Photomechanical Materials, Composites, and Systems: Wireless Transduction of Light into Work

�

� �

�

Photomechanical Materials,Composites, and Systems

�

� �

�

Photomechanical Materials,Composites, and Systems

Wireless Transduction of Light into Work

Edited by Timothy J. White

�

� �

�

This edition first published 2017© 2017 John Wiley & Sons, Ltd.

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, ortransmitted, in any form or by any means, electronic, mechanical, photocopying, recording orotherwise, except as permitted by law. Advice on how to obtain permission to reuse material fromthis title is available at http://www.wiley.com/go/permissions.

The right of Timothy J. White to be identified as the author(s) of this work has been asserted inaccordance with law.

Registered OfficeJohn Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030, USA

Editorial Office111 River Street, Hoboken, NJ 07030, USA

For details of our global editorial offices, customer services, and more information about Wileyproducts visit us at www.wiley.com.

Wiley also publishes its books in a variety of electronic formats and by print-on-demand. Somecontent that appears in standard print versions of this book may not be available in other formats.

Limit of Liability/Disclaimer of Warranty:The publisher and the authors make no representations or warranties with respect to the accuracyor completeness of the contents of this work and specifically disclaim all warranties; includingwithout limitation any implied warranties of fitness for a particular purpose. This work is soldwith the understanding that the publisher is not engaged in rendering professional services. Theadvice and strategies contained herein may not be suitable for every situation. In view of on-goingresearch, equipment modifications, changes in governmental regulations, and the constant flowof information relating to the use of experimental reagents, equipment, and devices, the reader isurged to review and evaluate the information provided in the package insert or instructions foreach chemical, piece of equipment, reagent, or device for, among other things, any changes in theinstructions or indication of usage and for added warnings and precautions. The fact that anorganization or website is referred to in this work as a citation and/or potential source of furtherinformation does not mean that the author or the publisher endorses the information theorganization or website may provide or recommendations it may make. Further, readers shouldbe aware that websites listed in this work may have changed or disappeared between when thisworks was written and when it is read. No warranty may be created or extended by anypromotional statements for this work. Neither the publisher nor the author shall be liable for anydamages arising here from.

Library of Congress Cataloguing-in-Publication Data

Names: White, T. (Timothy), editor.Title: Photomechanical materials, composites, and systems : wireless

transduction of light into work / edited by Timothy J. White.Description: Hoboken, New Jersey : John Wiley & Sons, Inc., [2017] | Includes

bibliographical references and index. | Description based on print versionrecord and CIP data provided by publisher; resource not viewed.

Identifiers: LCCN 2017001840 (print) | LCCN 2017012541 (ebook) | ISBN9781119123293 (Adobe PDF) | ISBN 9781119123286 (ePub) | ISBN 9781119123309| ISBN 9781119123309(cloth; pbk.) | ISBN 1119123305(cloth; pbk.)

Subjects: LCSH: Smart materials. | Polymers–Optical properties. |Polymers–Mechanical properties. | Nanocomposites (Materials)

Classification: LCC TA418.9.S62 (ebook) | LCC TA418.9.S62 P46 2017 (print) |DDC 620.1/9204295–dc23

LC record available at https://lccn.loc.gov/2017001840

Cover image: © eugenesergeev/GettyimagesCover design by Wiley

Set in 10/12pt Warnock Pro by SPi Global, Chennai, India

Printed in the United States of America

10 9 8 7 6 5 4 3 2 1

�

� �

�

v

Contents

List of Contributors xiPreface xv

1 A Historical Overview of Photomechanical Effects inMaterials, Composites, and Systems 1Toru Ube and Tomiki Ikeda

1.1 Introduction 11.1.1 Initial Studies of Photomechanical Effects in Materials 11.1.2 Research of Photomechanical Effects in Materials – 1950–1980 21.1.3 Research of Photomechanical Effects in Materials – 1980–2000 61.1.4 Photomechanical Effects Observed in Cross-Linked

Liquid-Crystalline Polymers – 2001–Present 91.1.5 Photomechanical Effects in Polymeric Materials and Composites

Systems since 2000 191.1.6 Classification 23

References 25

2 Photochromism in the Solid State 37Oleksandr S. Bushuyev and Christopher J. Barrett

2.1 Molecular Photoswitches in the Solid State 372.2 Molecular and Macroscopic Motion of Azobenzene

Chromophores 392.3 Photomechanical Effects 412.3.1 Photomechanical Effects in Amorphous Azo Polymers 422.3.2 Actuation in Liquid-Crystalline Polymers 432.3.3 Photosalient, Photochromic, and Photomechanical Crystals 492.4 Solid-State Photochromic Molecular Machines 542.4.1 Nanostructure Functionalization 552.4.2 Two-Dimensional Assemblies and Surface Functionalization 592.5 Surface Mass Transport and Phase Change Effects 622.6 Photochromic Reactions in Framework Architectures 65

�

� �

�

vi Contents

2.7 Summary and Outlook 68References 69

3 Photomechanics: Bend, Curl, Topography, and Topology 79Daniel Corbett, Carl D. Modes, and Mark Warner

3.1 The Photomechanics of Liquid-Crystalline Solids 813.2 Photomechanics and Its Mechanisms 823.2.1 Absorption, Photomechanics, and Bend Actuation 863.2.1.1 Photostationary Dye Populations and Mechanical Response 873.2.1.2 Dynamical Intensity and Dye Populations 883.2.1.3 Polydomain Photosolids 903.2.1.4 Photomechanics versus Thermal Mechanics upon Illuminating

