V. Brown, and J.R. Sambles, J. Appl. Phys. 86, 6682 (1999).71. Fuzi Yang, J.R. Sambles, Youmei Dong, and Hongjin Gao, J. Appl. Phys. 87, 2726 (2000).72. Fuzi Yang, Lizhen Ruan, and J.R. Sambles, J. Appl. Phys. 88, 6175 (2000).73. B.T. Hallam and J.R. Sambles, Phys. Rev. E 61, 6699 (2000).74. B.T. Hallam and J.R. Sambles, Liq. Cryst. 27, 1207 (2000).75. B. Hodder, J.R. Sambles, S. Jenkins, and R.M. Richardson, Phys. Rev. Lett., 85, 3181 (2000).76. M. Mitsuishi, S. Ito, and M. Yamamoto, J. Appl. Phys. 81, 1135 (1997).77. M. Mitsuishi, S. Ito, and M. Yamamoto, Appl. Phys. Lett. 69, 2199 (1996).78. E. Kretschmann, Opt. Commun. 26, 41 (1978).79. J.R. Sambles and N.J. Smith, Mol. Cryst. Liq. Cryst. 347, 37 (2000).80. N.J. Smith and J.R. Sambles, J. Appl. Phys. 85, 3984 (1999).81. N.J. Smith and J.R. Sambles, Mol. Cryst. Liq. Cryst. 347, 45 (2000).82. D.W. Berreman, J. Appl. Phys. 46, 3746 (1975).83. C.J. Gerritsma, C.Z. van Doorn, and P. van Zanten, Phys. Lett. A 48, 263 (1974).84. F. Nakano, H. Kawakami, H. Morishita, and M. Sato, Jpn. J. Appl. Phys. 19, 659 (1980).

Page 330: Nanotechnology and Nano-Interface Controlled Electronic Devices

Nanotechnology and Nano-Interface Controlled Electronic DevicesEditors: M. Iwamoto, K. Kaneto and S. Mashiko© 2003 Elsevier Science B.V. All rights reserved

CHAPTER 16

Explanation of the static and dynamicdirector orientation in thin nematic liquid

crystal films using deuterium NMRspectroscopy

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] Department of Chemistry and Southampton Liquid Crystal Institute, University of

Southampton, Highfield, Southampton, SO17 1BJ, UKE-mail: [email protected]

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3132. Theoretical background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3153. Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3204. Static director distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 321

4.1. Bistable director orientation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3214.2. Continuous director orientation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 323

5. Director dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3315.1. Director dynamics for the orthogonal geometry of B and E . . . . . . . . . . . . . . . . 3315.2. Director dynamics for the non-orthogonal geometry of B and E . . . . . . . . . . . 335

6. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 346Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 348References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 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 liquidcrystals because the spectra of specifically or fully deuteriated materials are rathersimple compared to the corresponding proton NMR spectra [1–5]. The quadrupolarsplitting for deuterons observed in the liquid crystal phase is related to the second rank

Page 331: Nanotechnology and Nano-Interface Controlled Electronic Devices

314 A. Sugimura and G.R. Luckhurst

orientational order parameter for the C–D bond direction and so deuterium NMR hasbeen widely used for the study of the orientational order of liquid crystals and theirphase transitions [2]. The quadrupolar splitting is also determined by the angle betweenthe director and the magnetic field. In consequence, deuterium NMR spectroscopy isproviding to be a powerful method with which to investigate the director orientationand its distribution, as well as the director dynamics in nematic [6–17] and smecticliquid crystals [18–20]. In more recent years deuterium NMR spectroscopy, combinedwith continuum theory, has been applied successfully to investigate the static directordistribution in thin nematic liquid crystal cells, with different film thicknesses anddifferent surface anchoring strengths, subject to both magnetic and electric fields [11–13]. Deuterium NMR spectroscopy has also been employed to investigate the dynamicdirector alignment process in a thin nematic film following the application or removalof an electric field [14–17]. This technique has the added advantage that the presenceof the magnetic field of the spectrometer ensures that during the electric field-inducedalignment the director rotates as a monodomain [13] which facilitates the analysis of theresults.

In this chapter, we describe some of our studies of the static director distribution inthin nematic liquid crystal cells with different film thicknesses and different surface an-choring strengths using a combination of deuterium NMR spectroscopy and continuumtheory. The nematic liquid crystal, 4-pentyl-d2-4′-cyanobiphenyl (5CB-d2) deuteriatedin the α-position of the pentyl chain, was confined between two glass plates with bothweak and strong anchoring conditions; the anchoring strengths were measured by usinga saturation voltage method [21]. A series of deuterium NMR spectra was acquired asa function of the applied electric field, which can be used to explore the director defor-mation. We also describe the application of deuterium NMR spectroscopy to investigatethe director dynamics in the same nematic liquid crystal (5CB-d2) confined betweentwo glass plates and subject to magnetic, B, and AC electric, E, fields. The cell wasset in the NMR probe with the electric field, whose direction is normal to the substratesurface, making an angle of about 45°, with the magnetic field. This experimentalgeometry allows a unique director motion in the alignment process. In the absence ofthe electric field the director for 5CB will align parallel to the magnetic field because thediamagnetic anisotropy, Δχ , is positive. When an electric field, which is strong enoughto overcome the magnetic torque, is applied then the director will make an angle withthe electric field since the dielectric anisotropy, Δε, is also positive. After the electricfield is switched off, the director will then move from being at an angle to the magneticfield to being parallel to it. The dynamics of the director relaxation can be followed bymonitoring the NMR spectrum during this alignment process, as a function of time. Thatis, the time dependence of the director orientation during and after the application of anelectric field is studied. This is possible because the NMR spectrum for a monodomainsample with one group of equivalent deuterons, having a negligible dipolar interaction,contains a simple quadrupolar doublet whose separation is determined by, among otherthings, 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 theoreticalbackground to both the NMR experiment and the continuum theory for the directordistribution; this section is also concerned with the determination of the surface

Page 332: Nanotechnology and Nano-Interface Controlled Electronic Devices

Static and dynamic director orientation in thin nematic liquid crystal films 315

anchoring energy. The NMR experiments are described in Section 3. The results forthe static director distribution are given in Section 4 where they are discussed in thecontext of the continuum theory analysis. The results for the director dynamics aredescribed in Section 5 where they are discussed on the basis of the hydrodynamictheory analysis. This allows the ratio Δε/Δχ , the diamagnetic anisotropy Δχ , thefield-induced relaxation times and the rotational viscosity coefficient to be determinedprovided Δχ 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, whichinteracts with the electric field gradient at the nucleus, to give a quadrupolar interactiontensor. This does not influence the number of lines in the deuterium NMR spectrum for anormal liquid because the rapid and random molecular motion averages the quadrupolarinteraction to zero. The NMR spectrum of a single deuteron, therefore, contains a singleline 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 rangeorientational order combined with the quadrupolar interaction of the deuterium nuclei.For a monodomain sample, in which the director is uniformly aligned, the NMR spec-trum consists of a single doublet (see Fig. 1b) this is also observed for sets of equivalentdeuterons provided the dipolar interaction is negligible in comparison with the linewidth.The separation, Δν, between the quadrupolar split lines is related to the componentsof the Saupe ordering matrix and the quadrupolar tensor, q. However, for aliphaticdeuterons 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 partiallyordered system (b).

Page 333: Nanotechnology and Nano-Interface Controlled Electronic Devices

316 A. Sugimura and G.R. Luckhurst

about the C–D bond direction, and then the quadrupolar splitting is given by [2]

Δν0 = 3

2qCDSCD, (1)

where SCD is the orientational order parameter for the C–D axis and qCD is thequadrupolar coupling constant. In this analysis the director is taken to be parallel tothe magnetic field. However, when the angle the director makes with the magnetic fieldchanges, the splitting will also change in a well-defined way. Thus if the director makesan angle, θ , with the magnetic field of the NMR spectrometer the quadrupolar splittingis given by [7]

Δν(θ) = Δν0 P2(cosθ), (2)

where Δν0 is the splitting when the director is parallel to the magnetic field (see Eq.1) and P2(cosθ) is the second Legendre polynomial. As the director moves away frombeing parallel to the field the splitting is predicted and observed to decrease, passthrough zero at the magic angle (θ = 54.74°) and then to increase to one half of theoriginal splitting, Δν0, when the director is orthogonal to the magnetic field. Strictlythe quadrupolar splitting changes sign at the magic angle but the sign of the splitting isnot directly available from the spectrum. The angular variation of the deuterium NMRspectrum predicted from Eq. 2 is illustrated in Fig. 2.

One of the prime advantages in the use of NMR spectroscopy to determine thedirector orientation is that the form of the spectrum is also influenced by the distributionof the director with respect to the magnetic field. In other words we can see fromthe spectrum whether the sample is a monodomain and if not the form of the directordistribution can be estimated given the aid of some theoretical prediction. This situationobtains because when the director is not uniformly aligned the observed spectrum isa weighted sum of the spectra from all director orientations provided the moleculardiffusion between different director orientations is slow on the NMR timescale, whichis 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 theobserved spectrum is given by

I (ν) =∫

L(ν, ν±(θ), T −12 )P(θ)sinθ dθ . (3)

In this expression L(ν, ν±(θ), T −12 ) denotes the Lorentzian shape of a spectral line

centered at either ν+(θ) or ν−(θ) and with a linewidth at half height of 2T −12 . The form

of the normalized Lorentzian lineshape is

L(ν, ν±(θ), T −12 ) = π−1T −1

2

/{(T −1

2

)2 + (ν − ν±(θ))2}

, (4)

where the angle dependent resonance frequency is

ν+(θ) = ν0 + (Δν0/2)P2(cosθ), (5)

with an analogous expression for the resonance frequency ν−(θ) for the other componentof the quadrupolar doublet. As an example of the spectrum observed when the directoris not uniformly aligned we consider the limiting case when the director is randomly

Page 334: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 themagnetic field. The value of θ is increased from 0 to 90° in steps of 2°. The simulated powder spectrum ofa 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. Thespectrum contains two dominant features at its center which come from the directorbeing perpendicular to the magnetic field and so have a splitting of approximatelyΔν0/2. The other feature is seen at the extremes of the spectrum and has a separationof approximately Δν0; these weak shoulders originate from the director being parallelto the magnetic field.

Deuterium NMR provides a very good method for investigating the director distribu-

Page 335: Nanotechnology and Nano-Interface Controlled Electronic Devices

318 A. Sugimura and G.R. Luckhurst

tion in a thin nematic cell subject to an electric field because of the sensitivity of thespectrum to the director orientation as indicated in Fig. 2. By varying the intensity of theelectric field, the total balance of the magnetic field, the electric field, the surface forceand the elastic torque can be controlled. In other words the application of an electric fieldmakes it possible to investigate the variation of the director distribution in a thin nematiccell with the electric field even in the presence of the strong magnetic field of the NMRspectrometer. However, to help interpret the deuterium NMR spectra, especially whenthe director is not uniformly aligned it is valuable to have some theoretical guidance asto the form of the director distribution. We shall, therefore, now consider the continuumtheory for the nematic director subject to both magnetic and electric fields as well assurface anchoring.

We begin with the total free energy, G, of the system in order to understand thedirector distribution in a thin nematic cell subject to a variety of torques. Here Gcontains four terms associated with elastic distortions, Gd, surface anchoring, Gs, andwith interactions of the nematic with the two fields, Gm and Ge, namely

G =∫

(Gd + Gm + Ge)dz + Gs, (6)

where the integral is over the z coordinate (corresponding to the direction normal tothe electrodes). The coordinate system defined by our experiments and used in thecalculations is shown in Fig. 2, in which the magnetic and electric fields are appliedparallel and normal to the electrodes, respectively. Because of the large difference in themagnitudes of the electric and magnetic susceptibilities of organic materials, the fieldenergy terms have different forms. For the magnetic field, the diamagnetic susceptibilityand its anisotropy are small (Δχ ≈ 10−6) so that the nematic does not significantlyperturb the applied magnetic field. In contrast for the electric field, both the permittivityand the permittivity anisotropy are large so that the electric field and the displacementcan be significantly altered by the presence of the nematic [22]. The contribution to thefree energy associated with the magnetic field and the diamagnetic anisotropy can bewritten as

Gm = −Δχ

2μ0(B · n)2, (7)

where μ0 is the permeability of free space and Δχ is the anisotropic diamagneticsusceptibility of the nematic. The free energy contribution due to the electric field hasthe analogous form

Ge = −ε0Δε

2(E · n)2, (8)

where ε0 is the permittivity of free space. The application of a voltage across the nematicfilm results in an electric displacement D,

Dα = ε0ε⊥Eα + ε0Δεnαnβ Eβ , (9)

where Δε = ε‖ − ε⊥, and ε‖ and ε⊥ are the principal values of the nematic dielectricsusceptibility tensor, parallel and perpendicular to the director, respectively. Assumingthat D and E vary only in the z direction, and neglecting the effects of space charge,

Page 336: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 fromEq. 9 as

Dz = [ε0ε⊥Eα + ε0Δε sin2 θ(z)]

E(z), (10)

where the dependence of the director orientation on the position in the cell is madeexplicit. The surface anchoring energy, GS, may be expressed as [23]

GS = − A

2(n · e)2. (11)

Here A is the surface anchoring strength and e denotes the easy axis or the anchoringdirection as given by de Gennes [24].

From the coordinate system defined in Fig. 2 the components of the director n, theelectric field, E, and the magnetic field, B, can be written as

n = (cosθ(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 theunified surface anchoring energy, leads to the basic equations from which to calculatethe director distribution θ(z) (see equations (18), (19), (22), and (23) in Ref. [25]). Theycan be expressed as

f (θ(z))d2θ(z)

dz2+ 1

2

d f (θ(z))

(dθ(z)

dz

)2

=[

Δχ

2μ0B2 − ε0Δε

2E2(z)

]sin2θ(z), (15)

f (θ(z))dθ(z)

dz

∣∣∣∣z=0

= A−

2sin2(θ0− − θ0−), (16)

f (θ(z))dθ(z)

dz

∣∣∣∣z=�

= − A+

2sin2(θ0+ − θ0+), (17)

⎧⎪⎪⎪⎪⎨⎪⎪⎪⎪⎩

d

dz

[ε0ε⊥ + ε0Δε sin2 θ(z)

]E(z) = 0,

V =�∫

0

E(z)dz,(18)

where

f (θ(z)) = K1 cos2 θ(z)+ K3 sin2 θ(z), (19)

and K1 and K3 are the splay and bend elastic constants, respectively, � is the thicknessof the nematic film, V is the voltage applied across the cell, and θ0− and θ0+ are thepretilt angles of the surface directors at z = 0 and z = �, respectively; these angles canchange depending on the torques acting on the surface directors. The angles made bythe easy axis at z = 0 and z = � are fixed; they are denoted by θ0− and θ0+, respectively.

Page 337: Nanotechnology and Nano-Interface Controlled Electronic Devices

320 A. Sugimura and G.R. Luckhurst

A− and A+ are the surface anchoring strengths at z = 0 and z = �, respectively. For oursystem we set θ0− = −θ0+ and A = A− = A+. The torque balance equations at bothsurfaces (see Eqs. 16–17), give boundary conditions to solve the torque balance equationin 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 filmas a function of the applied voltage.

Many of the parameters occurring in the continuum theory analysis and needed tocalculate the director distribution are already known. However, the determination of thesurface anchoring energy, A, presents more of a challenge. There is a formal relationshipbetween the anchoring strength and the saturation voltage, Vs, at which the directorbecomes completely homeotropic; that is, the director across the entire nematic filmincluding the boundary layers is oriented along the field direction. This, for a nematicfilm with a uniform director orientation, gives a formal relationship between Vs and A[21], namely

A = Vs√

ε0ΔεK3

�tanh

(Vs

2

√ε0Δε

K3

). (20)

The value of A can, therefore, be determined by measuring the saturation voltage. Asexpected from its definition, Vs can be determined precisely for the condition of zerooptical retardation and then the unified surface anchoring energy can be estimated fromEq. 20. Because the optical retardation is inversely proportional to the applied voltagein the high voltage regime, the intersection of the extrapolated line for the values of themeasured optical retardation and the horizontal inverse voltage axis gives the saturationvoltage, Us.

3. Experimental

The nematogen used for our studies of the static director distribution as well as for itsdynamics was 5CB-d2, which had been specifically deuteriated in the α-position of thepentyl chain. This was prepared using a procedure described elsewhere [26] but withthe reduction of the ketyl group performed using lithium aluminium deuteride ratherthan the hydride. A thin nematic sandwich cell was prepared. The glass plates werecoated with transparent In2O3 to act as electrodes. The cell was held together by aspecial glue which is stable in the presence of the cyanobiphenyls and which can becured using UV radiation for a few minutes. The saturation voltage method [21] wasemployed to measure the surface anchoring strength, A, at the interface of 5CB withthe substrate surface. All of the measurements were made at different temperatures inthe nematic phase. The arrangement of the nematic cell in the NMR spectrometer isshown schematically in Fig. 3. The spectra were recorded using a JEOL Lambda 300spectrometer, which has a magnetic flux density, B, of 7.05 T. The spectra were obtainedusing a quadrupolar echo sequence, with a 90° pulse of 7.7 μs and interpulse delay of40 μs. The post delay time was 30 ms. The number of free induction decays (FID) usedto produce spectra with good signal-to-noise varied from 2,000 to 10,000 depending on

Page 338: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 nematiccell was held in the NMR probe head so that the electric field, whose direction is normalto the substrate surface, makes an angle with the magnetic field. An amplifier and afunction 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. Bychanging the intensity of the electric field, the director orientation rotates in a planedefined by B and E, since the diamagnetic anisotropy and the dielectric anisotropy of5CB 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 thenematic cell of 100 μm thick was not treated so that the effective anchoring strengthwas essentially zero. The nematic cell was arranged in the NMR probe head so thatthe glass plates and the rubbing direction were aligned parallel to the magnetic field. Inour experiment the projection of the easy axis onto the glass surface is parallel to themagnetic field. The fine adjustment of the cell alignment was carried out by switchingthe electric field on while the sample was in the nematic phase and rotating the cell usingthe 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 electricfield (this uncertainty in the splitting means that the electric field can be arranged to beorthogonal 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 remainsparallel to the magnetic field giving rise to the large quadrupolar splitting. Then whenthe potential is about 60 V, corresponding to a threshold value of the electric field, a

Page 339: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 100μm nematic cell with untreated glass surfaces.

weak quadrupolar doublet appears in the spectrum with a splitting which is essentiallyhalf that of the original splitting. This shows that it comes from the director beingorthogonal to the magnetic field and so parallel to the electric field. As the potential isincreased so the intensity of the quadrupolar doublet associated with the director parallelto the electric field grows at the expense of that originating from the director parallel tothe magnetic field. When the potential is 70 V the director would seem to be completelyaligned parallel to the electric field. This voltage dependence of the quadrupolar splittingcan be understood by considering Eq. 15, which applies when B and E are orthogonalto each other, since the surface anchoring strength (Eqs. 16–17) is zero in this thickcell and so director deformation due to surface effects can be ignored. Accordinglythere is no director deformation in the nematic film and the director distribution shouldbe uniform giving a monodomain in the film. That is, the left hand side of Eq. 15can be set to be zero and this gives two kinds of solution. One is that the directoris oriented parallel to the magnetic field. This gives a spectrum with the maximumsplitting, which is shown by the lines with large dashes in Fig. 4. The other is withthe director aligned parallel to the electric field and this gives a spectrum with halfthe value of the splitting in zero electric field; this is indicated by the lines with smalldashes in Fig. 4. However, as we have noted there is a narrow voltage range over which

Page 340: Nanotechnology and Nano-Interface Controlled Electronic Devices

Static and dynamic director orientation in thin nematic liquid crystal films 323

Table 1

The nematic cells investigated

Cell no. Thickness TNI (K) Anchoring strength� (μm) A (×10−4 J/m−2)

Group 1 weak anchoring (A) 7.1 305.0 0.930(B) 36.8 305.2 0.194(C) 56.7 305.2 0.101

Group 2 strong anchoring (D) 7.6 305.3 34.6(E) 37.8 305.3 17.2(F) 56.7 305.0 4.20

both splittings are found at the same time; these two doublets appear at V = 62 V. Thesimultaneous appearance of the two doublets is, we believe, due to the inhomogeneityin 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 overthe entire cell; we are, therefore, using the nominal value of the spacer given by thesupplier. If the thickness variation over the entire cell is say from 100 to 105 μm, theelectric field strength over the nematic film has a gradient sufficiently large to give thetwo 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 thatdescribed in the previous subsection. Six thin nematic sandwich cells with differentthicknesses and anchoring strengths were prepared. The In2O3 coated glass surfaces ingroup 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 directoralignment. The surface anchoring strength at the interface of 5CB-d2 and the substratesurface for each cell used in the present experiments was measured using the saturationvoltage method described in Section 2. The surface anchoring strengths were foundto be in the range of 1.01–9.30 × 10−5 J/m2; this corresponds to the so-called weakanchoring condition. The surfaces of the cells in group 2, with (D) 7.6 μm, (E) 37.8μm, and (F) 56.7 μm thick cells, listed in Table 1, were coated with a thin polyimidefilm and also rubbed unidirectionally in a parallel manner to produce a uniform planardirector alignment. The surface anchoring strength was found to be in the range of4.20–34.6×10−4 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 area guide to the eye. It is obvious from these results that the surface anchoring strengthincreases with decreasing nematic film thickness, as previously reported [26,27]. Thenematic–isotropic transition temperature, TNI, for the nematic in each of the cells of thetwo 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 Kbelow TNI for the cells in the two groups

Page 341: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 thickand (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 weakanchoring are shown in Fig. 6 (A) (7.1 μm thick), (B) (36.8 μm thick), and (C) (56.7μm thick). The number of FIDs used to obtain spectra for each cell was 20,000. Thevoltage dependence of the spectra shows that with increasing electric field strength thequadrupolar splitting is reduced, passes through zero and then increases again to a value

Page 342: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 thickand (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 directororientation changes more or less continuously from being parallel to the magnetic fieldto being orthogonal to it, as the electric field grows. This change of the quadrupolarsplitting follows the pattern illustrated in Fig. 1 in accord with Eq. 2. It is of interestto note that the spectra are essentially, but not exactly, the same for the same electricfield strengths irrespective of the cell thickness, as expected from theory. Also as theelectric field is increased, the lines appear to broaden necessarily leading to a decreasein the signal-to-noise ratio. For higher electric fields, however, the lines sharpen againresulting in an improved signal-to-noise ratio. Thus the experimental results indicate acontinuous change in the director orientation with increasing electric field and a slightbroadening of the director distribution. This broadening is especially apparent in thecenter of the range when the director is at angle of 45° to the magnetic field, as expectedtheoretically [28].

The deuterium NMR spectra measured for the nematic cells with a strong anchoringare shown in Fig. 7 (D) (7.6 μm thick), (E) (37.8 μm thick), and (F) (56.7 μmthick). 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 issimilar 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 spectragiven in Figs. 6 and 7 change continuously with increasing electric field whereas for thethick cell with zero anchoring strength only two discrete bistable director orientations

Page 343: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 themagnetic field. The z1-axis is taken as being normal to the glass plate. The magnetic field (B), electricfield (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 substratesurface.

are observed. As we mentioned in the Introduction, the quadrupolar splitting is directlyrelated to the director orientation. That is, the experimental results in Figs. 6 and 7 showthat the director rotates continuously with increasing electric field from being parallel tothe magnetic field to being orthogonal to it and hence parallel to the electric field. It is tobe anticipated that the director deformation within the nematic film should be taken intoconsideration for the cells (A)–(F) with surface treatment because the surface anchoringaffects the director distribution in the nematic film. In other words the elastic torque inEq. 15 must be considered in order to understand the variation in the director orientationacross the cell with strong surface anchoring. However, it is generally expected for thethicker cell and zero anchoring strength that the director deformation is limited to thevicinity of the substrate surface and that the director orientation is constant over almostthe 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 thinkthat one reason for the observed continuous rotation of the director is associated with theexperimental geometry. As we have mentioned in the previous section, it is importantfor the observation of the bistable director orientation to set the cell precisely in theNMR 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 fieldexactly perpendicular to the magnetic field and so we consider here the more generalexperimental geometry given in Fig. 8 in which the electric field makes an arbitraryangle α with the magnetic field. Then the general torque-balance equation for the bulknematics, given in Eq. 15, should be changed to

f (θ(z))d2θ(z)

dz2+ 1

2

d f (θ(z))

(dθ(z)

dz

)2

= −Δχ

2μ0B2 sin2(θ(z)+α)− ε0Δε

2E2(z)sin2θ(z). (21)

Page 344: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 willbe uniform and so the elastic terms on the left hand side will vanish. Then the solution toEq. 21 will give a continuous variation in the director orientation with increasing electricfield for all the angles α between the two fields except 90°. When the surface forcescannot be ignored it is necessary to solve the torque balance equation, numerically, aswe shall see.

In general the director orientation is not expected to be uniform over the entirenematic film because of surface anchoring. Then for this non-uniform state the observedNMR spectrum is a weighted sum of spectra from all director orientations as we sawin Section 2. The voltage dependence of the deuterium NMR spectra shown in Figs. 6and 7 reflect the director distribution throughout the sample, which is determined by acombination of the unified surface anchoring, elastic, magnetic, and dielectric energies.However, the NMR spectra and their variation with the applied voltage suggest that thedirector is more or less uniformly aligned across the nematic film. Given the unexpectednature of this result we have sought to confirm it by simulating the spatial variationof the director. This is obtained theoretically via the numerical solution of Eqs. 16–18and 21 as a function of the electric field. In the numerical calculation of the directordistribution, Eqs. 16 and 17 give the boundary conditions to solve the torque balanceequation 21. The pretilt angles, θ0+ and θ0−, of the surface directors in Eqs. 16 and 17are variable because they can be changed by the external fields, both B and E. Thatis, Eqs. 16–18 and 21 have to be solved in a self-consistent manner. In our calculationsthe values B = 7.05 T, θ0+ = 6°, θ0− = −6°, Δχ = 1.5 × 10−6 [29,30], Δε = 6.75(ε⊥ = 7.16), K1 = 3.44×10−12 N, K3 = 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 valuesof α were determined from the experimental results. Figs. 9 (D), (E), and (F) showtypical examples of the simulation results for the director distribution θ(z/�) across theentire 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 atop-hat shape different from the bowler-hat shape, which is well-known as the directorprofile for the case with only an electric field. To emphasise the major effect of themagnetic 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 directororientation undergoes a significant variation across the film which we describe as thebowler-hat distribution. In both the top-hat and the bowler-hat distributions there isclearly an asymmetry which results from the easy axis at the surfaces making angleswhich are opposite in sign. It may be helpful to discuss the origin of the major differencein the shape of the director profile before starting our main discussion on the directordistribution. When a constant magnetic field is applied parallel to the substrate surface,the surface director tends to reduce its pretilt angle, θ0±. In other words, the pretiltangle of the surface director is also a function of the magnetic field strength. This isa completely different situation to that without the magnetic field. This means that astrong deformation energy is localized near the substrate surface in the initial directordistribution before the application of the electric field to the cell. This feeds back tothe bulk torque balance equation 21 and the boundary conditions, that is, Eqs. 16 and17 result in showing a top-hat shape for the director profile. Now we return to our

Page 345: Nanotechnology and Nano-Interface Controlled Electronic Devices

328 A. Sugimura and G.R. Luckhurst

Fig. 9: Typical continuum theory predictions for the director distribution θ (z/�) across the entire nematicfilm as a function of the electric field. Figures (D), (E), and (F) correspond to the director distributionpredicted for the strong anchoring cells (D), (E), and (F), respectively.

main discussion. The director is found to be aligned almost parallel to the magneticfield for low applied voltages. The pretilt angle of the director at the surface increasesslightly with increasing electric field, because of the strong anchoring condition. As

Page 346: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 butessentially continuous extent from being parallel to the magnetic field. In the bulk of thenematic the director is more or less uniformly aligned at a constant angle with respectto the magnetic field. This angle increases with increasing field until it reaches itslimiting value of α. Near the substrate surface the director orientation changes rapidlyand to an extent which increases with the electric field. However, as the thickness ofthe nematic film increases so the relative distance at the interface where the directorchanges its orientation becomes smaller. Similar numerical calculations provide thedirector distribution θ(z/�) over the entire nematic film for the weak anchoring cells asa function of the electric field. These simulations predict that the director distributionacross the film is almost homogeneous except in the vicinity of the interface. Theassociated NMR spectrum should, therefore, consist of a single quadrupolar doublet.It seems clear both for weak and strong anchoring that the inhomogeneity of thedirector distribution near the interface causes the broadening of the spectral lines andthe associated decrease of the signal-to-noise ratio. In order to test this interpretationquantitatively, it is necessary to simulate the deuterium NMR spectra directly by usingthe theoretical director distributions.

The NMR spectrum can be simulated from a knowledge of the director distributionas we saw in Section 2. However, since we know the director orientation for discretepositions across the cell it is convenient to replace the integral representation of thelineshape in Eq. 3 by the sum

I (ν) =1∑

z/�=0

L(ν,ν±(θ(z/�)), T −1

2

). (22)

Here θ(z/�) is the angle made by the director with respect to the magnetic fieldfor a particular slice in the nematic film at z/�. The deuterium NMR spectra havebeen simulated in this way from the theoretical results for the different directordistributions θ(z/�) and Eq. 22 as a function of the voltage for each cell. The numberof slices used was 200 within the nematic film. The value of the linewidth parameter,T −1

2 , was determined from the observed spectra to be 1 kHz and independent ofthe 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. Thesimulated spectra are shown in Figs. 10 (A), (B), and (C) for the weak anchoringcells, and Figs. 11 (D), (E), and (F) for the strong anchoring cells. In these figures thesolid lines indicate the simulated spectra and the dashed lines show the experimentalspectra which were taken from those shown in Figs. 4 and 5 in order to comparethe simulated results with experiment. The simulated spectra at different voltages arefound to be in surprisingly good agreement with the experimental spectra. The onlysignificant exception to this is for the spectra with the director close to the magicangle where the lines in the experimental spectra are observed to be broader than thosepredicted. To illustrate the possible origins of such discrepancies we consider the resultsfor cell A shown in Fig. 10 where the agreement between theory and experiment isat its worst. Cell A is 7.1 μm thick so that the amount of 5CB-d2 that it contains issmall which makes it difficult to obtain spectra with good signal-to-noise ratios. This

Page 347: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 thickand (C) 56.7 μm thick with weak anchoring as a function of the electric field strength. The solid linesindicate 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 solidlines 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 sharpand it is easy to process the FID leading to the spectrum. However, for electric fieldsof 0.79 and 0.90 V/μm the spectral baseline is not flat which results from the lowspectral intensity. At 0.75 V/μm (see Fig. 10) the lines in the experimental spectrum

Page 348: Nanotechnology and Nano-Interface Controlled Electronic Devices

Static and dynamic director orientation in thin nematic liquid crystal films 331

have broadened and in consequence the baseline is severely distorted. The origin ofthe 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 NMRspectrum when the director makes an average angle of 45° with the magnetic field.This occurs because the quadrupolar splitting is especially sensitive to small changesin the angle θ between the magnetic field and the director for this orientation (seeEq. (2)). It should be stressed, however, that the width of the spectral lines when theelectric field is 0.75 V/μm although large would only require a broadening of severaldegrees to account for it. The origin of the increased breadth of the director distributionmay result from the balance between the dielectric and magnetic torques at this pointso that other less controllable factors could influence the director distribution. Similarcomments can be made for the results obtained with cell E where the disagreementbetween 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 quiteremarkable agreement between the experimental spectra and those simulated using thedirector distribution obtained from continuum theory for quite different film thickness,potentials and surface anchoring strengths. In other words the combination of deuteriumNMR spectroscopy and continuum theory gives us a good understanding of the directordistribution 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 electrodeswere covered with polyimide its surface was rubbed unidirectionally in a parallelmanner to produce a uniform director orientation. The surface anchoring strength atthe interface of 5CB-d2 with the substrate surface is 1.2 × 10−4 J/m2 corresponding tostrong anchoring. All of the measurements were made at 303 K. The nematic cell washeld in the NMR probe head so that the glass plates and the rubbing direction werealigned parallel to the magnetic field (see Fig. 2). In this experiment the projection ofthe easy axis onto the glass surface was parallel to the magnetic field. An amplifierand a function generator were used to provide a 10 kHz sinusoidal ac electric field tothe cell. This frequency is sufficient to provide a time resolution of 0.1 ms during theturn-on dynamics measurements. On applying or removing the electric field, the directororientation rotates in a plane defined by B and E.

Figs. 12 (A) and (B) show schematically the pulse sequences for the observation ofthe spectra during (A) the turn-on and (B) turn-off processes. The triangular symbols inthe 2H observation pulse sequences indicate data acquisition during the free inductiondecay. ta is the time during which the external voltage is applied, tr is the time allowedfor 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 directorrelaxation was monitored at several values of ta between 0 and 3 ms following theapplication of the electric field (60.5 V). For the turn-off process, the director relaxation

Page 349: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 2H observation pulse sequences denote data acquisition, ta is the time duringwhich the external voltage is applied, tr is the time allowed for director relaxation, tD is the dead time of thereceiver 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 Vwas applied for 3 ms to obtain the initial director alignment orthogonal to the magneticfield.

Page 350: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 measuredby changing the application time, ta, for the voltage.

The deuterium NMR spectra recorded during the turn-on and turn-off processes areshown in Figs. 13 and 14, respectively. In the turn-on process, the spectra show that thealignment process is complete within 3 ms when the quadrupolar splitting is reducedto half of its initial value (the director is now parallel to the electric field which isorthogonal to the magnetic field). Following the application of the electric field thedirector remains parallel to the magnetic field for up to 0.7 ms before it changes itsorientation, i.e., there is an induction period of 0.7 ms for the turn-on process. Then afterthis induction period the lines begin to develop an asymmetric and powder-like shape asthe director moves away from being parallel to the magnetic field. This major broadeningof the spectral lines for times up to 1.1 ms or more shows the presence of a broad directordistribution 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 achievedan orientation with θ ≈ 54.74°. After this the director appears to move now more or lessas a monodomain until at t = 3 ms it is completely aligned along the electric field andorthogonal to the magnetic field. One other feature of the alignment process in this cellwhich is worth pointing out is the presence of a significant fraction of the sample withthe director parallel to the magnetic field for a fairly long time (up to 1.1 ms).

Page 351: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 measuredat a time tr after the voltage was turned off.

In the turn-off process, on the other hand, see Fig. 14, changes in the directororientation are observed at 1 ms after the removal of the electric field. For times shorterthan 1 ms, the director is observed to remain in its initial orientation parallel to theelectric field. In other words, and as expected, the turn-off process also exhibits theexpected induction period. At t = 1 ms a second quadrupolar splitting, with somewhatbroader lines, is observed which corresponds to the director oriented parallel to themagnetic field. Furthermore, the spectrum at t = 1 ms shows that the original lines atθ = 90° become broader and there is an increase in the spectral intensity in the regionbetween these two lines. This indicates a spread in the director distribution betweenthe extremes of 90° and 0°. Indeed as time evolves there is an increase in the spectralintensity in the region between the parallel and the perpendicular features of the spectrashowing 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 timeas the director distribution alters to give more of the director parallel to the magneticfield. The spectrum recorded after 2.5 ms is of interest as it shows a maximum spectralintensity corresponding to an intermediate director orientation between 0° and 90°. Aftera further 0.2 ms the intensity of the parallel features has increased at the expense of the

Page 352: Nanotechnology and Nano-Interface Controlled Electronic Devices

Static and dynamic director orientation in thin nematic liquid crystal films 335

perpendicular while at the same time the splitting originating from intermediate directororientations has increased showing that the angle between the director and the magneticfield has decreased. These processes continue until after 7 ms the director is almostuniformly aligned parallel to the magnetic field (although there are still two very weakpeaks coming from the director at θ = 90°). In summary the director starts uniformlyaligned perpendicular to the magnetic field then passes through a regime in which itsorientational distribution is very broad before becoming aligned parallel to the magneticfield. Such changes are consistent with those observed for other systems where there isa degeneracy in the alignment pathway for the director [10].

The alignment process for the change in the director orientation following the turn-onand turn-off of the electric field is clearly very complex. The director does not realignas a monodomain and this is perhaps to be expected either because of the degeneracy inthe alignment pathway or because of the surface alignment. In addition the alignmentprocess appears to be qualitatively and quantitatively different for the turn-on andturn-off process presumably because of the difference in the magnitude of the fieldtorques responsible for the alignment. Analysis of the spectral lineshapes which indicatea non-uniform distribution of the director can, in principle, yield not only informationabout 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 electrodeswere 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 themeasurements were made at different temperatures in the nematic phase of 5CB. Thenematic cell was held in the NMR probe head so that the electric field, whose directionis normal to the substrate surface, makes an angle, α ≈ 45°, with the magnetic field (seeFig. 8). Fig. 15 shows the spectrum recorded without an electric field at 295 K; thequadrupolar splitting, Δν0, is 54.2 kHz. The weak feature in the centre of the spectrumoriginates from a small amount of the isotropic phase of 5CB-d2 possibly dispersed inthe glue holding the cell together. The final adjustment of the cell alignment to ensurethat the electric field makes an angle of essentially 45° with the magnetic field wascarried out by switching on a large electric field (100 V) and rotating the cell by a fewdegrees, clockwise or counter-clockwise, using the goniometer of the spectrometer untila 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 themagnetic field). However, the electric field will not be at exactly 45° with respect to themagnetic field because the value of 100 Vrms is insufficient to align the director parallelto the electric field. An amplifier and a function generator were used to provide a 10 kHzsinusoidal 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 previousSubsection 5.1 (see Fig. 12a and b). For the turn-on process the director relaxation wasmonitored at several values of ton between 0 and 25 ms following the application of the

Page 353: Nanotechnology and Nano-Interface Controlled Electronic Devices

336 A. Sugimura and G.R. Luckhurst

Fig. 15: The deuterium NMR spectrum of 5CB-d2 recorded without an electric field at 295 K, thequadrupolar splitting frequency, Δν, 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 atseveral values of toff in the range 0 to 25 ms. An electric field of 50 V was applied forta = 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 295K are shown in Fig. 16a and b, respectively. In the turn-on process, the quadrupolarsplitting decreases and then saturates with time after about 3 ms. In the turn-offprocess, the quadrupolar splitting increases because the director moves from being atapproximately 30° to the magnetic field to being parallel to it. The spectra recordedin the fast time region for the turn-on and turn-off processes contain weak oscillatoryspectral features associated with the director rotation during the acquisition for thefree 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 notethat the oscillations appear on the inside of the quadrupolar doublet during the turn-onprocess but on the outside of this doublet for the turn-off process. They therefore givethe sense of the director motion. The time resolved director orientation can be easilydetermined from our experimental results contained in the quadrupolar doublets becauseof 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 slightbroadening of the spectral lines [33], then an appropriate form of the time dependenceof the director orientation can be determined relatively easily. This is a good assumptionsince there is no elastic deformation in the nematic slab used in our experiments. Theobserved quadrupolar splitting can then be used together with Eq. (2) and the value ofthe splitting when the director is parallel to the magnetic field to determine the directororientation.