Photosolids 913.3 A Sketch of Macroscopic Mechanical Response in LC Rubbers and

Glasses 923.4 Photo- and Heat-Induced Topographical and Topological

Changes 973.5 Continuous Director Variation, Part 1 973.6 Mechanico-Geometric Effects, Part 1 1003.7 Continuous Director Variation, Part 2 1003.8 Continuous Director Variation, Part 3 1033.9 Mechanico-Geometric Effects, Part 2 1063.10 Director Fields with Discontinuities–Advanced Origami! 1073.11 Mechanico-Geometric Consequences of Nonisometric

Origami 1103.12 Conclusions 110

References 112

4 Photomechanical Effects in Amorphous and SemicrystallinePolymers 117Jeong Jae Wie

4.1 Introduction 1174.2 Polymeric Materials 1194.3 The Amorphous Polymer State 1194.4 The Semicrystalline Polymer State 1214.5 Absorption Processes 1244.6 Photomechanical Effects in Amorphous and Semicrystalline

Azobenzene-Functionalized Polymers 1264.6.1 Influence of Crystallinity on Photomechanical Response of

Polyimides 1264.6.2 Backbone Rigidity 1284.7 Molecular Alignment 1324.8 Annealing and Aging 138

�

� �

�

Contents vii

4.9 Sub-Tg Segmental Mobility 1424.10 Cross-Link Density 1454.11 Concluding Remarks 146

References 148

5 Photomechanical Effects in Liquid-Crystalline PolymerNetworks and Elastomers 153Timothy J. White

5.1 Introduction 1535.1.1 What Is a Liquid Crystal Polymer, Polymer Network, or

Elastomer? 1535.1.2 How Are Liquid-Crystalline Polymer Networks and Elastomers

Prepared? 1545.1.2.1 Polysiloxane Chemistries 1545.1.2.2 Free Radical or Cationic Photopolymerization 1575.2 Optically Responsive Liquid Crystal Polymer Networks 1595.2.1 Historical Overview 1595.2.2 Photochromic and Liquid Crystalline 1625.2.3 Photomechanics 1645.3 Literature Survey 1655.3.1 Photomechanical Effects in Polysiloxane Materials and

Analogs 1655.3.2 Photomechanical Effects in Poly(meth)acrylate Materials and

Analogs 1665.4 Outlook and Conclusion 169

References 171

6 Photomechanical Effects in Polymer Nanocomposites 179Balaji Panchapakesan, Farhad Khosravi, James Loomis, and Eugene M.Terentjev

6.1 Introduction 1796.2 Photomechanical Actuation in Polymer–Nanotube

Composites 1806.3 Fast Relaxation of Carbon Nanotubes in Polymer Composite

Actuators 1866.4 Highly Oriented Nanotubes for Photomechanical Response and

Flexible Energy Conversion 1916.4.1 Highly Oriented Nanotubes/Nanotube Liquid Crystals 1916.4.2 Photomechanical Actuation of Oriented Nanotube

Composites 1976.4.3 Relaxation Behavior of Nanotube–Liquid Crystal Elastomers 2006.5 Photomechanical Actuation Based on 2-D Nanomaterial

(Graphene)–Polymer Composites 205

�

� �

�

viii Contents

6.6 Applications of Photomechanical Actuation inNanopositioning 213

6.6.1 Principle of GnP/Elastomer Photothermal Actuation 2146.6.2 Photomechanical-Actuation-Based Nanopositioning System 2186.6.3 GNP/PDMS Actuator Fabrication and Characterization 2186.6.4 Nanopositioner System Integration 2196.6.5 Kinetics of Photothermal Nanopositioners 2216.6.6 Useful Displacement versus Maximum Displacement 2226.6.7 Accuracy and Resolution 2236.7 Future Outlook 224

Acknowledgments 225References 225

7 Photomechanical Effects in Photochromic Crystals 233Lingyan Zhu, Fei Tong, Rabih O. Al-Kaysi, and Christopher J. Bardeen

7.1 Introduction 2337.2 General Principles for Organic Photomechanical Materials 2347.3 History and Background 2347.4 Modes of Mechanical Action 2407.4.1 Partial Reaction and Bimorph Formation 2407.4.2 Complete Transformation and Crystal Reconfiguration 2417.5 Photomechanical Molecular Crystal Systems 2427.5.1 Intramolecular Photochemical Reactions 2427.5.1.1 Ring-Opening/Closing Reactions 2427.5.1.2 Photoisomerization 2447.5.1.3 Photodissociation 2477.5.2 Intermolecular Photochemical Reactions 2487.5.2.1 [2 + 2] Photodimerization 2487.5.2.2 [4 + 4] Photodimerization 2507.5.3 Nonequilibrium Charge Distribution and Electronic Heating 2577.6 Future Directions 2607.6.1 Reaction Dynamics in Molecular Crystals 2607.6.2 New Materials 2617.6.3 Interfacing Molecular Crystals with Other Objects 2627.7 Conclusion 264

Acknowledgments 264References 264

8 Photomechanical Effects in Piezoelectric Ceramics 275Kenji Uchino

8.1 Introduction 2758.2 Photovoltaic Effect 2768.2.1 Principle of the Bulk Photovoltaic Effect 277

�

� �

�

Contents ix

8.2.1.1 “Bulk” Photovoltaic Effect 2778.2.1.2 Experimental Setup 2798.2.1.3 Current Source Model 2798.2.1.4 Voltage Source Model 2828.2.2 Effect of Light Polarization Direction 2858.2.3 PLZT Composition Research 2868.2.4 Dopant Research 2878.3 Photostrictive Effect 2888.3.1 Figures of Merit 2888.3.2 Materials Considerations 2898.3.3 Ceramic Preparation Method Effect 2908.3.3.1 Processing Method 2908.3.3.2 Grain Size Effect 2908.3.3.3 Surface/Geometry Dependence 2918.4 Photostrictive Device Applications 2948.4.1 Displacement Amplification Mechanism 2948.4.2 Photo-Driven Relay 2958.4.3 Micro-walking Machine 2958.4.4 “Photophone” 2978.4.5 Micro-propelling Robot 2978.5 Concluding Remarks 299