According to continuum theory [34], we should consider the one-dimensional

Page 354: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 for5CB-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 spectrain our study appear to be dominated by a single doublet which allows us to determinethe director orientation associated with an essentially monodomain sample. In thisanalysis, 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 thedirector orientation is given, for the turn-on process, by the torque-balance equation [35]which for a monodomain nematic [14] is

γ1dθ(t)

dt= −Δχ

2μ0B2 sin2θ(t)+ ε0Δε

2E2 sin2(α − θ(t )). (23)

The solution of Eq. 23 is obtained analytically as [14]

θ(t) = θ∞ + tan−1 [tan(θ0 − θ∞)exp(−t/τ )], (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

cos2θ∞ = Um +Ue cos2α(U2

m +2UmUe cos2α +U2e

)1/2 , (25)

Page 355: Nanotechnology and Nano-Interface Controlled Electronic Devices

338 A. Sugimura and G.R. Luckhurst

where

Um = Δχ

2μ0B2, Ue = ε0Δε

2E2,

are the strengths of the magnetic and electric interactions, respectively. The relaxationtimes for the turn-on (τon) and turn-off (τoff) processes are then

τon = (τ−2m +2τ−1

m τ−1e cos2α + τ−2

e

)−1/2, (26)

and

τoff(≡ τm) = γ1/(2Um), (27)

τe = γ1/(2Ue), (28)

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 = R2 sin2 2θ∞R2 −2R cos2θ∞ +1

, (29)

Um

Ue= B2

μ0ε0 E2

(Δχ

Δε

)= 1√

R2 −2R cos2θ∞ +1, (30)

where

R = τoff/τon. (31)

Eqs. 29 and 30 give the values of α and Δχ/Δε, respectively, by using τon, τoff and θ∞.It is also evident from Eqs. 27 and 30 that Δχ and γ1 can only be determined if Δε isknown from an independent experiment.

Fig. 17 shows the temporal variation in the ratio of the quadrupolar splittings,Δν(t)/Δν0, 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 theturn-on process the director rotates from the initial angle θ0 = 0° for Δν0 = 54.2 kHzand then aligns at the limiting angle, θ∞, of 29.8° for Δν∞ = 34.1 kHz (see Eq. 2), thelimiting value of Δν(t ) as t tends to infinity. The limiting value of θ∞ was obtainedby using Δν∞ and Eq. 2. In the turn-off process, the time dependence of the directororientation was obtained in the same way and is shown in Fig. 17. The director rotatesback parallel to the magnetic field and the time taken for the alignment process isslower in the turn-off process than in the turn-on process. This is clearly apparent fromthe results listed in Table 2 but is not so obvious from the different time dependencesof the quadrupolar splittings shown in Fig. 17. The large difference in the relaxationtimes for the turn-on and turn-off processes follows from the theory. Thus the relaxationtime, 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 itsorientation, α, with respect to the magnetic field (see Eq. 26). In our experiments α isapproximately 45° so that the cross term in the denominator is negligible, accordinglythe quadratic term in τe simply adds to that in τm thus increasing the denominator andso decreasing the relaxation time τon in comparison with τoff, as we find experimentally.

Page 356: Nanotechnology and Nano-Interface Controlled Electronic Devices

Static and dynamic director orientation in thin nematic liquid crystal films 339

Fig. 17: The time dependence of the ratio, Δν/Δν0, for (a) the turn-on and (b) the turn-off processesdetermined from the deuterium NMR spectra of 5CB-d2 shown in Fig. 16a and b at 295 K, in whichthe experimental value, Δν0 = 54.2 kHz (see Fig. 2), has been used. For the turn-on process, the directoraligns towards the limiting value (θ∞ = 29.8°) with time. For the turn-off process, θ relaxes back fromθ0 = 28.8° to θ∞ = 0°. The solid lines in (a) and (b) are the best fits to Δν/Δν0 = 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 theturn-off.

From the experimental values of Δν0 and Δν∞ combined with Eqs. 2 and 24, thevalues of the two relaxation times, τon and τoff, were obtained by fitting the ratio of thequadrupolar splittings as a function of time for the turn-on and turn-off processes at eachtemperature. The solid lines in Fig. 17 show the best-fit lines giving the values of τon andτoff to be 0.766 ms and 1.54 ms, respectively. We have chosen to fit the time dependenceof the quadrupolar splittings rather than that for the director orientation calculable from

Table 2

The temperature dependence of the quadrupolar splitting, Δν0, the ratio, Δν∞/Δν0, of the quadrupolarsplittings with and without the electric field, the turn-on, τon, and turn-off, τoff, relaxation times and theangle, α, between the electric and magnetic fields

T (K) Δν0 (kHz) Δν∞/Δν0 τr (ms) τd (ms) α (°)

295 54.2 0.630 0.766 1.54 44.7297 52.1 0.632 0.672 1.34 44.7299 49.6 0.636 0.594 1.18 44.6301 46.4 0.636 0.514 1.02 44.6303 44.6 0.627 0.443 0.88 44.7303.5 42.6 0.636 0.424 0.84 44.7304 41.2 0.637 0.404 0.80 44.5304.5 39.5 0.636 0.384 0.76 44.7305 37.6 0.634 0.348 0.70 44.7305.5 34.7 0.623 0.314 0.62 44.7

Page 357: Nanotechnology and Nano-Interface Controlled Electronic Devices

340 A. Sugimura and G.R. Luckhurst

Table 3

The temperature dependence of the dielectric anisotropy, Δε [29], the diamagnetic anisotropy, Δχ , therotational viscosity coefficient, γ1, the ratio of the anisotropic susceptibilities, Δχ/Δε, and the ratio of thediamagnetic anisotropy to the quadrupolar splitting, Δχ/Δν0

T (K) Δε Δχ (×106) γ1 (Pa s) (Δχ/Δε) (×107) (Δχ/Δν0) (×1011 kHz)

295 11.4 1.17 0.0710 1.03 2.15297 11.2 1.16 0.0614 1.04 2.22299 10.8 1.13 0.0526 1.04 2.27301 10.2 1.07 0.0432 1.05 2.31303 9.40 0.973 0.0339 1.04 2.23303.5 9.19 0.962 0.0319 1.05 2.26304 8.95 0.940 0.0297 1.05 2.29304.5 8.68 0.909 0.0273 1.05 2.30305 8.40 0.865 0.0239 1.03 2.30305.5 8.15 0.846 0.0207 1.04 2.44

it because the splitting constitutes the primary experimental data and the absolute errorassociated with each point is more or less constant. The values for these relaxation timesat other temperatures were obtained in the same manner and are listed in Table 2. Anaccurate value for the angle α between the magnetic and electric fields was calculatedby 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, Δχ/Δε, via Eq. 30, the results for thisimportant 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 wastaken as 306 K. It is immediately apparent that the ratio is independent of temperaturewithin 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 Δχ/Δε and Δχ/Δν0 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.

Page 358: Nanotechnology and Nano-Interface Controlled Electronic Devices

Static and dynamic director orientation in thin nematic liquid crystal films 341

with temperature. This may have resulted from the fact that Δε and Δχ were measuredin independent experiments unlike our results which are determined directly as the ratiofrom the NMR experiment. Although the results reported by Bunning et al. [37] arealmost certainly in error, it is of interest to consider their explanation for the temperaturedependence. They noted that mesogenic molecules are not cylindrically symmetricso that the contribution of the biaxial ordering of the molecule to the anisotropicsusceptibilities must be included. In general the anisotropy Δ A (= A‖ − A⊥) of somepartially averaged second rank tensorial property, A, is given by

Δ A = (3/2)Szz A′zz + (3/4)(Sx x − Syy)(A′

x x − A′yy). (32)

Here A′ is the traceless tensor set in a molecular frame and S is the Saupe orderingmatrix given in the same frame. For convenience the molecular frame xyz is usuallytaken to be the principal axis system for the ordering matrix; then Szz is the major orderparameter and (Sx x − Syy) is the biaxial order parameter and measures the differencein the ordering of the axes orthogonal to z, the axis which approximates closest to themolecular symmetry axis. The ratio, Δχ/Δε, is then given by

Δχ

Δε= Szzχ

′zz + (1/2)(Sx x − Syy)(χ ′

x x −χ ′yy)

Szzε′zz + (1/2)(Sx x − Syy)(ε′

x x − ε′yy)

, (33)

where the influence of internal electric fields have been ignored and ε′ is in essencean effective traceless dielectric tensor for a perfectly oriented system. Since the majorand the biaxial order parameters have different temperature dependences it is to beexpected that the ratio, Δχ/Δε, will, in general, be temperature dependent even thoughχ ′ and ε′ are usually temperature independent. The simplest way to ensure that Δχ/Δε

is temperature independent is if the biaxial terms are small in comparison with themajor terms. Certainly for 5CB (Sx x − Syy) is extremely small in comparison withSzz [38] and so provided the biaxialities in the associated molecular quantities arenot abnormally large then the biaxial contributions to Δχ/Δε (see Eq. 33) can beneglected leaving the major order parameter which will cancel and the ratio will beindependent of temperature as we find. This independence would also be expected if thebiaxial order parameter was linear in Szz which can occur for temperatures close to thenematic–isotropic transition which is the region in which our measurements are made.It seems likely that both factors contribute to the temperature independence which wehave observed.

The angle α between the electric and magnetic fields can also be determined fromour 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 aninternal check on our experiments and analysis for once the sample was positioned inthe probe head of the NMR spectrometer this was not changed for the studies of thesample at different temperatures. Accordingly α should not change with temperaturewhich is in accord with our results.

In order to calculate the values of Δχ and γ1 from our results we need the valueof Δε at the same temperature. For this we have used the dielectric anisotropiesdetermined by Dunmur and Miller [29]. The resultant temperature dependences of Δχ

Page 359: Nanotechnology and Nano-Interface Controlled Electronic Devices

342 A. Sugimura and G.R. Luckhurst

Fig. 19: The dependence of γ1 and Δχ for 5CB-d2 on the reduced temperature, T/TNI. The open circles andsolid line indicate the temperature dependences of Δχ 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 thepresent work (see Table 3) and the literature [31], respectively.

and γ1 are listed in Table 3 where Δε used in the calculation is given. Fig. 19 showsthe dependence of Δχ and γ1 of the reduced temperature. The open circles and solidline indicate the temperature dependence of Δχ from the present work and the literature[30,39], respectively. The open and closed triangles indicate the temperature dependenceof γ1 from the present work and the literature [31], respectively. In Fig. 19 the dashedlines for the present work are only a guide for the eye. We note that our valuesof Δχ and γ1are in relatively good agreement with those obtained by other studies.The temperature dependence of Δχ will be discussed later. Of course, analogousinformation concerning the director dynamics could be obtained by other methods, suchas capacitance and transmitted light measurements. However, it is difficult to determineboth the diamagnetic anisotropy and the rotational viscosity at the same time. On theother hand, using deuterium NMR measurements, the diamagnetic anisotropy and therotational viscosity are both determined using the simple relation between the relaxationtimes for the turn-on and turn-off processes. In addition, deuterium NMR spectroscopypresents the possibility of being able to explore the spatial distribution of the director innematic cells as a function of time during the turn-on and turn-off processes. This is ofspecial importance because it allows us to see if the director is uniformly distributed sothat the hydrodynamic theory which we have used is strictly applicable.

We now turn to a consideration of the temperature variation of the diamagneticanisotropy. As we have noted Δχ should be linear in the second rank orientational orderparameter for the effective symmetry axis. We can check this because the quadrupolarsplitting, Δν0, is also proportional to the same order parameter provided the moleculebehaves as if it is cylindrically symmetric which as we have seen is likely to be thecase. Accordingly the ratio, Δχ/Δν0, should be temperature independent. Its value for

Page 360: Nanotechnology and Nano-Interface Controlled Electronic Devices

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. Thedashed 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 ofthe electric field to that of the magnetic field (θ∞ < α). For Um ≥ Ue (e.g. Um/Ue = 1.2), θ∞ tends as beingsatisfied with the condition of θ∞ < α/2.

each temperature is listed in Table 3 and shown in Fig. 18 as a function of the reducedtemperature. The data clearly show that Δχ/Δν0 is indeed independent of temperature.

We also discuss the physical meaning of Eq. 30 with Eq. 29. These equationsgive Um sin2θ∞ = Ue sin2(α − θ∞), which is equivalent to that derived from the timeindependent torque balance equation for the same coordinate system (see Fig. 8). Forthe critical case of α = 90°, it gives two solutions of θ = 0° and θ = 90° independentof the intensities of B and E . This prediction has been confirmed experimentally for amonodomain nematic using deuterium NMR spectroscopy [12]. For α �= 90°, we finda relationship between α and θ∞ which depends on Um/Ue. Fig. 20 shows typicalexamples of the calculated α dependence of θ∞ for different values of Um/Ue, inwhich the dashed line shows the linear relationship between α and θ∞ in the limit ofan infinite electric field, that is Um � Ue. The curve for Um/Ue = 0.6 corresponds tothe 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 ifa higher electric field would be applied (Um/Ue = 0.2), θ∞ �= 45°for α = 45°. In orderto obtain θ∞ = 45° for α = 45°, the condition of Um � Ue is required. The simultaneousapplications of both the magnetic and electric fields to the nematic film, means thatthe director is not oriented parallel to the electric field (θ∞ �= α), but deviates fromthis 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

Page 361: Nanotechnology and Nano-Interface Controlled Electronic Devices

344 A. Sugimura and G.R. Luckhurst

have revealed several important features. First the widths of the spectral lines aresymmetric and relatively sharp which indicates that the director distribution is narrowand does not change during the alignment process. In other words the nematic is alignedas a monodomain. This is important because it allows the simplified torque-balanceequation 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 orientationpredicted by theory is in good agreement with the experimental results. In an attemptto understand why the theory which neglects the elastic energy and surface anchoringis so effective in explaining our results we have calculated the time dependence ofthe director orientation across the cell without ignoring these two effects. Accordingto the hydrodynamic theory, we should consider the time dependence of the one-dimensional director distortion across the nematic film. Here we introduce an angle,φ(z1, t ) (= θ(z1, t ) + 90° −α), made by the director with the glass plate, and a newcoordinate system, in which the z1-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 theextended torque-balance equation [24]

γ1∂φ(z1, t )

∂ t= f (φ)

∂2φ

∂z21

+ 1

2

∂ f (φ)

∂φ

(∂φ

∂z1

)2

+ Δχ

2μ0B2 sin2(α +φ)+ ε0Δε

2E2(z1)sin2φ, (34)

f (φ) = K1 cos2 φ + K3 sin2 φ. (35)

An initial director distribution, φ(z1,0), and the boundary conditions are needed tocalculate the time-dependence of the director orientation numerically using Eq. 34. Inour first attempt to calculate the time evolution of the director distribution we have usedEq. 34 without any further assumptions . The torque balance equations at both surfacesare given by Eqs. 16 and 17. These equations give the boundary conditions needed tosolve the torque balance equation in the nematic film and the variation of the electricfield across the film (see Eq. 18).

Numerical solution of Eqs. 16–18 and 34–35 gives the one-dimensional directordistortion across the nematic film as a function of time. In our calculations the valuesB = 7.05 T, � = 56.1 μm, φ0 = 4° at z1 = 0, φ0 = 4° at z1 = �, A = 1.0 × 10−7 J/m2,V = 50 V, α = 44.7° (see Table 2), Δχ = 1.17 × 10−6 (see Table 2), γ1 = 0.0714 Pa s(see Table 2), K1 = 8.42×10−12 N [31], K3 = 9.96×10−12 N [31], and Δε = 11.4 [39]at 295 K were used. In the calculations the initial director distribution for the turn-onprocess (t = 0 ms in Fig. 21a) is obtained as that after t = 0.1 ms using Eqs. 16–17 and34–35 for V = 0. The initial director distribution for the turn-off process (t = 0 ms inFig. 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 forthe turn-on process and 21b for turn-off process show typical examples of the simulationresults for the director distribution θ(z1/�) across the entire nematic film as a function oftime. These predicted distributions for the variation of the director orientation with timethroughout the nematic film show an interesting form. In Fig. 21 (a) most of the director

Page 362: Nanotechnology and Nano-Interface Controlled Electronic Devices

Static and dynamic director orientation in thin nematic liquid crystal films 345

Fig. 21: Typical examples of the simulated director distribution θ (z1/�) across the entire nematic film as afunction 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 nematicfilm the director orientation in the vicinity of the glass surfaces deviates slightly fromthis angle because of the weak surface anchoring. The director orientation increaseswith time and saturates to θ∞ at t = 7.1 ms. It is apparent from this calculation that thedirector is almost completely oriented as a monodomain as found by the experiments. Inthe turn-off process the director is also oriented with time as a monodomain as shown

Page 363: Nanotechnology and Nano-Interface Controlled Electronic Devices

346 A. Sugimura and G.R. Luckhurst

in Fig. 21b. Most of the director (> 99%) is oriented at θ = 28° after 0.1 ms. Thedirector orientation angle decreases with time and reaches θ∞ = 0 at t = 7.1 ms. Wehave also compared the time dependence of the director orientation obtained from thecomplete 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 agreementwith the predictions of the simplified hydrodynamic theory. These complete simulationresults for the spatial and temporal variation of the director distribution support theanalysis of the director dynamics for a thin nematic slab with weak boundaries using thetorque-balance equation 23 for a monodomain. This behaviour is entirely analogous tothat which we see in Section 4.2 for the static director distribution across a nematic cellfor varying field strengths.

The ability of the time resolved director distribution enables us to calculate thespectra 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 spectrumshown in Fig. 15. The simulated deuterium NMR spectra corresponding to these timeresolved 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 ina very good agreement with the experimental spectra shown in Fig. 16a and b. Thusalthough the recorded spectra contain weak oscillatory spectral features associated withthe director rotation during the acquisition of the FID [32], the quadrupolar splittingsof the calculated spectra at each time show a complete coincidence with those of theexperimental spectra. It is apparent from this significant agreement that the physicalparameters used in the simulations reflect the reliability of the values of the nematic forthe physical parameters, especially for the magnetic anisotropy and rotational viscositycoefficient obtained from the experiment. This means that the hydrodynamic theory canaccount for the experimental results and so the reorientation of the director towardsequilibrium follow exactly that predicted by Eq. 23.

6. Conclusion

Deuterium NMR spectroscopy has been employed to investigate the spatial directordistribution in a thin nematic film. The dependence of the quadrupolar splitting on theelectric field strength has been measured for nematic cells, with no surface forces, aswell as with weak and strong anchoring. When there are no surface forces we find fororthogonal fields that, at a certain critical value of the electric field strength, a suddenchange of the director orientation from being parallel to the magnetic field to beingparallel to the electric field. This behaviour can be explained by the absence of elasticdeformations in the thin nematic film and the critical orthogonal relationship betweenthe opposing fields.

For cells with weak anchoring, the director orientation changes continuously frombeing parallel to the magnetic field to being orthogonal to it. One cell with stronganchoring showed not only a continuous change in the director orientation as for theweak anchoring cells and the other strong anchoring cells but also the appearance ofa quadrupolar doublet due to the presence of a small, but significant, fraction of the

Page 364: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 ofthe glass plates. We have interpreted all of these results using continuum theory in termsof changes in the director distribution due to the presence of surface anchoring, elastic,electric and magnetic energies. The electric field dependence of the quadrupolar splittingcan 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 thedirector pretilt at the surface causes the director to be uniformly aligned across thenematic film which is an especially novel result. Our experimental results and theircomparison with theory show that deuterium NMR provides a very valuable techniquewith which to investigate the director distribution in thin liquid crystal samples.

Deuterium NMR spectroscopy has also been used to investigate the director dynamicsin the nematic slab, 56.1 μm thick, of 5CB with weak anchoring and time-resolvedNMR spectra have been obtained. The nematic cell was held in the NMR probe sothat the electric field makes an angle of α = 44.7° with the magnetic field. Thisexperimental geometry allowed us to avoid the degeneracy of the realignment pathwayfor the director found for larger angles. A series of deuterium NMR spectra, obtainedusing a quadrupolar echo sequence, was acquired as a function of time. When theelectric field, whose intensity is controlled so that the director makes a non-zero anglewith the magnetic field is applied to the nematic film, the director moves from beingparallel to the magnetic field to being at an angle θ with respect to it. After theelectric field is switched off, the director relaxes back to being parallel to the magneticfield. In this way a non-equilibrium director orientation with respect to the field canbe created without causing flow to occur within the nematic sample. Deuterium NMRspectra were recorded during the turn-on and the turn-off realignment processes as afunction of time. We have studied the time dependence of the director orientation for theturn-on and turn-off processes at different temperatures in the nematic phase. From themeasured relaxation times, we have determined the rotational viscosity coefficient andthe diamagnetic anisotropy of 5CB at different temperatures. Both of these quantitieswere found to be in reasonably good agreement with values reported in the literature.The ratio of the anisotropic susceptibilities, Δχ/Δε, was found to be independent oftemperature in agreement other but not all experimental studies; it can be understoodin terms of the effective cylindrical symmetry of the 5CB molecules. The time andspatial variation of the complete director dynamics were predicted by a hydrodynamictheory and found to be in good agreement with our experimental results as well as thepredictions of the simple theory in which elastic deformations and surface anchoringare ignored. The present results indicate that deuterium NMR spectroscopy providesa valuable technique with which to investigate the turn-on and the turn-off alignmentprocesses as a function of time in thin nematic slabs.

Deuterium NMR spectroscopy has also been used to investigate the dynamic directoralignment process following the application or removal of an electric field in the nematicliquid crystal subject to orthogonal magnetic and pulsed electric fields [15]. The timedependence of the NMR spectra has been measured for a thin nematic film. The directorwas found to be rotated from being orthogonal to the magnetic field to being parallelto it when the electric field, whose direction was essentially perpendicular to that ofthe magnetic field, was switched off. The experimental spectra show that the director

Page 365: Nanotechnology and Nano-Interface Controlled Electronic Devices

348 A. Sugimura and G.R. Luckhurst

is uniformly aligned orthogonal to the magnetic field in the initial state and that thedirector uniformly oriented parallel to this field in the final state. The intermediatestates are observed to correspond to a wide director distribution. In the realignmentpathway, the directors can rotate equally probably clockwise or counterclockwise andthis degeneracy induces a backflow effect. This can be included within the theory byusing 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 balanceequation including a time dependent viscosity torque term with an effective rotationalviscosity. For the experimental spectra found during the passage of the director frombeing orthogonal to being parallel to the magnetic field, however, the spectral linesare observed to be broader than those predicted. The domain walls resulting from thedegeneracy in the director realignment could cause the failure of the theory to predictthe broad director distribution. Although this particular investigation reveals the lack ofa good theoretical analysis for the director distribution, deuterium NMR spectroscopyis clearly providing a very valuable technique with which to investigate the directordistribution in nematic films both dynamic and static.

Acknowledgements

This work was supported by a Scientific Grant-in Aid from the Japan Society for thePromotion of Science (JSPS) and was carried out as an Anglo-Japanese joint researchproject of the International Exchange program supported by The Royal Society andJSPS.

References

1. J.C. Rowell, W.D. Phillips, L.R. Melby, and M. Panar, J. Chem. Phys. 43, 3442 (1965).2. G.R. Luckhurst, J. Chem. Soc. Faraday Trans. 2 84, 961 (1988).3. A.F. Martins, P. Esnault, and F. Volino, Phys. Rev. Lett. 57, 1745 (1986).4. M. Winkler, D. Geschke, and P. Holstein, Liq. Cryst. 17, 283 (1994).5. M. Bender, P. Holstein, and D. Geschke, J. Chem. Phys. 113, 2430 (2000).6. R. Stannarius, G.P. Crawford, L.C. Chien, and J.W. Doane, J. Appl. Phys. 70, 135 (1991).7. S.M. Fan, G.R. Luckhurst, and S.J. Picken, J. Chem. Phys. 101, 3255 (1994).8. 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).9. E. Ciampi, J.W. Emsley, G.R. Luckhurst, and B.A. Timimi, J. Chem. Phys. 107, 5907 (1997).

10. P. Esnault, J.P. Casquilho, F. Volino, A.F. Martins, and A. Blumstein, Liq. Cryst. 7, 607 (1990).11. 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).12. G.R. Luckhurst A. Sugimura, and B.A.Timimi, Mol. Cryst. Liq. Cryst. 347, 297 (2000).13. G.R. Luckhurst, T. Miyamoto, A. Sugimura, T. Takashiro, and B.A. Timimi, J. Chem. Phys. 114,

10493 (2001).14. C.J. Dunn, G.R. Luckhurst, T. Miyamoto, A. Sugimura, and B.A. Timimi, Mol. Cryst. Liq. Cryst.

347, 167 (2000).15. G.R. Luckhurst, T. Miyamoto, A. Sugimura, and B.A. Timimi, Thin Solid Films 393, 399 (2001).

Page 366: Nanotechnology and Nano-Interface Controlled Electronic Devices

Static and dynamic director orientation in thin nematic liquid crystal films 349

16. G.R. Luckhurst, T. Miyamoto, A. Sugimura, and B.A. Timimi, Mol. Cryst. Liq. Cryst. 347, 167(2000).

17. G.R. Luckhurst, T. Miyamoto, A. Sugimura, and B.A. Timimi, J. Chem. Phys. 117, 5899 (2002).18. G.R. Luckhurst, Mol. Cryst. Liq. Cryst. 347, 121 (2000).19. G.R. Luckhurst, T. Miyamoto, A. Sugimura, and B.A. Timimi, Mol. Cryst. Liq. Cryst. 347, 147

(2000).20. G.R. Luckhurst, T. Miyamoto, A. Sugimura, and B.A. Timimi, Mol. Cryst. Liq. Cryst., in press.21. A. Sugimura, T. Miyamoto, M. Tsuji, and M. Kuze, Appl. Phys. Lett., 72, 329 (1998).22. H.J. Deuling, E.Guyon, and P.Pieranski, Solid State Comm. 15, 277 (1974).23. B. Jérôme, Rep. Prog. Phys. 54, 391 (1991).24. P.G. de Gennes and J. Prost, The Physics of Liquid Crystals, 2nd ed. (Oxford University Press,

Oxford, 1993).25. A. Sugimura, G.R. Luckhurst, and Z. Ou-Yang, Phys. Rev. E 52, 681 (1995).26. G. Barbero and G. Durand, J. Appl. Phys. 67, 2678 (1990).27. A.A. Sonin, The Surface Physics of Liquid Crystals, (Gordon and Breach, 1995).28. S.A. Brooks, G.R. Luckhurst, and G.F. Pedulli, Chem. Phys. Lett. 11, 159 (1971).29. D.A. Dunmur and W.H. Miller, J. Physique 40, 361 (1979).30. B. Bahadur, Liquid Crystals – Applications and Uses, Vol.1, (World Scientific, 1990) p.156.31. K. Skarp, S.T. Lagerwall, and B. Stebler, Mol. Cryst. Liq. Cryst. 60, 215 (1980).32. A.M. Kantola, G.R. Luckhurst, A. Sugimura, and B.A. Timimi, Mol. Cryst. Liq. Cryst., in press.33. G.R. Luckhurst, T. Miyamoto, A. Sugimura, and B.A. Timimi, J. Chem. Phys. 116, 5099 (2002).34. F.C. Frank, Disc. Faraday Soc. 25, 19 (1958).35. 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. 197.36. Hp. Schad, G. Baur, and G. Meier, J. Chem. Phys 71, 3174 (1979).37. J.D. Bunning, D.A. Crellin, and T.E. Faber, Liq. Cryst. 1, 37 (1986).38. J.W. Emsley, G.R. Luckhurst, and C.P. Stockley, Mol. Phys. 44, 565 (1981).39. D.A. Dunmur, M.R. Manterfield, W.H. Miller, and J.K. Dunleavy, Mol. Cryst. Liq. Cryst. 45, 127

(1978).

Page 367: Nanotechnology and Nano-Interface Controlled Electronic Devices

This Page Intentionally Left Blank

Page 368: Nanotechnology and Nano-Interface Controlled Electronic Devices

Nanotechnology and Nano-Interface Controlled Electronic DevicesEditors: M. Iwamoto, K. Kaneto and S. Mashiko© 2003 Elsevier Science B.V. All rights reserved

CHAPTER 17

MDC–SHG spectroscopy of organicmonolayer 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. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3512. Analysis of polarization of monolayers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353

2.1. Monolayers with C∞-symmetry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3532.2. Spontaneous polarization and orientational order parameter . . . . . . . . . . . . . . . . 3552.3. Non-linear polarization and orientational order parameter . . . . . . . . . . . . . . . . . . 357

3. MDC and SHG measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3593.1. Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3593.2. Experimental system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 362

4. MDC–SHG spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3624.1. Detection of phase transition in C∞v-symmetry and orientational order . . . . . 3624.2. Detection of untilting to tilting phase transition in monolayer . . . . . . . . . . . . . . 365

5. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 373References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 374

1. Introduction

Many organic materials that are interesting in terms of electronics have been synthesizedand discovered during the past few decades [1,2]. One of the most remarkable achieve-ments is the Nobel Prize in Chemistry 2000 awarded to Heeger [3], MacDiarmid [4] andShirakawa [5], for their contribution to the discovery and development of conductingpolymers. In the hope of observing novel and useful electrical and optical properties,many investigations have been carried out to fabricate organic devices. Plastic solarcells, flexible-type field effect transistors (FETs), electroluminescent (EL) devices andso 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

Page 369: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 byAviram and Ratner [7], with which many experiments have been carried out towards thefruition of molecular diodes [8]. The idea of electron tunneling rectifier junctions basedon electron tunneling via molecular energy states has also called much attention [9]. Fur-thermore, many investigations on the preparation of conducting molecular wires usingcarbon nanotubes, multiporphyrins and others have been carried out [10,11]. Similarly,many state-of-the-art techniques for successful preparation of tunneling spacers usinginsulating thin films such as polyimide have also been developed [12]. As mentionedabove, many organic materials including conductors, semiconductors and insulators,have been synthesized and many efforts have been made to fabricate organic devicesutilizing the specific properties of organic molecules. However, these are no longersufficient. To find further new ways to molecular electronics, the relationship betweenthe function and structure of molecules must be clarified on the molecular level. A verysimple but fascinating way is to use monolayers for this study, because monolayers showvarious interesting behaviors that are not seen in bulk materials. It is known that mono-layers 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-chemicalproperties of monolayers at the air–water interface have been a research subject forphysicists, chemists and biologists since the discovery of the technique for the formationof floating monomolecular films by Langmuir [13]. For the past several decades, alongwith the development of a variety of experimental techniques including scattering,spectroscopic, electrical and optical techniques, the dynamical features of monolayersystems and their structures have been examined by many researchers [14–16].

However, the relationship between the structure of monolayers and the function ofmonolayers must be clarified from the viewpoint of molecular electronics. In otherwords, the dielectric property, optical property, electronic conduction, and others mustbe clarified on the nanometer scale for the breakthrough toward the 21st-centuryelectronics.

From the theoretical side, monolayer systems are viewed as quasi 2D systems andtheir states are characterized by the geometrical configuration and by the orientationaldistribution of the constituent molecules. For example, the state of floating monolayerscomposed 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 whichexpresses the molecular configuration describing the positional distribution of the headsof molecules on the water surface [18,19], and the other is the orientational orderparameter which expresses the orientational distribution of rod-like polar moleculesnormal to the water surface. Among these parameters, the orientational order parameteris very important, because this parameter specifies the shape of rod-like molecules andit 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 an-alyzed and it is expressed using orientational order parameters defined by Legendrepolynomials. The specific properties of organic monolayers such as spontaneous po-larization and non-linear polarization are interpreted using these parameters. Then the

Page 370: Nanotechnology and Nano-Interface Controlled Electronic Devices

MDC–SHG spectroscopy of organic monolayer film 353

importance of the development of an experimental method used for the determination ofthe orientational order parameters is stressed [16,20].

In Section 3, Maxwell displacement current (MDC)–optical second harmonic gener-ation (SHG) measurement used for the determination of orientational order parametersand the detection of phase transitions of a floating monolayer is mainly discussed. InSection 4, some experimental results on 4′-n-octyl-4-cyanobiphenyl (8CB) and fattyacid are shown. In this chapter, discussion is confined to Langmuir-monolayers, but theknowledge obtained from this study will be useful for a better understanding of mono-and multilayer films on solid surface.

2. Analysis of polarization of monolayers

As mentioned in Section 1, until now, a variety of experimental methods includingscattering, spectroscopic, electrical and optical techniques have been developed, andthey have been employed to study the molecular structure and orientational phase transi-tion of monolayers at the air–water interface [14,15]. Using X-ray diffraction, Brewsterangle microscopy (BAM) and others, monolayer structure, texture of monolayers, andso on have been elucidated step by step. However, the physico-chemical properties of2D systems have not been fully discussed yet from the viewpoint of dielectric physics.

Obviously, the dielectric property of monolayers is dependent on the polarizationof the monolayers. Thus experimental methods that allow dielectric polarization phe-nomena 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 spontaneouspolarization and non-linear dielectric polarization phenomena, respectively, and theyare generated only from polar non-centrosymmetric monolayers, although these are notgenerated from isotropic bulk materials. For a better understanding of the generationof MDC and SHG, it is convenient to start from the analysis of dielectric polarizationof monolayers. For this analysis, a monolayer composed of rod-like polar molecules isconsidered and a monolayer with C∞-symmetry is mainly discussed, but very importantinsight into the physics of monolayers can be obtained. Further, this analysis is easilyextended to the case of monolayers with Cs-symmetry, whose director makes an angleθt 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 permanentdipole moment μ along their molecular long axis, and an anisotropic electronicpolarizability, 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 moleculehas a molecular second-order susceptibility, α

(2)M . The water surface and the floating

monolayer film are considered as an infinite plane. The coordinate system is chosen insuch a way that the monolayer plane is parallel to the x–y plane and the monolayernormal falls along the positive z-axis, as shown in Fig. 1a. The molecule at theorigin is facing the water surface, and it tilts with an angle θ . The molecular area

Page 371: Nanotechnology and Nano-Interface Controlled Electronic Devices

354 A. Tojima et al.

z

θA

x

l

untilting monolayer

(b)

(a)

z (zm)z'

x

y

φ

θm

d

n y'

x'

,

Molecule

α⊥

α//

d = (sinθ cosφ, sinθ sinφ, cosθ)

m = (0, 0, 1), n = (0, 0, 1)

m

Fig. 1: Molecule of untilting monolayer with C∞v-symmetry. (a) The molecules in monolayers are assumedto be uniaxial, depicted by an ellipsoid of rotation for the electronic polarizability. (b)The molecules in anuntilting 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. Thecoordinate system (x ′, y ′, z′) is attached to the dipolar molecule at the origin with thez′-axis pointing toward the molecular long axis,

→d . Thus the direction of the molecule’s→

d is expressed as (sinθ cosφ, sinθ sinφ, cosθ) in the coordinate system (x , y, z). Asthe molecule considered here is rod-shaped, the interaction with the water surface

Page 372: Nanotechnology and Nano-Interface Controlled Electronic Devices

MDC–SHG spectroscopy of organic monolayer film 355

should be independent of the azimuthal angle φ. The molecular average direction calledthe director, →m, is expected to be normal to the water surface. That is, the directorpoints in the direction →n = (0,0,1) in the coordinate system (x , y, z), indicating that themonolayer has C∞-symmetry (Fig. 1a). To specify the average orientation of moleculeson the water surface, the macroscopic alignment parameters known as the orientationalorder parameters are used [17]. These parameters describe the alignment origin ofthe collective properties of molecules in monolayers [29,30]. The universal definitionSn ≡ 〈Pn(cosθ)〉 can be adopted as orientational order parameters to express monolayersystems with C∞-symmetry. Here Pn(cosθ) is the nth Legendre polynomial, and 〈 〉denotes thermodynamic average over all molecular directions.

It is apparent that the orientational order parameter S2 ≡ 〈P2(cosθ)〉 = 〈 3cos2 θ−12 〉

corresponds to the Maier–Saupe order parameter introduced by Tsvetkov [31,32] intoliquid crystals (LCs). For organic monolayers with C∞-symmetry, the orientational or-der parameter S1 ≡ 〈P1(cosθ)〉 = 〈cosθ〉 makes more sense to specify the characteristicproperty of monolayer films due to the symmetry breaking at the interface. Similarly,the orientational order parameter S3 ≡ 〈P3(cosθ)〉 = 〈 5cos3 θ−3cosθ

2 〉 specifies the charac-teristic property of monolayer films. As such, these two order parameters can be used todescribe the specific dielectric property of monolayer films with C∞-symmetry [16,20].

Suppose a monolayer is subjected to an external electric field E, the dielectricpolarization of each molecule is expressed as [24,33]

m = μ+↔α

(1) · E +α(2) : E E +·· · , (1)

where ↔α

(1) is the linear electronic polarizability and α(2) is the second-order suscep-tibility. Discarding the non-linear terms higher than second order, the polarization ofmonolayers, P , is described as

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 maincontributors to the MDC and SHG are 〈μ〉 and 〈α(2)〉, respectively. In the following, wefocus on 〈μ〉 and 〈α(2)〉.

2.2. Spontaneous polarization and orientational order parameter

Suppose the orientational distribution of constituent rod-like polar molecules in themonolayer film obeys the Boltzmann distribution rule, then the average of the dipolemoment components is given by [17,34]

〈μi 〉 =∫∫

μi f (φ,θ)sin θ dφdθ , (3)

where f (φ,θ) is the distribution function and it is proportional to exp(−W/kBT ). HereW 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 Boltzmannconstant, and T is the absolute temperature. μi (i = x , y, z) is the i-direction componentof the permanent dipole μ.

Page 373: Nanotechnology and Nano-Interface Controlled Electronic Devices

356 A. Tojima et al.

As the direction of the permanent dipole moment μ coincides with the molecularlong axis, μ = →

μ is expressed as μ = μ→d . When an external electric field E =→

E = (Ex , Ey , Ez) is applied to the monolayer film, the dipolar molecule at the originexperiences this electric field. The electric interaction energy We between the dipole andthis external field is given by

We = −∫

E · dm ≈ −μ→d · →

E . (4)

It is reasonable to neglect the interaction energy part between the external field andthe induced dipoles due to the electronic polarization ↔

α(1) and α(2) given by Eq. 1,

because they are proportional to E2i or E3

i (i = x , y, z) and as a whole only affect thethird-order or fourth-order non-linearity, and they are usually very small. Due to theelectric interaction We expressed by Eq. 4, the dipolar molecules prefer to reorient alongthe direction of the applied field. Using the Onsager approximation [35], we obtain thedielectric polarization due to the orientational polarization Pd = →

Pd = (Px , Py , Pz) as

Pd = Ns〈m〉= NsμS1

→n + Nsμ2

3kT

↔S ori · →

E , (5)

where

S1 = 〈P1(cosθ)〉 = 〈cosθ〉 =∫∫

cosθ f (φ,θ) sinθ dφdθ , (6)

is the orientational order parameter when E = →E = 0, and

↔S ori is a tensor defined as

↔S ori =

⎛⎝1− S2 0 0

0 1− S2 00 0 1−3S2

1 +2S2

⎞⎠ (7)

with

S2 = 〈P2(cosθ)〉 =⟨

3cos2 θ −1

2

⟩=∫∫ (

3cos2 θ −1

2

)f (φ,θ)sin θ dφdθ . (8)

↔S ori describes the influence of the orientational distribution of the polar molecules inmonolayer 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 spontaneouspolarization and the second term is the orientational dipolar polarization due tothe external electric field. It is also found from Eq. 7 that the contribution of theorientational dipolar polarization becomes μ2/3kT if S1 = S2 = 0, i.e.,

↔S ori = ↔

I . Thisis nothing but the Debye–Langevin equation, which expresses the orientational polar-ization in isotropic bulk states [33,35]. In other words, a monolayer shows specificdielectric property originating from the non-zero orientational order parameters, S1 andS2 [16,20].

In the MDC measurement, an MDC generated across monolayers is monitored inclosed-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 thismeasurement, and the orientational order parameter S1 is determined from the MDC

Page 374: Nanotechnology and Nano-Interface Controlled Electronic Devices

MDC–SHG spectroscopy of organic monolayer film 357

over the entire range of molecular areas generated by monolayer compression. For abetter understanding of orientational order parameter S1 of monolayers, it is instructiveto discuss a special case where the molecular orientational motion is restricted withinthe angle 0 < θ < θA = arcsin

√A/A0 (A0 = πl2, l: the length of long axis of polar

molecules) with θt = 0, due to the hardcore repulsive interaction among molecules asshown in Fig.1b. Using S1 �= 0, Pz is given by

Pz = NsμS1. (9)

This situation appears when the molecular area A is smaller than A0. On theother hand, when the molecular area A is greater than A0, the electrostatic Coulombattractive force working between molecules and the water surface makes a significantcontribution, and the molecules lie on the water surface. The interaction energybetween dipolar molecules and water is only dependent on θ and it is given byWm(θ) = −(μ2/16πε0εml3 cosθ)[(εw − εm)/(εw + εm)], where ε0, εm and εw are thepermittivities of free space, monolayer and water, respectively. With the apparentBoltzmann distribution function f (θ) ≈ exp[−Wm(θ)/kBT ], Wm(θ) becomes −∞ inthe 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 transitionfrom planar to orientational alignment phase happens at the molecular area A0 duringmonolayer compression [17], and S1 and Pz are expected to change smoothly at thistransition point. As MDC is generated due to the change of Pz , MDC is effective for thedetection of the phase transition.

2.3. Non-linear polarization and orientational order parameter

Whenever the dielectric property of monolayers is discussed, the non-linear polarizationis inevitably involved, as the centrosymmetry is broken at the interface [24,26]. Thishigher order polarization arises from the non-linear electronic polarizability 〈↔α (2)〉, thenon-linear additional electronic polarizability produced by the local-field molecularinteraction, and the non-linear orientation-induced dipole moment as a result of theinteraction between the dipole moment and the external field. Usually, the contributionof these non-linear polarizations is rather small, but this situation changes under somespecial conditions. The non-linear electronic polarization originating from the quantuminteraction between electrons and the external electric field is quite large, and can bedetected as SHG signals. In SHG signals, the main contributor is the second-ordernon-linear electronic polarizability 〈↔α (2)〉. The non-linear polarization is approximatelywritten as

P (2) = χ (2) : E E = Ns〈↔α (2)〉 : E E. (10)

The second-order non-linear susceptibility (SOS) χ (2) ≡ [χi j k] is related to thesecond-order non-linear electronic polarizability of molecules, expressed by the tensorα(2) ≡ [αi ′ , j ′,k ′], by χi j k = Ns〈T i ′ j ′k ′

i j k 〉αi ′ j ′k ′ . T i ′ j ′k ′i j k describes the coordinate transformation

between the molecular coordinates (x ′, y ′, z′) and the laboratory coordinates (x , y, z) inthe case of a monolayer with C∞-symmetry. There are many ways used to transformbetween laboratory and molecular frames, such as the irreducible tensor approach with

Page 375: Nanotechnology and Nano-Interface Controlled Electronic Devices

358 A. Tojima et al.

the Wigner rotation matrix [36]. For the unification of the notation for the transformationperformed in this chapter, the usual Euler rotation matrix R(φ,θ ,ψ) = [Ri ′

i ] [37] isadopted. The transformation T i ′ j ′k ′

i j k is expressed by T i ′ j ′k ′i j k = Ri ′

i R j ′j Rk ′

k with

Ri ′i =

⎛⎝cosφ cosψ − sinφ cosθ sinψ , −cosφ sinψ − sinφ cosθ cosψ , sinφ sinθ

sinφ cosψ +cosφ cosθ sinψ , −sinφ sinψ +cosφ cosθ cosψ , −cosφ sinθ

sinθ sinψ , sinθ cosψ , cosθ

⎞⎠ . (11)

In the expression of Eq. 10, is not taken into account the local field correction factorssuch as the Lorentz factor which arises due to the surface, but also, significantly, due todipole–dipole interaction with neighboring molecules, where the latter will be obviouslyconcentration dependent. However, this simplification does not lose underlying physicshere, because one way of making the local field correction is to introduce a factor in theright-hand side of Eq. 10 [24,26].

From the SHG measurement, one can deduce χ (2), but to obtain α(2), one has to haveknowledge of 〈T i ′ j ′k ′

i j k 〉. After a lengthy calculation of the 93 components for obtainingχi j k [38], the complete expression for the macroscopic SOS tensor χ (2) of a monolayerwith C∞-symmetry on a material surface is derived as function of the componentsof the second-order non-linear electronic polarizability α(2) and the orientational orderparameters Sn ≡ 〈Pn(cosθ)〉 [17,29,30]. The non-linear polarization P (2) = →

P N is foundto be given by the sum of the polarization

→P N

ch, associated with the chirality of themonolayer, and the polarization

→P N

ach associated with the non-chirality of the monolayer.These two polarization are expressed in a vectorial form as [27,38]

→P N

ch = 12 s14[(

→E · →n )(

→F ×→n )+ (

→F · →n )(

→E ×→n )]+ 1

2 a14(→E ×→

F)

+ 12 (a36 −a14)→n · (

→E ×→

F)→n , (12)→P N

ach = (s33 − s15 − s31)(→n · →E)(→n · →

F)→n + s31(→E · →

F)→n

+ 12 s15[(→n · →

F)→E + (→n · →

E)→F]+ 1

2 a15(→E ×→

F)×→n . (13)

Here→E and

→F are external electric fields, and 7 independent non-zero elements are

given by

s14 = Ns

2S2(σ14 −σ25) = χ123 +χ132,

s15 = Ns

5(S1 − S3)(2σ33 −σ32 −σ31)+ Ns

10(3S1 +2S3)(σ24 +σ15) = χ131 +χ113,

s31 = Ns

10(S1 − S3)(2σ33 −σ24 −σ15)+ Ns

10(4S1 + S3)(σ32 +σ31) = χ311,

s33 = Ns

5(S1 − S3)(σ32 +σ31 +σ24 +σ15)+ Ns

5(3S1 +2S3)σ33 = χ333,

a14 = Ns

3(λ14 +λ25 +λ36)+ Ns

6S2(λ14 +λ25 −2λ36) = χ123 −χ132,

a15 = Ns

2S1(λ15 −λ24) = χ131 −χ113,

a36 = Ns

3(λ14 +λ25 +λ36)+ Ns

3S2(2λ36 −λ14 −λ25) = χ312 −χ321, (14)

Page 376: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 tensorα(2) in the same way as the conventional contracted notation of (si j ) and (ai j ) definedfrom 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 ofmonolayers, whereas s15, s31, s33 and a15 specify the non-chiral property of monolayers.Eventually, from the view point of order parameters, S2 characterizes the orientationalorder of monolayers related to the molecular chirality, whereas S1 and S3 the ori-entational order of monolayers related to the molecular non-chirality. In the case ofmonolayers 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]

→P N = (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 orderparameters S1 and S3.

3. MDC and SHG measurements

3.1. Method

Fig. 2 shows a schematic diagram of MDC measurement coupled with SHG measure-ment, 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 connectedto each other through an electrometer. The induced charge on electrode 1 changes inaccordance with the orientational motion of molecules on the water surface and thechange of surface density of the molecules. In the MDC measurement, monolayers arecompressed with aid of two moving barriers. MDC flows through the closed circuit (seeFig. 3). The charge induced on electrode 1 suspended in air due to the spontaneous

A

Q1

M DCSHGIω I2ω

L

+

-water Electrode 2

Ep

EsElectrode 1

Fig. 2: Experimental setup for MDC and SHG measurements.

Page 377: Nanotechnology and Nano-Interface Controlled Electronic Devices

360 A. Tojima et al.

γR

pin

γT

sin

Eo(2ω)

s2ω

Eo(2ω)p2ω

ωE ( )in

xy

z

Reflected SHIncident light

Transmitted SH

A+

-

electrode 2

electrode 1

water

barrier

p2ω

s2ω

θo

θo

θin

δ

Fig. 3: Molecules and optical arrangement. The monolayer is compressed in the x-direction. The reflectedand transmitted light can be detected.

polarization Pz , given by Eq. 9, is expressed as [16,42]

Q1 = −Pz B/L −Cφs, (16)

where B is the working area of electrode 1, C is the capacitance between electrode1 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 watersurface in the range of low surface pressure due to the electrostatic Coulomb interactionworking between the polar molecules and water, the order parameter of the monolayeris determined over the entire range of molecular area.

For example, for monolayers with C∞v-symmetry, the orientational order parameterS1 defined in Eq. 9 is determined, assuming S1 = 0 in the range of low surface pressurebefore monolayer compression. Furthermore, it is clear from Eq. 16 that the MDC flowsdue to the change in the polarization of monolayers, and thus phase transitions such asthe transition from planar isotropic phase to polar orientational phase of monolayers areeasily detected.

On the other hand, optical second-harmonic (SH) light is generated from non-centro-symmetric monolayers by laser irradiation. This SHG is due to the quantum interactionbetween 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 fieldcorrection factors such as Lorentz factor are not taken into account, the SH intensity inthe direction →e out depends on the term →e out · →

P N , where →e out is the unit vector.From Eq. 15, it is shown that the SHG signal is obviously dependent on →e out · →n and

Page 378: Nanotechnology and Nano-Interface Controlled Electronic Devices

MDC–SHG spectroscopy of organic monolayer film 361

→E · →n . Thus the SHG measurement is helpful for the detection of phase transitions suchas the tilting to untilting (t–U) phase transition.

For example, if the monolayer is in the planar isotropic phase, an SH signal shouldnot be generated because of

→P N = 0, i.e., →e out ·→

P N = 0, whereas p-polarized radiation isexpected 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 bedetermined from the SHG measurements [16,27,28], as will be described below.

Fig. 3 shows the experimental arrangement. θin and θo represent the incident andoutput angles, respectively. The angles δ and γ represent the polarized angles of incidentlight and output SH light, respectively. The input light is the sum of s- and p-polarizedwaves and it is expressed as

→E in = Sω

→s in + Pω→p in, (17)

where →s in and →p in are unit vectors for s- and p-polarized waves, and Sω and Pω

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, sinθin), respectively. Further, the angle ofthe polarizer is chosen as δ for the input light (see Fig. 3), Sω and Pω are given bySω = Ein sinδ and Pω = Ein cosδ, respectively.

Similarly, the output light is the sum of the s- and p-polarized waves. Taking intoaccount these, the output light intensity I2ω is found to be proportional to →e out(2ω) · →

P N .Thus by choosing appropriate angles δ and γ , the orientational order parameters can bedetermined [22,27,28].

For example, in the special case that the second-order susceptibility α(2) is dominatedby a single component, i.e. αz′z′z′ along the molecular long axis, the SH intensitygenerated from a monolayer with C∞v-symmetry is given by

I2ω ∝ |(AS1 + BS3)2| · I 2ω, (18)

where A and B are obviously functions of the angles θin, θo, δ and γ . A and B are givenby

A = −2sinθin cosθin cosθo cos2 δ cosγ +2sinθin sinδ cosδ sinγ

+ cos2 θin sinθo cos2 δ cosγ + sinθo sin2 δ cosγ +3sin2 θin sinθo cos2 δ cosγ , (19)

and

B = 2sinθin cosθin cosθo cos2 δ cosγ −2sinθin sinδ cosδ sinγ

− cos2 θin sinθo cos2 δ cosγ − sinθo sin2 δ cosγ +2sin2 θin sinθo cos2 δ cosγ . (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 shouldbe noted here that a similar expression as given by Eq. 18 can be obtained even whenthe local field correction factors such as Lorentz factor are taken into consideration inthe analysis. That is, we can have the similar prediction that the orientational orderparameters S1 and S3 can be obtained [40].

Page 379: Nanotechnology and Nano-Interface Controlled Electronic Devices

362 A. Tojima et al.

3.2. Experimental system

The experimental setup consists of Langmuir-trough equipped with a two-electrodearrangement for the MDC measurement and optical measurement arrangement with aQ-switched Nd : YAG laser (e.g., wavelength 0.532 μm, pulse duration <7 ns, repetitionpulse rate ≤15 Hz) for the SHG measurement (see Figs. 2 and 3). The rectangular-shape Langmuir-trough (600 mm × 150 mm in length and width, 10 mm in depth) iscomposed of polytetrafluoroethylene (PTFE) and it is filled with pure water (electricalresistivity >17 M� cm).

One transparent silica glass plate is attached to the bottom of the Langmuir trough forthe SHG measurement. In this experimental system, reflected and transmitted SH signalsfrom 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 distanceof 1 mm. Electrode 2 is a gold-wire (1 mm ∅ and 500 mm) and it is immersed in thewater. Electrodes 1 and 2 are connected to each other through an electrometer (Keithley617) whose internal electrical resistance is negligibly small.

For the SHG measurement, the Q-switched laser irradiates onto the monolayer at anintensity 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 aWilhelmy 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 fattyacid molecules used here. The permanent dipole moments of methyl and cyano-groupsin 8CB molecules make the contribution to the generation of an MDC, whereas the

8CB

CNC8H17

COOH

fatty acid (octadecanoic acid)Fig. 4: 4′-n-octyl-4-cyanobiphenyl (8CB) and fatty acid molecules.

Page 380: Nanotechnology and Nano-Interface Controlled Electronic Devices

MDC–SHG spectroscopy of organic monolayer film 363

40 60 80 100 1200

2

4

Sur

face

pres

sure

[mN

/m]

Area per Molecule [Å2]

0

50

100

Dis

plac

emen

tcu

rren

t[fA

]

0

200

400

600

Dip

ole

mom

ent

[mD

]

0

0.5

1

1.5

Ref

lect

edS

Hin

tens

ity[a

.u.]

p-ss-s

0

5

10

15

Tra

nsm

itted

SH

inte

nsity

[a.u

.]

3 124

p-ps-p

40 50 60 70 800

0.5

1

Fig. 5: A typical example of SHG and MDC of 8CB monolayers during the course of monolayercompression. p-polarized SHG and MDC are generated at the same molecular area. Compression speed4.2 Å2 min−1 molecule−1.

electronic polarization of the biphenyl group with alkyl and cyano-groups makes thecontribution to the SHG. As monolayers are placed on the water surface, the contributionof the hydrophilic part is diminished owing to the screening effect of the water with ahigh dielectric constant (≈ 78). As a result, although the dipole moment of methyl-groupis very small in comparison with that for the cyano group, the long alkyl chain parts alsomake a contribution to the generation of MDC, in a manner to that seen in fatty acidsmonolayers [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 asindicated in the figure, such as p–p, s–p, etc.

From bottom to top, the surface pressure–area, MDC–area, the vertical componentof dipole moment–area, reflected SH intensity–area, and transmitted SH intensity–area

Page 381: Nanotechnology and Nano-Interface Controlled Electronic Devices

364 A. Tojima et al.

are plotted. As we can see in the figure, the MDC and SHG signals are generated inaccordance with the change of molecular area. The isotherms can be divided into fourregions. In region 1, where the surface pressure is immeasurably low, MDC is verysmall and almost zero. Similarly the SH intensity is very small. These results suggestthat the 8CB molecules lie on the water surface and they are randomly distributed. Inother words, the monolayer in region 1 is in the planar and isotropic state, i.e., Pz givenby 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 bymonolayer compression, possibly due to the standing up of 8CB molecules on the watersurface. As the MDC is generated due to the change of spontaneous polarization Pz , theMDC 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 thedipole moment of 8CB molecules given by μS1 is calculated. The dipole moment–area isotherm in Fig. 5 represents the relationship between μS1 and the moleculararea A (≡ 1/Ns). It is found that S1 gradually increases with increasing monolayercompression, and finally saturates at the end of region 2. The SHG is observed forthe p-polarized output light, whereas it is not observed for the s-polarized outputlight. These results can be explained using Eq. 15. Briefly, for the s-polarized waveirradiation of monolayers, the external electric field

→E is given by

→E in = Sω

→s in, with→s in = (−1,0,0) in the laboratory frame (see Eq. 17). Substituting

→E in into Eq. 15,

→P N is

calculated as→P N = s31S2

ω→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 = s31S2

ω→n · →s out = 0 and

→P N · →pout =

s31S2ω

→n · →pout �= 0, respectively. Similarly, for the p-polarized wave irradiation, theexternal electric field

→E is given by

→E in = Pω

→p in with →p in = (0,cosθin, sinθin) in thelaboratory frame (see Eq. 17). Substituting

→E in into Eq. 15,

→P N is calculated as→

P N = (s33 − s15 − s31)P2ω(→n · →p in)2→n + s31 P2

ω(→p in · →p in)→n + s15 P2ω(→n · →p in)→p in.

Thus it is concluded that the s- and p-polarized radiation are proportional to→P N · →s out = 0 and

→P N · →pout �= 0, respectively, because of →n · →s out = 0, →n · →pout =

→n · →p in = sinθin �= 0, and →p in · →pout = cos(θin − θo) �= 0, →p in · →p in = 1. This theoreticalprediction agrees well with the results observed in region 2, where p-polarized waveoutput 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 monolayercompression, and non-linear second-order polarization expressed by Eq. 15 is inducedby 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 thatdomains with Cs-symmetry randomly distribute on the water surface and it looks likea C∞v-symmetry owing to the randomly distributed domains. If the domains make asignificant contribution to the generation of MDC, MDC flows even in region 1 dueto the condensation of domains possessing the spontaneous polarization [42], but suchMDC 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 makesthe contribution of the generation of MDC on the water surface, whereas the second-order non-linear electronic polarization of the biphenyl group with alkyl and cyano

Page 382: Nanotechnology and Nano-Interface Controlled Electronic Devices

MDC–SHG spectroscopy of organic monolayer film 365

groups makes the contribution to the SHG. From the experiment shown in Fig. 5, it isclear 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 longalkyl are located in line, stand up by monolayer compression.

In region 3, the surface pressure increases smoothly but MDC flow is steady atfurther compression. SHG signals are also generated steady, where the transmitted andreflected light are generated steady, and the intensities of these transmitted and reflectedsignals somewhat increase by compression, due to the condensation of molecules. Theseresults indicate that the orientational order is nearly saturated in this region and onlythe density of molecules increases by monolayer compression. As the MDC and SHGsare proportional to the spontaneous polarization expressed by Eq. 9 and non-linearpolarization given by Eq. 15, respectively, they are only proportional to the density ofmolecules when the orientational order does not change by monolayer compression. Asa result, only very small increases of MDC and SHG are seen in region 3 by monolayercompression. This increase is obviously not so drastic in comparison with the increaseof surface pressure.

At the end of region 3, MDC decreases abruptly by further compression, whereas theSH signal is generated in a way similar to that in region 3. These experimental resultssuggest that the transition from one layer to an interdigitated three-layer is inducedin region 4 during monolayer compression [43]. Briefly, the spontaneous polarizationPz 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 tothe change of the polarization Pz . Further the non-linear polarization given by Eq. 15does not change, indicating that the density of molecules does not change due to thistransition. 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 therelationships given by Eqs. 9 and 18. In the SHG measurement, the optical arrangementwas set to satisfy the relationship A = 0 or B = 0 for transmitted and reflected SHlight. In region 1, S1 and S3 of SHG are very small although fluctuations are observedowing to the formation of domains on the water surface. In region 2, S1 and S3 increasegradually and reach their maximum. In region 3, they are nearly constant about 0.5. Incontrast, S1 of MDC is nearly zero in region 1, and it increases monotonically in region2. S1 is about 0.5 in region 3. From these results, it is postulated that molecules lying ona water surface in the range of immeasurably low surface pressure gradually stand up bymonolayer 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 monolayerswith C∞v-symmetry in U-phase.

As mentioned in Section 2, using a variety of refined experimental methods coupled

Page 383: Nanotechnology and Nano-Interface Controlled Electronic Devices

366 A. Tojima et al.

40 60 80 100 1200

2

4

Sur

face

pres

sure

[mN

/m]

Area [Å2/molecule]

0

50

100

Dis

plac

emen

tcu

rren

t[fA

]

0

0.5

1

S1,

MD

C

0

0.5

1S

1,S

HG

0

0.5

1

S3,

SH

G

3 124

Fig. 6: Order parameters of 8CB monolayer determined from the MDC–SHG measurement. Compressionspeed 4.2 Å2 min−1 molecule−1.

with π–A measurement [13,14], both in-plane 2D order of the hydrophilic polar heads ofthe amphiphiles and out-of-plane orientation order of their hydrophobic long axes havebeen recognized. Among various kinds of Langmuir-monolayers, the most fundamentalone is fatty acids, and their monolayers have been extensively explored. The orderingstructures of all phases of fatty acid monolayers measured by many researchers havebeen summarized (see, e.g., fig. 32 and table I in Ref. [15]). Among the determinedphase 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 L′

2(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 andL′

2 phase the long molecular axes are tilting [tilting phase (t)] uniformly to a nearestneighbor (NN) and a next nearest neighbor (NNN), respectively, and the polar headsof in-plane structures appear in a distorted hexagonal lattice, a stretching along the

Page 384: Nanotechnology and Nano-Interface Controlled Electronic Devices

MDC–SHG spectroscopy of organic monolayer film 367

a

d0

a/cos

m,n

A0

mn

A0/cos

U-phase T-phase

d0

(A0)

(a) (b)

βθt

θt

θt

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 beendiscussed in association with the generation of MDC and SHG.

As has been pointed out by Sirota [15,44], in the condensed phases where the longaxes are like cylinders and they are closely packed, a tilt directly causes a distortion ofthe projection of the hexagon on the surface (Fig. 7). As such, the in-plane geometryrelation of closely packed monolayers on the water surface is represented only by atilting order parameter, the tilt angle θt from the water surface normal. The generation ofMDC 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 thetilt angle θt.

Fig. 7a and b show model structures of U-phase (untilted phase), and t-phase (tiltingphase, tilting toward a nearest neighbor (NN)), respectively. At the condensed U-phase,uniaxial rod-like molecules with a permanent dipole moment μ along the molecularlong axis are orthogonally and hexagonally packed. The average direction of moleculardirector →m is parallel to the unit vector →n . β(As) = arcsin

√As/πl2 is the maximum

angle between the molecular long axis and director →m. Here l is the length of the rod-likemolecules and As is the molecule area with the relation As = 2d2

0/√

3 (Fig. 7a).On the other hand, at the condensed t-phase, the director →m is tilted uniformly to

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

Page 385: Nanotechnology and Nano-Interface Controlled Electronic Devices

368 A. Tojima et al.

way that the tilting causes the stretching of the projected lattice cell in the tilt directionwith a ratio of 1/cosθt. Thus, the U → t transition is considered microscopically asa geometric distortion of the orthogonally hexagonal orientation of the molecules bycompetition between the 2D packing entropy and the interaction between molecules inbulk and interface, where the tilting causes the stretching of the projected lattice cell onwater surface but not the section normal to →m, as shown in Fig. 6. Here the interactionis described using the Lennard-Jones (L-J) potential and others. Interestingly, the tiltcauses 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 amphiphilicmolecule in the U state is assumed as FU = W U − T SU. The internal energy W U =3(−C/(2d0/

√3)6 + D/(2d0/

√3)12) describes the L-J potential with rmin = (2D/C)1/6

and Wmin = −3C2/4D where rmin is the inter-molecular distance when W U reaches itsminimum Wmin defined by a characteristic temperature TLJ as Wmin = −kTLJ. Here k isthe Boltzmann constant. The free energy of entropy −T SU = −kT ln A0 − kT ln2π(1−cosβ(A0)) contains both contributions from in-plane entropy (first term) and chain-orientation entropy (second term) of the molecule. From Fig. 7, in the t-phase the en-tropy energy changes to −T St = −kT ln(A0/cosθt)−kT ln2π(1−cosβ(A0)) and, witha lengthy derivation [45], the L-J potential becomes W t = −C(

√3cosθt/2d0)6∑3

i=1(1−sin2(φt + iπ/3)sin2 θt)−3 + D(

√3cosθt/2d0)12∑3

i=1(1− sin2(φt + iπ/3)sin2 θt)−6 whereφt 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 = sinθt. Discardingthe higher terms, the free energy change ΔF = F t − FU from untilted phase to tiltingphase in amphiphile monolayers is approximately expanded as [45,50]

ΔF ≈ Gη2 + Bη4 − Eη6 (21)

where

G = −1

2kT +3(1− ξ )ξkTLJ,

B = 3

8(13ξ −4)ξkTLJ − 1

4kT ,

E = 1

6kT − kTLJ

ξ (26−89ξ )

4− 1

8kTLJξ cos6 θt

(14ξ cos6 θt −5

)(10− cos6φt), (22)

with ξ = (rmin/(2d0/√

3))6. This theory predicts the phase diagram in the relevantexternal parameters space, i.e. T and d0. ΔF yields the most important result: WhenB > 0, i.e. T/TLJ < 39

2

(ξ − 2

13

)2 − 613 , the minimum of ΔF is achieved at η = 0 (U

phase) for G > 0 (i.e. T/TLJ < −6(ξ − 1

2

)2 + 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

T/TLJ = 3

2

13η2 −4

1+η2

(ξ +2

1−η2

13η2 −4

)2

−6

(1−η2

)2(13η2 −4

)(1+η2

) , (23)

and U → t phase transition occurs.

Page 386: Nanotechnology and Nano-Interface Controlled Electronic Devices

MDC–SHG spectroscopy of organic monolayer film 369

It is interesting here to note that this model predicts that the tilting phase (the lowsymmetry 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 thechain-orientation entropy: At high temperature the tilt causes an increase of L-J potentialbut decreases the energy of orientation entropy.

As mentioned earlier, the tilting causes a distortion of the projection of the hexagonon the surface, i.e., the increase in molecular area, and thus leads to the increase inentropy of monolayer systems. As a result, the free energy change ΔF gives the abovementioned result.

Given these, it is interesting to show that the MDC experimental technique is helpfulto detect such t → U transition predicted by the relationship between free energy ΔFand η.

First, it is necessary to calculate the orientational order parameter S1,z = 〈cosθ〉 formonolayers in tilting phase [17], for analyzing the MDC across monolayers due tothe U → t phase transition, where the induced charge on the suspended electrode isproportional 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 (sinθ cosφ, sinθ sinφ, cosθ). That is,

the molecule tilts with an angle of θ from the normal direction (z-direction) to the watersurface (x–y plane), and orients with an azimuthal angle π/2−φ from the y-axis.

In another way,→d is written as (sinΘ cosΦ, sinΘ sinΦ, cosΘ) in the coordinate

system (xm, ym, zm), where the zm-axis is attached to the axis of director →m, it tilts withan angle θt from the z-axis in the z–x plane. Thus the following relation is satisfiedbetween the two coordinate systems, (x , y, z) and (xm, ym, zm):

cosθ = −sinΘ sinΦ sinθt + cosΘ cosθt. (24)

Therefore the orientational parameter S1,z is written asS1,z = 〈cosθ〉

= −〈sinΘ sinΦ〉sinθt +〈cosΘ〉cosθt. (25)

Here, 〈sinΘ sinΦ〉 and 〈cosΘ〉 are two orientational order parameters. Therefore thepolarization Pz in the monolayer normal direction, corresponding to the 1st term ofEq. 5, is given by

Pz = NsμS1,z

= Nsμ〈cosθ〉= −Nsμsinθt〈sinΘ cosΦ〉+ Nsμcosθt〈cosΘ〉, (26)

If the interaction between the molecular dipole and water surface is discarded, thedistribution f (Θ) is isotropic in the U and t phases and it is not modified by thetilting process. Thus among the two orientational order parameters, the latter one justcorresponds to the orientational order parameter S1 of monolayers in single monolayersin untilted phase (θt = 0, see Eq. 5).

Looking at the experimental arrangement for MDC measurement, a charge Q isinduced on electrode 1 and it is given by [23,17]

Q = − NμS1,z

L, (27)

Page 387: Nanotechnology and Nano-Interface Controlled Electronic Devices

370 A. Tojima et al.

z

x

n

m

molecule

θt

water

tilting monolayer

x

zzm

xm

y

Molecule

Φ

φ

θ

θtΘm

d

d=(sinΘcosΦ, sinΘsinΦ, cosΘ) =(sinθcosφ, sinθsinφ, cosθ)

(ym)

(b)

(a)

Fig. 8: Molecules in tilting monolayer with a tilting angle of θt. The molecules in monolayers are assumedto be uniaxial.

using Eq. 16 under the assumption φs = 0. Here N is the number of molecules underelectrode 1 and it is given as N = B/A. B is the working area of the electrode, L isthe distance between suspended electrode and water surface, and A is the moleculararea given by A = A0/cosθt in the tilting phase (see Fig. 7). Therefore the MDC iscalculated as

I = Bμ

L

d

dt

(S1,z

A

). (28)

Page 388: Nanotechnology and Nano-Interface Controlled Electronic Devices

MDC–SHG spectroscopy of organic monolayer film 371

20 30 40 500

20

40

60

Sur

face

Pre

ssur

e[m

N/m

]

Area per Molecule [Å2]

Compression speed 20 mm/min (11Å2/min/molecule)

-200

204060

Dis

plac

emen

tC

urre

nt [

fA]

0

100

200

Dip

ole

Mom

ent

[m

D]

4 Region 123

Fig. 9: MDC due to tilting to untilting (t → U) phase transition observed for fatty acid monolayer.

Substituting the relation A = A0/cosθt into Eq. 28, we obtain

I = γ1Bμ

L

d

dA0

(cos2 θt

〈cosΘ〉A0

− cosθt sinθt〈sinΘ sinΦ〉

A0

),

with γ1 = dA0

dt. (29)

It is found from Eq. 29 that the MDC changes at the molecular area A = A0, where thetilt angle θt changes due to the U → t transition, especially due to the contribution of thefirst 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 theair–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 previousstudy. Fig. 9 shows the experimental result, where surface-pressure–area (π–A) andMDC–A, and the vertical component of dipole moment–area (μS1z–A), proportional toS1z–A isotherms, were plotted from bottom to top, respectively. At a molecular area of20 Å2 (region 2 → 3), MDC changes, i.e., first decreasing and then increasing duringmonolayer compression. The decrease is due to the condensation of molecules, whereasthe increase is due to the concerned U → t phase transition.

For simplicity, in the case where molecules can randomly orient over all orientational

Page 389: Nanotechnology and Nano-Interface Controlled Electronic Devices

372 A. Tojima et al.

directions θ < θA, due to the hard-core repulsive interaction working among rod-likemolecules, 〈cosΘ〉, 〈sinΘ sinΦ〉 and A0 are given by (cosθA +1)/2, 0 and πl2

0 sinθA,respectively [17]. Thus the MDC is approximately given as

I = γ1Bμ

L

d

dA0

⎛⎝cos2 θt

1+√

1− (A0/πl20)2

A0

⎞⎠ , with γ1 = dA0

dt. (30)

Therefore the generation of MDC is expected in the positive direction due tothe t → U transition, i.e., θt → 0, as A0 decreases by monolayer compression, ifμ > 0. The increase in the MDC observed in Fig. 9 supports this prediction well. Inpractical observation, the MDC changes abruptly at the transition. This seems contraryto Eq. 30, in which everything varies smoothly at the transition. It is speculated thatthe compression procedure is dynamically anisotropic and must create a small butfinite θt tilt even in the U phase. Furthermore, it is experimentally and theoreticallyshown that the angle of the molecular long-axis from the normal direction to thewater 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 pointchanges with a ratio of cosθc > 0.76. In other words, the molecular area changesabout ΔA (= A0(1/cosθt − 1)) < 6.0 Å2 from θt = 0 to θt = θc at the molecular areaof around 20 Å2 where the MDC changes due to this transition. This prediction istrue as shown in region 3 (see Fig. 9). Furthermore, for θt = 0 one finds from Eq. 30I = −(γ1 Bμ/L A2

0)(1 + 1/√

1− (A0/πl20)2), so that for a negative pulse γ1 = dA0/dt

occurring in the t → U transition by compression the MDC appears as a positive pulseas shown in Fig. 9. As discussed above, the t → U phase transition is detectable byMDC measurement, and the generation of MDC due to this transition can be explainedby a model using one parameter, η.

The fatty acid monolayer is not active at laser irradiation, whereas 8CB monolayeris SH active. In this sense it is interesting to discuss the non-linear polarization ofmonolayers in tilting phase.

As mentioned earlier, the spontaneous polarization is characterized using two orienta-tional order parameters y10 = 〈cosΘ〉 and y11 = 〈sinΘ cosΦ〉 when the director is tiltedfrom the monolayers with C∞v-symmetry. If we assume that the uniaxial symmetryof the molecular distribution around →m (Fig. 7b) is held in both tilting and untiltingphases due to the repulsive interaction between the molecules of the monolayer and theattractive interaction between dipolar molecules and the interfaces [17], the orientationdistribution function f (Φ,Θ) of molecules in the monolayer becomes independent ofΦ, i.e., y11 = 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

→P N = (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 Ex , Ey , andEz (see equations (1) and (2) in Ref. [14]), we can easily check that the non-linearpolarization given by Eq. 31 represents the non-linear polarization of monolayers with

Page 390: Nanotechnology and Nano-Interface Controlled Electronic Devices

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,the ten non-zero SOS components (s ′

pq) of monolayers with Cs-symmetry are calculatedas

s ′11 = (s31 + s15)sinθt cos2 θt + s33 sin3 θt

s ′12 = s31 sinθt

s ′13 = s31 sin3 θt + (s33 − s15)sinθt cos2 θt

s ′15 = (2s33 −2s31 − s15)sin2 θt cosθt + s15 cos3 θt

s ′24 = s15 cosθt

s ′26 = s15 sinθt

s ′31 = s31 cos3 θt + (s33 − s15)cosθt sin2 θt

s ′32 = s31 cosθt

s ′33 = s33 cos3 θt + (s31 + s15)cosθt sin2 θt

s ′35 = (2s33 −2s31 − s15)cos2 θt sinθt + s15 sin3 θt (32)

If the interaction between the water surface and molecules makes a significantcontribution, probably, this is true in an actual monolayer system, the distributionfunction f (Φ,Θ) should be dependent on both Θ and Φ, and thus the monolayerchanges symmetry from C∞v to Cs-symmetry. From Eq. 24, the angle θ is dependent onΦ and Θ if θt �= 0. In this case, the spontaneous polarization of monolayers should bedescribed using two order parameters, y10 = 〈cosΘ〉 and y11 = 〈sinΘ cosΦ〉, and thenon-linear polarization of monolayers should be described using six parameters, y10 =〈cosΘ〉, y11 = 〈sinΘ cosΦ〉, y30 = 〈(5cos3 Θ −3cosΘ)/2〉, y31 = 〈(3sinΘ(5cos2 Θ −1)cosΦ)/2〉, y32 = 〈15(1− cos2 Θ)cosΘ cos2Φ〉, and y33 = 〈15sin3 Θ cosΦ〉 [45].

However, the merit of the MDC measurement coupled with SHG measurement forthe detection of the phase transition of monolayers [27,28,22] has already been shownby simply using the polarization approximately given by Eq. 31, in which only twoorder parameters S1 and S2 are used [27,51]. By using SHG measurement, the phasetransition from C∞v → Cs has been confirmed in 4-heptyloxy-4′-cyanobiphenyl (7OCB)monolayers [52]. Further we may conclude that the interaction among molecules canbe discussed using aforementioned six order parameters determined by MDC–SHGmeasurement.

5. Summary

The dielectric polarization of organic monolayers at the air–water interface has beenanalyzed, assuming that the monolayers have a C∞-symmetry. The formula of polar-ization of organic monolayers is derived and it is expressed using orientational orderparameters. It was revealed that Maxwell displacement current (MDC) measurementcoupled with optical second harmonic generation (SHG) measurement is helpful for thedetermination of these orientational order parameters as well as for the detection ofphase transitions. Monolayers of 4′-n-octyl-4-cyanobiphenyl (8CB) and fatty acids at

Page 391: Nanotechnology and Nano-Interface Controlled Electronic Devices

374 A. Tojima et al.

the air–water interface were examined during monolayer compression and the molecularmotion of monolayer was discussed on the basis of the orientational order parametersdetermined in the experiment.

References

1. Special issue on Functional Organic Materials for Devices, J. Mat. Chem. 9, 1853 (1999).2. Organic Thin Films, Materials Chemistry Discussion No. 2, J. Mat. Chem. 10, 1 (2000).3. A.J. Heeger, Rev. Mod. Phys. 73, 681 (2001).4. A.G. MacDiamid, Rev. Mod. Phys. 73, 701 (2001).5. H. Shirakawa, Rev. Mod. Phys. 73 713 (2001).6. C.J. Brabec, N.S. Sariciftci, and J.C. Hummelen, Adv. Funct. Mater. 11, 15 (2001).7. A. Aviram and M.A. Ratner, Chem. Phys. Lett. 29, 277 (1974).8. C. Joachim, J.K. Gimzewski, and A. Aviram, Nature 408, 541 (2000)9. C.M. Fischer, M. Burghard, S. Roth, and K. v. Klitzing, Appl. Phys. Lett. 26, 3331 (1995).

10. Y. Zhang and S. Iijima, Phys. Rev. Lett. 82, 3472 (1999).11. H.S. Nalwa, Supramolecular Photosensitive and Electractive Materials, (Academic Press, San Diego,

2001).12. 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. 815–884.13. G.L. Gaines, Insoluble Monolayers at the Liquid–Gas Interfaces (Interscience, New York, 1966).14. A. Ulman, Characterization of Organic Thin Films, (Butterworth-Heimann, Boston, 1995).15. V.M. Kaganer, H. Möhwald, and P. Dutta, Rev. Mod. Phys. 71, 779 (1999).16. M. Iwamoto and C.X. Wu, The Physical Properties of Organic Monolayers (World Scientific,

Singapore, 2001).17. A. Sugimura, M. Iwamoto, and Z.C. Ou-Yang, Phys. Rev. E 50, 614 (1994).18. D.M. Taylor and G.F. Bayes, Phys. Rev. E 49, 1439 (1994).19. R.E. Collin, Field Theory of Guided Waves (McGraw-Hill, New York, 1960) Ch. 12.20. M.Iwamoto, IEICE Trans. on Electronics E83-C, 1062 (2000).21. M. Iwamoto et al., Nature (London) 353, 645 (1991).22. M. Iwamoto, Y. Majima, H. Naruse, T. Noguchi, and H. Fuwa, J. Chem. Phys. 95, 8561 (1991).23. M. Iwamoto and Y.Majima, J. Chem. Phys. 94, 5135 (1991).24. Y.R. Shen, The Principles of Nonlinear Optics (Wiley, New York, 1984).25. N. Blöembergen and P.S. Pershan, Phys. Rev. 128, 606 (1962).26. T.F. Heinz, in Nonlinear Surface Electromagnetic Phenomena, edited by H.E. Ponath and G.I.

Stegemen (Elsevier Science, New York, 1991), pp. 397–398.27. A. Tojima, T. Manaka, and M. Iwamoto, J.Chem.Phys. 115, 9010 (2001).28. A. Tojima, Y. Matsuo, R. Hiyoshi, T. Manaka, Y. Majima, and M. Iwamoto, Thin Solid Films 393, 86

(2001).29. S. Chandrasekhar, Liquid Crystals (Cambridge Univ. press, London, 1977).30. P.G. de Gennes, The Physics of Liquid Crystals (Clarendon, Oxford, 1991).31. A. Saupe, Z. Naturforsch. 19a, 161 (1964).32. V. Tsvetkov, Acta Physicochim. (USSR) 16, 132 (1942).33. C. Kittel, Introduction to Solid State Physics (Wiley, New York, 1974).34. C.X. Wu, W. Zhao, M. Iwamoto, and Z.C. Ou-Yang, J. Chem. Phys. 112, 10548 (2000).35. H. Fröhlich, Theory of Dielectrics (Oxford Univ. Press, New York, 1958).36. G.R. Luckhurst and G.W. Gray, The Molecular Physics of Liquid Crystals (Academic Press, London,

1979).37. M.E. Rose, Elementary Theory of Angular Momentum (Wiley, New York, 1957).38. M. Iwamoto, C.X. Wu, and Z.C. Ou-Yang, Chem. Phys. Lett. 325, 545 (2000).39. C.C. Wang, Phys. Rev. 178, 1457 (1969).