References 300

9 Switching Surface Topographies Based on Liquid CrystalNetwork Coatings 303Danqing Liu and Dirk J. Broer

9.1 Introduction 3039.2 Liquid Crystal Networks 3049.2.1 Photoresponsive Liquid Crystal Networks 3079.2.2 Photoinduced Surface Deformation 3079.2.3 Photoinduced Surface Deformation Preset by Patterned Director

Orientation 3119.2.4 On the Mechanism of Surface Deformation 3189.3 Conclusions 322

References 322

10 Photoinduced Shape Programming 327Taylor H. Ware

10.1 One-Way Shape Memory 32910.1.1 Photothermal 33110.1.2 Photochemical 33610.2 Two-Way Shape Memory 34310.2.1 Photothermal 344

�

� �

�

x Contents

10.2.2 Photochemical 35310.3 Summary and Outlook 358

References 358

11 Photomechanical Effects to Enable Devices 369M. Ravi Shankar

11.1 Introduction 36911.2 Analog Photomechanical Actuators 37111.3 Discrete-State (Digital) Photomechanical Actuators 37311.3.1 Binary Actuators 37411.3.2 Latency of Binary Actuators and Repetitive Actuation 37511.3.3 Multistable Implementations 38011.3.4 Beyond Bistable, Buckled Rods 38411.4 Photomechanical Mechanisms and Machines 387

References 388

12 Photomechanical Effects in Materials, Composites, andSystems: Outlook and Future Challenges 393Timothy J. White

12.1 Introduction 39312.2 Outlook and Challenges 39312.2.1 Breadth and Depth 39312.2.2 Beyond Bending: Mechanics Implementations 39412.2.3 Harvesting and Harnessing Light 39612.2.4 Speed is Limited 39612.2.5 Systems Design and Implementation 39812.2.6 Applications 39812.2.6.1 Optical Elements 39812.2.6.2 Morphing Shapes and Surfaces 40012.2.6.3 Actuation 40012.3 Conclusion 401

References 401

Index 405

�

� �

�

xi

List of Contributors

Rabih O. Al-KaysiDepartment of Basic Sciences,College of Science and HealthProfessionsKing Saud bin Abdulaziz Universityfor Health SciencesRiyadhSaudi Arabia

and

Ministry of National Guard HealthAffairsKing Abdullah International MedicalResearch CenterRiyadhSaudi Arabia

Christopher J. BardeenDepartment of ChemistryUniversity of California, RiversideRiverside, CAUSA

Christopher J. BarrettDepartment of ChemistryMcGill UniversityMontrealCanada

Dirk J. BroerDepartment of ChemicalEngineering and ChemistryInstitute for Complex MolecularSystemsTechnical University of EindhovenEindhovenNetherlands

Oleksandr S. BushuyevDepartment of ChemistryMcGill UniversityMontrealCanada

Daniel CorbettSchool of Chemical Engineering andAnalytical ScienceThe University of ManchesterManchesterUK

Tomiki IkedaResearch and Development InitiativeChuo UniversityTokyoJapan

�

� �

�

xii List of Contributors

Farhad KhosraviSmall Systems Laboratory,Department of MechanicalEngineeringWorcester Polytechnic InstituteWorcester, MAUSA

Danqing LiuDepartment of ChemicalEngineering and Chemistry, Institutefor Complex Molecular SystemsTechnical University of EindhovenEindhovenNetherlands

James LoomisDepartment of MechanicalEngineeringUniversity of AucklandAucklandNew Zealand

Carl D. ModesCenter for Studies in Physics andBiologyThe Rockefeller UniversityNew York, NYUSA

Balaji PanchapakesanSmall Systems Laboratory,Department of MechanicalEngineeringWorcester Polytechnic InstituteWorcester, MAUSA

M. Ravi ShankarDepartment of IndustrialEngineeringUniversity of PittsburghPittsburgh, PAUSA

Eugene M. TerentjevCavendish LaboratoryDepartment of PhysicsUniversity of CambridgeCambridgeUK

Fei TongDepartment of ChemistryUniversity of California, RiversideRiverside, CAUSA

Toru UbeResearch and Development InitiativeChuo UniversityTokyoJapan

Kenji UchinoInternational Center for Actuatorsand TransducersElectrical Engineering and MaterialsResearch InstituteThe Pennsylvania State UniversityUniversity Park, PAUSA

�

� �

�

List of Contributors xiii

Taylor H. WareDepartment of BioengineeringThe University of Texas at DallasRichardson, TXUSA

Mark WarnerCavendish LaboratoryDepartment of PhysicsUniversity of CambridgeCambridgeUK

Timothy J. WhiteDayton, OHUSA

Jeong Jae WieDepartment of Polymer Science andEngineeringInha UniversityIncheonSouth Korea

Lingyan ZhuDepartment of ChemistryUniversity of California, RiversideRiverside, CAUSA