Page 392: Nanotechnology and Nano-Interface Controlled Electronic Devices

MDC–SHG spectroscopy of organic monolayer film 375

40. A. Tojima, T. Manaka, and M. Iwamoto, Rev. Sci. Instrum., to be published.41. J.A. GiordMaine, Phys. Rev. 138, A1599 (1965).42. M. Iwamoto, T. Kubota, and M.R. Muhamad, J. Chem. Phys. 102, 19368 (1995).43. J. Xue, C.S. Jung, and M.W. Kim, Phys. Rev. Lett. 69, 474 (1992).44. E.B. Sirota, Langmuir 13, 3849 (1997).45. M. Iwamoto and Z.C. Ou-Yang, J. Chem. Phys. 117, 7705 (2002).46. T. Kihara, Rev. Mod. Phys. 25, 831 (1953).47. K.M. Aoki, Y. Tabe, and H. Yokoyama, Mol. Cryst. Liq. Cryst. 367, 2979 (2001).48. J.L. Barrat and L. Bocquet, Phys. Rev. Lett. 82, 4671 (1999).49. See, for example, N. Israelachivichi, Intermolecular and Surface Forces (Academic Press, London,

1992).50. M. Iwamoto, T. Manaka, A. Tojima, and Z.C. Ou-Yang, Chem. Phys. Lett, 359, 169 (2002).51. M. Iwamoto, Y. Mizutani, and A. Sugimura, Phys. Rev. B 54, 8186 (1996).52. M. Iwamoto et. al., submitted.

Page 393: Nanotechnology and Nano-Interface Controlled Electronic Devices

This Page Intentionally Left Blank

Page 394: Nanotechnology and Nano-Interface Controlled Electronic Devices

Nanotechnology and Nano-Interface Controlled Electronic DevicesEditors: M. Iwamoto, K. Kaneto and S. Mashiko© 2003 Elsevier Science B.V. All rights reserved

CHAPTER 18

Light-driven dynamic controls innano-hybrid materials

Takahiro Seki *

Chemical Resources Laboratory, Tokyo Institute of Technology, 4259 Nagatsuta-cho, Midori-ku,Yokohama, Japan

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3772. Photoalignment of polysilane chain by Az monolayer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 378

2.1. Procedures and preparative conditions of polysilane film . . . . . . . . . . . . . . . . . . . 3782.1.1. System and principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3782.1.2. Photoalignment procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3792.1.3. Effects of film thickness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 381

2.2. Optimum conditions of the azobenzene monolayer . . . . . . . . . . . . . . . . . . . . . . . . . 3822.2.1. Lateral packing density . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3832.2.2. Tail length . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3832.2.3. Micro-patterning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 384

3. Facilitated photoinduced mass migration by nano-hybridization . . . . . . . . . . . . . . . . . . . . 3853.1. Binary component hybrid materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 385

3.1.1. Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3853.1.2. SRG Formation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3873.1.3. Thermal Properties and SRG Stability . . . . . . . . . . . . . . . . . . . . . . . . . 390

3.2. Soft Crosslinkable Liquid Crystalline Polymer. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3913.2.1. Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3913.2.2. SRG formation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3913.2.3. Crosslinking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 392

4. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 393Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 394References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 394

1. Introduction

In biological systems, nano-scale elements such as proteins, DNA, and lipid membranesare assembled in very sophisticated ways so that they exert various functions for life.

* Present address: Department of Applied Physics, Graduate School of Engineering, NagoyaUniversity, Chikusa, Nagoya 464-8603, Japan.

Page 395: Nanotechnology and Nano-Interface Controlled Electronic Devices

378 T. Seki

This could be the ultimate example for the utility of hybridization. In materials scienceand technology, hybridization of materials at nanometer scales is of key essence inmany aspects for technological innovations, for example, creation of nano-composite forhigh modulus materials, smart and intelligent materials that find themselves a suitableresponse to an environmental stimulus, high performance optical and electrical devices.These materials and devices strongly rely on the hybridization and interfacial controlsbetween different materials.

This chapter introduces our on-going two topics on light-responsive, smart, softmaterial systems in which the interface and hybridization play the essential roles inthe generation of functions. In both subjects, the dynamic photofunctions are producedby the trans/cis isomerization of azobenzene (Az). In the former part, photoresponsiveazobenzene monolayers that are capable of controlling the orientation of polymer chainsare mentioned. It is revealed that the in-plane orientation of polysilane (silicon catenatedpolymer) backbone is controlled through transfer from a pre-oriented Az monolayerby linearly polarized light (LPL). In the latter part, the photoinduced mass migrationin soft liquid crystalline Az polymers will be introduced. By nano-hybridization withlow molecular mass liquid crystal (LC) molecules, the efficiency of mass migration isgreatly 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 principleThe surface-mediated photoalignment has recently become an important technology inliquid crystalline materials [1,2]. The orientational control of polymer chains by suchphotochemical procedure is an alluring and challenging target because it may providenew technologies for micro-processing of polymer films. Here, the photoalignmentbehavior of (polydi-n-hexylsilane) (PDHS) by an azobenzene monolayer (6Az10-PVA)is introduced [3–5]. Illustrative representations of the system and chemical structureof materials are shown in Fig. 1. Use of a polymeric material for the Az monolayeris of particular significance for the following reasons. First, the reversible trans/cisphotoisomerization reaction readily proceeds due to the amorphous nature of themonolayer. Second, the polymeric monolayer is mechanically and environmentally sorobust 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 preferen-tially in the perpendicular direction to the polarization plane of actinic light with respectto the in-plane component. This effect is called the photoinduced optical anisotropyor Weigert effect (Fig. 2) [6,7]. The transition moment of Az is nearly parallel to thedirection of the long axis of the rod-like molecule. When the randomly oriented Azunits are irradiated LPL, the Az units whose long axis are in the parallel direction arepreferentially excited. This situation causes the reorientation of the Az side chains toa non-excitable direction, namely orthogonal to the light polarization direction. In thepresent case, the light illumination at both wavelengths (365 and 436 nm) is performed

Page 396: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 the6Az10-PVA monolayer after irradiation with LPL at 436 nm. In this typical example thedichroic ratio (DR = [Abs(⊥)−Abs(‖)]/[Abs(⊥)+Abs(‖)]) is 0.43. This large in-planeanisotropy cannot be obtained with simple irradiation with 436 nm LPL from the initialtrans state. The higher in-plane orientation is only obtained after non-polarized 365 nmlight is pre-irradiated to the Az monolayer. This may be ascribed to the fact that thecis-Az film is more fluid [8], having larger motional freedom. This can lead to efficientmolecular reorientation.

2.1.2. Photoalignment procedureA spin-cast film of PDHS (Mw = 2.5 × 104, film thickness = 45 nm) from a hexanesolution is subsequently prepared onto this photooriented Az monolayer. Use of hexane

Page 397: Nanotechnology and Nano-Interface Controlled Electronic Devices

380 T. Seki

Fig. 2: Schematic illustration of photoinduced optical anisotropy by linearly polarized light in photochromicmolecular 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 absorptionspectra taken with perpendicular and parallel vector of the probing light.

as the cast solvent does not damage or destroy the photoinduced orientation of theAz monolayer. Storage of the PDHS film in the dark at room temperature for 2 daysallows sufficient crystallization of this polymer giving the absorption peak around 360nm. As shown in Fig. 4a, the film shows no preferential in-plane orientation just afterthe solvent casting. After crystallization the PDHS film exhibited a strong in-planeanisotropic nature (Fig. 4b). Since the transition moment of the Si backbone is alongthe backbone direction, the polarized absorption spectra indicates that the Si mainchain is aligned perpendicular to the polarization plane of the actinic light. The aligneddirection 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

Page 398: Nanotechnology and Nano-Interface Controlled Electronic Devices

Light-driven dynamic controls in nano-hybrid materials 381

Fig. 4: Photoalignment behavior of PDHS film on the LPL-irradiated 6Az10-PVA monolayer. (a) Immedi-ately after solvent cast, (b) after first crystallization of PDHS, (c) after annealing, and (d) after successivesecond crystallization of PDHS.

Az monolayer in an epitaxial manner. The orientational order of PDHS film wasfurther enhanced upon annealing and successive cooling. Upon heating, the backboneconformation adopts the helical gauche state (320 nm absorption) with enhanced thein-plane orientational order (Fig. 4c). Successive crystallization leads to a more highlyoriented PDHS film (Fig. 4d).

2.1.3. Effects of film thicknessSince the in-plane-component orientation of the Az monolayer is transferred to thePDHS chain in the contacting region, the orientational order of the entire film shouldbe dependent on the film thickness [4]. PDHS films with varied thickness from 10to 100 nm are prepared on the pre-irradiated Az monolayer. Fig. 5 shows the orderparameter (S) as a function of film thickness for the PDHS films of the lower molecularweight (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. Theorder parameter is enhanced after the second crystallization for the lower molecularweight sample (triangles in Fig. 5a). The orientational order is enhanced upon annealingand the S for thicker films became comparable with those of the thinner ones. It isreasonably assumed that the influence of the photooriented Az layer does not reach

Page 399: Nanotechnology and Nano-Interface Controlled Electronic Devices

382 T. Seki

Fig. 5: Orientational order parameter (S) of photoaligned PDHS film on the 6Az10-PVA monolayer as afunction of film thickness. The circle and triangle represent data after the first and second crystallization,respectively. Mws of PDHS are 2.5×104 (a) and 1.4×106.

the overall thickness only after the first crystallization. However, the lateral packingcorrelation of the disordered part is improved by heating, leading to an enhancement oforientational order at least within the thickness less than 100 nm. This interpretation isrationalized by AFM topographic observations. In the case of the high molecular weightmaterial, the photoalignment of PDHS is considerably suppressed (Fig. 5b). When thePDHS film becomes thicker than 30 nm, essentially no alignment in the PDHS chain isinduced.

2.2. Optimum conditions of the azobenzene monolayer

Much knowledge has been accumulated on the LPL induced molecular reorientationin photochromic LB films (mostly Az systems) [9–12]. However, little attention hasbeen paid as to what extents the lateral packing density and structural modification ofthe molecular structure affect the photoorientation behavior. Changes in the molecularstructure at the alkylene spacer and alkyl tail attached to the Az part have been foundto influence the molecular cooperativity with contacting liquid crystal molecules in thecommand surface systems [13,14].

Page 400: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 functionof occupying area of Az unit. The occupying area is controlled by pre-mixing of the spreading solutionscontaining trans- and cis-6Az10-PVA.

2.2.1. Lateral packing densityThe lateral packing density of an LB monolayer can be readily changed by the positionof the moving barrier on the water surface. The molecular occupying area (Aoc) canbe varied by changing the surface pressure for transfer [15]. An alternative way tocontrol Aoc is to vary the mixing ratios of the trans/cis isomers of Az in the spreadingsolution. Since the cis-Az monolayer occupies much larger areas than the trans one atlow pressures, deposition of mixed monolayers at a fixed low surface pressure (5 mNm−1) allows large variations of Aoc.

The data of S for the PDHS film on the photooriented Az monolayer are shown inFig. 6 together with the DR data of the Az monolayer [5]. Here, the area is changed bymixing the two isomerized states at different ratios. A good correlation was obtainedbetween 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 lengthThe length of tail part is anticipated to influence the order of photoinduced anisotropybecause the length of the alkyl chain greatly affects the molecular packing. In the presentsystem, the Az monolayer is anchored to the hydrophilic substrate surface via the polarPVA backbone and the tail part is positioned to the outermost surface. PDHS shouldinteract directly with the tail part of the Az side chain. In this context, an exploration onthe tail effect is of great interest.

DR of the Az layer after photoorientation (436 nm LPL, 3.0 J cm−2) and the resultingS of PDHS (thickness = 25 nm ) as a function of the carbon number of the tail partare shown in Fig. 7. In all cases, the PDHS backbone is commonly aligned parallel tothe direction of the Az monolayers, namely perpendicular to the polarization directionof the actinic light. As obviously shown, the highest DR of Az monolayers and S ofPDHS are obtained for the Az monolayer having the C8 tail [5]. DR and S are 0.56

Page 401: Nanotechnology and Nano-Interface Controlled Electronic Devices

384 T. Seki

Fig. 7: DR of Az monolayers (open) and S of PDHS films (closed) as a function of tail length (m) in themAz10-PVA monolayer.

and 0.68, respectively, for the C8 tail. The magnitude of the DR has a good relationshipwith the degree of spectral shift. The monolayer showing the larger hypsochromic shiftof λmax provides the larger DR. Thus, the higher orientational order is obtained for themonolayer containing more amounts of H-aggregated Az in the common lateral densityof 0.4 nm2 per Az unit.

2.2.3. Micro-patterningA great benefit to apply the photoprocess is the feasibility of micro-patterning. Fig. 8shows polarized optical microscopic images of a locally photoaligned PDHS film as abent 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. Theirradiation to the Az monolayer is first performed in the above conditions through a pho-tomask 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 at45° indicates the clear switching of emergence of disappearance of the line. This factindicates that the bright line area is birefringent and the PDHS chains are uniaxiallyaligned. The dark parts do not change its low transmission by rotation, indicative ofthe random orientation of the polymer at micrometer levels. Thus, the micro-patterningof polymer alignment at 6 μm resolution can be successfully achieved [5]. In thephotoprocess, basically any desired pattern is applicable. It would be impossible orlaborious to attain such locally addressed orientation via mechanical or flow processes.

Page 402: Nanotechnology and Nano-Interface Controlled Electronic Devices

Light-driven dynamic controls in nano-hybrid materials 385

Fig. 8: Polarized optical microscopic images of a microscopically photoaligned DHS film (thickness = 25nm) 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 cooperativityin dynamic processes observed in solvent cast films. The essentials have much incommon to monolayer systems [16,17].

3.1.1. BackgroundAz polymers are potentially useful as materials for reversible holographic informationstorage and photonic devices [18–22]. Surface relief grating (SRG, regular topologicalsurface modification) formed via the irradiation with an interference pattern of coherent

Page 403: Nanotechnology and Nano-Interface Controlled Electronic Devices

386 T. Seki

Fig. 9: Schematic illustration of photoinduced surface relief gratings (SRG) and an typical example of theSRG structure in 6Az10-PVA/5CB nano-hybrid film observed by AFM.

light has been demonstrated only recently [23–25] and is perhaps the most attractingtarget in the current research of Az polymers (Fig. 9). A great deal of data has beenaccumulated quite rapidly due to its basic phenomenological interest [26–30] and alsoto attractive technological applications [30]. This process has particular technologicaladvantages since (i) it offers a facile, all-optical and single step fabrication processthat does not require a wet development procedure, and (ii) the surface topology iserasable by application of circularly polarized light or heating above the glass transitiontemperature (Tg), which realizes the repeatable utilization. It is of no doubt that the SRGis formed via large-scale polymer chain migration, however, the precise mechanism isstill the subject of intensive investigation.

Page 404: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 polymersincluding the side-chain and main-chain types. A novel approach using a binarycomponent 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 liquidcrystal (LC) molecule, 4′-pentyl-4-cyanobiphenyl (5CB) (Fig. 10).

3.1.2. SRG FormationThe spin-cast films (50–100 nm thickness) are prepared from chloroform solutionsdissolving 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-polarizedUV (365 nm) light in advance to attain a cis-rich photoequilibrated state (UV lighttreatment). The interference Ar ion laser beam can be obtained simply by mixinga 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.

Page 405: Nanotechnology and Nano-Interface Controlled Electronic Devices

388 T. Seki

Fig. 12: (a) First-order diffraction efficiency of the spin-cast films of pure 6Az10-PVA (diamonds) and anequimolar 6Az10-PVA/5CB hybrid film ( f = 0.5) with (circles) and without (triangles) UV light treatmentas 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 reflectedHe–Ne laser beam (488 nm) as a function of the total exposure energy. This figurecontains the data obtained with the pure 6Az10-PVA film (diamonds), the hybrid filmof 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 efficiencyis observed at an early exposure stage (circles). The diffraction efficiency reaches amaximum 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 surfacegeometrical modulation. Fig. 12b and c show AFM images of the hybrid films exposedwith 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 nmdepth fluctuations with no regular periodicity. The SRG structure is clearly generated

Page 406: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 exactcoincidence with the interference pattern of the light intensity. It is noteworthy that suchquick topological induction proceeds at room temperature. At 5.0 s exposure, the surfacemodulation becomes more clear and provides a regularly spaced sinusoidal structure(12c). The depth from the peak to the bottom (Δh) is ca. 70 nm, which is comparable tothe initial film thickness. The photo-modulated structure is stable and unchanged at leastfor a month at room temperature. The surface undulation can be erased by non-polarizedUV light irradiation at 200 mJ cm−2 that is nearly sufficient for the attainment of thecis-rich photoequilibrated state at room temperature. Heating at 100°C (isotropic phaseof the hybrid film) for 30 min also deletes the surface structure. For the pure 6Az10-PVAfilm in the same procedure with UV light treatment, no significant surface undulation isformed 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 nmAr-ion laser beam and (ii) the plasticization of the film in the cis-Az form. The trans-to-cis photoisomerization of Az leads to an increase in the film fluidity of 6Az10-PVA asshown by the microscopic observation [8]. Judging from the general knowledge that theSRG is formed below Tg of polymer films [30], the surface undulation should be formedalong with stiffening of the hybrid film due to the back isomerization of Az to the transform by 488 nm illumination.

Fig. 13 indicates the most essential feature of the binary system. This figure displaysthe diffraction efficiency (closed circles) and the surface modulation depth (open circles,Δh) 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, whichcorresponds to the stoichiometry of two 5CB molecules per one Az unit. Below andabove this ratio the diffraction efficiency rapidly decreases. The profile of Δh almostfollows that of the diffraction efficiency to give the maximum depth (100 nm) atf = 0.67.

Fig. 13: The first-order diffraction efficiency (closed) and surface modulation depth (open) of 6Az10-PVA/5CB hybrid film at various molar fraction of 5CB and 6Az10-PVA.

Page 407: Nanotechnology and Nano-Interface Controlled Electronic Devices

390 T. Seki

Fig. 14: DSC diagrams of pure 6Az10-PVA and 5CB and their binary mixtures.

3.1.3. Thermal Properties and SRG StabilityDifferential scanning calorimetric (DSC) profiles for the pure materials and the mixturesare shown in Fig. 14 [32]. The DSC trace of pure 6Az10-PVA ( f = 0.0) contains twoendothermic peaks at 71 and 98°C. The higher temperature endotherm is associated withthe clearing transition. On cooling from the isotropic phase, pure 6Az10-PVA developedfocal conic and schlieren textures which are characteristic of smectic phase. The DSCprofile of pure 5CB ( f = 1.0) shows two endothermic peaks at 23 and 35°C whichcorrespond 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 indicatesthat the mixtures behave like a newly formed hybridized liquid crystalline polymer. Athigher 5CB molar ratios ( f > 0.67), DSC diagrams provide additional two peaks inthe low temperature region (indicated by arrows), which are enlarged with increasingthe 5CB content. These endothermic peaks should arise from pure 5CB, and thus thephase separation is suggested above f = 0.67. The Langmuir monolayer experimentsalso show that the co-spreading of these materials on water provides the lateral phaseseparation above f = 0.67 as proven from the surface pressure–area isotherms, UV-visible spectroscopic data, and Brewster angle microscopic observation [16]. In boththe bulk and monolayer, the 6Az10-PVA can accommodate two 5CB molecules per Azside chain unit without phase separation, and above this criterion, the phase separationstarts to occur. The film transport seems to be strongly assisted by the cooperativeself-assembling process.

Page 408: Nanotechnology and Nano-Interface Controlled Electronic Devices

Light-driven dynamic controls in nano-hybrid materials 391

Upon heating the SRG inscribed film ( f = 0.5), the diffraction efficiency increasedgradually until 55°C, and then shows a sudden decrease. The critical temperature inthe Az/LC hybrid film is nearly in accord with the strong endothermic peak at 47°Cin the DSC curve (Fig. 14). Elevating temperature up to 114°C almost completelydiminished the light diffraction. After heating to this level, the surface becomes highlyflat with some small hollows with a depth comparable to the initial film thickness. Thetopological 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. MotivationAnother important requirement for SRG material is the shape stability in terms of long-term storage and durability at higher temperatures. The stability can be improved whenone employs amorphous polymers with high Tg [33,34] or liquid crystalline polymershaving high Ti (transition temperature to isotropic state) [35]. However such materialsmarkedly reduce the mass mobility.

A unique soft and crosslinkable SRG polymer is developed by Zettsu et al. [36] Thepolymer involves an oligo(ethylene oxide) (EO) side chain (6Az10-PE4.5, Fig. 15) in-stead of incorporation with LC molecule. After the surface relief structure is formed, thepolymer is then subjected to chemical crosslinking via formalization (acetal formationwith formaldehyde) between the hydroxyl group at the terminus of EO.

3.2.2. SRG formation6Az10-PE4.5 adopts a liquid crystalline state at room temperature where light exposureexperiments are performed. The SRG structure is formed also at very low dose levelsaround 100 mJ cm−2. The photo-modulated structure is adequately stable as far as it iskept at room temperature. The surface undulation could be erased by heating up to theisotropization temperature above 80°C (see below), and regenerated at the same positionmany times.

Fig. 15: Chemical structure of soft and crosslinkable liquid crystalline Az polymer (6Az10-PE4.5) thatshows rapid photoinduced mass transfer.

Page 409: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 squaresindicate data for non-crosslinked and crosslinked films, respectively. The initial first-order diffractionefficiency was ca. 3%.

3.2.3. CrosslinkingWhen the inscribed film is exposed to a mixed vapor of formaldehyde and hydrogenchloride for 24 h at room temperature, the formalization reaction (acetal formation)proceeds. This procedure chemically links two hydroxyl groups located at the terminusof EO unit to yield a crosslinked polymer network (Fig. 16). Fig. 17 shows the changesin diffraction efficiency (normalized) on heating for the inscribed 6Az10-PE4.5 filmbefore and after formalization. Before formalization, the diffraction efficiency reducesdrastically around 80°C, which is in exact accord with the isotropization temperatureof this liquid crystalline polymer. In sharp contrast, the formalized SRG film retains

Page 410: Nanotechnology and Nano-Interface Controlled Electronic Devices

Light-driven dynamic controls in nano-hybrid materials 393

the diffraction efficiency up to 240°C without appreciable reduction. The 6Az10-PE4.5film after formalization reaction becomes insoluble in tetrahydrofuran or chloroformboth of which are good dissolving solvents for the untreated 6Az10-PE4.5. The SRGstructure disappears by heating at 100°C for the untreated film whereas the formalizedfilm exactly preserves the surface modulation pattern even at 250°C without anydamage. The 6Az10-PE4.5 is in an isotropic state above ca. 100°C, however, thisdoes not cause a reduction in diffraction efficiency. The process employed here canbe 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, incompensation for requirement of vast amounts of exposure energy. Light doses requiredfor 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. Thesoft 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 longirradiation time spans reaching to several ten minutes with an Ar ion laser beam ofmoderate 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 asholographic optical recording, waveguide formation, liquid crystal aligning etc. [30]. Onthe other hand, for the rapid migrating polymer systems described here, utilization ofmotional functions may be possible, for example, application to soft micro actuators andmicro-patterning of other guest materials by means of mass transfer.

4. Summary

This chapter described two sets of light-driven dynamic material systems. In theorientational transfer system from the Az monolayer to the polysilane film, nano-hybridization at the two-dimensional interface should play important roles for theeffective control. Here, the monolayer modifications at nanometer levels such assubtle lateral density and tail length crucially alter the control behavior. Probably thecooperative interactions between the alkyl tail of the Az unit in the monolayer andthe hexyl substituent in the polysilane are essentially involved in this process. Thecooperative effect is quite obvious in the photoinduced mass transfer process observedfor the hybrid films. The effective transport is attained only when the LC molecule ora soft segment is incorporated. It is emphasized that the nano-hybridization providesnew functions in soft materials. For the present cases, high segmental flexibility andmolecular mobility is realized while molecular orientation is retained in the nano-hybridized states. This situation can be found typically in biological systems, wherewe find a number of processes and structures that inspire us for materials science andtechnology.

Page 411: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 greatcooperation 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 Ministryof Education, Culture, Sports, Science and Technology of Japan.

References

1. K. Ichimura, Chem. Rev. 100, 1847 (2000).2. M. O’Neill and S.M. Kelly, J. Phys. D: Appl. Phys. 33, R67 (2000).3. T. Seki, K. Fukuda, and K. Ichimura, Langmuir 15, 5098 (1999).4. K. Fukuda, T. Seki, and K. Ichimura, Macromolecules 35, 2177 (2002).5. K. Fukuda, T. Seki, and K. Ichimura, Macromolecules 35, 1951 (2002).6. T. Todorov, N. Tomova, and L. Nikolva, Opt. Commun. 47, 123 (1983).7. K. Anderle, R. Birenheide, M.J.A. Werner, and J.H. Wendorff, Liquid Cryst. 9, 691 (1991).8. T. Seki, H. Sekizawa. S. Morino, and K. Ichimura, J. Phys. Chem. B 102, 5313 (1998).9. M. Barnik, V.M. Kozenkov, N.M. Shtykov, S.P. Palto, and S.G. Yudin, J. Mol. Electron. 5, 53 (1989).

10. S. Yokoyama, M. Kakimoto, and Y. Imai, Langmuir 10, 4594 (1994).11. M. Schönhoff, L.F. Chi, H. Fuchs, and M. Löshe, Langmuir 11, 163 (1995).12. R. Wang, L. Jiang, T. Iyoda, D.A. Tryk, K. Hashimoto, and A. Fujishima, Langmuir 12, 2052 (1996).13. T. Seki, Supramol. Sci. 3, 25 (1996).14. T. Seki, K. Ichimura, R. Fukuda, and Y. Tamaki, Kobunshi Ronbunshu 52, 599 (1995).15. T. Seki, R. Fukuda, T. Tamaki, and K. Ichimura, Thin Solid Films, 243, 675 (1994).16. T. Ubukata, T. Seki, and K. Ichimura, J. Phys. Chem. B 104, 4141 (2000).17. T. Ubukata, T. Seki, S. Morino, and K. Ichimura, J. Phys. Chem. B 104, 4148 (2000).18. M. Eich, J.H. Wendorff, B. Reck, and H. Ringsdorf, Makromol. Chem. Rapid. Commun. 8, 59 (1987).19. S. Ivanov, I. Yakovlev, S. Kostromin, V. Shibaev, L. Läsker, J. Stumpe, and D. Kreysig, Makromol.

Chem. Rapid. Commun. 12, 709 (1991).20. J. Stumpe, L. Müller, and D. Kreysig, Makromol. Chem. Rapid. Commun. 12, 81 (1991).21. A. Natansohn, P. Rochon, J. Gosselin, and S. Xie, Macromolecules 25, 2268 (1992).22. T. Ikeda and O. Tsutsumi, Science 268, 1873 (1995).23. P. Rochon, E. Batalla, and A. Natansohn, Appl. Phys. Lett. 66, 136 (1995).24. D.Y. Kim, S.K. Tripathy, L. Li, and J. Kumar, Appl. Phys. Lett. 66, 1166 (1995).25. P.S. Ramanujam, N.C.R Holme, and S. Hvilsted, Appl. Phys. Lett. 68, 1329 (1996).26. J. Kumar, L. Li, X.L. Jiang, D.Y. Kim, T.S. Lee, and S.K. Tripathy, Appl. Phys. Lett. 72, 2096 (1998).27. C.J. Barrett, P. Rochon, and A. Natansohn, J. Chem. Phys. 109, 1505 (1998).28. P.S. Ramanujam, M. Pedersen, and S. Hvilsted, Appl. Phys. Lett. 74, 3227 (1999).29. K. Sumaru, T. Yamanaka, T. Fukuda, and H. Matsuda, Appl. Phys. Lett. 75, 1878 (1999).30. 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).31. T. Ubukata, T. Seki, and K. Ichimura, Adv. Mater. 12, 1675 (2000).32. T. Ubukata, T. Seki, and K. Ichimura, Colloids Surfaces A 113, 198 (2002).33. T. Fukuda, H. Matsuda, N. Viswanathan, S.K. Tripathy, J. Kumar, T. Shiraga, M. Kato, and H.

Nakanishi, Synth. Met. 102, 1435 (1999).34. T. Fukuda, H. Matsuda, T. Shiraga, T. Kimura, M. Kato, N. Viswanathan, J. Kumar, and S.K.

Tripathy, Macromolecules 33, 420 (2000).

Page 412: Nanotechnology and Nano-Interface Controlled Electronic Devices

Light-driven dynamic controls in nano-hybrid materials 395

35. L. Andruzzi, A. Altomare, F. Ciardelli, R. Solaro, S. Hvilsted, and P.S. Ramanujam, Macromolecules32, 448 (1999).

36. N. Zettsu, T. Ubukata, T. Seki, and K. Ichimura, Adv. Mater. 13, 1693 (2001).

Page 413: Nanotechnology and Nano-Interface Controlled Electronic Devices

This Page Intentionally Left Blank

Page 414: Nanotechnology and Nano-Interface Controlled Electronic Devices

Part E

Fabrication and CharacterizationTechnology

Page 415: Nanotechnology and Nano-Interface Controlled Electronic Devices

This Page Intentionally Left Blank

Page 416: Nanotechnology and Nano-Interface Controlled Electronic Devices

Nanotechnology and Nano-Interface Controlled Electronic DevicesEditors: M. Iwamoto, K. Kaneto and S. Mashiko© 2003 Elsevier Science B.V. All rights reserved

CHAPTER 19

Solvent-induced morphology innano-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 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3991.1. Carbon nanotube brushes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4001.2. Templated semiconductor oxide nanotubes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 401

2. Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4012.1. Preparation of honeycomb structure in CNT–polymer composite films . . . . . . 4012.2. Preparation of aligned semiconductor oxide nanotubes . . . . . . . . . . . . . . . . . . . . . 4012.3. Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 401

3. Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4023.1. CNT–polymer composite films . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4023.2. Templated oxide nanotubes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 405

4. Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 407Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 409References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 410

1. Introduction

As devices grow smaller and approach the nanoscopic scale, conventional fabricationtechniques (pick-and-place, wafer-to-wafer transfer, etc.) will become obsolete. Forcessuch as gravity become virtually irrelevant as other interactions (surface tension, vander 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 ultimatemorphology in condensed phases of such objects – the relative spatial positioningof the objects – can be manipulated by controlling the attributes of the solvents

Page 417: Nanotechnology and Nano-Interface Controlled Electronic Devices

400 B. Cheng et al.

Fig. 1: SEM image of an aligned multi-walled carbon nanotube “brush” prepared via microwave plasma-enhanced chemical vapor deposition (from Ref. [14]; scale bar 3 μm).

from which they are harvested. Herein we will show that the morphology of twoseemingly unrelated nano-structures – arrays of multi-walled carbon nanotubes andsol–gel-templated semiconductor-oxide nanotubes – share strikingly similar responsesto 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 surfaceof silicon impregnated with a dispersion of catalyst particles [2]. The resulting arrayof upright CNTs, each anchored to the silicon substrate at a catalyst site and eachapproximately 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 andedge of the CNT brush.) We affect a dramatic change in the morphology of the brushfrom 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 surfaceof the CNT brush. The thickness of the final dry polymer film is designed to be lessthan the height of the brush “bristles” – the length of the CNTs. The presence ofwater vapor during the polymer solution evaporation causes micro-droplets of water tocondense on the cold evaporating solution surface, and the water surface tension causessevere structural changes in the array of CNTs – namely, coalescence of the CNTs intocrop-circle-like defects as these water micro-droplets dry. The mechanical properties ofthis honeycomb reinforcement in thin films impregnated with polymer are examined.Also, field emission data from the honeycomb morphology was contrasted with datafrom the original CNT brush.

Page 418: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 In2O3 and Ga2O3 [4]. After thesynthesis of the nanotubes the excess oxide is ground off the exposed surface of thetemplate – a microporous alumina substrate comprised of nearly hexagonally packedcylindrical pores (200 nm i.d.) spanning the template thickness (50 μm). The othersurface 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 bundledmorphology due to the high surface tension of the water and the associated forcesduring the drying process. This morphology can be affected by exchanging the aqueouswashing 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 onoxidized silicon substrates in advance [5]. Dilute polymer solutions (e.g., polystyrene intoluene, polystyrene in benzene etc.) were slowly evaporated on the top surface of thealigned CNT brush. The residual solvent was removed by heating at 50°C in vacuum. Ahoneycomb morphology was induced in the brush when the evaporation was carried outin the presence of a moist airflow. The moisture content of the airflow was controlledby 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 porousalumina 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 templateby immersing in a sol of metal ions for a specific time. The sol-containing membranewas dried and annealed in air at elevated temperatures: 973 K for 12 h (In2O3), and 773K for 12 h (Ga2O3). The oxide-containing membrane was bonded to a substrate and theexposed surface was polished with a 13 micron sandpaper. The array of quasiparallelsemiconductor-oxide nanotubes was isolated by dissolving the alumina template with a6 M aqueous NaOH solution. The surplus of NaOH can be removed by washing withwater or ethanol prior to drying the array.

2.3. Measurements

Field emission measurements were carried out in a vacuum chamber (10−7 Torr) atroom temperature. The experimental setup is a point-plane type [6]. CNT brushes with

Page 419: Nanotechnology and Nano-Interface Controlled Electronic Devices

402 B. Cheng et al.

and without the honeycomb morphology were used as the cathode. The brushes wereattached to a stainless steel substrate by silver paste. In this configuration, the CNTscomprising a brush were placed with their long axes perpendicular to the substrate. Astainless rod was used as the anode. The distance between the emitting brush surface andthe end of the rod was determined by first lowering the rod to the sample until electricalcontact was observed, then lifting the rod to a desired distance from the cathode surface.

SEM measurements were conducted using a JEOL JEM-6300 scanning electronmicroscope. The samples were attached to a SEM stub with conductive tape, and a verythin 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 rangefrom −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 plasma-enhanced CVD has been described in detail [5]. Fig. 1 shows a typical SEM image ofaligned 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) beforechoosing dilute (0.5%) polystyrene in toluene. The final thickness of polymer filmcan be designed to be less than the height of the CNTs by controlling the amountand the concentration of the polymer casting solution. Our initial goal was to slowlyevaporate the polymer solution in order to produce a thin (∼1 μm) polymer film at thesubstrate–CNT brush interface.

The objective of adhering a polymer to the CNT brush was originally pursued in anattempt to merely produce a stronger bond between the CNT “bristles” of the brushand the silicon substrate in an attempt to produce more robust arrays for field emissionapplications. However, after the solvent evaporated, we discovered that the entire brushcould be lifted from the substrate intact in the form of a flexible thin film with the CNTsnormal to the film surface. Fig. 2 shows a part of a partially peeled composite polymerfilm with the CNT brush embedded in it. The film appears contiguous and probablyresults from the evaporating polymer solution going to the substrate and wetting thesubstrate via capillary aided transport between the ∼100 nm-spaced CNTs. Since theCNTs are embedded in the polymer film with their axes normal to the film surface, itmay be possible to pattern the underside of such a flexible CNT-containing polymerfilm with a conductor and thereby achieve patterned emission from a flexible composite.This fabrication methodology may provide an alterative to patterned growth of CNTnanostructures from patterned catalyst sites on rigid substrates [7].

When a polystyrene–toluene solution is evaporated on the top surface of the alignedCNT brush in the presence of moisture, crop-circle-like defects formed among the brushbristles (see Fig. 3). The defect distribution is not homogeneous. This result shows

Page 420: Nanotechnology and Nano-Interface Controlled Electronic Devices

Solvent-induced morphology in nano-structures 403

Fig. 2: SEM image of part of a peeled-up, flexible polystyrene composite film with the CNT brushmorphology 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 intoluene is evaporated on the brush surface (scale bar 1 mm).

Page 421: Nanotechnology and Nano-Interface Controlled Electronic Devices

404 B. Cheng et al.

that moisture may play a key role to the formation of defects. Currently there is aconsiderable effort underway to understand the mechanism of “breath-figure” formation– the two-dimensional ordered microdroplets of water that form on the surface ofevaporating organic liquids in the presence of water vapor. These two-dimensionalstructures 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 apolystyrene–toluene solution. The resulting ordered array of holes in the polymer filmwas formed by evaporating in the presence of moisture with a forced airflow across thesolution surface. A hexagonally packed array of holes – replicas of the ordered 2D arrayof condensed water microdroplets which sinks into the less-dense, evaporating polymersolution – is evident in the resulting polymer film [8].

Such observations prompted us to evaporate polystyrene–toluene solutions on theCNT brush surface in the presence of water vapor. The final result is the formationof honeycomb nano-structure in the brush with the honeycomb cell dimensions relatedto the humidity and velocity of moist airflow during the evaporation process. Thehoneycomb morphology of a CNTs–polystyrene composite film on its silicon substrateis shown in Fig. 4. At high magnification, the SEM pictures show that many CNTscomprise the walls of the honeycomb cells. We speculate that the high surface tensionof water droplets (that have condensed on the brush surface during the polymer castingprocedure) causes the CNT brush bristles to coalesce and collapse into this uniqueultra-structure when the water dries.

We have studied the field emission and mechanical properties of this novel polymercomposite. Fig. 5 displays typical current–voltage (I–V ) characteristics correspondingto Fowler–Nordheim behavior for the original CNT brush (Fig. 5a) and the honeycombcomposite (Fig. 5b). The maximum current we measured in the original CNT brush isabout 20 μA whereas for composite honeycomb structure it is about 24 μA. No signif-icant difference was observed in the field emission results between these two samples.

Carbon fibers are routinely used as reinforcements for polymeric matrices in theaircraft and sports industries because of their high modulus and low density. For similarreasons – a high Young modulus (over 1 TPa) and high strength (tensile strength,200 GPa) [9] – carbon nanotubes may have potential for applications requiring high-modulus 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 themechanical properties of composites. We have contrasted the Young modulus of a thinpolystyrene film reinforced with the CNT honeycomb nanostructure with the modulusof pure polystyrene films of comparable dimensions (Fig. 6). The reinforced compositefilm is prepared by evaporating a second more concentrate polystyrene solution on thehoneycomb morphology. SEM pictures of the honeycomb nanostructure impregnatedwith polymer show the distribution of CNTs in the polystyrene matrix. The CNTs arenot parallel to one another within in the honeycomb wall; the solidifying polystyrenematrix during the second solution evaporation does not destroy the integrity of the tubes.Good interfacial bonding between polystyrene and nanotubes appears to be present inthe SEM image and may account for the good load transfer from the polystyrene matrixto the nanotubes via the interfacial shear stresses and thereby explain the enhancedmodulus of the reinforced film.

Page 422: Nanotechnology and Nano-Interface Controlled Electronic Devices

Solvent-induced morphology in nano-structures 405

Fig. 4: SEM images of honeycomb structures of polystyrene composite films with CNTs; (a) the top surfaceview of large area of honeycomb structure (scale bar 10 μm); (b) view of part of the area of he honeycombstructure (10 μm); (c) one enlarged hole with CNTs clearly visible within the walls (1 μm); (d) a tilted viewof the honeycomb morphology (10 μm).

3.2. Templated oxide nanotubes

When using the sol–gel templating method to prepare semiconductor oxides in the formof an array of aligned nanotubes, the synthesis includes purification steps wherein wateror ethanol is used to wash away the surplus of NaOH employed used to dissolve thealumina template. After the NaOH is completely removed, the resulting array of tubularoxides is dried in air at room temperature. Our findings show that it is difficult to get avery good aligned array of individual, free-standing nanotubes. Fig. 7 shows the usualsolvent-induced morphology of semiconductor nanotubes. Most nanotubes are eithercollapsed entirely (Fig. 7a), or bundled together (Fig. 7b). Other groups have reportedsimilar results [12]. This morphology appears to be the result of solvent induced-forces (a surface tension effect) when the nanotube array is dried in air following thedissolution of the template.

In order to avoid this collapsed/bundled morphology, a low surface tension solventmust be used in the final stage of processing. The surface tension of liquid CO2 (0.005N/m) is much less than that of H2O (0.072 N/m) and ethanol (0.024 N/m), and critical

Page 423: Nanotechnology and Nano-Interface Controlled Electronic Devices

406 B. Cheng et al.

200 400 600 800 1000 1200 1400

0

4

8

12

16

20Z

14μm

238 μmC

urre

nt /

μA

Voltage / V

0.000 0.002 0.004 0.006 0.008 0.010

- 4 0

- 3 8

- 3 6

- 3 4

- 3 2

- 3 0

- 2 8

- 2 6

- 2 4

- 2 2

- 2 0

Z

ln(I

/ V2)

1 / V

200 400 600 800 1000 1200 1400

0

4

8

12

16

20

24

Z

14μm

238μm

Cur

rent

/ μ

A

Voltage / V

0.002 0.004 0.006 0.008 0.010 0.012

- 4 0

- 3 6

- 3 2

- 2 8

- 2 4

decreasing Z

ln(I

/ V

2)

1 / V

(a)

(b)

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 polystyrenecomposite film with a honeycomb morphology.

Page 424: Nanotechnology and Nano-Interface Controlled Electronic Devices

Solvent-induced morphology in nano-structures 407

-120 -100 -80 -60 -40 -20 0 20 40 60 80 100

0.0

5.0x107

1.0x108

1.5x108

2.0x108

2.5x108

3.0x108 CNT and polymer composites

Pure polymer

You

ng's

Mod

ulus

(Pa)

Temperature(oC)Fig. 6: A comparison of the Young modulus for the CNT-reinforced honeycomb morphology of polystyrenecomposite film with that of a comparable pure polystyrene film.

point drying is a well-established technique [13]. Comparisons of the solubility ofsome reagents in liquid CO2, suggest that if ethanol is used to wash away the surplusof NaOH, the ethanol can be readily exchanged with liquid carbon dioxide. Afterthe NaOH was completely removed after the template dissolution step, the templatednanotube array, wet with ethanol, was transferred to a high-pressure stainless cell forexchange with CO2. Several exchanges with liquid CO2 insured that the array waseventually immersed in pure liquid CO2. A reduction of the pressure of CO2 yielded adry array wherein the morphology of the nanotubes exhibited little of the collapsed orbundled morphology. Fig. 8 shows the morphology of nanotubes harvested from liquidCO2. Fig. 8a shows the SEM image of the top surface of the aligned array of nanotubesand Fig. 8b shows the SEM image of a cross section of the array. Individual nanotubesare evident confirming that the morphology of these arrays can be dramatically alteredby the solvent from which they are dried (compare with Fig. 7).

4. Concluding remarks

Solvent-induced phenomena sometimes have undesirable influences on the morphologyof carefully designed nanomaterials. We have shown that the CNT brush morphologycan be converted into a honeycomb nanostructure on silicon substrates by evaporating

Page 425: Nanotechnology and Nano-Interface Controlled Electronic Devices

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. Thiscomposite CNT honeycomb morphology exhibits a combination of good electrical andmechanical 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 ofwater-induced capillary forces within the brush during drying. The same forces appearto be the origin of the collapsed or bundled morphology exhibited by sol–gel-templatedprocessing of semiconductor nanotube arrays. Liquid and supercritical CO2 drying ofnanostructures can eliminate the effect of capillary (or interfacial tension) forces on the

Page 426: Nanotechnology and Nano-Interface Controlled Electronic Devices

Solvent-induced morphology in nano-structures 409

Fig. 8: SEM images of semiconductor nanotubes developed in liquid CO2; (a) the top surface view (scalebar 10 μm); (b) a cross section view (scale bar 10 μm); neither views show collapsed nor bundledmorphologies.

final morphology and enable the unique properties and potential applications of thesenovel 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

Page 427: Nanotechnology and Nano-Interface Controlled Electronic Devices

410 B. Cheng et al.

measurements. These findings are based upon work supported in part by the NASAGrant NAC-1-2301, the STC Program of the National Science Foundation underAgreement No.CHE-9876674, and instrumentation used in this research is supported bythe Office of Naval Research MURI Grant N00014-98-1-0597.

References

1. H.O. Jacobs, A.R. Tao, A. Schwartz, D.H. Gracias, and G.W. Whitesides, Science 296, 323 (2002).2. C.J. Lee, J. Park, S.Y. Kang, and J.H. Lee, Chem. Phys. Lett. 323, 554 (2000).3. J.C. Hulteen and C.R. Martin, J. Mater. Chem. 7, 1075 (1997).4. B. Cheng and E.T. Samulski, J. Mater. Chem. 11, 2901 (2001).5. H. Cui, O. Zhou, and B.R. Stoner, J. Appl. Phys. 88, 6072 (2000).6. W. Zhu, L. Bower, O. Zhou, G. Kochanski, and S. Jin, Appl. Phys. Lett. 75, 873 (1999).7. B.Q. Wei, R. Vajtai, Y. Jung, J. Ward, R. Zhang, G. Ramanth, and P.M. Ajayan, Nature 416, 495

(2002).8. M. Srinivasarao, D. Collings, A. Philips, and S. Patel, Science 292, 79 (2001).9. 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).10. H.D. Wagner, O. Lourie, Y. Feldman, and R. Tenne, Appl. Phys. Lett. 72, 188 (1998).11. L.S. Schadler, S.C. Giannaris, and P.M. Ajayan, Appl. Phys. Lett. 73, 3842 (1998).12. B.A. Hernandez, K.-S. Chang, E.R. Fisher, and P.K. Dorhout, Chem. Mater. 14, 480 (2002).13. D.L. Goldfarb, J.J. de Pablo, and P.F. Nealey, J. Vac. Sci. Technol. B 18, 3313 (2000).14. C. Bower, W. Zhu, S. Jin, and O. Zhou, Appl. Phys. Lett. 77, 830 (2000).

Page 428: Nanotechnology and Nano-Interface Controlled Electronic Devices

Nanotechnology and Nano-Interface Controlled Electronic DevicesEditors: M. Iwamoto, K. Kaneto and S. Mashiko© 2003 Elsevier Science B.V. All rights reserved

CHAPTER 20

Polarons in conjugated polymer and itscomposite with fullerene

Kazuhiro Marumoto and Shin-ichi Kuroda

Department of Applied Physics, Nagoya University, Chikusa-ku, Nagoya 464-8603, Japan

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4112. Experimental details . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4133. Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 414

3.1. Light-induced electron spin resonance (LESR) and excitation spectra ofphotogenerated polarons in regioregular poly(3-alkylthiophene) (PAT) . . . . . . 414

3.2. LESR and excitation spectra of photogenerated polarons in regioregularPAT–C60 composite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 416

3.3. Microwave saturation of LESR in regioregular PAT–C60 composite: spin–lattice and transverse relaxation time of photogenerated polarons . . . . . . . . . . . 420

3.4. Transient response of LESR in regioregular PAT–C60 composite . . . . . . . . . . . . 4233.5. Dependence of LESR in regioregular PAT–C60 composite on the light

excitation intensity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4254. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 426Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 427References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 427

1. Introduction

Quasi-one-dimensional conducting nondegenerate conjugated polymers have been in-vestigated extensively because of a wide variety of interesting physical properties suchas 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 electronicsystems. Polarons carry spin and charge and are also expected to play important roles inmanifesting the functions such as high conductivity, electroluminescence, etc. of thesematerials. However, the nature of the polaronic state is not yet completely clear.

Photoinduced charge separation in polymeric semiconductors is important becauseits study not only contributes to understanding of the basic photoexcited states in theseone-dimensional semiconductors, but also contributes to the development of efficient

Page 429: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 opticaland photovoltaic devices. This charge separation is attributed to photoinduced electrontransfer between polymer and C60, which forms positive and negative polarons on thepolymer chain and C60 molecule, respectively. Recently, it has also been shown that suchefficient charge separation occurs in blends of PPV derivatives and cadmium selenidenanocrystals [15,16].

Light-induced electron spin resonance (LESR) is a direct microscopic method fordetecting and studying the photogenerated polarons, and has been employed to clarifythe polaronic state in the typical electroluminescent polymer PPV and its derivatives[17–26] and oligothiophenes [27]. For the PAT–C60 composites using regiorandomPATs, the LESR spectra and microwave power dependence of the LESR intensity havebeen studied [9,11], in addition to observations of the photoluminescence quenching andthe remarkable enhancement of the photoconductivity [5,7,8]. Recently, regioregularPATs have been synthesized, and electronic absorption, X-ray diffraction, and crossedpolarizing micrograph studies show that the cast films of the regioregular PATs areself-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 eVlower than that of the regiorandom PATs (2.1 eV) [28]. The typical conductivity andcharge carrier mobility of the regioregular PATs in thin films is greatly improvedand are three orders of magnitude larger than those of the regiorandom PATs, whichis due to self organization of the regioregular PATs, having a lamella structure withtwo-dimensional conjugated sheets formed by interchain stacking [29–32]. Moreover,gate-induced superconductivity in the regioregular PAT film of a field-effect device hasbeen observed very recently, and it is reported that the appearance of superconductivityseems to be closely related to the self-assembly properties of the polymer [33]. Thecharge separation process in the PAT–C60 composites is expected to be greatly affectedby such regioregularity of PATs, however, PAT–C60 composites using regioregularPATs have not been studied. Moreover, excitation spectra, transient response uponirradiation by light, and excitation-light intensity dependence of the LESR signals ofthe PAT–C60 composites have not been reported so far, which would provide importantinformation concerning the mechanism of the charge separation and recombination ofthe photogenerated polarons.

In the sections that follow, we report on LESR studies of photogenerated polarons ina regioregular PAT–C60 composite, in addition to a pure regioregular PAT, using variablephotoexcitation energy [34,35]. A remarkable enhancement of the LESR signals inthe excitation spectrum is observed compared with the case of the pure regioregularPAT, which is consistent with the enhancement of the photoconductivity at around1.8 eV where an optically forbidden transition of C60 occurs. Prompt and persistentLESR signal components are observed, and the excitation-light intensity dependenceshows that the prompt contribution increases monotonically with increasing excitation-

Page 430: Nanotechnology and Nano-Interface Controlled Electronic Devices

Polarons in conjugated polymer and its composite with fullerene 413

light intensity while the persistent contribution is found to be excitation intensityindependent. The comparison of the present LESR results with those of PPV and itsderivatives, oligothiophenes, the regiorandom PAT–C60 composites, and the compositesof 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 uniformmixing. Cast films of the pure regioregular PAT8 and the PAT–C60 composite witheach a thickness of approximately 6 μm were fabricated inside ESR sample tubes.ESR measurements were performed with a Bruker E500 X-band spectrometer witha microwave cavity with optical windows down to liquid-He temperature using anOxford ESR900 gas-flow cryostat. The absolute magnitude of the g value was calibratedutilizing an NMR gaussmeter for obtaining the static magnetic-field strength and amicrowave frequency counter. A JASCO SM-5 light source with a 300 W xenon lampwas used to provide excitation in the range 300–1100 nm (1.1–4.1 eV) at power levelsup to 2 mW/cm2 with a spectral width of 10 nm. The light was delivered by an opticalfiber to the quartz sample tube. For the excitation spectrum, the light intensity wasadjusted to give the same photon flux at each wavelength.

The LESR experimental procedure consisted of the following sequence [14]: (i) scanthe ESR spectrum of the nonilluminated sample; (ii) scan the ESR spectrum under lightillumination; (iii) turn off the illumination and scan the ESR spectrum; and (iv) warmthe 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).

Page 431: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 switchingoff of the excitation light does not lead to the disappearance of the ESR signals atlow temperature. To eliminate the ESR spectrum completely, the sample had to bewarmed up to room temperature (annealing). This step was performed every time whilemeasuring the dependence of the temperature, microwave power, excitation-light energy,and excitation-light intensity. The dark and annealed ESR signals were confirmed to bethe same every time. As default definition of the term LESR we choose the light-onsignal corrected by subtracting the dark signal. Finally, we distinguish between theprompt LESR signal (light-on minus light-off) and the persistent one (light-off minusdark).

3. Results and discussion

3.1. Light-induced electron spin resonance (LESR) and excitation spectra ofphotogenerated polarons in regioregular poly(3-alkylthiophene) (PAT)

The upper curves in Fig. 2 show the observed first derivative ESR spectra of thePAT8 under dark condition (dotted line) and 700 nm illumination (solid line) at 60K with a microwave power of 0.06 mW. The lower curve in Fig. 2 shows the LESRspectrum obtained by the subtraction of the two curves mentioned above. The g valueis 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 spectrumfrom that under 700 nm illumination at 60 K.

Page 432: Nanotechnology and Nano-Interface Controlled Electronic Devices

Polarons in conjugated polymer and its composite with fullerene 415

to trapped polarons, possibly generated by oxygen doping or thermal excitation duringthe synthesis, as observed in PPV [17] and oligothiophenes [27]. The spin concentrationin the dark condition is obtained as 1 spin per 3.5 × 105 PAT-molecular unit, whichis 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 toless 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 consistentwith the great improvement of the typical conductivity and charge carrier mobility ofthe regioregular PATs in thin films [29–32]. In the cases of two dialkoxy derivativesof 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 ESRintensity of the PAT8 is enhanced considerably upon irradiation by light, and theLESR intensity is approximately 14 times as large as the dark ESR intensity. The darkESR and LESR spectra are almost the same and have asymmetric line shapes. Thepeak-to-peak linewidth (ΔHpp) of the ESR signal of PAT8 is approximately 3.2 G. TheLESR signal of the PAT8 becomes undetectable above approximately 240 K due tothe 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 thePAT–C60 composite discussed later. The ΔHpp of the LESR signal does not depend onthe temperature, except for the low temperature region where the saturation occurs.

The excitation spectrum of the LESR signal provides important information concern-ing the mechanism of the charge separation. Fig. 3 shows the variation of the normalizedpeak-to-peak LESR intensity (Ipp) with the photon energy of the incident light for thePAT8. The measurements were performed with a microwave power of 0.06 mW at 60K. 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 absorptionspectrum of the PAT8.

Page 433: Nanotechnology and Nano-Interface Controlled Electronic Devices

416 K. Marumoto and S.-i. Kuroda

The absorption measurement was performed at room temperature using a thin filmcast on a glass substrate. The absorption spectrum is qualitatively consistent with thatof poly(3-hexylthiophene) [28]. The excitation spectrum of the LESR signal shows amonotonic and rapid increase above 2.5 eV up to 4.1 eV, despite of the lower opticaldensity, as observed in the PPV and its derivatives [20–25]. This behavior indicates thatthe observed spins are photogenerated polarons whose generation efficiency increasesabove the optical absorption peak due to excitons, because weakly bound electronsand holes created can be more readily dissociated into polarons. In the cases of PPVderivatives (CN-PPV and MEH-PPV), the excitation spectrum has a clear threshold atan energy of 2.7–2.8 eV in CN-PPV and approximately 3 eV in MEH-PPV, which ishigher than the peak energy of the optical absorption of approximately 2.5eV in bothpolymers [23–25]. The difference in the energy of the optical absorption peak and thethreshold of the LESR excitation spectrum provides the exciton binding energy, and thevalue 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 intensityat around 1.77 eV (700 nm) as shown in Fig. 3. The LESR spectral line shape at 700nm almost agrees with that at 300 nm, suggesting that the light-induced spin speciesare also polarons. Even with low photon energy, charge separation such as electrontransfer between the molecule and oxygen or bipolaron dissociation may occur due tothe existence of defects. Similar increases have been observed in PPV [20–22] andoligothiophenes [27], and a correlation between such peak and the existence of the spinconcentration in the dark condition is suggested. Another explanation for the peak ataround 1.77 eV may be given by the DeVore theory, which presents a similar peak in theshape of photoconductivity spectral distribution curves by a theoretical analysis basedupon the effects of surface and volume recombination of the charge carriers liberated bythe light [37]. Measurements of the photoconductivity spectra and comparison with theLESR excitation spectra are needed in order to study this peak in more detail, which arenow in progress and will be published elsewhere [38].

3.2. LESR and excitation spectra of photogenerated polarons in regioregularPAT–C60 composite

The upper curves in Fig. 4 show the observed first derivative ESR spectra of thePAT–C60 composite under dark condition (dotted line) and 700 nm illumination (solidline) at 60 K with a microwave power of 0.06 mW. The lower curve in Fig. 4 showsthe 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 areobserved. The obtained g value of g1 = 2.002 and g2 = 1.999 correspond to positiveand negative polarons on PAT8 and C60, respectively, which is consistent with thoseof the regiorandom PAT–C60 composites [9,11]. The composites of PPV derivativeand C60 derivative or C60 have been also reported to show two LESR signals withg values of g = 2.0000–2.0025 and g = 1.9955–1.9995 for low- and high-field ESRsignals, which were assigned to the positive polaron P+ on the conjugated polymerbackbone and the C−

60 radical, respectively [6,9,14]. The observed polaron spins in thedark condition are considered to be attributed to the ground-state electron transfer in the

Page 434: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 700nm illumination (solid line) at 60 K. Lower curve: LESR spectrum of the PAT–C60 composite obtained bysubtracting 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 thepure PAT8. The spin concentration in the dark condition is obtained as 1 spin per2.2×105 PAT-molecular unit, in other words, 1.4×1016 spin/g, which is more than 17times lower than that of the regiorandom PAT–C60 composites [11]. In the case of theregiorandom PAT–C60 composites, the enhancement of spin density upon C60 dopingwas more modest than that by using conventional dopants such as BF−

4 and AsF−6

[11,39]. The ESR intensity of the PAT–C60 composite is enhanced considerably uponirradiation by light, and the LESR intensity is approximately 10 times as large as thedark ESR intensity. The dark ESR and LESR spectra are almost the same and haveasymmetric line shapes. Since the LESR signals of PAT8 (g1) and C60 (g2) overlap witheach other, it is difficult to obtain a precise half-amplitude linewidth (ΔH1/2) from theexperimental results directly. Hereafter, we define the peak-to-peak linewidths (ΔHpp)of the LESR signals of PAT8 and C60 as an interval of the magnetic field between amaximum and a minimum of the first derivative LESR spectrum around g1 and g2,respectively. The ΔHpp of the LESR signals of PAT8 and C60 are approximately 3.1 Gand 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 0.06mW under 300 nm illumination. Similar temperature dependence of the LESR signalswas also observed under 700 nm illumination. The ΔHpp of the LESR signals doesnot depend on the temperature, except for the low temperature region where saturationoccurs. The LESR signals become undetectable above approximately 200 K due tothe 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 theESR signals due to the longer spin–lattice relaxation rate. The LESR signal intensity

Page 435: Nanotechnology and Nano-Interface Controlled Electronic Devices

418 K. Marumoto and S.-i. Kuroda

Fig. 5: Temperature dependence of the LESR intensity of the PAT–C60 composite. The filled circles andopen squares denote the signal intensity of PAT8 and C60, respectively.

of PPV and its derivatives also shows a similar temperature dependence as mentionedabove [22,23,25]. The LESR signal of PAT8 tends to saturate at higher temperature thanthat of C60, indicating that the spin–lattice relaxation time of the polaron spins of PAT8is longer than that of C60, which is consistent with the results of the microwave-powerdependence of the LESR intensity as discussed later.

The excitation spectrum of the LESR signals in the PAT–C60 composite is shown inFig. 6, which presents the variation of the normalized Ipp with the photon energy ofthe incident light for the PAT–C60 composite. The measurements were performed witha microwave power of 0.06 mW at 60 K. The data are plotted by using the signal ofPAT8 (g1). For the comparison, the previously reported absorption spectrum (solid line)[39] and photocurrent spectrum (dotted line) [8] of the PAT–C60 composites are alsoshown in Fig. 6. The samples utilized for the absorption and photocurrent measure-ments 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 PAT18are presented here, other composites using regiorandom PATs with different alkyl-chainlength have been reported to exhibit similar characteristics [7]. The photoconductivityof the PAT–C60 composites is remarkably enhanced upon C60 doping due to the photoin-duced electron transfer at around 1.8 eV where the optically forbidden transition of C60

(hu → t1u) occurs in the composites. It has been reported that the enhancement of thephotoconductivity in the regiorandom PAT18–C60 composites increases monotonicallywith increasing the concentration of C60 up to 10 mol% [8]. The excitation spectrumof the LESR signal of PAT8 (g1) shows a remarkable enhancement at around 1.77eV (700 nm) and a monotonic increase above 2.5 eV up to 4.1 eV. The excitationspectrum of the LESR signal of C60 (g2) shows almost the same behavior as that ofPAT8. There is almost no excitation-energy dependence of the LESR line shapes. The

Page 436: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 dottedlines show the absorption spectrum of PAT6–C60 composite (C60: 5 mol%) and photocurrent spectrum ofPAT18–C60 composite (C60: 5 mol%) for the comparison, respectively.

enhancement at around 1.8 eV is clearly shown by comparing the LESR excitationspectrum of the PAT–C60 composite with that of the pure PAT8 shown in Fig. 3. If themagnitude of the LESR data of the PAT8 and PAT–C60 composite films is normalizedusing the same photon flux for the light intensity, the same sample weight for the PAT8polymer, and the same film dimension, the LESR magnitude of the PAT–C60 compositeis enhanced by a factor of 3.9 compared with that of the pure PAT8 at 700 nm, althoughit is enhanced by a factor of 1.1 at 300 nm. The enhancement of the LESR signals ataround 1.8 eV is qualitatively similar to the enhancement of the photocurrent spectrumon the whole, which is consistent with the photogeneration of the polarons due to thephotoinduced electron transfer. However, the detailed structures of the excitation spectraof the LESR and photocurrent around the peaks are quantitatively different from eachother. The peak energy of the excitation spectrum of the LESR is slightly lower thanthat of the photocurrent, and considerable photogenerated signals are observed in thelower photon-energy region even at around 1 eV. One possible explanation for thedifferences of the excitation spectra between the LESR and photocurrent is given byconsidering aggregation effects of the polymer chain. The studies of PPV derivative(MEH-PPV) films have presented strong evidence that interchain species do form inconjugated polymer films, and that the degree of interchain interactions can be con-trolled by varying the solvent and polymer concentration of the solution from whichthe 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 improvedfield-effect mobility [29]. This planarity leads to extension in the conjugation lengthhaving strong electron–phonon coupling leading to a significant correlation between theconformation and electro-optical properties [42]. The photocurrent spectrum shown in

Page 437: Nanotechnology and Nano-Interface Controlled Electronic Devices

420 K. Marumoto and S.-i. Kuroda

Fig. 6 has been measured using the regiorandom PAT–C60 composite, and the planarconformation in such regiorandom PAT–C60 composite may be different from that inthe regioregular PAT–C60 composite. Therefore, the regioregularity of PATs may affectthe interaction between PAT and C60 and cause the differences of the excitation spectrabetween the LESR and photocurrent mentioned above. For another possibility to explainthe differences of the excitation spectra, the difference of the measurement conditionssuch as the effect of electric field may be considered, as mentioned in the discussion ofthe transient response later. Anyway, these differences on the aggregation effects and/ormeasurement conditions may affect the charge-separation processes between PAT andC60 in the PAT–C60 composite. Further measurements of the absorption, LESR and pho-toconductivity using the same sample seem to be needed to clarify the charge-separationprocesses and the photoexcitation of the polarons in more detail.

3.3. Microwave saturation of LESR in regioregular PAT–C60 composite: spin–latticeand transverse relaxation time of photogenerated polarons

To obtain more information about the photogenerated polarons at around 1.8 eV wherethe enhancement occurs, the microwave-power dependence, the transient responseupon irradiation by light, and the excitation-light intensity dependence of the LESRsignals due to the photogenerated polarons have been studied. In this section, themicrowave saturation of the LESR in regioregular PAT–C60 composite and spin–latticeand transverse relaxation time of photogenerated polarons are presented below. Fig. 7shows the microwave-power dependence of the LESR signals at 60 K under 700 nmillumination. In the low microwave-power region, the signal intensity of PAT8 is largecompared with that of C60. At a microwave power of 200 mW, however, the signal ofPAT8 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 observedin the composites of PPV derivative and C60 derivative or C60 [14]. The ΔHpp of thesignal of C60 increases from 2.1 G to 5.3 G as the microwave power increases from2 μW to 200 mW for the saturation effect. The definition of the ΔHpp used here ismentioned above. The ΔHpp of the signal of PAT8 seems to decrease from 3.4 G to1.3 G as the microwave power increases from 2 μW to 60 mW because of the overlapbetween 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 magneticfield as the microwave power increases, and therefore the true ESR linewidth of thesignal of PAT8 is considered to increase as the microwave power increases for thesaturation effect. Since the ESR linewidth considerably depends on the microwavepower, the double integrated LESR intensity is estimated for the individual ESR signalsand used in the following discussion of the microwave-power dependence.

Fig. 8 shows the microwave-power dependence of the double integrated LESRintensity of the PAT–C60 composite at 60 K under 700 nm illumination. The filled circlesand open squares denote the signal intensity of PAT8 (g1) and C60 (g2), respectively.When the microwave power increases, the LESR intensity of PAT8 saturates, showinga 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

Page 438: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 nmillumination.

experimentally available microwave-power range up to 200 mW. These features aresimilar to those of the regiorandom PAT–C60 composites [11] and the composites ofPPV derivative and C60 derivative [14]. In the latter case, the temperature variationof the saturation curve for the low-field LESR signal due to the positive polarons hasbeen measured between 90 K and 200 K. The microwave power corresponding to thesaturation maximum increases from around 0.2 mW to 0.7 mW as temperature increasesfrom 90 K to 120 K, and finally the saturation maximum cannot be reached with themicrowave power available above 150 K [14]. The field for the saturation maximumis related to the spin–lattice relaxation time T1 by T1 ∝ 1/(T2γ

2e H 2

1max). Here H1max

is the microwave-field amplitude at the sample where the maximum in the saturationcurve occurs, T2 is the transverse relaxation time, and γe is the electronic magnetogyricratio, respectively. The LESR signal of PAT8 can be saturated at around 0.6 mW (60K), whereas that of C60 does not saturate under the same conditions up to a powerwhich is 330 times higher. Therefore, the T1 of the polaron spins of PAT8 is at leastapproximately 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.

Page 439: Nanotechnology and Nano-Interface Controlled Electronic Devices

422 K. Marumoto and S.-i. Kuroda

Fig. 8: Microwave-power dependence of the double integrated LESR intensity of the PAT–C60 composite at60 K under 700 nm illumination. The filled circles and open squares denote the signal intensity of PAT8 andC60, respectively.

A Jahn–Teller-type distortion on the fullerene ball would split the T1u level, and result intwo closely spaced energy levels available for the polaron spins of C60 [43,44]. Thermalaveraging 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-powerdependence.

The ESR linewidth of PAT8 signal of the PAT–C60 composite is governed byinhomogeneous broadening due to the hyperfine interaction between the polarons andprotons bonded to the carbon in the system. In the case of an inhomogeneouslybroadened ESR line, the T2 cannot be determined directly by using the ΔHpp. Hereafter,the continuous-wave saturation method considering the inhomogeneous broadening ofthe ESR linewidth is utilized to determine the T2 and T1 of the present system accordingto the standard procedures by Castner [45]. The concept of a spin packet is neededto describe the inhomogeneous broadening. A spin packet is composed of all spinshaving the same static field to within a field interval determined by the dephasingtime T2. The inhomogeneous broadening producing the static width is larger thanthe individual spin-packet width. The ratio ΔHL/ΔHG of the Lorentzian spin-packetwidth to the inhomogeneous Gaussian width measures the degree of inhomogeneousbroadening, and is expressed using a parameter a as a = ΔHL/ΔHG. The quantity ais determined by using the ratio H1(VR1/2 Upper)/H1(VR1/2 Lower), where VR1/2 Upper andVR1/2 Lower are 1/2 the maxima of the ESR intensity VR for a given saturation curve, andthe H1(VR1/2 Upper) and H1(VR1/2 Lower) are corresponding microwave fields. The ratioH1(VR1/2 Upper)/H1(VR1/2 Lower) ≈ 50 is obtained from the experimental result shown inFig. 8, and then the a is estimated as a ≈ 0.045 with the help of figure 3 in Ref. [45]. TheT2 of the homogeneous Lorentzian line is defined as T2 = 1/(γeΔHL) = 1/(aγeΔHG).The ΔHG is estimated approximately by using ΔHG = √

2ln2ΔHpp/2, and therefore

Page 440: Nanotechnology and Nano-Interface Controlled Electronic Devices

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

T2γ 2e H 2

1/2

, (1)

where H1/2 is H1 at the onset of saturation. The intersection of the linear signal belowsaturation with the horizontal line drawn tangent to the maximum of the saturation curvelocates what we shall call H1/2 Uncorrected. The multiplicative correction factor whichrelates it to the true H1/2 is a known function of a and ranges from 1 for a equalzero to 2 for the completely homogenous case (for the homogenous saturation caseH1/2 corresponds to VR max) [45]. Using the values of T2 and H1/2 Uncorrected, the T1 isestimated 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 compositeupon irradiation by repeated light (700 nm) pulses with different time widths. The datawere recorded using the lower field peak of the PAT8 signal (g1) with a microwavepower of 0.06 mW at 100 K. The photoresponse is found to have two parts: a promptpart and a persistent one. That is, upon irradiation by light, the LESR signal increasesstepwise then almost saturates. When irradiation ceases, the LESR signal decreases bysome amount stepwise, and then the remaining part decreases very slowly. The samebehavior 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 PAT8signal (g1) at 100 K.

Page 441: Nanotechnology and Nano-Interface Controlled Electronic Devices

424 K. Marumoto and S.-i. Kuroda

and deep traps in the PAT–C60 composite, respectively. The time response of the promptcomponent is considerably faster than that of PPV and its derivative [22,23]. The decaytime of the remanent component is of the order of several hours, which is extremelylong compared with that of the prompt component. It is difficult to fit these decayprocesses by double exponential functions. These features of the transient response aresimilar to those of the photoconduction of the regiorandom poly(3-octadecylthiophene)(PAT18)–C60 composite [46]. However, there are some differences between the transientresponses of the LESR and photoconduction. In the LESR transient response, almostno slowly rising component after the stepwise increase upon irradiation by light isobserved, and the magnitude of the remanent component is almost independent of thepulse width (duration) of the irradiation light. These features are inconsistent with thoseof the photoconduction of the PAT18–C60 composite, where the slowly rising componentincreases as the pulse width increases and so does the persistent photoconductivity. It isconsidered that these features in the photoconduction may originate in the accumulationof carriers contributing to the persistent photoconductivity [46]. The accumulation ofcarriers may be caused by the effect of the electric field that is not applied in theLESR measurements, which may cause the differences mentioned above. Fig. 10 showsthe transient responses of the prompt LESR component for the PAT8 signal at thetemperatures 60 K, 100 K and 120 K. As temperature increases, the magnitude of theprompt LESR component as well as the persistent one decreases. The decay time to halfvalue of the prompt LESR component also decreases from 18 s (60 K) to 2.1 s (120K). Similar tendency is also observed in the transient response of the prompt LESRcomponent 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 and120 K.

Page 442: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 transientresponse, 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 excitationintensity

The excitation-light intensity dependence of the double integrated LESR intensityfor the prompt and persistent LESR signals in the PAT–C60 composite is shown inFig. 11a and b, respectively. The filled circles and open squares denote the signalintensity of PAT8 (g1) and C60 (g2), respectively. These measurements were performedunder 700 nm illumination at 60 K with a microwave power of 0.06 mW. Theprompt 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 theintensity of the exciting light at 60 K: the prompt components (a), and the persistent components (b). Thefilled circles and open squares denote the signal intensity of PAT8 and C60, respectively.

Page 443: Nanotechnology and Nano-Interface Controlled Electronic Devices

426 K. Marumoto and S.-i. Kuroda

integrated ESR intensity) and increases monotonically as excitation-light intensityincreases for both the low- and high-field LESR lines of PAT8 (g1) and C60 (g2),respectively. Both prompt components show a ∼ I 0.27

exc. power dependence on theintensity of excitation light. Almost the same behavior for the excitation-light intensitydependence is also observed at 100 K. This dependence shows a saturation behaviorwith respect to the excitation-light intensity rather than the bimolecular type (I 0.5

exc.),which is inconsistent with the results of the composites of PPV derivative and C60

derivative [14]. The present results suggest quadrumolecular-type recombination amongthe photogenerated positive and negative polarons, because the excitation-light intensitydependence of the quadrumolecular-type recombination is I 0.25

exc. , which is consistent withthe experimental results. The LESR transient responses shown in Fig. 10 have been alsoexplained preliminary by using calculated curves based on the quadrumolecular-typerecombination at various temperatures, which shows that the recombination rate ofthe photogenerated polarons obtained by the fitting calculation increases monotonicallyas temperature increases. Further experimental studies on the sample dependence, theC60 concentration dependence, and the effects of the regioregularity of PATs usingregiorandom PATs are needed in order to study this power dependence on the excitation-light intensity in more detail. These studies are now in progress. Anyway, the promptLESR data clearly shows the monotonic increase on excitation-light intensity, whichmeans that the photogeneration of the polarons due to the photoinduced electrontransfer between PAT8 and C60 increases as the excitation-light intensity increases. Forthe persistent component, it is nearly independent of the intensity of the previouslyapplied light within the light-power range 0.038–0.61 mW/cm2. Similar excitation-lightintensity dependence is also observed at 100 K. This result is consistent with that of thecomposites of PPV derivative and C60 derivative [14], which is related to the deep trapsdue 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 withfullerene (C60) having 5 mol% C60 concentration have been investigated by means of thelight-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 dueto the photogenerated polarons shows a monotonic increase above 2.5 eV, despite of thelower optical density, because of the dissociation of weakly bound electrons and holescreated into polarons. For the PAT–C60 composite, the transient two LESR signals areobserved, which come from the photogenerated positive and negative polarons on PAT8and C60 due to the photoinduced electron transfer between PAT8 and C60, respectively.A remarkable enhancement of the LESR signals in the excitation spectrum at around1.8 eV is observed, which is consistent with the enhancement of the photoconductivityat around 1.8 eV where the optically forbidden transition of C60 occurs. Spin–latticerelaxation time of the polaron spins of PAT8 is obtained considering inhomogeneousbroadening of the ESR linewidth, which is at least approximately 18 times longer thanthat of C60. The prompt and persistent LESR signal components are observed, and

Page 444: Nanotechnology and Nano-Interface Controlled Electronic Devices

Polarons in conjugated polymer and its composite with fullerene 427

the excitation-light intensity dependence shows that the prompt contribution increasesmonotonically as the excitation-light intensity increases while the persistent contributionis 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 valuablediscussions and comments. We also would like to thank N. Takeuchi and T. Ozakifor their cooperation in the LESR measurements. This work is supported by NEDOInternational Joint Research Program, 99MB1 ‘Nonlinear Excitations in MolecularElectronic Materials: Detection, Control and Device Application’.

References

1. A.J. Heeger, S. Kivelson, J.R. Schrieffer, and W.-P. Su, Rev. Mod. Phys. 60, 781 (1988).2. H.G. Kiess (Ed.), Conjugated Conducting Polymers, Springer Series in Solid-State Sciences 102,

(Springer, Berlin-Heidelberg, 1992).3. 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).4. 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).5. S. Morita, A.A. Zakhidov, and K. Yoshino, Solid State Commun. 82, 249 (1992).6. N.S. Sariciftci, L. Smilowitz, A.J. Heeger, and F. Wudl, Science 258, 1474 (1992).7. K. Yoshino, S. Morita, T. Kawai, H. Araki, X.H. Yin, and A.A. Zakhidov, Synth. Met. 55–57, 2991

(1993).8. K. Yoshino, X.H. Yin, S. Morita, T. Kawai, and A.A. Zakhidov, Solid State Commun. 85, 85 (1993).9. L. Smilowitz, N.S. Sariciftci, R. Wu, C. Gettinger, A.J. Heeger, and F. Wudl, Phys. Rev. B 47, 13835

(1993).10. G. Yu, J. Gao, J.C. Hummelen, F. Wudl, and A.J. Heeger, Science 270, 1789 (1995).11. 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).12. 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).13. E.K. Miller, K. Lee, K. Hasharoni, J.C. Hummelen, F. Wudl, and A.J. Heeger, J. Chem. Phys. 108,

1390 (1998).14. 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).15. N.C. Greenham, X. Peng, and A.P. Alivisatos, Phys. Rev. B 54, 17628 (1996).16. D.S. Ginger and N.C. Greenham, Phys. Rev. B 59, 10622 (1999).17. S. Kuroda, T. Noguchi, and T. Ohnishi, Phys. Rev. Lett. 72, 286 (1994).18. S. Kuroda, K. Murata, T. Noguchi, and T. Ohnishi, J. Phys. Soc. Jpn. 64, 1363 (1995).19. S. Kuroda, Int. J. Mod. Phys. B 9, 221 (1995).20. S. Kuroda, K. Murata, Y. Shimoi, S. Abe, T. Noguchi, and T. Ohnishi, in: Materials and Measure-

ments in Molecular Electronics, Springer Proceedings in Physics 81, edited by K. Kajimura and S.Kuroda (Springer, Tokyo, 1996), p. 256.

21. K. Murata, S. Kuroda, Y. Shimoi, S. Abe, T. Noguchi, and T. Ohnishi, J. Phys. Soc. Jpn. 65, 3743(1996).

22. K. Murata, Y. Shimoi, S. Abe, S. Kuroda, T. Noguchi, and T. Ohnishi, Chem. Phys. 227, 191 (1998).

Page 445: Nanotechnology and Nano-Interface Controlled Electronic Devices

428 K. Marumoto and S.-i. Kuroda

23. S. Kuroda, K. Marumoto, H. Ito, N.C. Greenham, R.H. Friend, Y. Shimoi, and S. Abe, Chem. Phys.Lett. 325, 183 (2000).

24. S. Kuroda, K. Marumoto, N.C. Greenham, R.H. Friend, Y. Shimoi, and S. Abe, Synth. Met. 119, 655(2001).

25. S. Kuroda, K. Marumoto, Y. Shimoi, and S. Abe, Thin Solid Films 393, 304 (2001).26. S. Kuroda, K. Marumoto, N.C. Greenham, R.H. Friend, Y. Shimoi, and S. Abe, Mol. Cryst. Liq.

Cryst. 371, 159 (2001).27. K. Marumoto, N. Takeuchi, S. Kuroda, R. Azumi, and M. Matsumoto, Synth. Met. 119, 549 (2001).28. T.-A. Chen, X. Wu, and R.D. Rieke, J. Am. Chem. Soc. 117, 233 (1995).29. Z. Bao, A. Dodabalapur, and A.J. Lovinger, Appl. Phys. Lett. 69, 4108 (1996).30. H. Sirringhaus, N. Tessler, and R.H. Friend, Science 280, 1741 (1998).31. 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).