�

� �

�

xv

Preface

Transduction of energy is pervasive within our modern society – examplesinclude the conversion of chemical energy to power the motion of an auto-mobile, harvesting wind to provide electric power to our homes, or capturingsolar radiation to power a communications satellite. The focus of this book isthe transduction of light (photons) into a mechanical output.Photomechanical effects in materials or composites are a subcategory of the

broader class commonly referred to as stimuli-responsive or “smart” materi-als. The focus of this book is on materials and composites that are sensitive tolight as the input energy stimulus. Light is compelling as an input energy sourcefor many reasons. Foremost of these reasons is the potential for speed. Youngstudents around the world are taught that nothing moves faster than light – itis the speed limit that defines our universe. Daily, we rely on the transmissionof light over long distances, which is a distinguished method for wireless andremote control of a system or subsystem in a device. Light can also be readilymanipulated to be polarized (linear or circular) as well as complex and evolvingpolarization vortices. Synthetic light, generated by lasers or LED, is increasinglydiverse inwavelength, spanning theUV to the infrared at ever-increasing powerlevels. All the aforementioned properties can very easily be turned on or off,reoriented, or spatially varied.These variations allow for a unique and unprece-dented level of control in generating distinguished mechanical responses. Putsuccinctly, light is a “smart” stimulus for “smart” materials.As detailed by the international collection of authors assembled here,

photomechanical effects in materials or material composites have beenobserved since ancient times in the various versions of the sundial. More than100 years ago, the American inventor Alexander Graham Bell was captivatedin part by the aforementioned properties of light and focused years of researchinto the “photophone,” after his earlier invention of the telephone. Seminalpapers that appeared in the 1960s and 1970s initiated a renaissance in the topicwhich has steadily grown into the practicing research community of today. In2016, more than 900 papers were published using the term “photomechanical”(or variants thereof )!

�

� �

�

xvi Preface

As will be evident throughout the book – photomechanical effects inmaterials and composites are a complex interplay of light, photochemistry,polymer chemistry and physics, and mechanics. Due to the breadth of thefundamental subject matter, the book begins with three introductory chapters.Chapter 1, by Ikeda and Ube, gives a high level and introductory survey ofthe generally topic to emphasize the historical evolution of the topic and tellthe unfolding story of the development and employment of these materials.Subsequently, Chapter 2 by Bushuyev and Barrett details the basics of pho-tochromism in the solid state. The foundational chapters are completed witha contribution from Corbett, Modes, and Warner, detailing the interplay ofphotochemistry and mechanics, with specific emphasis on anisotropic andpatterned material systems prepared from liquid-crystalline polymers.Thereafter, the book transitions into detailed treatments of the subclasses

of photomechanical materials including conventional polymers (Chapter 4by Wie), liquid-crystalline polymer networks and elastomers (Chapter 5 byWhite), crystalline solids (Chapter 7 by Bardeen and coauthors), and ceramics(Chapter 8 by Uchino) as well as a chapter on photomechanical effectsin nanocomposites (Chapter 6 by Panchapakesan, Khosravi, Loomis, andTerentjev). The book concludes with chapters detailing cross-cutting topics ofrecent interest including photoinduced topographical features (Chapter 9 byLiu and Broer), shape programming (Chapter 10 by Ware), actuating devices(Chapter 11 by Ravi Shankar), and an outlook (Chapter 12 by White).I am forever grateful to the wonderful collection of authors for taking their

time and spending their expertise on the chapters that follow. I would beremiss not to thank the editorial staff at Wiley for their help and assistancein navigating an endeavor such as this. Most of all, I thank my wife Jaymieand children Avery, Micah, and Beckett for their sacrifice in allowing for thisproject to go forward in my personal time away from an already overscheduledand full work week.I andmany of the authors of this book believe that these materials are quickly

defining and finding unique potential application opportunities. It is my hopethat this bookwill captivate aspiring scientists and peers in other research com-munities to join in this pursuit to further realize the promise that has captivatedso many for so long.

Tim WhiteDayton, OH

�

� �

�

1

1

A Historical Overview of Photomechanical Effectsin Materials, Composites, and SystemsToru Ube and Tomiki Ikeda

Research and Development Initiative, Chuo University, Tokyo, Japan

1.1 Introduction

Photomechanical effects in materials are a topic of considerable recentresearch. Many papers are continually appearing in top-ranked journalsreporting novel materials, demonstrations of distinctive mechanical outputs,and initial demonstrations of device utility. This book is a comprehensivereview of the material development, fundamental science (photochemistry,optics, and mechanics), and application of photomechanical effects in mate-rials. This chapter provides an overview of the historical development of thesimple yet captivating idea of photomechanical energy conversion inmaterials.In this way, the reader will have a general awareness of the interrelated natureof the topics and themes discussed throughout the subsequent chapters.

1.1.1 Initial Studies of Photomechanical Effects in Materials

Historians might argue that the first implementation of photomechanicaleffects in materials was the invention of the sundial by the ancients. It isinarguable, however, that humankind has sought to harvest this plentifulresource. Many of these pursuits have found their inspiration in nature inwhich countless species have adapted to use and leverage light-inducedmotility (photomechanical effects) to harvest more energy (sunflower), protectsensitive leaves (circadian rhythm plants), or even camouflage (chameleon,cephalopods).The emergence of the potential utility of photomechanical effects in the

modern era can largely be attributed to the famous American inventor Alexan-der Graham Bell and his work in the late 1800s [1]. After Bell invented thepractical telephone, he shifted his focus on the development of a photophoneto enable communication without the necessity of a conducting wire between

Photomechanical Materials, Composites, and Systems: Wireless Transduction of Light into Work,First Edition. Edited by Timothy J. White.© 2017 John Wiley & Sons, Inc. Published 2017 by John Wiley & Sons, Inc.