32. K. Kaneto, W.Y. Lim, W. Takashima, T. Endo, and M. Rikukawa, Jpn. J. Appl. Phys. 39, L872(2000).

33. J.H. Schön, A. Dodabalapur, Z. Bao, Ch. Kloc, O. Schenker, and B. Batlogg, Nature 410, 189 (2001).34. 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.

35. K. Marumoto, N. Takeuchi, T. Ozaki, and S. Kuroda, Synth. Met. 129, 239 (2002).36. M. Onoda, Y. Manda, S. Morita, and K. Yoshino, Jpn. J. Appl. Phys. 31, 2265 (1992).37. H.B. DeVore, Phys. Rev. 102, 86 (1956).38. H. Ito, K. Hayashi, T. Ozaki, K. Marumoto, and S. Kuroda, unpublished.39. S. Morita, A.A. Zakhidov, T. Kawai, H. Araki, and K. Yoshino, Jpn. J. Appl. Phys. 31, L890 (1992).40. T.-Q. Nguyen, I.B. Martini, J. Liu, and B.J. Schwartz, J. Phys. Chem. B 104, 237 (2000).41. S.S. Pandey, S. Nagamatsu, W. Takashima, and K. Kaneto, Jpn. J. Appl. Phys. 39, 6309 (2000).42. P. Barta, F. Cacialli, R.H. Friend, W.R. Salaneck, M. Zagorska, and A. Pron, Synth. Met. 101, 296

(1999).43. D. Dubois, M.T. Jones, and K.M. Kadish, J. Am. Chem. Soc. 114, 6446 (1992).44. J. Stinchcombe, A. Pénicaud, P. Bhyrappa, P.D.W. Boyd, and C.A. Reed, J. Am. Chem. Soc. 115,

5212 (1993).45. T. G. Castner, Phys. Rev. 115, 1506 (1959).46. K. Yoshino, T. Akashi, K. Yoshimoto, S. Morita, T. Kawai, and A.A. Zakhidov, Jpn. J. Appl. Phys.

34, L127 (1995).

Page 446: Nanotechnology and Nano-Interface Controlled Electronic Devices

Nanotechnology and Nano-Interface Controlled Electronic DevicesEditors: M. Iwamoto, K. Kaneto and S. Mashiko© 2003 Elsevier Science B.V. All rights reserved

CHAPTER 21

Characterization of semiconductor surfaceswith 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 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4291.1. History of AFM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4301.2. Principle of NC-AFM Measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4311.3. Principles of spatial resolutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 433

1.3.1. Expected vertical resolution of NC-AFM . . . . . . . . . . . . . . . . . . . . . . . 4331.3.2. Expected lateral resolution of NC-AFM . . . . . . . . . . . . . . . . . . . . . . . . 433

1.4. Contact point and noncontact region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4342. Spatial resolution of high performance NC-AFM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 436

2.1. Vertical resolution of home-built NC-AFM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4372.2. Lateral resolution of home-built NC-AFM. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4382.3. Atomic displacement around missing-dimer defects in Si(100)2×1 surface . 439

3. Functions of NC-AFM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4423.1. Confirmed functions of high performance NC-AFM . . . . . . . . . . . . . . . . . . . . . . . 4423.2. Three-dimensional mapping of atomic force . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4433.3. Discrimination of atomic force mechanisms and atom species . . . . . . . . . . . . . . 4443.4. Control of atomic force and atom position . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4473.5. Electrostatic force imaging of local charges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4483.6. Atomic imaging of contact potential difference and capacitance . . . . . . . . . . . . 4493.7. Atom manipulation using NC-AFM based on a mechanical method . . . . . . . . 449

3.7.1. Vertical manipulation based on a mechanical method . . . . . . . . . . . . 4503.7.2. Lateral manipulation based on a mechanical method . . . . . . . . . . . . 451

4. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 452References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 452

1. Introduction

The scanning tunneling microscope (STM) is an atomic tool based on an electricmethod that measures the tunneling current between a conductive tip and a conductive

Page 447: Nanotechnology and Nano-Interface Controlled Electronic Devices

430 S. Morita and Y. Sugawara

surface. On the other hand, the atomic force microscope (AFM) is a unique atomic toolbased on a mechanical method that measures the deflection of a cantilever due to theforce between the cantilever tip and a sample surface. Since the invention of AFM in1986 by G. Binnig et al. [1] the AFM has rapidly developed into a powerful surfaceanalysis technique on micro/nanoscales and even on atomic/molecular scales, becausethe AFM has the following characteristics: (1) it has true atomic resolution, (2) it canmeasure atomic force (so-called atomic force spectroscopy), (3) it can observe eveninsulators, and (4) it can measure mechanical responses such as elastic deformation. Thenoncontact 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] succeededto obtain the lattice image of a graphite surface in the contact mode. After that, theAFM was believed to be an atomic resolution microscope which could measure even inambient atmosphere. Nevertheless, Pethica and Oliver [3] pointed out that the contactAFM could not observe atomic defects such as atom vacancy and also that the normalload for usual AFM measurement awfully exceeded the load limit of a single atombecause of the strong repulsive force between a tip apex and a sample surface. Thereforethe contact area for AFM measurement should be larger than a single atom, as shownin Fig. 1a, and so the contact AFM has not true atomic resolution. In connectionwith 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 becauseof the strong repulsive force. (b) Noncontact AFM (NC-AFM) is nondestructive and can observe even anatomic point defect if weak attractive force can be detected as a frequency shift of the mechanical oscillationof the cantilever.

Page 448: Nanotechnology and Nano-Interface Controlled Electronic Devices

Characterization of semiconductor surfaces with noncontact atomic force microscopy 431

destroyed by scanning from the monoatomic step lines at the loading force of onlyapproximately 1 nN, although they succeeded to take a step line image with atomicresolution in UHV at 4 K using the contact AFM. Their best results were obtainedwhen the total force acting on the cantilever was approximately −1 nN (attractive).Furthermore, Ohnesorge and Binnig [6] succeeded to obtain true atomic resolution ofcalcite in an aqueous environment by the AFM through repulsive and attractive forcesand resolved atomic-scale kinks, representing point-like defects, along monoatomic steplines. Atomic resolution that showed STM-like reliability was achieved, particularly inthe attractive-force regime in an aqueous environment.

Nevertheless, to demonstrate true atomic resolution of AFM more clearly, atomicallyresolved imaging of both Si(111)7×7 and atomic defects with STM-like reliabilityhave been claimed. In 1995 Giessibl [7] and Kitamura and Iwatsuki [8] succeededin NC-AFM imaging of Si(111)7×7 surface with atomic resolution in UHV at roomtemperature (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-likereliability, and then Sugawara et al. [11] finally achieved observation of defect motion ofatomic vacancy on an InP(110) cleaved surface. In 1997 Bammerlin et al. [12] reportedhow true atomic resolution on NaCl(001) in UHV can be systematically obtained usingNC-AFM. Thus using the FM detection method, the NC-AFM in the attractive regimein 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 Δν ofmechanical resonant oscillation of a cantilever due to the weak attractive force betweena tip apex atom and a sample surface atom, as shown in Figs. 1b and 2. Under theconditions 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,Δν and A0 are the mechanically free oscillation frequency, the mechanical oscillation frequency of thecantilever under weak attractive force, frequency shift Δν = ν −ν0, and oscillation amplitude, respectively.

Page 449: Nanotechnology and Nano-Interface Controlled Electronic Devices

432 S. Morita and Y. Sugawara

Fig. 3: Schematic diagram of the NC-AFM using the FM detection method. This system has three feedbackloops.

force interaction (|∂ f/∂z| � k), the frequency shift Δν can be approximated as

Δν = (ν0/2k)

(−∂ f

∂z

). (1)

Here k, f and −∂ f/∂z are the spring constant, the tip–sample force and the forcegradient, respectively. However, most of the measurements were executed under thecondition of large oscillation amplitude. Then, Δν is approximated as the geometricmean of the interaction energy V and tip–sample force f = −∂V/∂z, i.e.,

Δν ∝√V f . (2)

The NC-AFM uses three feedback loops as shown in Fig. 3. In contrast, the STMmeasurement under constant tunneling current uses only one feedback loop for STMtopography. The first feedback loop of the NC-AFM is used to oscillate the cantileverat its mechanical resonant oscillation frequency ν under weak attractive force. Thesecond feedback loop is used to maintain oscillation amplitude A0 (switch 1) orthe excitation voltage to oscillate (switch 2) constant. The last feedback loop thatincludes the FM demodulator circuit is used to obtain NC-AFM topography underconstant frequency shift condition (Δν = constant). Thus the electronic circuit for theNC-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 verydifficult task. On the other hand, the STM can easily detect very small tunneling currentwith 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 tipfor the STM. Hence building/floor vibration will easily perturb the tip–sample distanceof the NC-AFM and increase noise in the signal. As a result, achievement of high-S/Nmeasurement in the NC-AFM is very difficult compared with the STM.

Page 450: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 ofthe NC-AFM. Here, we will simply deduce approximated equations of the vertical andlateral resolutions of the NC-AFM.

1.3.1. Expected vertical resolution of NC-AFMHere the value g(z) measured with the NC-AFM was assumed to be proportionalto exp(−z/L), where z and L are the tip–sample distance and the decay lengthof the measured value, respectively. This means that, by increasing the tip–sampledistance 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 thatthe 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 distancecontrollable by the feedback loop is δz, the vertical resolution can be defined asδz. This also means that δg(z) is equivalent to g(z) − g(z + δz). Thus we can obtainthe relation δg(z) = g(z)/h = g(z) − g(z + δz) = g(z)[1 − exp(−δz/L)]. Then from1/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 assumingδz/L � 1, that is, h � 1 Eq. 3 is approximated as

δz ∼= L/h. (4)

Eq. 4 suggests that the scanning probe microscope (SPM) as well as the NC-AFM canobtain better vertical resolution by decreasing the decay length L, i.e., by making thetip–sample distance dependence of the measured value stronger, and by increasing thesignal-to-noise ratio h [13].

1.3.2. Expected lateral resolution of NC-AFMBy assuming that the radius R of tip curvature is much larger than the tip–sampledistance z0, i.e., R � z0, the lateral resolution δx is approximately determined by thevertical resolution δz as shown in Fig. 4a, because the ambiguity of tip–sample distanceis δz. From Fig. 4a, we can obtain the relation δz = R(1− cosθ) and δx = R sinθ . Fromthese equations, we can derive the equation of lateral resolution

δx ∼= [δz(2R − δz)]1/2. (5)

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 oftip curvature, and by improving the vertical resolution δz, i.e., by decreasing the decaylength 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–sampledistance z0, i.e., R � z0, the lateral resolution δx is approximately determined by the

Page 451: Nanotechnology and Nano-Interface Controlled Electronic Devices

434 S. Morita and Y. Sugawara

Fig. 4: (a) Schematic simple one-dimensional model of tip–sample configuration in case of R � z0 [13]. (b)Schematic simple one-dimensional model of tip–sample configuration in case of R � z0 [13].

vertical resolution δz as shown in Fig. 4b, because the ambiguity of tip–sample distanceis δz. From Fig. 4b, we can obtain the relation (z0 + δz)2 = z2

0 + δx2. From this relation,we can derive the equation of lateral resolution

δx ∼= [δz(2z0 + δz)]1/2. (7)

By assuming z0 � δz, we can rewrite Eq. 7 as

δx ∼= [2z0δx]1/2. (8)

Eq. 8 suggests that we can obtain better lateral resolution by decreasing the tip–sampledistance z0, and by improving the vertical resolution δz, i.e., by decreasing the decaylength 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 andnoncontact regions. Fig. 5 shows simultaneously measured approaching curves ofthe frequency shift and the oscillation amplitude in the constant-excitation mode (switch2 in Fig. 3). From the oscillation amplitude curve in Fig. 5, we determined the contactpoint z = 0 nm where the oscillation amplitude begins to decrease due to cycliccontact by decreasing the tip–sample distance [14]. Then we assigned the contact and

Page 452: Nanotechnology and Nano-Interface Controlled Electronic Devices

Characterization of semiconductor surfaces with noncontact atomic force microscopy 435

Fig. 5: Simultaneously measured frequency shift and oscillation amplitude curves. Sample is Zn-dopedp-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 noncontactregion at first increases slowly then quickly by decreasing the tip–sample distance. Thesmallest decay length was 0.16 nm. In general, the decay length takes the smallest valuejust before the contact point as shown in Fig. 5. This phenomenon also suggests that therather strong repulsive force due to cyclic contact works below the contact point andsuppresses the increase of the absolute value of the frequency shift by decreasing thetip–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 threedifferent tip–sample distances: (a) z ∼ 0.4 nm, (b) z ∼ 0.1 nm and (c) z ∼ 0.08 nmbefore the contact point as shown in Fig. 5. Fig. 6a, b and c show measured NC-AFMimages 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. 6ameasured at z ∼ 0.4 nm, we found that the NC-AFM image at the rather far distanceshows only a large-scale vague contrast, perhaps, due to defects, but no atomic-scalecontrast. On the other hand, from Fig. 6b measured at z ∼ 0.1 nm, we found that theNC-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 ∼ 0.08nm, we found that the NC-AFM image just before contact becomes very clear. Thusthe distance dependence of the NC-AFM image is very strong. This result qualitativelyagrees with the expectation predicted by the tip–sample distance dependence of thelateral resolution given by Eq. 8. On the other hand, in the contact region shown inFig. 5, we could not stably obtain the atomic point-defect image because of the strongcontact between the tip and sample surfaces. Therefore, we conjectured that true atomicresolution could be experimentally achievable only just before the contact point.

Page 453: Nanotechnology and Nano-Interface Controlled Electronic Devices

436 S. Morita and Y. Sugawara

Fig. 6: Distance dependence of the frequency shift image and cross-sectional line profile along white linesindicated 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 performanceSTM 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 andlateral 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 High performance NC-AFM

Spatial resolution Applications Spatial resolution Applications(rms) (rms)

Vertical Lateral Vertical Lateral

Commercial 1 pm 10 pm STS/metal atom 10 pm 100 pm semiconductorSPM 0.01 Å 0.1 Å observation 0.1 Å 1 Å atom/ionic atom

Home-built 100 fm 1 pm atom 1 pm 10 pm observation of metalSPM 0.1 pm 0.01 Å manipulation/ 0.01 Å 0.1 Å atom/3D-measurement

0.001 Å electron waves and control of atomicon metal surface force, etc./inelastic STM

Developing 10 fm 100 fm true observation 100 fm 1 pm atom and moleculehome-built 0.01 pm 0.1 pm of metal atom? 0.1 pm 0.01 Å manipulation?SPM 0.001 Å 0.001 Å

Page 454: Nanotechnology and Nano-Interface Controlled Electronic Devices

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) filmevaporated on Si(111)7×7 surface [16]. Scan area is 80 nm × 80 nm. Cross-sectional line profile obtainedalong 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 × 2nm. (d) NC-AFM image of the atomically flat surface of Ag(111) [16]. Cross-sectional line profile obtainedalong 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 anda 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 ofan atomic layer of Ag(111) surface. Fig. 7c shows a high-resolution NC-AFM imagewith two Ag(111) terraces separated by an atomic step [16]. The trigonal patternthat 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

Page 455: Nanotechnology and Nano-Interface Controlled Electronic Devices

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) hydrogenpassivated Si(100)2×1:H monohydride surface [17]. Cross-sectional line profiles are obtained along thewhite 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]. Thecross-sectional line profile of Fig. 7d obtained along the white line clearly shows latticespacing of 0.28 ± 0.01 nm of Ag(111) surface. Thus we can observe even metal atomsusing atomic force, in spite of the nearly homogeneous and hence site-independentsurface on atomic force due to free-electron screening. It should be noted that thecross-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 aSi(100)2×1:H monohydride surface. Si(100)2×1 surface has tilted dangling bondsand hence is reactive, while Si(100)2×1:H surface is terminated by hydrogen atoms andhence 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 thehydrogen termination effect on the NC-AFM imaging of Si(100)2×1:H surface can beclarified 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〈Δν〉 = −30 Hz [17]. Paired bright spots (imaged dimer) constituting rows with a 2×1symmetry were clearly observed. Further, the distance between paired bright spots is0.32±0.01 nm. As shown in the cross-sectional line profile of Fig. 8a, the change of thefrequency shift along a white dotted line in the NC-AFM image of Fig. 8a was estimated

Page 456: Nanotechnology and Nano-Interface Controlled Electronic Devices

Characterization of semiconductor surfaces with noncontact atomic force microscopy 439

to be about 9 Hz. Fig. 8b shows the atomic resolution image of hydrogen terminatedsilicon [Si(100)2×1:H monohydride] surface measured at 〈Δν〉 = −11 Hz. Paired brightspots (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 agreeswith the distance between hydrogen atoms on monohydride surface, i.e., 0.352 nm. Asshown in the cross-sectional line profile of Fig. 8b, the averaged change of the frequencyshift 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 brightspots with an accuracy of ±0.01 nm (= 10 pm). This result roughly shows that ourhome-built NC-AFM has nearly 10 pm lateral resolution.

From such precise measurements of the distances between paired bright spots, wecan 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 surfacestructure models. In the case of Si(100)2×1 surface, we found that the bright spotsof Fig. 8a do not image the silicon atom site, because the distance 0.32 ± 0.01 nmbetween bright spots forming the dimer structure is larger than the distance betweenpaired silicon atoms of a dimer structure model (maximum distance between alternatingupper Si atoms on an asymmetric dimer structure is 0.292 nm). This result suggeststhat chemical bonding interaction strongly works between the tilted dangling bond outof the silicon dimer and the dangling bond out of silicon tip apex. As a result, a dimerstructure with a larger distance than that between paired silicon atoms of a dimer isobtained. Besides, as shown in the cross-sectional line profile of Fig. 8a, the frequencyshift rapidly increases around imaged dimers. This also suggests that a strong chemicalinteraction occurs between dangling bonds.

On the other hand, the bright spots of Fig. 8b seem to be located at hydrogen atomsites on Si(100)2×1:H surface, because the distance between bright spots 0.35 ± 0.01nm forming a dimer structure approximately agrees with the distance between the pairedhydrogen atoms, i.e., 0.352 nm. As a result, each bright spot of the NC-AFM imagecorresponds 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 bya hydrogen atom, and the hydrogen atoms on the top most layer do not have a strongchemical reactivity as a silicon atom on Si(100)−2×1 surface. Therefore, the interactionbetween the hydrogen atom on the top most layer and the silicon tip apex may be weakchemical or physical interaction without strong directional dependence, in contrast tothe 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 dimersaround a missing-dimer defect in a Si(100)2×1 surface. As shown in Fig. 9 indicatedby the white arrow, the NC-AFM clearly images the missing-dimer defect. The thinbroken lines indicate the centerlines of the dimers. By careful investigation, we foundthat the lateral position of the dimer is displaced toward the missing-dimer defect alongthe dimer row. In general, the dimers adjacent to missing-dimer defects exhibit subtlevariations in height, length and lateral position depending on the size and structure

Page 457: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 thedimer 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:Hsurfaces as a function of distance from a defect. Noise level is ±0.01 nm. The regular distance betweenadjacent dimers (lattice constant) is 0.384 nm.

of the defect. Some of these varying properties, such as the height, are related to thescanning direction of the NC-AFM, and the variation can be explained as overshootdue to feedback error. However, most of the dimers adjacent to missing-dimer defectsexhibited nearly identical lateral displacement along the dimer row, irrespective ofthe scanning direction. As shown in Fig. 10, the first and second dimers adjacent tothe missing-dimer defects in Si(100)2×1 surface are displaced. The figure shows theaverage lateral displacement of dimers with respect to the regular distance betweenadjacent dimers of 0.384 nm, which corresponds to the lattice constant along the dimer

Page 458: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 cleansurface 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, andthe second neighbouring dimer is displaced by ca. 10%. The third and fourth dimersare displaced only within the experimental error (±0.01 nm). In contrast, as shown inFig. 10, the dimers adjacent to missing-dimer defects in the Si(100)2×1:H monohydridesurface are not displaced.

The corresponding models of the plan structures of the Si(100)2×1 and Si(100):Hsurfaces 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, cutting4 covalent bonds between 2 Si atoms to form one Si dimer in the first layer and leaving4 Si atoms in the second layer. These 4 dangling bonds then form 2 Si dimers in thesecond layer to reduce the number of dangling bonds, yielding the rebonded structurefor 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 theattractive force between paired Si atoms in the lower dimers pulls the upper dimerstoward the missing-dimer defect. If the length of the lower dimer is the same as that ofthe upper dimers (asymmetric, 0.230 nm; symmetric, 0.234 nm), the first neighbouringupper 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 consistentwith 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-dimerdefect agrees well with the previous results obtained by STM at low negative samplebias and by local density approximation (LDA) computer modeling [18]. Although theLDA model and STM image calculated for high negative sample bias clearly predictedthe lateral displacement of the first neighbouring Si dimer toward the missing-dimerdefect in agreement with the present NC-AFM results, the practical spatial resolutionof STM was insufficient to verify the predictions [18]. Further LDA simulation takingthe lateral displacement of the second neighbouring dimer into consideration will berequired in order to explain the lateral displacement of the second neighbouring dimer

Page 459: Nanotechnology and Nano-Interface Controlled Electronic Devices

442 S. Morita and Y. Sugawara

((4) in Fig. 3(a)) in terms of atomic stress and atomic relaxation around the dimer. In thecase of the Si(100)2×1:H monohydride surface, 4 dangling bonds in the second Si layerleft by the missing-dimer defect are each terminated by hydrogen atoms (Fig. 11b). Thistermination appears to eliminate the stress that would otherwise result in displacementof the Si dimers around the missing dimer.

This is the first NC-AFM measurement that has provided a direct visualization of theatomic displacement around missing-dimer defects. Our home-built NC-AFM has beenshown 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 straininduced by the missing-dimer defects extends up to two dimers from the defect, withdirect observations of the displacement of the first and second dimers adjacent to amissing-dimer defect. NC-AFM is therefore a highly useful method for making directmeasurements of atomic displacements and strain with subatomic resolution, allowingthe 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. Asshown 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 samplesurface atom B, the NC-AFM can measure atomic force between the tip apex atom Aand the sample surface atom B three-dimensionally. From three-dimensional mapping ofatomic force, we can deduce the strength, distance dependence and angular dependenceof atomic force between A and B. Then we can investigate the mechanism of the atomicforce. If there are two kinds of atoms B and C on the sample surface, as shown inFig. 12b, we can discriminate atom species B and C from three-dimensional mappingof atomic force, which clarifies the difference of the strength, distance dependence andangular dependence of atomic forces between A and B, and A and C. Besides, by placinga suitable atom D at the tip apex instead of the atom A, we can control atomic forcebetween the tip apex atom and the surface atom B. Furthermore, by moving the tip apextoward the sample surface, we can increase the strength of the attractive force and we canpull up the surface atom. In principle, using such a strong attractive force or moderatelyweak repulsive force, we can mechanically manipulate individual atoms/molecules andcan 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 ourgroup. Existence of measured samples means confirmation of corresponding functions.Table 2 shows that our home-built NC-AFM can observe compound semiconductorsurfaces such as InP(110) [10,11] and GaAs(110) [15] cleaved surfaces, silicon surfacessuch as Si(111)7×7 [19] and Si(100)2×1 [17] clean surfaces, metal surfaces such asAg(111) [16] and Cu(111) [20], metal oxide surfaces such as TiO2 [21] with atomicresolution. Our home-built NC-AFM can observe molecules such as adenine [22] and

Page 460: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 tipapex atom A and a sample surface atom B. (b) Discriminations of force mechanisms between A and B, andA 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 thetip apex instead of atom A.

thymine [23], self-assembled monolayers such as alkanethiol [23], and DNA with highresolution [24].

3.2. Three-dimensional mapping of atomic force

As shown in Table 2, our NC-AFM can map atomic force three-dimensionally. Thereare 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 suchas A, B and C as shown in Fig. 13a. This kind of three-dimensional (3D) mappingof 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 site-dependence of frequency shift curves such as A, B and C as shown in Fig. 13b. Thiskind of three-dimensional mapping of atomic force on an atomic scale was achieved onSi(111)7×7 surface [26].

Fig. 14 shows one example of three-dimensional mapping of atomic force obtainedby measuring the tip–sample distance dependence of the NC-AFM images on oxygenadsorbed Si(111)7×7 surface. At relatively far distance, only several bright spotsindicated by white open circles can be seen clearly, as shown in Fig. 14a. It shouldbe noted that at this distance the 7×7 structure cannot be observed clearly. On the

Page 461: Nanotechnology and Nano-Interface Controlled Electronic Devices

444 S. Morita and Y. Sugawara

Table 2

Functions of high performance NC-AFM

Measured samples by NC-AFM (Osaka University)

Observation InP(110) Si(111)7×7 Ag(111) TiO2* Adenine* Alkanethiol*(atom/molecule) GaAs(110) Si(100)2×1 Cu(111)* Si(100)2×1:H Thymine* DNA*

3D-Measurement Si(111) Si(111) oxygen(atomic force) 7×7

√3×√

3–Ag adsorbedSi(111)7×7

Discrimination Si(111) Si(111) Si(111) oxygen(atomic force) 7×7

√3×√

3–Ag 5√

3×5√

3–Sb adsorbedSi(111)7×7

Discrimination Si(111) oxygen Si(111) Si(100)(atom) 5

√3×5

√3–Sb adsorbed 7×7–Ge 2×1–Ge

Si(111)7×7

Control oxidation Ag adsorbed Sb adsorbed Si(100) Si(100)(atomic force) of Si tip Si tip Si tip 1×1:2H 1×1:2D

Control Si(100) Si(100)(atom position) 1×1:2H 1×1:2D

Manipulation Si(111) Si(111) Si(111)(atom) 7×7

√3×√

3–Ag 5√

3×5√

3–Sb

Assembly(atom by atom)

Point charge n+-GaAsimaging (110)

CPD and dC/dz Si(111) Si(111)image 7×7 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 atomicforce shows clear difference in distance dependence of atomic forces exerted fromoxygen 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 canconclude that two force mechanisms work between the Si tip and oxygen adsorbedSi(111)7×7 surface. One is a kind of long-range force, which works well even atrelatively far distance. The other is a kind of short-range force, which works stronglyonly at near distance. The short-range force contributes to NC-AFM imaging of the7×7 structure, that is, Si adatoms, as shown in Fig. 14b. Hence the short-range force isthe covalent bonding force between dangling bonds of Si adatoms and the Si tip apexatom, and has a critical distance dc, below which the covalent bonding force worksstrongly, as shown in Fig. 15a and b. On the other hand, at relatively far distance,

Page 462: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 forceusing 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 andC.

different atoms from Si adatoms seem to be imaged by the long-range force. Oxygenadsorbed Si(111)7×7 surface has only two kinds of atoms, that is, Si atoms and oxygenatoms. Therefore, we conjectured that, at relatively far distance, oxygen atoms wereimaged 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 beelectrostatic force between O− ion and Si tip. Thus using three-dimensional mapping ofatomic force, we can discriminate mechanisms of atomic forces between the tip and thesample surface, and also we can discriminate atom species on the sample surface.

We will show another example of discrimination of atom species. We fabricatedSi(111)7×7–Ge mixed surface as shown in Fig. 16. On this surface, Si adatoms andGe adatoms coexist. As a result, using the clean Si tip, covalent bonding force worksbetween the Si tip apex atom and adatoms of Si and Ge. We found two kinds of brightspots forming a 7×7 structure, as shown in Fig. 16. One kind of bright spots is brighterthan the other kind of bright spots. These two kinds of bright spots seem to correspond

Page 463: Nanotechnology and Nano-Interface Controlled Electronic Devices

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) neardistance. 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 MLand 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 toGe adatoms. Covalent bond energies for Si–Si bond and Si–Ge bond are 2.32 eV and2.12 eV, respectively. Therefore, the covalent bonding force for Si adatoms will be largerthan that of Ge adatoms. This speculation qualitatively agrees with the experimentalresults. Thus, our home-built NC-AFM can clearly discriminate Si adatoms and Ge

Page 464: Nanotechnology and Nano-Interface Controlled Electronic Devices

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–Gebond is only ca. 10%.

We also succeeded to discriminate the difference of strength of the covalent bondingforce for Si adatoms with a dangling bond and for Sb adatoms with a lone pairon 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 andenables 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 forcemechanism between the tip apex atom and the surface atom. This kind of atomic forcecontrol was demonstrated on Si(111)7×7 surface by using an oxidized Si tip instead ofthe clean Si tip [19,26], on Si(111)

√3 ×√

3–Ag surface by picking up an Ag atom onthe 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 anotherkind of atomic force control, which also enables us to control atom position. Thiskind 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 [ordeuterium] atoms due to the proximity effect and originates a 1×1 canted structurewhihc increases the distance between adjacent hydrogen atoms, i.e., reduces therepulsive force strength. Fig. 17a, b and c show NC-AFM images of Si(100)1×1:2Dsurface 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

Page 465: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 cleanSi 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 thenear distance of Fig. 17b, the NC-AFM showed a 2×1 pattern. Moreover, at the closedistance of Fig. 17c, the NC-AFM again showed a 1×1 pattern. Thus, the NC-AFMon Si(100)1×1:2D surface showed a pattern change from 1×1, through 2×1, to 1×1similar 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 consideredthe effect of attractive force between the tip apex atom and the nearest deuterium atomon the surface structure. At the relatively far distance, the magnitude of the attractiveforce between the tip apex atom and the nearest deuterium atom is smaller than thatof the repulsive force between adjacent deuterium atoms. In this case, the original1×1 canted structure is stable even under NC-AFM measurement. Then NC-AFM willobserve 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 nearlyequivalent to that of the repulsive force. As a result, the attractive force will pull upthe nearest deuterium atom, and will play the role of trigger to induce self-organized2×1 novel structure. Thus NC-AFM will observe the tip-induced self-organized 2×1novel structure as shown in Fig. 17b. Further, at the close distance, the magnitude of theattractive force becomes larger than that of the repulsive force. As a result, the attractiveforce will pull up the nearest deuterium atoms one by one. This vertical motion of eachdeuterium atom due to pulling up seems large enough compared with the corrugationof the self-organized 2×1 novel structure, and hence tip-induced 1×1 novel structurewill be observed by the NC-AFM. Thus NC-AFM can control atomic force and atomposition 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 theelectrostatic force using the noncontact AFM, at first we eliminated the contributiondue to the electrostatic force by applying a sample voltage which compensates the

Page 466: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 bethe dominant force mechanism, because the contribution due to the electrostatic forcewill be eliminated. Then, to image the noncontact AFM topography due to the vander Waals force, we closed the feedback loop to maintain a constant frequency shiftunder this sample voltage and imaged the feedback voltage by scanning with the tip. Onthe other hand, by applying a large positive sample voltage or a large negative samplevoltage, we enhanced the electrostatic force between tip and sample surface. For theelectrostatic force imaging, we opened the feedback loop under a large positive samplevoltage or a large negative sample voltage, and imaged the change of the frequency shiftbetween two alternate sample voltages by scanning with the tip. This method enables usto simultaneously image the atomic corrugation and the charge of atomic point defectson n+-GaAs(110) cleaved surface [29]. It made clear that there are atomic point defectswith 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 willchange the NC-AFM topography, and hence change the magnitude of the electrostaticforce. Therefore, we should compensate the contact potential difference with atomicresolution. This means that we should combine Kelvin probe force microscopy (KFM)with NC-AFM (KFM/NC-AFM) and that we should simultaneously observe the NC-AFM topography and the CPD image of the point charge.

We succeeded in simultaneous imaging of the NC-AFM topography, the CPD imageand 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 topographyclearly showed a Si(111)7×7 pattern. Thus charge distribution and capacitance changeare important even on an atomic scale. KFM measurement compensates contact potentialdifference and hence electrostatic force on an atomic scale, while it originates excessterms of (dC/dz)V 2

AC/4 in the DC term and a 2ω component in the frequency shift. Here,VAC is the modulation amplitude of the alternating bias potential VAC sinωt between thetip and sample. Therefore, it should be noted that the NC-AFM topography using thismethod still includes an excess term of (dC/dz)V 2

AC/4 which should be subtracted.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 CPDof Si adatoms and Sb adatoms on an atomic scale, and enabled us to discriminate Siadatoms 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 micro-scope based on the electric method that measures the tunneling current between the tip

Page 467: Nanotechnology and Nano-Interface Controlled Electronic Devices

450 S. Morita and Y. Sugawara

apex atom and surface atom. Therefore the STM has inherent limitations such as STMcannot observe an insulator surface, cannot manipulate atoms and molecules on insulatorsurfaces, and cannot directly measure atomic force or force related information.

On the other hand, AFM is the microscope based on the mechanical method thatmeasures the atomic force or force related information between the tip apex atomand the surface atom, and has potentials for insulator surface observation with atomicresolution, manipulation of atoms and molecules on insulator surface, and measurementof atomic force or force related information on an atomic scale.

3.7.1. Vertical manipulation based on a mechanical methodNC-AFM succeeded in a kind of atom manipulation at room temperature such asextraction of a single Ag atom from Si(111)

√3 × √

3–Ag sample surface [27], andextraction 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 (LT-NC-AFM), as shown in Fig. 18. After NC-AFM imaging, such as in Fig. 18a, wemechanically moved the Si(111)7×7 sample surface toward the tip apex Si atom upto 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 signsof atom extraction, we again moved the sample surface toward the tip apex Si atomfurther. By repeating such procedure, we succeeded to extract a single Si adatom on theSi(111)7×7 surface mechanically. Fig. 18a and b are NC-AFM images before extractionand after extraction, respectively. White open circles show a missing Si adatom, whichis a marker of the site. By comparing Fig. 18b with Fig. 18a, it is clear that a corner Siadatom 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 samplesurface. 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 adatomextraction.

Page 468: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 firstan attractive force works between Si tip apex atom and Si adatom, (c) then a repulsive force works and cutsthree 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 adatomextraction shown in Fig. 19, the tip apex Si atom with a dangling bond will cut the threeSi–Si covalent bonds between the Si adatom and the lower three Si atoms. However, theextraction process such as in Fig. 19 will increase the number of dangling bonds. Thusthe Si extraction process also seems to be complicated. One possible explanation is thatcutting of three Si–Si covalent bonds happened under the repulsive force as shown inFig. 19c. This means that the Si adatom was not extracted due to the pull-up processindicated in Fig. 19b, but due to the push-out process as indicated in Fig. 19c and d. Inthis process, mechanical energy due to atomic indentation under weak repulsive forcewill increase the atom potential, and may finally cut the three Si–Si covalent bonds andpush out a single Si adatom. Then, the pushed out Si atom may adsorb at the top of thetip apex, the side of the tip or on the sample surface.

3.7.2. Lateral manipulation based on a mechanical methodUsing NC-AFM based on the mechanical method, we sequentially extracted Si adatomson a Si(111)7×7 sample surface up to 5 Si adatoms. During this procedure, weaccidentally succeeded to laterally manipulate Si adatoms. Fig. 20a and b are NC-AFMimages before lateral manipulation and after lateral manipulation, respectively. Whiteopen circles indicate a manipulated Si adatom. By comparing Fig. 20b with a, it isclear that a Si adatom, indicated by white open circles, was laterally manipulated by themechanical method. This lateral manipulation may occur if two covalent bonds betweenthe manipulated Si adatom and lower two Si atoms were cut, and then this Si adatomformed two covalent bonds with two different lower Si atoms with dangling bonds.

Page 469: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 lateralmanipulation of a Si adatom and (b) after lateral manipulation of a Si adatom. Scan areas are 9.2 nm × 9.2nm.

4. Summary

Our home-built NC-AFM roughly has a vertical resolution of 1 pm and a lateralresolution of 10 pm. Using this NC-AFM, we can observe metal atoms and can measureatomic forces three-dimensionally on an atomic scale. Besides, we can discriminatemechanisms of atomic force and species of atoms. Further, we can control atomic forceand atom position. Moreover, recently, we succeeded to manipulate a single atom bymechanical contact. In this process, mechanical energy due to atomic indentation underweak repulsive force will increase the atom potential and may finally push out a singleatom. We should explore near contact and close contact phenomena in more detail. Thenthis will enable us to assembly atom by atom, which may lead to construction of novelnanomaterials and nanodevices.

References

1. G. Binnig, C.F. Quate, and Ch. Gerber, Phys. Rev. Lett. 56, 930 (1986).2. G. Binnig, Ch. Gerber, E. Stoll, T.R. Albrecht, and C.F.Quate, Europhys. Lett. 3, 1281 (1987).3. J.B. Pethica and W.C. Oliver, Phys. Scri. T 19, 61 (1987).4. G. Binnig, Ultramicroscopy 42–44, 7 (1992).5. F.J. Giessibl and G. Binnig, Ultramicroscopy 42–44, 281 (1992).6. F. Ohnesorge and G. Binnig, Science 260, 1451 (1993).7. F.J. Giessibl, Science 267, 68 (1995).8. S. Kitamura and M. Iwatsuki, Jpn. J. Appl. Phys. 34, L145 (1995).

Page 470: Nanotechnology and Nano-Interface Controlled Electronic Devices

Characterization of semiconductor surfaces with noncontact atomic force microscopy 453

9. T.R. Albrecht, P. Grütter, D. Horne, and D.Rugar, J. Appl. Phys. 69, 668 (1991).10. H. Ueyama, M. Ohta, Y. Sugawara, and S. Morita, Jpn. J. Appl. Phys. 34, L1086 (1995).11. Y. Sugawara, M. Ohta, H. Ueyama, and S. Morita, Science 270, 1646 (1995).12. 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).13. S. Morita and Y. Sugawara, Appl. Surf. Sci. 140, 406 (1999).14. H. Ueyama, Y. Sugawara, and S. Morita, Appl. Phys. A 66, S295 (1998).15. S. Morita, M. Abe, K. Yokoyama, and Y. Sugawara, J. Cryst. Growth 210, 408 (2000).16. S. Orisaka, T. Minobe, T. Uchihashi, Y. Sugawara, and S.Morita, Appl. Surf. Sci. 140, 243 (1999).17. K. Yokoyama, T. Ochi, A. Yoshimoto, Y. Sugawara, and S. Morita, Jpn. J. Appl. Phys. 39, L113

(2000).18. J.H.G. Owen, D.R. Bowler, C.M. Goringe, K. Miki, and G.A.D. Briggs, Surf. Sci. 341, L1042 (1995).19. T. Uchihashi, Y. Sugawara, T. Tsukamoto, M. Ohta, and S. Morita, Phys. Rev. B 56, 9834 (1997).20. K. Yokoyama, T. Ochi, T. Uchihashi, M. Ashino, Y. Sugawara, N. Suehira, and S. Morita, Rev. Sci.

Instrum. 71, 128 (2000).21. M. Ashino, T. Uchihashi, K. Yokoyama, Y. Sugawara, S. Morita, and M.Ishikawa, Appl. Surf. Sci.

157, 212 (2000).22. T. Uchihashi, T. Okada, Y. Sugawara, K. Yokoyama, and S. Morita, Phys. Rev. B 60, 8309 (1999).23. 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).24. T. Uchihashi, M. Tanigawa, M. Ashino, Y. Sugawara, K. Yokoyama, S. Morita, and M. Ishikawa,

Langmuir 16, 1349 (2000).25. T. Minobe, T. Uchihashi, T. Tsukamoto, S. Orisaka, Y. Sugawara, and S. Morita, Appl. Surf. Sci. 140,

298 (1999).26. S. Morita, Y. Sugawara, K. Yokoyama, and T. Uchihashi, Nanotechnology 11, 120 (2000).27. K. Yokoyama, T. Ochi, Y. Sugawara, and S. Morita, Phys. Rev. Lett. 83, 5023 (1999).28. S. Araragi, A. Yoshimoto, N. Nakata, Y. Sugawara, and S. Morita, Appl. Surf. Sci. 188, 272 (2002).29. Y. Sugawara, T. Uchihashi, M. Abe, and S. Morita, Appl. Surf. Sci. 140, 371 (1999).30. K. Okamoto, Y. Sugawara, and S. Morita, Appl. Surf. Sci. 188, 381 (2002).

Page 471: Nanotechnology and Nano-Interface Controlled Electronic Devices

This Page Intentionally Left Blank

Page 472: Nanotechnology and Nano-Interface Controlled Electronic Devices

Nanotechnology and Nano-Interface Controlled Electronic DevicesEditors: M. Iwamoto, K. Kaneto and S. Mashiko© 2003 Published by Elsevier Science B.V.

CHAPTER 22

Transport and photocarrier generation inpoly(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. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4552. Sample preparation and experimental setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4583. Optical absorption in HT- and NR-PATn . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4604. J –V Characteristics of Al/PATn/Au diodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 461

4.1. Rectification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4614.2. Depletion layer in the Schottky junction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4624.3. New estimation method of the thickness of the depletion layer . . . . . . . . . . . . . 4634.4. Potential profile of diodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 464

5. Carrier mobilities in PATn . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4665.1. TOF mobilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4665.2. Negative field dependence of photocarrier mobility due to non-uniform fields 4685.3. FE mobilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 471

6. Photovoltaic effects and mechanism of carrier generation . . . . . . . . . . . . . . . . . . . . . . . . . . 4747. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 476Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 477References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 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 beenturned into active elements from passive materials in electronic devices. The conductingpolymers were novel materials until they became the Nobel polymers in 2000. Theydemonstrate a variety of functions such as semiconductors, metals and even switchingproperty between semiconductor and metal upon oxidation and reduction [6]. Theunique features, which originate from the one dimensional π-electron systems along thepolymer chain, are the high anisotropy in carrier transport and the excellent photore-

Page 473: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 oxidizedstate can be utilized as a lightweight electrical conductor, antielectrostatic shield andcapacitor. The semiconductive property due to the localized π-electrons in the neutralstate is of noting from the view point of high potential for use in diodes with efficientlight emission [8], in solar cells [9] and also thin film transistors. Other interestingfeatures of organic and polymer devices are flexibility of the laminate structure and lowcost of the fabrication, which cannot be achieved by conventional semiconductors.

The varieties of conducting polymers as shown in Fig. 1 are categorized for their

Page 474: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 thatoxidize or decompose easily at ambient conditions. The former and its derivatives arepoly(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 conductingpolymers are insoluble in common organic solvents, therefore it is very difficultto 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 inorganic solvents such as chloroform [10], dichloromethane, etc.

Polythiophene and its derivatives are studied intensively for active layers of photo-and electronic devices [7,10–12], since they are stable against oxidation in ambient con-ditions. 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 andtail to tail. When the coupling position is precisely controlled at the HT position, theregioregularity is enhanced as shown in Fig. 2. It has been shown that the HT-PATsexhibit larger carrier mobility compared to those of NR-PATns by one or two orders ofmagnitude [11,12].

The photovoltaic effect, which is the reversed phenomenon to light emitting, isanother important function in optical and electrical conversion devices. The photo-voltaic effect results from photocarrier generation and migration toward the respectiveelectrode. In inorganic semiconductors, the mechanism of photocarrier generation iswell 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 workfunctions between the contacting materials. On the other hand, in organic materials

Page 475: Nanotechnology and Nano-Interface Controlled Electronic Devices

458 K. Kaneto et al.

the band theory is not fully applicable. This is due to the fact that organic materialsare discontinuously assembled with inert and isolated molecules and condensed withthe weak Van der Waals force. Namely, the carrier transport in organic materials isgoverned by intermolecular hopping or conduction in a narrow band. Furthermore,photocarrier generation also is not favorable in organic materials. However, it is in-teresting to note that quite high intrinsic conversion efficiencies as high as 25% inorganic solar cells are demonstrated. For instance, the dye sensitized solar cell orGrätzel type cell shows an overall efficiency of close to 10% and the solid state ofa poly(p-phenylene vinylenes)/fullerene system is exceeding 3.2%. Taking the purityand crystallinity of organic materials in comparison with that of inorganic materialsinto consideration, the intrinsic efficiency of organic photocells should be quite large.The potential of organic materials for electronic devices is high and, therefore, it isworthwhile to study the mechanisms of photocarrier generation and transport in thenear-future devices.

In this chapter, studies of electrical transport and photovoltaic properties of diodeswith the structure Al/PAT films/Au are presented. These characteristics are examinedas a function of alkyl chain length and regioregularity. The results show that the carriertransport and photocarrier generation are strongly influenced by the regioregularity aswell as alkyl chain length, and are discussed with regard to the role of the depletionlayer 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 ofChen and Rieke et al. [13,14], as reported in our previous papers [10–12,15]. Also werethey obtained commercially and purified further as described below. The ratios of HTcoupling contents are more than 98% as determined by 13C 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 tetraaceticacid to remove chloride and ferric ions followed by Soxhlet extraction of low-weightimpurities and oligomers with methanol and hexane [18]. Thus obtained HT- andNR-PATn were dissolved in chloroform at concentrations of 0.7 ∼ 4.0 wt% dependingon the film thickness. The solutions were used for drop cast or spin-coating on glasssubstrates.

For the measurement of optical absorption spectra, the films on glass substrateswere prepared by spin coating. The chloroform solution was dropped onto the glasssubstrate, which was spun at 1000 rpm for 30 s and then switched to 4000 rpm for10 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-570spectrophotometer. Diodes [10,11,15] with a structure of indium tin oxide (ITO) orAu/PATn/Al, as shown in Fig. 3, were fabricated. A 18 mm2 ITO glass was washedwith acetone and ethanol for 10 min with sonication and etched to leave electrodes witha width of 1 mm using a solution of hydrochloric acid and Zn powder for 1 min followedby rinsing with toluene and ethanol. The ITO on the substrate was polished with 0.05

Page 476: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 Clean56 (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 invacuum for 24 h, the ITO substrates were subjected to spin-coating with PATn. Al andAu 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 byannealing in a similar manner to the sample prepared for the absorption measurement.The typical thickness (�) of PATn films prepared by spin coating was approximately500 nm, except those for the measurement of thickness dependence. The top Au orAl electrodes were also deposited on PATn film with a thickness of 30 ∼ 40 nm invacuum. The transparency of Al electrode with a thickness of 40 nm was approximately25% [15]. Current–voltage (I–V ) characteristics of the cells were measured using aKeithley 6517 electrometer in air atmosphere or in vacuum. Temperature of the diodewas 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 of0.6 ns and energy of 70 μJ/pulse under the external bias (Va) across the electrodes ofsandwich cells. The transit times (Ttr) of photocarriers during the migration in PATn filmwere 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/VaTtr. (1)

The FE mobility was obtained by fabricating an FE transistor [12]. An n+-Si waferwith a conductivity of approximately 100 S/cm and (100) surface was used as thesubstrate and gate electrode. The substrate was oxidized to form SiO2 gate insulatorwith a thickness of 400 nm. The capacitance (Ci) of the gate insulator was approximately8.4 nF/cm2. Comb shaped Au–Cr electrodes on the SiO2 surface served as the sourceand 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 activechannel 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 parameteranalyzer at ambient atmosphere. The gate (VG) and drain (VD) voltages were measured

Page 477: Nanotechnology and Nano-Interface Controlled Electronic Devices

460 K. Kaneto et al.

against the source potential at the earth potential of 0 V. The general formula of draincurrent, ID,

ID = μCiW

L

[(VG − VT)VD − V 2

D

2

], (2)

was employed [18] to analyze the data, where μ and VT are FET mobility and thethreshold voltages. To evaluate FET mobilities,

ΔID

ΔVG= μCiW

LVD, (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 us-

ing a lock-in amplifier by the integrated constant photon source system. Monochromaticlight 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 resistanceof 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 peaksand shoulders except for HT-PAT18, which shows blue shift in the peak and no structurein comparison with others. This fact indicates that the electronic structures of PAT4∼12films 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.

Page 478: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 notso strong as to shift the electronic energy level by the optical detection. On the otherhand, 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 morerandomness with longer side chains. It is noted that the absorption coefficients decreasewith increasing alkyl chain length [12], indicating that the absorption coefficients aredetermined with the density of π-electrons. Namely, the side chain simply separatesmain chains and makes the film bulky. These results are important to account for thealkyl 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/HT-PATn/Al, n = 4, 6, 10, 12 and 18 films are shown in Fig. 5. The positive bias to theITO side against Al is the forward bias. Excellent rectification with a rectification ratioof more than 103 is obtained. These J–V characteristics of the diode is given by Eq. 4based 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 idealityfactor 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 insandwich cells of ITO/HT-PATn/Al.

Page 479: Nanotechnology and Nano-Interface Controlled Electronic Devices

462 K. Kaneto et al.

increase of current at forward bias. The best is unity. In our diodes n was 1.7 ∼ 3.2depending on the preparation procedure. The result in Fig. 5 clearly indicates thatperformance of the diodes is better for shorter alkyl chain length. The smallest oneof n = 1.7 was obtained when Al was first deposited on glass substrate, immediatelyfollowed by spin coating with HT-PAT4 and then Au deposition. The detailed survey offabrication 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 isdetermined by the bulk conductivity (mobility and concentration of carriers) of the PATfilms and also the contact resistance between Au and PATn. The shoulder observed atlow forward bias is conjectured to result from defects of the Schottky junction such aspin holes and impurities.

4.2. Depletion layer in the Schottky junction

The depletion layer in Schottky diodes plays the most important role in photocarriergeneration as well as rectification. The thickness and the potential profile of the depletionlayer are subjects to investigate. Taking the fact that an Au (or ITO)/PAT contact showsohmic behavior, the rectification results from the Schottky junction between the PATfilm and Al electrode [7,16,17]. Therefore, PAT films are conjectured to be a p-typesemiconductor due to oxidation with a trace of acceptors or oxygen in air. Accordingto the semiconductor theory, the depletion layer is formed by electron transfer fromAl (n-type metal with low work function) to acceptors in PATn film with a large workfunction. This results in negative space charge and a strong electric field at the depletionlayer. It should be also noted that the resistance of the depletion layer increases to thatof an insulator. The number of electrons transfered from Al to PATn film is equilibratedby the difference of the work functions (Vw or the diffusion potential). The thickness ofdepletion layer (d) depends [27] on the reciprocal density of acceptors (Na) and is givenby Eq. 5,

d =[

2εrε0

eNa(Vw − Va)

]1/2

, (5)

where a homogeneous distribution of acceptors is assumed. Eq. 5 indicates that thedepletion layer vanishes when Va reaches Vw, followed by the onset of forwardcurrent. 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 ratioof 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 rectificationcharacteristics in J–V curves. However, the performance of NR-PATn diodes wasinferior 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 = εrε0/dper unit area appears and can be expressed by

1

C2= 2

VW − Va

εrε0eNA[cm4/F2]. (6)

Page 480: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 configurationsAu/HT-PAT6/Al and Al/HT-PAT6/Au.

Although the model is derived from a conventional semiconductor, the experimentaldata 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 biasaccording 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/Aland Al/HT-PAT6/Au. The difference of the samples is that the bottom electrode, whichis represented by the left hand side, was first deposited, then followed by spin coatingand deposition of the top electrode. Namely, in the latter the Al electrode might beoxidized during the spin coating of PAT and covered with insulating Al2O3. Using therelative dielectric constant [20] of εr = 2.8 estimated from the capacitance measurementin 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 wasobtained 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/Aland (b) Al/HT-PHT/Au, respectively, from the gradients of the curves using Eq. 6. Thedensity of acceptors in HT-PHT is in the order of 1016 cm−3 and coincides well withthose estimated from the measurement of field effect transistor [12]. The bias voltages atthe intercept with the horizontal axis give rise to the VW being approximately 0.8±0.1V, 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 upondeposition of Al on the top of the film with a deposition rate of 0.5–1.0 nm/min. Theincrease of resistance is due to the formation of an insulating depletion layer in theHT-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 theresistance of depletion layer is sufficiently larger compared to the bulk region, the

Page 481: Nanotechnology and Nano-Interface Controlled Electronic Devices

464 K. Kaneto et al.

Fig. 7: The change of conductance in PAT6 upon deposition of Al. The inset shows the structure of the testcell.

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 afterdeposition of Al. It is supposed that the first increase in resistance is due to the formationof a depletion layer and the second gradual increase is due to the diffusion or migrationof Al into the PATn film. Typically the first abrupt increase of resistance gives thethickness 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 inthe 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 semi-conductor region depending on the resistances. Therefore, it is supposed that the biasis applied mainly to the depletion layer since the resistance of the depletion layer ismuch larger than that of the bulk region. The potential profile along a cross sectionof the Al/HT-PAT6/Au diode has been measured directly using probing tips mountedon a micromanipulator with the accuracy of 3 nm. Fig. 8a shows a schematic drawingof the measurement system for the potential profile. A planer type of Schottky diodewas fabricated, because a sandwich type cell is spatially too thin to map the potentialprofile. The Pt/Ir tip prepared for scanning tunneling microscopy was used as thepotential probing tip and touched the surface of the PAT film between the Al and Au

Page 482: Nanotechnology and Nano-Interface Controlled Electronic Devices

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/Audiode. (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 theAu electrode. The potential at each position in the PATn film was picked up with theprobing tip.

The curves in Fig. 8b show the potential profile along a cross section of the Al/HT-PAT6 film/Au diode. A steep jump was observed within 3 μm from the Al electrodeand 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 approximatelyproportional to the bias voltage. On the other hand, the potential at the depletion layerunder forward bias did not increase proportional to the bias and the potential gradient atthe bulk region increases with increasing bias. For forward bias, however, a large voltagedrop at the depletion layer, which should vanish at a bias of Va > Vw, is observed andcould be due to some barrier originating from Al oxide or inherent contact resistancebetween metal and PAT film. In fact, there is appreciable contact resistance at theinterface between PAT film and Au electrode also. For the detailed carrier transportbetween organic materials and metals not only Schottky type but also ohmic type contactis an important factor in fabricating better performance devices.

Page 483: Nanotechnology and Nano-Interface Controlled Electronic 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, Auand Al side of the diode at various bias fields and polarities. The carrier sign, placeof carrier generation and carrier mobilities were investigated. The largest photocurrentwas observed when the laser was irradiating the positively biased Al electrode againstthe 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, thephotocarriers are generated effectively in the PAT film near the Al electrode or thedepletion 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 castand spin coated HT-PAT10 films, respectively. In the photoresponse, a clearer kinkassociated with the transit time of photocarriers was observed in cast film than in spincoated film, indicating that the carrier migration in cast film is less dispersive thanthat of spin coated film. The dispersive transport results from photocarriers drifting tothe opposite electrode by hopping via multiple traps distributed energetically and/orspatially. Generally the HT-PATn films with ordered structure show a clearer inflectionpoint due to less multiple trapping levels [21]. The log–log plot of the photoresponsehas 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 linesbefore and after the inflection point of the photocurrent. It is found that the inflectionpoints change to obscure with increasing alkyl chain length.

The field dependence of transit times is shown in Fig. 10 for various spin coatedPATn films, in which the field in the film is assumed to be uniform along the filmdirection. 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.

Page 484: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 coatPATn 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. Photocarriermobilities μ evaluated from Eq. 1 are shown in Fig. 11 as function of bias. As will bementioned later, it is noted that the mobilities at the lower fields below 5 × 105 V/cmshow 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 mobilityin HT-PAT films and similar behavior to RD-PAT12 film. The field dependence ofmobilities 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 originallyaccounts for the positive field dependence in the manner of lnμ ∝ E1/2, which is basedon the increased probability of carriers detrapping from ionic impurities at the higherfield. And also the polaron hopping model explains the positive field dependence with

Page 485: Nanotechnology and Nano-Interface Controlled Electronic Devices

468 K. Kaneto et al.

the relationship of lnμ ∝ E . Especially, the negative field dependence of the mobilitiesin 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

μ = μ0 exp

(− E0

kT

)exp

(E1/2

β/T −γ

), (8)

where μ0 is the pre-exponential factor of mobility, E0 is the activation energy, β isthe constant given by the Poole–Frenkel model and γ is an empirical parameter. WhenT > β/γ , the field dependence of mobilities is negative. Bässler [21] proposed a modelto explain the negative field dependence of the mobility by a Monte Carlo simulation ofrandom walk of carriers in a Gaussian distribution of the density of states in trappingsites, 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 takea detour at lower fields, resulting in the negative field dependence of the mobility.Analytically, these models account well for the experimental data for the negativefield dependence of the mobility by choosing parameters appropriately. However, thephysical meaning of the chosen parameter is not clear. The positive field dependences ofmobilities 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 todistinguish them.

The temperature dependence of the TOF mobilities [15] gives as activation energiesof 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 resultindicates that the activation energy results from the hopping between the π-conjugatedmain chains and is unlikely to be due to the intrachain π-conjugation, since the HT-PATn is supposed to be extended to the whole length of the individual chain. It is alsofound that the activation energy scarcely depends on the field, which contradicts theprediction 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 thedepletion layer [27], because of the larger resistance compared to that in the bulkregion as discussed in the previous section. A simplified model of the non-uniformfield distribution for a cell with a depletion layer is shown in Fig. 12, where the spacecharges at the depletion layer are ignored because of the first-order approximation. Theconductivities of the depletion layer and bulk region are assumed to be constant as σd

and σb, respectively. Hence, the fields at the depletion layer (Ed) and the bulk region(Eb) are constant and given by

Ed = σbVa

σbd +σd(�−d), (9)

Eb = σdVa

σbd +σd(�−d). (10)

Page 486: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 aconstant, the transit times of carriers at the depletion layer (Td) and the bulk region (Tb)are given by Td = d/μEd and Tb = (� − d)/μEb, respectively. Eventually, the transittime (Tr) is expressed by Eq. 11,

Tr = Td + Tb = d2 + (σd/σb +σb/σd)(�−d)d + (�−d)2

μVa. (11)

One may remind that the rectification ratio above the onset is roughly proportional tothe 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 thatEq. 11 can be simplified as

Ttr �[

1+ σb

σd

d

]�2

μVa. (12)

The transit time dominantly originates from the carrier transit at the bulk region. At1 < (σb/σd)(d/�), the Ttr should be roughly proportional to V −1/2

a , since d ∝ V 1/2a ,

as predicted in Fig. 10 at lower fields. Plots of TtrVa vs. V 1/2a are shown in Fig. 13,

indicating a fairly good linear relationship. The straight lines are fitted to the data with alinear least-squares fitting. The intercepts of the lines, TtrVa at V 1/2

a → 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, hencethe 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 (2.7μm) films, respectively.

The evaluation of TOF mobilities for the negative field dependence is based on theSchottky junction with the conditions σb � σd, a field independent mobility and, moreimportantly, that the carrier generation occurs at the depletion layer. It has already beenmentioned that the carrier generation takes place at the depletion layer with the help ofthe large built-in field and reversed bias in our previous paper [8,17]. This can also beconfirmed by plotting of the number of generated charges with the laser light againstthe thickness of the depletion layer (d or V 1/2

a ). The number of generated charges is

Page 487: Nanotechnology and Nano-Interface Controlled Electronic Devices

470 K. Kaneto et al.

Fig. 13: Plots of TtrVa vs. V 1/2a under the assumption of a non-uniform field distribution in a Schottky

junction cell.

Fig. 14: Dependence of number of photocarriers ItrTtr vs. V 1/2a .

obtained by integrating the current against time in the iph vs. t curve of Fig. 9, orapproximately evaluated by ItrTtr, where Itr is the current at the transit time. Fig. 14shows fairly good dependence of ItrTtr vs. V 1/2

a , indicating that the photocarriers aregenerated at the depletion layer, expanding with increasing the reversed bias, V 1/2

a . It isinteresting to note that the ItrTtr vs. V 1/2

a curves of all PAT films fall on approximately asingle line. It is confirmed that the photocarriers are actually generated at the depletionlayer that is expanding with the reversed bias. It should be reminded again that theproposed model of the non-uniform field distribution is based on σb � σd, or thelarger conductivity. In fact, the negative field dependence has been scarcely observed inregiorandom PAT films with lower conductivity [11,20].

Page 488: Nanotechnology and Nano-Interface Controlled Electronic Devices

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-PAT4film at room temperature. The curves are the typical characteristics [10,18] of an FEtransistor, showing signs of saturation of ID at high VD due to the pinch off of thechannel. The enhancement of ID by the negative gate bias indicates that a positivecarrier (holes) is favorably induced in the PATs channel and coincides with the resultsof TOF measurements. The on/off ratio of 15 at VG = 30 V and −30 V for VD = −10V is relatively small, which could result from the rather thick PAT film employed. Thechannel conductivities (σ ) of PAT films were evaluated from the curve at VG = 0 and atlow VD and are summarized in Table 1 with the data of NR-PAT6 and NR-PAT12 forcomparison. The typical conductivity of HT-PAT4 was 1.4×10−6 S/cm, which is largerthan those of HT-PAT18 and NR-PATn by three orders of magnitude.

I ′D–VG curves for various VD are shown in Fig. 16, where the I ′

D was obtained bysubtracting ID at VG = 0 from the raw data of ID. The mobilities are estimated fromthese curves according to Eq. 3 from the gradient at the linear part of the curvesand 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 1.4×10−6 1.0×10−2 8.9×1014

HT-PAT6 1.2×10−6 2.9×10−3 2.6×1015

HT-PAT10 1.2×10−6 9.1×10−4 8.2×1015

HT-PAT12 9.1×10−7 7.9×10−4 7.2×1015

HT-PAT18 6.1×10−9 5.0×10−6 7.5×1015

NR-PAT6 4.8×10−9 4.3×10−6 7.9×1015

NR-PAT12 1.6×10−9 2.8×10−7 3.6×1016

Page 489: Nanotechnology and Nano-Interface Controlled Electronic Devices

472 K. Kaneto et al.

Fig. 16: I ′D–VG curves for various VD, where the I ′

D was obtained by subtracting ID at VG = 0 from the rawdata of ID.

Fig. 17: (a) Mobilities of PATn films and (b) conductivities and carrier densities as a function of the alkylchain length.

N = σ/eμ. The dependencies of conductivity and carrier density on alkyl chain lengthare shown in Fig. 17. It is apparent that the mobilities are larger for the films withshorter alkyl chain length. The mobility in HT-PAT4 of 1.0×10−2 cm2/V s is noticeablylarger than those of the other HT-PAT films and about three times larger than that ofNR-PAT4 film reported by Ohmori et al. [28]. This fact is explained by the assumptionthat carrier migration is consisting of two components, namely, of the transport in mainchains within the effective π-conjugation and of hopping between main chains. Thecarrier transport in the main chain is supposed to be much faster than that of inter-chainhopping. Therefore the carrier transport is predominantly controlled by the interchainhopping, 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 PATfilms as shown by the curves in Fig. 17 and the conductivities are primary determined

Page 490: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 withshorter alkyl chains is not known, however, may be related to impurities in films. Theexceptionally low conductivity due to low mobility in HT-PAT18, NR-PAT6 and NR-PAT12 results from the randomness of main chains. The small mobility of HT-PAT18,

Page 491: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 entanglementof too long side chains. The energetic diagram for the randomness of the main chainis schematically shown in Fig. 18. In HT-PATn the π-conjugation extends to the wholelength of the polymer chain and the carriers transport as if in a single band. On the otherhand, in NR-PATn the π-conjugated region plays the role of the potential barriers forcarrier transport within a main chain.

Recently, Bao et al. [29] and Sirrighaus et al. [30] have reported FE mobilitiesin HT-PAT4 of 0.045 cm2/V s and 0.1 cm2/V s, respectively, which were attained bycareful purification and surface treatment of SiO2. It is noted, as depicted in Fig. 19, thatthe FE mobilities generally agree with the TOF mobilities [15]. In fact the FE mobilityrepresents the carrier transport along the film surface, whereas the TOF mobility isthe transport perpendicular to film surface. Therefore, the agreement of FE and TOFmobilities 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 photonto carrier estimated from photocurrent in Al/HT-PHT/Au diode for the illumination atthe Al and Au side. The action spectrum for the light illuminated at Al side is similarto the absorption spectrum at the absorption maximum around 2.5 eV. On the otherhand, the action spectrum for the illumination at the Au side is quite different from theabsorption spectrum. The results indicate that photocarriers are generated dominantly atthe junction of Al and HT-PATn films. This idea is supported by taking the penetrationdepth of light (120 nm: inverse of absorbance at the peak of HT-PATs films [7]) and thefilm thickness of 200 nm into account. Namely, upon the illumination at the Au side

Page 492: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 inAl/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 thedepletion layer, resulting in opposite spectra to that of Al side illumination. It shouldbe noted that a large Q.E. is observed at a photon energy above 3 eV, indicating that adifferent 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 Auand 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.

Page 493: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 forthinner films, and the Q.E. spectra came to be similar to those for illumination at theAl side. The result indicates that the thickness of the photocarrier generation layer isless than 80 nm. The thickness dependence of the Q.E. for the illumination at the Alside 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 carriergeneration, and the thickness of the photocarrier generation layer is less than 80 nmor equal to the depletion layer. The carrier generation mechanism is summarized asshown by the schematic drawing in Fig. 22 for illumination at the Al side. The lightgenerates excitons, which dissociate into a electron and hole pair by the assistance ofthe strong internal field. Electrons transfer to the Al electrode and holes migrate to theAu electrode through the bulk region, resulting in a photovoltaic effect. The migrationof 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 andphotocarrier generation at the Schottky junction. These functions of optoelectronic

Page 494: Nanotechnology and Nano-Interface Controlled Electronic Devices

Transport and photocarrier generation in poly(3-alkylthiophene) and metal junctions 477

devices are found to originate from the depletion layer with nanometric thicknessand space charge. The thickness of the depletion layer is estimated by a couple ofmethods to be several 10 nm. It is mentioned that the photocarriers are generated mosteffectively in the depletion layer with the assistance of the strong built-in field. Carriermobilities which have been evaluated by means of TOF and FE are discussed from theviewpoint of regioregularity and effect of alkylchain length. The authors are planning tofabricate a diode with high performance of the photovoltaic effect or solar cells basedon poly(3-alkylthiophene).

Acknowledgements

The authors express their sincere thanks to Dr. S.S. Pandey, Mr. S. Nagamatsu forassistance in carrying out experimental works and discussion. This study is supportedby a Grant-in-Aid for Scientific Research from the Ministry of Education, Science andSport and Culture, Japan. The authors express their sincere thanks to Fukuoka IST foruse of the experimental facility.

References

1. H. Shirakawa, T. Ito, and S. Ikeda, Makromol. Chem. 179, 1565 (1978).2. C. Chiang, M. Druy, S. Gau, A. Heeger, E. Louis, A. MacDiarmid, Y. Park, and H. Shirakawa, J. Am.

Chem. Soc. 100, 1013 (1978).3. A.F. Diaz, K. Kanazawa, and G.P. Gardini, J. Chem. Soc. 100, 1013 (1978).4. K. Kaneto, K. Yoshino, and Y. Inuishi, Jpn. J. Appl. Phys. 21, L567 (1982).5. K. Kaneto, K. Yoshino, and Y. Inuishi, Jpn. J. Appl. Phys. 22, L412 (1983).6. 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).7. K. Kaneto and W. Takashima, Curr. Appl. Phys. 1, 355 (2001).8. P. Barta, F. Cacialli, R.H. Friend, and M. Zagorska, J. Appl. Phys. 84, 6279 (1998).9. M. Granstrom, K. Petritsch, A. Arias, A. Lux, M. Anderson, and R. Friend, Nature 395, 257 (1998).

10. S. Pandey, W. Takashima, S. Nagamatsu, and K. Kaneto, IEICE Trans. Electron. E83-C, 1088 (2000).11. W. Takashima, S.S. Pandey, T. Endo, M. Rikukawa, and K. Kaneto, Curr. Appl. Phys. 1, 90 (2001).12. K. Kaneto, W. Lim, W. Takashima, T. Endo, and M. Rikukawa, Jpn. J. Appl. Phys. 39, L872 (2000).13. T. Chen, X. Wu, and R.D. Rieke, J. Am. Chem. Soc. 117, 233 (1995).14. R.S. Loewe, S.M. Khersonsky, and R.D. McCullough, Adv. Mat. 11, 250 (1999).15. K. Kaneto, K. Hatae, S. Nagamatsu, W. Takashima, S. Pandey, K. Endo, and M. Rikukawa, Jpn. J.

Appl. Phys. 38, L1188 (1999).16. M. Rikukawa, Y. Tabuchi, K. Ochiai, K. Sanui, and N. Ogata, Thin Solid Films 327, 469 (1998).17. M. Ahlskog, J. Paloheimo, H. Stubb, P. Dyrklev, M. Fahlman, O.Inganas, and M. Anderson, J. Appl.

Phys. 76, 893 (1994).18. D.K. Schroeder, Semiconductor Material and Device Characterization (John Wiley and Sons, New

York, 1990).19. S.S. Pandey, W. Takashima, S. Nagamatsu, T. Endo, M. Rikukawa, and K. Kaneto, Jpn. J. Appl. Phys.

39, 94 (2000).20. S.S. Pandey, S. Nagamatsu, W. Takashima, and K. Kaneto, Jpn. J. Appl. Phys. 39, 6309 (2000).21. H. Bässler, Phys. Rev. 175, 15 (1993).22. H. Scher and E.W. Montroll, Phys. Rev. B 12, 2455 (1975).

Page 495: Nanotechnology and Nano-Interface Controlled Electronic Devices

478 K. Kaneto et al.

23. J. Frenkel, Phys. Rev. 54, 647 (1938).24. E. Lebedev, T. Dittrich, V. Petrova-Koch, S. Karg, and W. Brutting, Appl. Phys. Lett. 71, 2686 (1997).25. W. Gill, J. Appl. Phys. 43, 5033 (1972).26. L. Schein, A. Rosenberg, and D. Glatz, J. Appl. Phys. 60, 25 (1986).27. M. Goodman, J. Appl. Phys. 34, 329 (1963).28. Y. Ohmori, H. Takahashi, K. Muro, M. Uchida, T. Kawai, and K. Yoshino, Jpn. J. Appl. Phys. 30,

L610 (1991).29. Z. Bao, A. Dodabalapur, and A.J. Lovinger, Appl. Phys. Lett. 69, 4108 (1996).30. H. Sirringhaus, N. Tessler, and R. Friend, Science 280, 1741 (1998).

Page 496: Nanotechnology and Nano-Interface Controlled Electronic Devices

Nanotechnology and Nano-Interface Controlled Electronic DevicesEditors: M. Iwamoto, K. Kaneto and S. Mashiko© 2003 Elsevier Science B.V. All rights reserved

CHAPTER 23

Thermochromic behaviorin 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. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4792. Preparation of samples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4803. Experimental details . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4824. Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 483

4.1. Large change of electrical conductivity of poly(3-alkylthiophene) at thesolid–liquid phase transition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 483

4.2. Large change of the absorption spectrum of poly(3-alkylthiophene) at thesolid–liquid phase transition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 485

4.3. Anomalous dependencies of photoluminescence of poly(3-alkylthiophene)on temperature and alkyl chain length . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 488

4.4. Solvatochromism in poly(3-alkylthiophene) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4914.5. Influence of hydrostatic pressure on electrical and optical properties of

poly(3-alkylthiophene) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4954.6. Thermochromism in poly(p-phenylene vinylene) derivatives . . . . . . . . . . . . . . . 501

5. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 507References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 507

1. Introduction

The Nobel prize in chemistry 2000 was awarded for the discovery and development ofconducting polymers by Professors Hideki Shirakawa, Alan G. MacDiarmid and AlanJ. Heeger. Conducting polymers with highly extended π-conjugated electron systems intheir main chains have been considered to be insoluble and infusible materials becauseof their stiff main chain. The difficulty of processing them has restricted the practicalapplications of conducting polymers. However, to shape conducting polymers into films

Page 497: Nanotechnology and Nano-Interface Controlled Electronic Devices

480 M. Onoda and K. Tada

or fibers, preparation methods, such as using solution-processable precursors [1] orintroducing long side chains [2,3], have been determined. In particular, the introductionof long side chains (e.g., alkyl or alkoxy) in the molecular structures makes it possiblefor the conducting polymers to be soluble in some solvents such as chloroform, andfusible by heating. By utilizing the solubility [2,4] and fusibility [5,6], these polymerscan be easily shaped into thin films by wet processes such as casting, spin-coatingand hot-press techniques. This is a great advantage from the viewpoint of applicationto electronic devices [7]. Their mechanical, electrical and optical properties dependhighly on their side chain length, moreover they exhibit unique characteristics suchas 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 beprepared by molding methods. For example, films and fibers can be prepared easily by ahot-press method and also by a melt-spinning method. These films and fibers indicatedlow conductivity as a characteristics of insulators but turns into metal upon exposure toiodine vapor and a conductivity as high as 10–100 S/cm can be obtained just in the caseof non-substituted samples. The fusibility can be also applied to connect or stick twoconducting polymer films, as conducting binder, by sandwiching powder or a film offusible polymer between two non-fusible conducting polymer films and pressing at hightemperature. This connected sample also demonstrated an insulator–metal transitionupon doping.

In this paper, electrical and optical properties of poly(3-alkylthiophene) in theliquid state, and their dramatic change at the phase transition are reported. Moreover,remarkable effect of hydrostatic pressure on electrical and optical properties of poly(3-alkylthiophene) and dependence of absorption spectra on solvent are discussed indetail.

On the other hand, poly(2,5-alkoxy-p-phenylene vinylene), ROPPV [11] is oneexample of soluble and fusible conducting polymers. The ionization potential of thisconducting polymer is lowered by the donor alkoxy chain, and hence its doped-statestability is extremely high [12]. Therefore, ROPPVs are expected to be promising candi-dates for organic electronic devices. Therefore, we report on the thermochromic behaviorof ROPPVs. The variation of properties with side chain lengths is also mentioned. Theresults 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 usingFeCl3 as a catalyst as shown in Fig. 1a [2,5,13]. That is, 3-alkylthiophene monomer andFeCl3 were dissolved in anhydrous chloroform at −10°C. After reaction the mixturewas kept at this temperature for 12 h. The polymer was precipitated by pouring thewhole reaction mixture into a large excess of methanol and isolated by filtration. Theraw product was washed in a Soxhlet apparatus with methanol to remove FeCl3. ThePATn are soluble in conventional organic solvents such as chloroform, tetrahydrofuran,dichloromethane, etc. and red in color.

Page 498: Nanotechnology and Nano-Interface Controlled Electronic Devices

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-dialkoxy-benzene monomer using KOt -Bu (potassium t -butoxide) as a catalyst as shown inFig. 1b [14]. That is, 2,5-dialkoxybenzene was prepared through the use of theWilliamson ether synthesis from phenol and alkylbromide. 1,4-bis(chloromethyl)-2,5-dialkoxybenzene monomer was obtained according to the procedure which is anadaptation of the method described by Wood and Gibson. A solution of KOt -Buin t-butanol was quickly added at room temperature under N2 atmosphere to astirred 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 thepolymer was precipitated by pouring the whole reaction mixture into a large excess ofmethanol and isolated by filtration. The raw product was washed in a Soxhlet apparatuswith methanol to remove KOt-Bu and dried under vacuum. The ROPPVn is solublein conventional organic solvents such as chloroform, tetrahydrofuran, dichloromethane,etc., and reddish orange in color. Here, the n of ROPPVn and PATn denotes the numberof carbon atoms in the alkoxy and alkyl chains. The side chain length dependence ofmelting points of ROPPVs and PATs is shown in Fig. 2. It is observed that the meltingpoints 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 conductingpolymers were fabricated on glass substrates from chloroform solution.

Page 499: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 electrom-eter. The optical absorption spectra of the spin-coated films were measured using aspectrophotometer (Hitachi U-3410). The bandgap energies of these conducting poly-mers 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 energyhν. The photoluminescence (PL) spectra of the casted films were obtained with amonochromator and a photomultiplier tube. A Xe lamp combined with monochromatorwas used as an excitation source. Each sample was set in the pressure chamber withoptical 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 anatmospheric low-energy photoelectron spectroscopy instrument (Rikenkeiki AC-1) [15].Fig. 3 shows the schematic diagram of this instrument. Monochromatic ultraviolet lightin the energy range of 3.4–6.2 eV is irradiated onto the sample in air. The photoelectronsemitted from the sample surface stimulate the ionization of the atmospheric oxygen, andthe number of ionized oxygen molecules is counted. These measurements were carriedout at temperatures ranging from room temperature to 200°C. For the measurementsof the optical absorption spectra and the PL spectra, the surface of the polymersspin-coated or casted on a glass plate were covered with another glass plate in order toavoid oxidation during heating. Prior to carrying out the measurements, the polymerswere heated to higher temperatures than their melting points to adhere them to the glasssurfaces.

Page 500: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 thesolid–liquid phase transition

Findings of solubility in some solvents and even fusibility at relatively low temperatureof different types of PATn have stimulated the extensive study of these series ofpolymers. 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 ashydrocarbon and rare gas liquids, but their electronic band scheme has not beenclarified. Especially, the electronic conduction in polymer liquids has not been studied.

Fig. 4 indicates the temperature dependence of the electrical conductivity of PAT12in both the solid and the liquid state. In the solid state near the premelting regionwith increasing temperature, some anomalous temperature dependence of the electricalconductivity was observed. In particular, at the phase transition from solid to liquidstate, the conductivity decreased dramatically in a step-wise manner, and again increasedwith an activation energy in the liquid state with increasing temperature. The step-wisedecrease could be related to a rapid change of either electron mobility or carrier densityat the phase transition. By contrast, during the cooling cycle, the conductivity increasedin a step-wise manner at the phase transition by the same magnitude as that during theheating stage. However, there existed a hysteresis in the conductivity observed in theheating and cooling cycles. That is, the temperature at which the conductivity increasedstep-wise in the cooling stage was lower than that in the heating stage. This couldbe related to a super-cooling effect when the temperature was lowered at a relativelyhigh speed (0.5°C/min). The activation energy of the conductivity in liquid state wascalculated to be about 0.7 eV.

A similar unique temperature dependence of the conductivity was also observed inother fusible PATn such as poly(3-docosylthiophene), PAT22, shown in Fig. 5 as an

Page 501: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 tobe about 0.9 eV in this case.

Page 502: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 thesolid–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 tem-peratures ranging from solid to liquid phase, respectively. Thermochromism in PATnfilms was previously reported [8]. A shift of the absorption peak to a higher energywith increasing temperature was found. However, the shift of the absorption edge wasrelatively small. In these figures, a large shift of the absorption peak but a small shift ofthe edge are also confirmed in the solid state.

Figs. 8 and 9 show the temperature dependence of the absorption peak and the edgeenergies of PAT12 and PAT22 as evaluated from Figs. 6 and 7, respectively. In thesefigures it is clearly indicated that the absorption peak in the solid phase shifts with thetemperature shift, but not the absorption edge. At the phase transition from solid toliquid, a very large absorption change was observed. That is, the absorption edge shiftedremarkably to a higher energy in the liquid phase. The temperature dependence of theabsorption 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 liquidphase.

Page 503: Nanotechnology and Nano-Interface Controlled Electronic Devices

486 M. Onoda and K. Tada

Fig. 7: The change of absorption spectra of PAT22 at various temperatures ranging from solid to liquidphase.

changes with melting. In the liquid state, the co-planarity of the neighboring thiophenerings is expected to be minimized. The resultant decreased conjugation length shouldgive rise to an increase of the bandgap. In the case of PAT12, the bandgap in the liquidphase was evaluated to be about 2.5 eV from the absorption spectrum. This is larger byabout 0.3 eV compared with that in the solid phase.