�

� �

�

2 1 A Historical Overview of Photomechanical Effects in Materials, Composites, and Systems

M

L

L

CS

H

PR

S

T

B

FR

LS

Figure 1.1 Schematic illustration of a photophone proposed by A. G. Bell. LS, light source;M, mirror; L, lens; H, heat absorber; S, sound; FR, flexible reflector; C, crystal; PR, parabolicreflector; B, battery; T, electroacoustic transducer.

a transmitter and a receiver (Figure 1.1). To accomplish this, Bell used acrystalline material (selenium) as a component of a receiver, which was con-nected in a local circuit with a battery and an electroacoustic transducer. Thesound emission changes depending on the state of light through a variationin resistance of selenium. The photophone Bell envisioned is the basis ofoptical communication and realized in recent times in practical applicationsenabled by the development of optical fibers and lasers [2]. Bell subsequentlyinvestigated nonelectronic photoresponsive receivers to make light audiblewithout the aid of electricity. He found that diaphragms of various substances(metals, rubbers, paper, etc.) produce sounds when irradiated with light. Thisphenomenon is explained in terms of a vibration of the diaphragm, which iscaused by a local, photoinduced temperature rise and a corresponding changein thermal expansion of the material. Recent examinations of photoacoustictomography extend upon this fundamental tenet pursued by Bell [3]. Accord-ingly, Alexander Graham Bell can be considered as the originator and “father”of photomechanical effects in materials in the modern era.

1.1.2 Research of Photomechanical Effects in Materials – 1950–1980

Stimuli-induced deformation of materials has attracted much attention sincethe 1950s. The most responsive form of these materials is a polymer gel, whichconsists of a cross-linked polymer network and solvent. Kuhn, Katchalsky,and coworkers demonstrated expansion and contraction of hydrogels con-taining carboxyl groups by successive addition of alkali and acid [4]. Thecarboxyl groups ionize and deionize depending on the pH, leading to thechange in intramolecular electronic repulsion and subsequent expansion andcontraction of polymer chains.This conformational change at amolecular scaleis translated to macroscopic deformation. Subsequently, various types of the

�

� �

�

1.1 Introduction 3

so-called smart materials have been developed, which deform when subjectedto stimuli such as heat, electricity, light, magnetic field, and humidity [5].Photoresponsive materials have potential advantages compared to these

other stimuli. Light is a comparably “smart” stimulus allowing for remoteand wireless controllability with spatial selectivity and also direct control ofresponsemagnitude via variation of intensity, wavelength, or even polarization.Initial research activities of photomechanical effects in polymeric materialswere undertaken in the 1960s. The general approach of these initial studiesremains largely unchanged today, focused on incorporating photoresponsivemoieties into polymeric or crystalline materials.By far, the most common approach to sensitizing polymeric materials to

light is to functionalize these materials with azobenzene. Azobenzene isa common dye molecule and widely known to photoisomerize between athermally stable trans isomer and a metastable cis isomer (Figure 1.2) [6].

UV

Vis

UV

Vis

UV

Vis

UV

Vis

(a)

(b)

(c)

(d)

R1

R2

R1

R1 R1

R2

R2

F2F2

F2F2

R1R1

F2F2

R5R5

R4R4R6 R6

R3R3

R2R2

R3R3

R2

R1 R2

NN

N

N+−O R2

NO2

R1

NO2O

O

O

O

O

OO

N N

SSS S

Figure 1.2 Typical photochromic molecules used to induce photomechanical effects:(a) azobenzene, (b) spiropyran, (c) fulgide, and (d) diarylethene.

�

� �

�


Generally, trans-azobenzene isomerizes to the cis isomer upon irradiationwith UV light, whereas cis-azobenzene reverts to the trans isomer uponirradiation with visible light or heating. The isomerization of azobenzeneproduces a variety of changes in properties such as molecular shapes andpolarity. Photochromic behavior and applications of azobenzene derivativeshave been actively studied since the isolation of the cis isomer in 1937 [7]. Thephotochemistry of azobenzene and other chromophores employed to generatephotomechanical effects is exhaustively detailed in Chapter 2.In 1967, Lovrien predicted that light energy could influence the confor-

mation of polymer chains if photochromic molecules such as azobenzenewere parts of polymers or bound to them [8]. In this seminal work, Lovrienproposed four strategies to achieve a conversion of light energy into mechan-ical energy. (i) Use of a polymer electrolyte solution containing azobenzenesin side chains (Figure 1.3a). trans-Azobenzenes in the side chains tend tocontract polymers by hydrophobic interaction. When irradiated with light,the hydrophobic interaction within the side chains decreases with trans–cisisomerization and results in a local expansion of the spacing of the polymerchains driven by Coulombic interaction. (ii) Use of solutions composed ofpolymer and azobenzene electrolytes (Figure 1.3b). In this approach, thepolymer chains are spaced by electronic repulsion between trans isomers,which Lovrien suggested would assemble on the chains. Upon trans–cisisomerization with light irradiation, the polymer chains could organize intoneutral coil conformation upon liberation of azobenzenes from chains. (iii)Incorporation of photoisomerizable groups in the backbone of polymerchains. (iv) Introduction of photoisomerizable cross-links so that light cangovern the distance between chains. Experimentally, Lovrein investigated thefirst two approaches: a polymer electrolyte solution containing azobenzenechromophores in the side chains and a polymer solution blended with azoben-zene electrolytes. In both systems, photoinduced changes in viscosity wereobserved. This effect is ascribed to the conformational change of the materialsystem, which was correspondingly amplified to macroscopic deformation orforce. Thereafter, van der Veen and Prins prepared a water-swollen polymergel containing a sulfonated azostilbene dye (chrysophenine) [9]. The presenceof cross-links enables the translation of microscopic changes in conformationinto macroscopic deformation of gels. These authors observed shrinkage asmuch as 1.2% upon irradiation with UV light.Photomechanical effects of dye-doped polymers were also observed in bulk

polymeric systems. Merian first reported the photoinduced deformation ofpolymer fibers containing photochromic molecules [10]. Azobenzene is acommon dye molecule, and in the course of using an azobenzene derivative todye hydrophobic fibers, Merian found that the dyed nylon fiber shrank about0.1% upon irradiation with light. He attributed this macroscopic dimensionalchange to the conformational change of the azobenzene moieties. Agolini

�

� �

�

1.1 Introduction 5

UV

Vis

(a)