Two modifications of the polymer main chain seem to be candidates as conformationchanges which reduce co-planarity: bending and/or twisting (torsion) of neighboringthiophene 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.

Page 504: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 ofthe influence of some conformation change with temperature triggered by the changeof conformation of alkyl group. That is, in the solid state at high temperatures of thepremelting region, the torsion of the neighboring thiophene rings should be initiated atseveral points along a polymer main chain. The separation (number of thiophene rings)between two neighboring torsioned points which corresponds to the conjugation lengthshould decrease at a high temperature and be randomly distributed. This results in theformation of regions of short conjugation whose lengths are widely distributed. Thedecrease of longer conjugated chains, and formation of shorter conjugation sections, canexplain the blue shift of the absorption peak. The fact that the absorption edge does notchange 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, astructure is observed. This may indicate the existence of some shorter conjugationlengths. However, this can be explained by a contribution of vibrational subbandsbecause 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 atthe phase transition. That is, in this case, the decrease of the carrier density in the liquidstate due to the increased bandgap may be the cause of the lower conductivity in theliquid state. However, electron scattering should also be enhanced in the premeltingand melting regions. The anomalous temperature dependence of conductivity observedin the premelting region could be related to electron scattering and correlated with theanomalous absorption peak shift in this range.

Fluctuation of the bandgap due to the existence of regions of various conjugationlengths and molecular motion of alkyl side chains in the premelting region could be thecause of strong carrier scattering.

The electronic band scheme in the liquid state has not been explained in detail sofar. 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 ultravioletlight. There are only a few papers in which the bandgap has been estimated in liquids

Page 505: Nanotechnology and Nano-Interface Controlled Electronic Devices

488 M. Onoda and K. Tada

and it has not been completely clarified whether the band model is still applicable inliquids or not. The present observation of a clear absorption edge, even in the liquidstate, and the similarity of the shape of the absorption spectra support the concept of theapplicability of electronic bands in liquids.

Optical studies of liquid conducting polymers are much easier compared withconventional liquids due to the small bandgap which corresponds to the energy ofvisible light. The electrical study is also simplified because the electronic carrier densityis relatively high due to the small band gap.

The authors propose the study of the properties of conjugated polymers in theliquid state (liquid conducting polymer) for the understanding of fundamental electronicprocesses in liquids.

4.3. Anomalous dependencies of photoluminescence of poly(3-alkylthiophene) ontemperature and alkyl chain length

Some of conducting polymers such as PATn have been found to be soluble in somesolvents and even fusible. In these new type of substituted conducting polymersvarious interesting phenomena such as thermochromism and anomalous temperaturedependence of electrical conductivity were found in a relatively low temperature range.In this section, we report detailed characteristics of anomalous temperature dependenceof photoluminescence of PATn film both below and above melting point. It is alsoindicated 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. Itshould be noted in this figure that with increasing temperature the emission peak shiftsto higher energy slightly and its intensity increases in the solid phase. Above the meltingpoint, it again decreased in the liquid phase.

Fig. 10: Photoluminescence of PAT12 at various temperatures.

Page 506: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 clearlyindicated in Fig. 11. That is, with increasing temperature the emission intensity increasesin the solid phase and after showing a maximum at the melting point it again decreasesin the liquid phase. Corresponding to the lower melting point, the photoluminescencemaximum intensity appears at lower temperature in PATn with longer alkyl chain length.

These temperature dependencies of the photoluminescence are quite anomalouscompared with those of ordinary inorganic and organic semiconductors and insula-tors, in which the photoluminescence intensity decreases with increasing temperaturesremarkably due to the increased probability of non-radiative recombination.

This anomalous behavior in PATn can be explained qualitatively in terms of con-formation change of polymer main chain with temperature and its influence on thedynamics of photoexcited species as follows.

By the excitation with the Ar ion laser of photon energy 2.54 eV, electrons and holesare excited in conduction and valence bands, and some of them should turn into negativeand positive polarons, respectively, in a very short time [16]. The photoluminescencein these conducting polymers is considered to be due to the radiative recombinationof these excited species. Therefore, when positive and negative excited species moveout from the excited region and separate, escaping from the recombination, the pho-toluminescence intensity should be suppressed. We have already reported that withincreasing temperature co-planarity of neighboring thiophene rings decreases due tothe introduction of torsion at some bondings between thiophene rings, resulting in thedecrease of effective conjugation length [17]. This can explain the thermochromism inPATn. Therefore, at higher temperature due to the decreased effective conjugation lengththe mobility of the excited species and their escape probability from the excited regionshould decrease, which can result in the enhancement of the probability of recombina-tion and the emission of the observed stronger photoluminescence. When the effect ofenhancement of recombination probability is larger than the increase of non-radiativerecombination at higher temperature, the enhancement of photoluminescence should be

Page 507: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 inten-sity in the solid phase was estimated to be around 0.1–0.2 eV. This should correspond tothe energy necessary for the introduction of the torsion angle with the activation energyof non-radiative recombination subtracted.

We already reported that the torsion angle does not change remarkably with tem-perature in the liquid phase contrary to in the solid state [18], though it is larger thanthat in the solid phase. Therefore, the effective conjugation length should not changeremarkably with temperature in the liquid state and the probability of confinement ofexcited species in the excited region should not change drastically with temperature. Insuch a case, the increased non-radiative recombination at higher temperature will resultin the suppression of photoluminescence intensity in the liquid phase.

It should be also pointed out that the existence of similar emission spectra inboth the liquid and solid phase clearly supports the applicability of the concept ofelectronic bands in the liquid phase. This observation of stronger photoluminescencein the system with shorter conjugation length reminds of the case of cis- and trans-polyacetylene films. That is, in cis-polyacetylene with a non-degenerated ground state inwhich the formation and migration of solitons is highly limited showed much strongerphotoluminescence compared with the case of trans-polyacetylene in which solitons canbe excited and migrate freely [19,20].

It should be also noticed in Fig. 11 that PATn with longer alkyl chains shows strongerphotoluminescence. Fig. 12 shows the dependence of peak photoluminescence intensityon the alkyl chain length. This can be also tentatively interpreted in terms of dynamicsof photoexcited species as follows. In PATn with longer alkyl chains, because of thebulky side chain effect, the interaction of neighboring polymer main chains should bereduced and the probability of the transfer of photoexcited species from a polymer mainchain to the neighboring one should be suppressed. In such a case the density of positiveand negative excited species in a chain will increase, resulting in the enhancement ofrecombination photoluminescence. That is, the interchain transfer of excited speciesshould also reduce on the quantum efficiency of photoluminescence. This also supports

Page 508: Nanotechnology and Nano-Interface Controlled Electronic Devices

Thermochromic behavior in novel conducting polymers at the solid–liquid phase transition 491

the idea that the photoluminescence originates in the recombination of excited specieson 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. Fusibilityof these polymers has also been observed. The color of the solution of these PATsis strongly dependent on temperature, which has been explained in terms of largeconformation change with solvent and temperature [17]. For example, the color ofthe solution of poly(3-docosylthiophene), PAT22, in anisole is red at 30°C (lowtemperature) and yellow at 50°C (high temperature). The blue shift of the absorptionat high temperature has been interpreted in terms of the decrease of the effectiveconjugation length due to the decrease of the co-planarity of thiophene rings triggeredby the conformation change of alkyl side chains. However, it is not clear even why PATis soluble in some solvents but not in other solvents. In this session, the dependence ofabsorption spectra on solvent is discussed in detail. That is, solubility was checked bydissolving 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 identicaltemperature [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 thatthe color of the solution depends on the solvents delicately.

Table 1 shows the solubility of PAT6 and PAT-22 in various solvents. There are lotsof good solvents which are common to both, but it is also clear that a nice distinctionlies 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 differencein color has been explained in terms of difference of effective conjugation length withsolvent 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 discussthis spectral difference in these yellow solutions. However, PAT6 in anisole was alsoincluded because the spectral width of the main peak is just similar to those of yellowsolutions. The relatively small spectral differences with solvent are discussed in terms ofTaft’s solvatochromic parameter π∗, which is a measure of solute chromophore–solventinteraction effects [22]. When the absorption maximum λmax is a linear function of π∗,the existence of interaction between solute chromophore and solvent is evidenced. TheTaft 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 theinteraction between solvent and polymer main chain exists in PATn with shorter alkylchain.

Solvation of the substituted alkyl group is considered, gradually, to be essential for

Page 509: Nanotechnology and Nano-Interface Controlled Electronic Devices

492 M. Onoda and K. Tada

Table 1

Solubility of PAT6 and PAT22 in various solvents

Solvent PAT6 PAT22

RT 50°C RT* RT 50°C RT*

n-Heptane � � � ∗ ◦ ◦ a

n-Hexane � � � ∗ ◦ ◦ a

Di-n-butyl ether � � � ∗ ◦ ◦ a

Carbon tetrachloride ◦ ◦ ◦ ◦ ◦ ◦p-Xylene ◦ a ◦ ◦ ◦ ◦ ◦1-Butanol ∗ ∗ ∗ ∗ ∗ ∗Tetrahydropyran ◦ ◦ ◦ ◦ a ◦ ◦Trichloroethylene ◦ ◦ ◦ ◦ ◦ ◦Toluene ◦ a ◦ ◦ ◦ a ◦ ◦1,4-Dioxane � � � ∗ ∗ ∗2-Butanone � � � ∗ ∗ ∗Anisole ◦ a ◦ a,b ◦ a ∗ ◦ a,b ◦ a

Chloroform ◦ ◦ ◦ ◦ ◦ ◦Dichloromethane ◦ a ◦ ◦ ∗ ◦ ◦N ,N-Dimethylaniline ◦ ◦ ◦ ◦ ◦ ◦◦ = 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.

Page 510: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 rolefor solubility.

Page 511: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 concludethat the solubility of PAT6 is not a simple function of either solvent dipolarity orpolarizability. It is also clear that solubility is higher in electron-rich solvents whichhave non-bonded electron pairs and low in electron-poor solvents. For example, thehalogenated compounds such as carbon tetrachloride (CCl4), chloroform (CHCl3),dichloromethane (CH2Cl2), which have six or more non-bonded electron pairs permolecule, are good solvents, but hydrocarbons without such electrons are poor solvents.That is, solvent electron donation is one of the important parameters for solubility ofPATn with relatively short alkyl chain length such as PAT6.

On the other hand, the corresponding Taft plot for PAT22 is a scatter diagram asshown in Fig. 16. The large difference in these two Taft plots suggests that solute–solventinteraction effects for PAT6 and PAT22 are very different. In PAT22, the interactionof alkyl side chain and solvent molecule becomes more important. Therefore, PAT22becomes soluble in linear hydrocarbon liquids. Also the interaction between polymermain chain and solvent becomes smaller in PAT22. In PAT22, long alkyl groups act asmolecular sieve and the polymer main chain is shielding the interaction with solventmolecules. Only solvent molecules which are small in size and also can penetratethrough 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 alsosolvent (donor)–polymer chain (acceptor) interaction. The spectral shift with solventis determined by both change of co-planarity of thiophene rings and solvent–polymerchain interaction. That is, the solvatochromic shifts of λmax in the spectra of PAT6and PAT22 solutions have been analyzed in terms of Taft linear solvation energyrelationships.

Page 512: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 ofpoly(3-alkylthiophene)

Remarkable changes of electrical and optical properties have been observed in PATnwith the hydrostatic pressure. We have already reported that the conductivity decreasesstep-wise at the phase transition from the solid to the liquid state during a heating stageat 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 thephase transition temperature with lower by 25°C in the cooling stage.

As also evident from Fig. 17, under hydrostatic pressure similar temperature depen-dencies of the conductivity in the solid and the liquid state and a step-wise change atthe phase transition were observed. It is clear from this figure that the phase transitiontemperature increases with hydrostatic pressure. In this case, measurements were carriedout during the cooling stage.

The pressure dependence of the phase transition temperature Tm is more clearlyindicated 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 theincrease of the bandgap and/or increase of carrier scattering.

Electrical conductivity in the solid phase increased with increasing pressure, asshown in Fig. 17. This is more clearly indicated in Fig. 19. This increase of conductivitywith pressure can be tentatively explained in terms of the decrease of carrier scatteringunder pressure, as is discussed later. On the other hand, in the liquid phase, electricalconductivity and its activation energy were not influenced remarkably. If the conductionin the liquid phase originated in ionic transport, the conductivity should be muchsuppressed due to the decreased free volume under hydrostatic pressure. The observedsmall pressure dependence of conductivity seems to support the interpretation of

Page 513: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 theliquid state both the bandgap and the carrier scattering are not influenced markedly bypressures in this range.

Fig. 20 shows the absorption spectral change of PAT12 film in the pressure cell withtemperature at 1 bar. As already reported by us, in the pre-melting region, the blue shiftof the absorption peak is more remarkable than the shift of the absorption edge [8]. Thiscan be explained by the introduction of torsioning of the bonding between thiophenerings at several points on the polymer main chain, with which shorter conjugation chainsare created and the number of original long conjugated chain decreases. At highertemperature, the number of torsioned bonds increases. This results in the shift of theabsorption peak to the higher energy side (blue shift) at high temperature. We have also

Page 514: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 remarkablywith 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 athigh temperature which has blue shifted, again shifts to the lower energy side and comes

Page 515: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 absorptionpeak at room temperature does not shift remarkably with application of hydrostaticpressure up to 3 kbar, in which range the remarkable change was observed at hightemperature. Only at high pressures, the absorption peak shows slight red shift. Theseresults indicate that the torsioning of bonds on the polymer main chain and the resultantdecrease of co-planarity of conjugations can be suppressed by the applied hydrostaticpressure. That is, under high hydrostatic pressure the conjugation length again becomeslong even at high temperature.

The slight red shift of the absorption peak at room temperature when applyingextremely high pressure may suggest that even at room temperature some torsionedbonds exist at 1 bar. The probability also exists that the one-dimensionality can beenhanced under high pressure.

The enhancement of electrical conductivity by pressure shown in Figs. 17 and 19 canbe interpreted as follows. The existence of various lengths of conjugated chains due tothe introduction of torsioning of bonds may result in the deformation of the electronicband scheme. That is, fluctuation of conduction and valence bands and bandgaps canbe created, which results in a strong carrier scattering in the pre-melting region. Thesuppression of torsioning of bonds at high pressure as observed by spectroscopicmeasurements can suppress the fluctuation of the energy band and carrier scattering.Therefore, electrical conductivity in the pre-melting region can be enhanced underhydrostatic pressure as observed.

The shift of absorption spectrum with temperatures is much less in liquid phaseas already mentioned. Therefore, the effect of hydrostatic pressure on electronic statesshould not be so remarkable compared with that in the pre-melting region. This may

Page 516: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 rangecovering both the solid and the liquid phase.

explain the small influence of hydrostatic pressure on the conductivity in the liquidphase, shown in Fig. 17.

Just the similar effect of hydrostatic pressure on the melting point, electricalconductivity and absorption spectra were also observed in other fusible PATn.

Fig. 23 shows the temperature dependence of the photoluminescence in PAT12 at anambient pressure (1 bar) in a range covering both the solid and the liquid phase. As isevident from this figure, the luminescence peak shifted to higher energy with increasingtemperature. The luminescence intensity in the solid phase increased with increasingtemperature; however, at higher temperatures beyond the melting point (117°C) thephotoluminescence intensity is again suppressed.

A blue shift to the higher-energy side of the absorption peak had already beenreported for PATn [8,23] and has been explained in terms of an increased bandgap athigher temperature due to the decrease of effective conjugation length on account ofthe introduction of torsions at bonds between neighboring thiophene rings. The increaseof photoluminescence in the solid phase with temperature can also be explained bythe introduction of torsion. That is, photoexcited species (such as electrons and holes,or negative and positive polarons) should be confined in the narrow excited regiondue to the decrease of effective conjugation length, resulting in the enhancement ofrecombination photoluminescence. On the other hand, the increased suppression ofphotoluminescence with increasing temperature in the liquid phase in which the torsionangle is considered to be markedly independent of temperature can be explained by theincreased non-radiative decay of excited species.

Fig. 24 shows emission spectra of PAT12 under various pressures at 150°C. It isclearly observable in this figure that the emission peak does not shift very noticeablybut its intensity increases markedly with pressure up to 3 kbar. However, above 3 kbar

Page 517: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 lowerenergy.

These results can be tentatively explained as follows by taking the change ofmolecular conformation with pressure into consideration. With increasing pressure at150°C, a transition from the liquid phase into the solid phase was clearly observedat around 2.5 kbar [24]. This means that the enhancement of photoluminescencewith pressure is a characteristic in the liquid phase. If the torsion angle decreasedmarkedly with pressure in the liquid phase, the photoluminescence intensity should beenhanced contrary to observation. However, it should be noted that at higher pressurethe difference of temperature (Td = 150 − Tm) from the melting point Tm markedlydecreased. Molecular motion is expected to become more violent at larger Td even ifthe mean torsion angle is kept nearly constant. In such a case, non-radiative decay ofphotoexcited species should be more effective. Therefore, with increasing pressure thephotoluminescence can be enhanced.

On the other hand, Fig. 25 shows the effect of hydrostatic pressure on the absorptionspectrum of PAT12 at 32°C. Only at high pressures does the absorption peak show aslight red shift. We have also confirmed that the effective conjugation length becomeslonger with hydrostatic pressure as shown in Fig. 25 in the solid phase. This isconsidered to be due to the decrease of the torsion angle. This is also consistent withthe 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 becomeshorter, which restricts the escape of the excited species from the excited region,resulting in the enhancement of the recombination photoluminescence. The shift of theabsorption peak to lower energy can also be satisfactorily explained by the increasedconjugation length.

Exactly the same behavior was also observed in other PATn with an alkyl chain

Page 518: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 becomenoticeable increases in the sample with shorter alkyl chains. The changes of thetorsion angle between neighboring thiophene rings are expected to be triggered by theconformation 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 varioustemperatures. The absorption peak and the absorption edge shift slowly to higher energy

Page 519: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 samplevaries from orange to yellow, i.e. thermochromism occurs. This temperature dependenceof the optical absorption spectrum of ROPPV16 is different from that of PAT16 shownin Fig. 26b, in particular, at the shift of the absorption edge. The bandgap energy ofPAT16 increases suddenly from about 2.0 eV to about 2.4 eV around the melting pointof 160°C. On the other hand, the bandgap energy of ROPPV16 increases graduallyfrom about 2.2 eV at room temperature to about 2.4 eV at 200°C. This difference ofbandgap energy shift is more clearly indicated in Fig. 27, which shows the temperaturedependence of bandgap energies of various ROPPVs and PATs. The side chain lengthdependence 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 betweenthe 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 theside chain length dependence is not observed distinctly. This means that although thebandgap energies of ROPPVs depend on the temperature, they are not correlated withthe solid–liquid phase transition.

The mechanism of thermochromism in a conducting polymer was proposed to beas follows [25]. With increasing temperature, the side chain changes its conformationdue to the trans–gauche transition and becomes bulky. The steric hindrance between thebulky 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 planaritywith increasing temperature, resulting in deceased effective π-conjugation length andincreased bandgap energy.

The difference in the thermochromic behavior between ROPPVs and PATs can betentatively interpreted in terms of the main chain structure which is schematically shownin Fig. 28 [10]. The phenyl rings composing the PPV-type main chain can rotate freelywith respect to the main chain without significant deformation of the main chain, as

Page 520: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 canbe gradually decreased with increasing temperature. On the other hand, the thiophenerings 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 straightline. This indicates that the planarity of the π-electron systems in PATs changes not inthe solid state but in the liquid state. Therefore, the π-electron systems in PATs losetheir planarity not gradually with increasing temperature at the solid state, but suddenlyat the solid–liquid phase transition.

Fig. 29a shows the PL spectrum of ROPPV16 measured at various temperatures. ThePL peak shifts slowly to higher energy corresponding to the increase of bandgap energywith increasing temperature. However, the PL intensity decreases monotonically withincreasing 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 inthe solid state is explained as follows [8,25]. With the increase in temperature, thepolymer main chain twists and the effective π-conjugation length decreases, resultingin enhancement of the confinement effect of the radiative-excited species such asexcitons and/or exciton polarons. Therefore, the probability of radiative recombinationis increased and the PL intensity is enhanced. However, the decrease of PL intensities ofROPPVs with increasing temperature cannot be explained by this mechanism.

As schematically shown in Fig. 30a, the distance between polymer main chainsis enlarged by the side chains. The side chains take trans conformation at lowertemperature, and gauche conformation near and above the melting point. The side chainis more bulky and shorter when it is in gauche than in trans. Therefore, with increasing

Page 521: Nanotechnology and Nano-Interface Controlled Electronic Devices

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.

Page 522: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 inFig. 30b. In this case, the dissociation probability of the excited species such as excitonsand/or exciton polarons through interchain charge transfer increases and the probabilityof radiative recombination decreases.

On the other hand, the twist of polymer main chain would enhance the overlapof π-electrons, as schematically shown in Fig. 31. Assume that there is a pair ofneighboring polymer main chains whose phenyl rings are perpendicular to each otherand therefore the π-electrons do not overlap each other, as shown in Fig. 31a. In sucha case, interchain charge transfer does not occur. However, with increasing temperaturethe polymer main chain starts to twist and the π-electron orbitals of the pair becomemore parallel to each other, as shown in Fig. 31b. Consequently, the dissociationprobability of the excited species through the interchain charge transfer increases andthe probability of radiative recombination decreases. Thermally excited phonons wouldalso increase the exciton dissociation probability.

In the above discussion, we only concentrated on the change of interaction betweenthe main chains. However, the interaction between the main chain and the oxygenatoms of alkoxy side chains would also play an important role, because the interactionessentially contributes to the electronic energy structure of ROPPV [1].

Page 523: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 andPATs. The ionization potentials of PATs change steeply from 4.7 eV to about 5.4 eVnear the melting point. Thus, the electronic states of PATs change significantly with thesolid–liquid phase transition. However, in ROPPVs the ionization potentials decreaseslightly with increasing temperature and the degree of decrease becomes larger withlonger alkoxy chains. For example, the ionization potential of ROPPV16 decreasesfrom about 5.4 eV at room temperature to about 5.1 eV at 200°C, while the ionizationpotential 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 notclear at this stage.

From the results mentioned above, the schematic energy diagrams of ROPPV16 andPAT16 are estimated as shown in Fig. 33. In ROPPV16 with increasing temperaturefrom room temperature to 200°C the bandgap energy increases gradually from about 2.2

Page 524: Nanotechnology and Nano-Interface Controlled Electronic Devices

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 about5.1 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 ionizationpotential increases from about 4.7 eV to about 5.4 eV.

5. Conclusion

A reversible absorption spectrum change of poly(3-alkylthiophene), PAT, films withtemperature and anomalous temperature dependence of photoluminescence have beenobserved. By the application of hydrostatic pressure the spectrum at high temperatureturns into that of low temperature. Drastic change of electrical conductivity and opticalproperties have been also found at the melting point. These results were explainedin terms of modification of polymer conformation and conjugation system at hightemperature, i.e. introduction of torsioning of some bonds between thiophene ringsand its suppression under hydrostatic pressure. To confirm this interpretation, timedependence of the photoluminescence in the picosecond range, Raman and infraredspectra measurements are necessary.

The solvatochromic shifts of λmax of spectra of poly(|3-hexylthiophene) and poly(3-docosylthiophene) solutions have been analyzed in terms of Taft linear solvation energyrelationships. Solubility of PAT is determined not only by alkyl group solvation but alsosolvent (donor)–polymer chain (acceptor) interaction. The spectral shift with solventis determined by both change of co-planarity of thiophene rings and solvent–polymerchain interaction.

We have observed the thermochromic behaviors of poly(2,5-dialkoxy- p-phenylenevinylene)s and compared the results with the properties of poly(3-alkylthiophene)s.The results are summarized as follows. The temperature dependencies of the opticalabsorption spectra of ROPPVs were different from those of PATs, in particular, at theshift of the absorption edge. In other words, a difference in temperature dependenceof the bandgap energy was observed. This difference in the thermochromic behaviorcan be tentatively interpreted in terms of the main chain structure. The PL intensitiesof ROPPVs decreased monotonically with increasing temperature. This was opposite tothe 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 distancebetween polymer main chains and the enhancement of the overlap of π-electronswith the twist of polymer main chain, in addition to thermally excited phonons. Theionization potentials of ROPPVs decrease slightly with increasing temperature and thedegree of decrease becomes larger with longer alkoxy chains, although the origin of thistendency is not clear at this stage.

References

1. I. Murase, T. Ohnishi, T. Noguchi, and M. Hirooka, Polym. Commun. 25, 327 (1984).2. R. Sugimoto, S. Takeda, H.B. Gu, and K. Yoshino, Chem. Express 1, 635 (1986).

Page 525: Nanotechnology and Nano-Interface Controlled Electronic Devices

508 M. Onoda and K. Tada

3. M. Sato, S. Tanaka, and K. Kaeriyama, J. Chem. Soc. Chem. Commun., 635 (1986).4. D.L. Elsenbaumer, K.Y. Jen, and R. Oboodi, Synth. Met. 15, 169 (1986).5. K. Yoshino, S. Nakajima, and R. Sugimoto, Jpn. J. Appl. Phys. 26, L1038 (1987).6. K. Yoshino, S. Nakajima, M. Fijii, and R. Sugimoto, Polym. Commun. 28, 309 (1987).7. D. Braun and A.J. Heeger, Appl. Phys. Lett. 58, 1982 (1991).8. K. Yoshino, S. Nakajima, D.H. Park, and R. Sugimoto, Jpn. J. Appl. Phys. 27, L716 (1988).9. M. Hamaguchi and K. Yoshino, Jpn. J. Appl. Phys. 33, L1478 (1994).

10. K. Tada, M. Onoda, and K. Yoshino, J. Phys. D Appl. Phys. 30, 2063 (1997).11. I. Murase, T. Ohnishi, T. Noguchi, and M. Hirooka, Polym. Commun. 26, 362 (1985).12. K. Yoshino and M. Onoda, Kobunshi Electronics (Polymer Electronics), (Corona, Tokyo, 1996),

pp. 18 (in Japanese).13. K. Yoshino, S. Nakajima, H.B. Gu, and R. Sugimoto, Jpn. J. Appl. Phys. 26, L2046 (1987).14. K. Yoshino and M. Onoda, Kobunshi Electronics (Polymer Electronics), (Corona, Tokyo, 1996),

pp. 117 (in Japanese).15. M. Onoda, Y. Manda, M. Yokoyama, R. Sugimoto, and K. Yoshino, J. Phys. Condens. Matter 1, 3859

(1989).16. K. Kaneto, F. Uesugi, and K. Yoshino, J. Phys. Soc. Jpn. 57, 747 (1988).17. K. Yoshino, D.H. Park, B.K. Park, M. Onoda, and R. Sugimoto, Jpn. J. Appl. Phys. 27, L1612 (1988).18. K. Yoshino, D.H. Park, B.K. Park, M. Onoda, and R. Sugimoto, Solid State Commun. 67, 1119

(1988).19. K. Yoshino, S. Hayashi, Y. Inuishi, H. Kato, and Y. Watanabe, Jpn. J. Appl. Phys. 21, L653 (1982).20. K. Yoshino, S. Hayashi, Y. Inuishi, K. Hattori, and Y. Watanabe, Solid State Commun. 46, 583

(1983).21. K. Yoshino and M. Onoda, Chemistry 44, 1989 (in Japanese).22. M.J. Kamlet, J.L. Abboud, M.H. Abraham, and R.W. Taft, J. Org. Chem. 48, 2877 (1983).23. K. Yoshino, S. Nakajima, D.H. Park, and R. Sugimoto, Jpn. J. Appl. Phys. 27, L454 (1988).24. K. Yoshino, K. Nakao, M. Onoda, and R. Sugimoto, Solid State Commun. 68, 513 (1988).25. K. Yoshino, Y. Manda, K. Sawada, M. Onoda, and R. Sugimoto, Solid State Commun. 69, 143

(1989).

Page 526: Nanotechnology and Nano-Interface Controlled Electronic Devices

Subject Index

1×1 canted structure, 447

Adenine, 442Adsorption, 50, 53Ag(111) film, 437Alkanethiol, 443Alkylsiloxane monolayer, 193Ambipolar diffusion, 86, 92, 93Amplified spontaneous emission (ASE), 43,

44Anatase, 90Antibiotic peptide, 217–220Atom extraction, 450Atom manipulation, 449Atomic displacement, 439Atomic force control, 447Atomic force microscopy (AFM), 200, 204,

214, 430Atomic imaging of contact potential differ-

ence and capacitance, 449Atomic relaxation, 442Atomic stress, 442Au(111) surface, 50–55, 57, 60Azobenzene, 378Azobenzene monolayer, 378, 382

Bond order, 447Breath-figure, 404Brookite, 90

C60, 101Calixarene, 6, 7Carbon dioxide, 401Carbon nanotubes, 400Cell membranes on semiconductors, 215Cell surface glycocalix, 193, 201, 209, 215,

226Charge resonance, 113Chemical “switching” of cell/surface inter-

actions, 207Chemical bonding interaction, 439

Composite, 402, 412, 416Computer simulation, 144, 145, 149Conducting polymer, 455, 456, 479, 482,

488, 502Conduction, 134–136, 139, 140, 145Conduction mechanism, 135, 136, 139, 154Conduction model, 135, 136, 140Constant-excitation mode, 434Contact point, 434Contact potential difference, 449Control of atomic force and atom position,

447Coulomb island, 32, 37CPD image of the point charge, 449Critical point drying, 407Cu(111), 442

Dangling bond, 438Decay length, 433Dendrimer, 41, 42Dendrite molecules, 32, 36, 38Dendrite polymer, 33Deuterium NMR, 313–316, 321, 324, 325,

329, 333, 336, 337, 342, 346Dimer structure, 439Dipole, 264Director distribution, 314, 316, 318–322,

325–327, 329, 331, 344, 346Director dynamics, 314, 315, 331, 335, 342,

346, 347Discrimination of atom species, 445Discrimination of atomic force mechanisms

and atom species, 444DNA, 443Dye-sensitized solar cells, 83

EFM/NC-AFM, 449Elastomeric membranes, 183, 186, 187Electrical transport, 458Electron-beam (EB), 12Electron beam (EB) lithography, 4

Page 527: Nanotechnology and Nano-Interface Controlled Electronic Devices

510 Subject Index

Electron beam (EB) nanolithography, 4Electron hopping, 97Electron spin resonance (ESR) measure-

ments, 413Electron-transfer, 85Electron transport, 86Electrostatic force imaging, 448Electrostatic force microscope (EFM), 449Extracellular matrices (ECMs), 201

F-doped SnO2, 88Fabrication process, 13Fick’s diffusion model, 88Field emission, 400Fill factor, 84Fluorine-doped TiO2, 95Fluorine doping, 95Focused-ion-beam (FIB), 12Focused-ion-beam chemical vapor deposition

(FIB-CVD), 13, 14Fowler–Nordheim behavior, 404Frequency shift of mechanical resonant os-

cillation, 431Fullerene (C60), 100 ,412, 413, 422, 426Fully guided mode, 274, 275, 288, 289Fully leaky guided mode (FLGM), 274, 293,

301–305, 306, 307–309Fully leaky waveguide, 276, 309Functions of NC-AFM, 442

GaAs(110) cleaved surfaces, 442Gallium arsenide (GaAs), 194Glycocalix model, 209, 214Glycolipid monolayers, 209, 210Grating-coupling, 283–286, 310Guided mode technique, 273, 274, 279, 289,

290, 301, 306–308, 310Guided wave mode, 117–119, 124

Half-leaky guided mode (HLGM), 274, 293–301, 299, 300

Half-leaky waveguide, 275, 309Helical peptide, 253, 254, 263Helical peptide on gold, 264Helical peptide SAM (self-assembled mono-

layer), 256, 265Helical peptides on gold, 257Helical peptides on subphase, 258Helical peptides on water, 258

High performance NC-AFM, 436, 442High performance STM, 436Hole transporting material, 168, 172Home-built NC-AFM, 436Honeycomb nano-structure, 400Hopping conduction, 137, 139Hopping conduction model, 153HSQ, 10

Imidazolium, 91, 97In situ field-effect measurement, 157, 158,

162, 164, 176Incident photon to current efficiency (IPCE),

85Indium arsenide quantum dots (InAs QDs),

198Information storage, 21–24InP(110) cleaved surface, 431Interfacial stress rheometer (ISR), 214Iodine exchange (Grotthuss-type) mecha-

nism, 97Ionic head groups, 183–185, 187Ionic liquids, 98

Kelvin probe force microscopy (KFM), 449KFM/NC-AFM, 449

Langmuir–Blodgett (LB) film, 23, 24Langmuir–Blodgett technique, 183Langmuir film, 352Langmuir-monolayer, 353, 366Laser dye, 41, 43, 47Laser emission, 41, 43–45Lateral manipulation, 451Lateral resolution of NC-AFM, 433, 438Light-induced electron spin resonance

(LESR), 412–414Limiting molar conductivity, 97Lipid bilayer, 207, 209, 215, 218, 220Lipid membrane, 202, 216–218, 220Lipid monolayer, 215, 223Liquid crystal, 313–315, 347Liquid crystal layer, 279, 280, 286, 288, 289,

291–294, 304, 306, 307Living systems, 191, 192, 196Lone pair, 447Long-range force, 444Low-molecular-weight gelators, 96

Page 528: Nanotechnology and Nano-Interface Controlled Electronic Devices

Subject Index 511

Low-temperature NC-AFM (LT-NC-AFM),450

Maxwell displacement current (MDC), 353,355, 356, 359, 362, 365

Mechanical method, 449Mechanical resonant oscillation frequency,

432Membrane charge sensor, 221, 222Mesoporous, 85Metal oxide surfaces, 442Metal surfaces, 442Micro-system parts, 16Missing-dimer defects, 439Model cell membranes, 192, 193, 199, 203,

205, 215Molecular assembly, 253, 254Molecular devices, 32Molecular orientation, 258Molecular photonics, 105–107Molecular reorientation, 379Molecular systems, 32Molten salts, 98Monolayer, 353, 356, 357, 359, 364, 365Morphology of glycolipids, 211

n+-GaAs(110) cleaved surface, 449Nanoimprint-lithography (NIL), 6–9Nanometer, 65, 66, 80Nanophase-separated structure, 239, 242Nanostructure, 49Native cell membranes, 193, 203, 205, 215,

218, 223, 226Nernst equation, 92Noncontact atomic force microscopy (NC-

AFM), 429Noncontact region, 434

Open-circuit photovoltage, 84Optical guided mode, 277Optical second harmonic generation (SHG),

353, 355, 357, 359, 360, 362, 364, 365Optical waveguide, 273, 274, 279, 282, 286,

287Optoelectronic devices, 65, 76Optoelectronics, 65, 76Organic electronic device, 158Organic light-emitting diode (OLED), 133–

136, 139, 144, 145, 153

Organic semiconductor films, 157, 167Organic semiconductor materials, 159Organic semiconductors, 158Organic thin film, 159Orientation of helical peptides, 254Orientational control, 378Orientational order, 381, 382, 384Oscillation amplitude, 432Oxygen adsorbed Si(111)7×7, 443Oxygen vacancies, 95

Passivation, 194, 196, 197Photoalignment, 378, 379, 382, 384Photoalignment procedure, 379Photocarrier generation, 455, 457–459, 461–

463, 465, 467, 469, 471, 473, 475–477Photoconversion efficiency, 84Photodiodes, 66–68, 72–76, 79, 80Photoexcitation, 412, 420, 426Photoinduced anisotropy, 383Photoinduced electrochromism, 106, 109,

111Photoinduced mass migration, 378, 385Photoinduced mass transfer, 391, 393Photoinduced migration, 393Photoinduced orientation, 380Photoinduced surface relief grating, 386Photon localization, 47Photoorientation, 383Photoorientation behavior, 382Photovoltaic effect, 457, 460, 474, 476, 477Photovoltaic properties, 458Physical interaction, 439Polarons, 411, 412, 415, 416, 420, 422, 425,

426Poly(3-alkylthiophene) (PAT), 412, 414, 455Poly(3-hexylthiophene) (PAT6), 415, 418Poly(3-octadecylthiophene) (PAT18), 418,

424Poly(3-octylthiophene) (PAT8), 413, 426Poly(methylsiloxane), 100Poly(phenylene oxide-co-2-allylphenylene

oxide), 100Polyimide, 23, 24, 32Polyiodide, 96Polymer, 65–80Polymer brushes, 205, 207Polypeptide-based LB films, 233, 236, 242,

244, 249

Page 529: Nanotechnology and Nano-Interface Controlled Electronic Devices

512 Subject Index

Polypyrrole (PPy), 100, 101Polysaccharide films, 202–205, 226Polysilane, 378Polysilane chain, 378Polysilane film, 378, 393Prism-coupling, 282, 283, 286, 287, 303Pull-up process, 451Push-out process, 451

Radius of tip curvature, 433Rebonded dimer structure, 441Rebonded structure, 441Reflection absorption spectroscopy (RAS),

257, 258, 261, 264Regiorandom, 470Regioregularity, 457, 458, 477Room temperature nanoimprint-lithography

(RT-NIL), 7, 8Roughness factor, 85Rutile, 91

Scanning probe microscope (SPM), 21Scanning tunneling microscope (STM), 429Schottky junction, 457, 462, 466, 468–470,

476Self-assembled monolayer (SAM), 193, 194,

256, 258, 265, 443Self-organized two-dimensional patterning,

235, 236, 242Semiconductor surfaces, 429Semiconductor-oxide nanotubes, 400Short circuit photocurrent density, 84Short-range force, 444Si(100)1×1:2D dideuteride, 447Si(100)1×1:2H dihydride surface, 447Si(100)2×1 clean surface, 438Si(100)2×1:H monohydride surface, 438Si(111)5

√3×5

√3–Sb surface, 447

Si(111)7×7, 431Si(111)

√3×√

3–Ag surface, 443Signal-to-noise ratio, 432Single electron tunneling (SET), 31, 32, 35,

38Single electron tunneling characteristic, 34,

38Sol–gel, 401Solid–liquid phase transition, 483, 485, 495,

502, 506

Solvatochromism, 491Space charge limited current (SCLC), 136,

140, 145Spatial light modulation, 107, 116, 117Spatial resolution of NC-AFM, 433Spectroscopy, 362, 365Stimulus–response coupling, 234, 249, 252Structural color formation, 233, 236, 242,

244, 249Supramolecular structure, 49, 50Surface anchoring, 314, 315, 318–320, 322,

323, 326, 327, 331, 335, 345, 347Surface anchoring energy, 320Surface engineering of GaAs, 199Surface relief grating, 385Surface relief structure, 391Surface states, 95Surface tension, 405Synthetic polymer brushes, 205

Template, 401Thermochromism, 480, 485, 488, 489, 502Thin-film transistor (TFT), 159, 160, 168Three-dimensional mapping of atomic force,

443Three-dimensional nanostructure, 12, 14Thymine, 443TiO2, 83, 442Tip–sample distance, 433Tip-induced 1×1 novel structure, 448Tip-induced self-organized 2×1 novel struc-

ture, 448Transparent conducting glass, 88Trapping model, 86

Vacuum evaporation, 158Van der Waals force, 449Vertical manipulation, 450Vertical resolution of NC-AFM, 433, 437Viscoelasticity of oligosaccharides, 214

X-ray photoelectron spectroscopy (XPS),196, 197

Young modulus, 404

z-derivative image of capacitance, 449ZEP520, 4–6