(b)

trans-Azobenzene cis-Azobenzene

− −

−

−

−−

−

−

−

−

− −

UVVis

−

−−

−

−

−

−

−

−−

− −

Figure 1.3 Systems for photoinduced deformation of polymer chains proposed by Lovrien.(a) Polymer electrolyte functionalized with azobenzene moieties. (b) Blend solutioncomposed of polymer and azobenzene electrolytes.

and Gay observed macroscopic deformation of about 0.5% and measuredphotogenerated stresses when azobenzene-functionalized polyimide filmswere exposed to light [11]. Smets and de Blauwe reported deformation ofpolymer networks containing spirobenzopyran as photochromic cross-linkers,confirming that photomechanical effects in polymeric materials are notlimited to azobenzene chromophores [12]. The photomechanical responseof polymeric materials and gels prepared from conventional morphologies(amorphous, semicrystalline) is detailed in Chapter 4.

�

� �

�


In these early examinations of photomechanical effects in polymeric systems,the correspondingmechanismwas solely ascribed to photochemical processes.However, heat generated by nonradiative deactivation process could also causemacroscopic deformations of thesematerials.The importance of photothermalcontributions was first elucidated byMatejka et al. [13].The rise in temperaturewas shown to causemacroscopic deformation ofmaterials due to dilation and achange in elasticmodulus.They carefully investigated the force induced by irra-diation with light under constant strain for a cross-linked copolymer of maleicanhydride and styrene, which contains azobenzene groups in the side chains.The time evolution of the generated force was found to correlate directly withtemperature rather than the isomerization of azobenzene.Thus, photothermalcontributions in these materials, composites, and systems must be considered.Photomechanical effects can also be realized through photoelectrical pro-

cesses within inorganic solids [14]. In 1966, Tatsuzaki et al. reported photoin-duced strain in a single crystal of SbSI, which shows photoconductivity andferroelectricity [15]. This behavior is attributed to the combination of pho-tovoltaic effect and converse piezoelectric effect. When ferroelectric materi-als are irradiated with light, a high voltage is generated, which considerablyexceeds the band gap energy. Subsequently, mechanical strain is induced dueto the converse piezoelectric effect. The photoinduced contraction of this classofmaterials is often called photostriction. Photomechanical effects in ferroelec-tric ceramics of lanthanum-modified lead zirconate titanate (PLZT) have beenextensively studied. In 1983, Brody demonstrated photoinduced bending of abimorph consisting of two PLZT ferroelectric layers with different remanentpolarization [16]. The bending of the material is caused by the expansion ofone layer and the contraction of the other. Uchino applied the photomechan-ical response of PLZT to micro-walking machines driven by light [17, 18], asdetailed in Chapter 7. The machine has two legs of bimorph of PLZT plates,which are fixed to a plastic board. When the legs are alternately irradiated withlight, the machine moves similarly to an inchworm. Photomechanical effectsof inorganic solids have also been observed in polar semiconductors (e.g., CdSand GaAs crystals) and nonpolar semiconductors (e.g., Si and Ge crystals) [14].

1.1.3 Research of Photomechanical Effects in Materials – 1980–2000

In the 1980s and 1990s, considerable effort focused on enhancing themagnitude of the photomechanical output of gels and dry polymers. Largedeformation of photoresponsive gels was reported by Irie and Kungwatchakun[19]. The authors’ strategy was to utilize photoinduced variation in long-rangeelectrostatic (repulsive) forces rather than employ the microscopic shapechanges accompanying the conformational change of chromophores such asazobenzene. Toward this end, polyacrylamide gels functionalized with triph-enylmethane leuco derivatives were employed. These derivatives dissociate

�

� �

�

1.1 Introduction 7

UVVis

(a)

(b) (c) (d)

H2C CH

CH3

CH3CN

CH

−CN

H2C

H3C

H3C CH3

CH3H3C

H3C

NN

N N+

Figure 1.4 Photoinduced bending of an acrylamide gel containing triphenyl methane leucodyes under an electric field. (a) Photochromism of triphenyl methane leucocyanide.(b) Before irradiation. (c) Under irradiation with UV light. (d) Under irradiation in the reverseelectric field to that in (c). (Irie [20]. Reproduced with the permission of American ChemicalSociety.)

into ion pairs upon irradiation with UV light (Figure 1.4). The electrostaticrepulsion between photogenerated charges led to substantial swelling ofpolyacrylamide gels. Photoinduced reversible bending of rod-shaped gels wasobserved under an electric field applied perpendicular to the rod [20]. Thebending is attributed to the inhomogeneous deformation of the gel, whichdepends on diffusion of free counter ions derived by an electric field. Anothernotable work exploring photoresponsive gel systems was detailed by Suzukiand Tanaka [21], where they employed a poly(N-isopropylacrylamide) gel,which is known to undergo a volume change by thermal phase transition [22].The authors incorporated chlorophyllin in the side chains as a light absorber.Upon irradiation with visible light, the gel collapsed due to phase transitioninduced by a photothermal effect.The enhancement of photomechanical effects in bulk polymer systems

was comparably limited in this time period. Although the photoinduced

�

� �

�


deformation was observed in various polymers containing photochromicmoieties in cross-links [23] or main chains [24], the magnitude of strainremained small (typically <1%). Through these studies, it was generallyconcluded that the photoinduced change in molecular shape associated withconformational changes of photochromic groups such as azobenzene tendsto be accommodated by the local motion of flexible polymer chains. Thus, tomore efficiently translate molecular level events into the desired macroscopicmechanical response, the molecular systems should be densely packed andwell organized.During this era, research of thesematerials extended intomonolayer systems,

which are restricted in two dimensions, and the change in molecular shapecan be readily transferred to macroscopic deformation. Various azobenzenepolymers form monolayers at air/water interfaces when organic solutions ofazobenzene polymers are spread on the water surface (Langmuir technique).Photomechanical effects in the monolayers of polymers containing azoben-zene moieties were first reported by Blair et al. in 1980 [25, 26]. They preparedmonolayers of polyamide with azobenzene moieties in the main chain. Theycompared surface pressure–area curves of polyamides under UV-irradiatedand dark conditions.TheUV-irradiatedmonolayer showed a reduction in area,suggesting that polymers aremore contracted in cis forms.The in situ areamea-surement under UV irradiation with constant surface pressure showed rathercomplicated behavior. Depending on the applied surface pressure, monolay-ers exhibited either contraction or expansion. This behavior was understoodto indicate that the conformation of polymer chains strongly depends on thepreparation processes of the monolayers. The polymers with azobenzene moi-eties in the side chains were investigated as well. Malcolm and Pieroni pre-pared monolayers consisting of polypeptides with azobenzene moieties in 40%of the side chains [27]. The samples contracted upon irradiation with UV light.They speculated that the more extended trans form occupies a larger area inthe air/water interface compared to the cis form. On the other hand, Menzelet al. used polypeptide with azobenzene moieties in all of the side chains [28].Themonolayers expanded upon irradiation with UV light, which is opposite tothe result of Malcolm and Pieroni. The trans–cis isomerization of azobenzenemoieties leads to a large increase in dipole moment and a high affinity for awater surface (Figure 1.5a). Therefore, azobenzene moieties move to the watersurface with trans–cis isomerization, resulting in the increase in surface areaper monomeric unit. These two examples clearly indicate that photomechan-ical response can be very sensitive to the architecture of polymers. Seki andcoworkers extensively studied monolayers of poly(vinyl alcohol)s containingazobenzene side chains [30]. Upon irradiation with UV light, the film exhib-ited a rapid threefold expansion from the original area [31]. The clear in situobservation of the photoinduced deformation of the monolayer was enabledby Brewster angle microscopy (Figure 1.5b) [29].

�

� �

�

1.1 Introduction 9

0 s 500 s

UV

UVVis

(a)

(b)

Figure 1.5 Photoinduced deformation of polymer monolayers containing azobenzene.(a) Schematic illustration. (b) Photoinduced expansion of a monolayer of poly(vinyl alcohol)with azobenzene side chain observed by Brewster angle microscopy. (Seki et al. [29].Reproduced with the permission of American Chemical Society.)

1.1.4 Photomechanical Effects Observed in Cross-LinkedLiquid-Crystalline Polymers – 2001–Present

Further improvement of photomechanical effects in bulk polymeric materials,composites, and systems was enabled by the introduction of alignment and ori-entation of the photoresponsive molecules. The alignment and orientation, asin liquid-crystalline (LC) systems, allows for amplification of small changes atthemolecular level to yield largemacroscopic deformations.The science under-lying the improvements detailed in this section is the result of the convergenceof advances in the preparation of cross-linked LC polymers and insights intothe photochemistry of azobenzene in polymeric materials that were separatelypursued in extensive research studies in the past decades.

�

� �

�


The development of cross-linked liquid-crystalline polymers (CLCPs)(interchangeably referred to as liquid crystalline polymer networks or LCNsthroughout the book) was a new horizon for the research and developmentof stimuli-responsive materials [32]. LCs have many unique properties asself-organized anisotropic materials [33, 34]. Without any treatment, LCstend to form microdomains: the orientation is locally ordered in each domainbut random among different domains (often referred to as polydomain). Inlow-molecular-weight LCs, the orientational axis (director) can be easilycontrolled by substrates coated with alignment layers such as polyimidesthat are rubbed to induce shallow groves as well as to induce orientationof the aromatic chains. The alignment of LC molecules can be switched byexternal stimuli such as electric and magnetic fields. The employment oflow-molar-mass LC materials in displays is largely attributable to the surfacealignment of LCs and sensitivity of LCs to align to fields [35]. Combinationof the anisotropies of LCs in polymers leads to fascinating materials knownas liquid-crystalline polymers (LCPs). Orientation of mesogens (rod- ordisc-like parts responsible for liquid crystallinity) is strongly coupled with theconformation of polymer main chains. In CLCPs, this relation is extended tothe macroscopic shape of materials as predicted by de Gennes in 1975 [36, 37].It should be noted that CLCPs is a general term referring to materials withglass transition temperature (Tg) below room temperature (elastomers) andabove room temperature (glassy) [32].Finkelmann et al. initially prepared elastomeric CLCPs of polysiloxanes

containing phenyl benzoate mesogens in the side chains with polydomainorientation [38]. Later, this same group oriented the mesogens within thematerial upon uniaxial stretching of the polydomain films, elucidating for thefirst time the coupling between the alignment of mesogens and macroscopicshape [39, 40]. Later, they prepared monodomain films with homogeneouslyaligned mesogens by the two-step cross-linking method now referred to asthe “Finkelmann method” (Figure 1.6a) [41]. Monodomain CLCPs exhibitlarge contractions along the director orientation upon heating through theLC–isotropic phase transition (Figure 1.7). The film restores to the originalshape upon cooling. The deformation ratio (strain) of 40% in the original workhas ultimately been enhanced to 400% by modification of the macromolecularstructure, specifically the inclusion of main-chain mesogenic units [42, 43].Other CLCPs chemistries have been examined, including polyacrylates andpolymethacrylates with side-chain mesogens, initially reported by Zentel andReckert [44]. Low-molar-mass LC monomers and the employment of surfacealignment methods to induce highly oriented polyacrylate CLCPs were furtherdeveloped by Broer et al. (Figure 1.6b) [45–47]. Commonly, CLCPs of thistype are prepared from mixtures of monofunctional and bifunctional acrylatemonomers that are photopolymerized at elevated temperatures where themixtures exhibit LC phases (referred to in this chapter as the in situ method).

�

� �

�

1.1 Introduction 11

(a)

(b)

SubstrateAlignment layer

MesogenInitiator

Polymer

Polymerization

Crosslinking

Crosslinking

Stretching

Figure 1.6 Preparation methods of CLCPs with aligned mesogens. (a) Two-stepcross-linking. (b) In situ polymerization.

In this approach, the orientation of LCmonomers during the polymerization ismemorized and retained after cross-linking. Countless CLCPs have been syn-thesized by the two-step cross-linking, and in situ polymerizationmethods andtheir thermomechanical properties have been thoroughly investigated [42].Concurrent to these developments in the fundamental materials chem-

istry of CLCPs, methods to induce alignment changes of dye moleculesand liquid crystals by light were subject to intense study in the 1980s and1990s [48]. Photoinduced (phototropic) phase transitions were observed inlow-molecular-weight LC systems doped with azobenzene (Figure 1.8a) [49].When irradiated with UV light in a nematic phase, the LC–isotropic phasetransition occurred with trans–cis isomerization of azobenzene moieties.The LC phase was restored by irradiation with visible light or heating. Thisphenomenon is based on the LC nature of azobenzene moieties. Rod-liketrans-azobenzene stabilizes LC phases, whereas bent cis-azobenzene disturbsthem. In low-molecular-weight systems, the photoinduced isotropic phasereadily returns to the LC phase due to either diffusion of cis isomers fromthe irradiated sites or fast cis–trans thermal back isomerization. On the

�

� �

�


Isotropic Nematic

Cool

Heat

115 °C 90 °C 20 °C

Figure 1.7 Contraction andextension of a CLCP film inducedby temperature change. (Ohmet al. [42]. Reproduced with thepermission of John Wiley andSons.)

other hand, polymeric systems functionalized with mesogens and azobenzenemoieties show more stable isotropic phases because the diffusion of the cisisomers is limited and the disordered states of chromophores remain evenafter the cis–trans thermal back isomerization [50, 51]. The photoinducedisotropic phase can be maintained stable for a long time (more than 10 years)below Tg of the polymer. Therefore, LCPs functionalized with azobenzeneshave been extensively studied as photomemory materials. Furthermore,polymers with azobenzene moieties in all monomer units were found toshow rapid photoinduced phase transition [52]. Thus, photoisomerizationof azobenzene molecules can be amplified into the alignment change of thewhole system. Photoinduced phase transition has been investigated for various

�

� �

�

1.1 Introduction 13

Figure 1.8Photomodulation ofmolecular alignment.(a) Photoinduced phasetransition.(b) Photoalignmentinduced by linearlypolarized light (LPL).(c, d) Surface-mediatedphotoalignment.

UV

Vis

LPL

LPL

LPL

LPL

UV

Vis

(a)

(b)

(c)

(d)

photochromic molecules such as spiropyran, fulgide, and diarylethene inaddition to azobenzene [48].A related but distinct method to generate orientation (and eventually pho-

tomechanical responses) in azobenzene materials employs linearly polarizedlight (LPL) typically in the blue-green region of the electromagnetic spectrum(Figure 1.8b). When irradiated with light in this portion of the electromagneticspectrum, trans-azobenzene molecules parallel to the polarization directionof the LPL are activated effectively while molecules perpendicular to thepolarization direction are largely insensitive to the incident LPL. Statisti-cally, trans-azobenzenes align perpendicular to the polarization direction ofthe actinic light after repeating the trans–cis and cis–trans (referred to astrans–cis–trans) cycles. This reorientation of the azobenzene chromophoreto LPL is sometimes referred to as the Weigert effect [48, 53]. Todorov et al.first succeeded in controlling the alignment of dyes embedded in polymer

�

� �

�


systems [54]. Anisotropic orientation of dyes was induced with high sensitivitywhen the dye-doped polymers are irradiated with LPL. In dye-doped systems,the photoinduced dichroism is readily retained even in the dark. Natansohnand coworkers created stable birefringence in amorphous polymers at ambienttemperatures below Tg by attaching the azobenzene moieties to polymers withcovalent bonds [55].During the course of the studies on alignment change of azobenzene-

containing polymers, a surprising optical effect was discovered in thin films.Separately but concurrently, Natansohn/Rochon [56] and Tripathy/Kumar[57] found that polymer thin films with azobenzene chromophores in the sidechains show spontaneous surface deformation under irradiation with visiblelight [58]. By exposing an azobenzene polymer thin film to the interferencepattern at room temperature (well below Tg), the films show sinusoidal surfacetopography with periodicity of hundreds of nanometers now referred to as asurface relief grating.The formation of patterns implies that themass transportat a microscopic level is induced by light.Another way to control alignment of molecules by light has been developed

by Ichimura and Seki [59, 60]. They prepared azobenzene monolayers on glasssubstrates by silane coupling agents and dispersed photoinert LC moleculeson the monolayers. Upon irradiation with light, the alignment of LCmoleculeschanged following the reorientation of surface azobenzenes (Figure 1.8c and d).Such photoactive layers are called a “command surface.” Using this technique,photoinducedmovement of liquid on coated substrates has been achieved [61].The development of methods to amplify microscopic changes into macro-

scopic deformation and to induce alignment changes of molecules by lightenabled the development of bulk polymer systems showing large photome-chanical effects [62–68]. In 2001, Finkelmann et al. incorporated photochromicmoieties into CLCPs to add a new function to the previous studies of thermo-mechanical effects in CLCPs [62]. They synthesized polysiloxanes containingphenyl benzoate mesogens and azobenzene moieties in the side chains andcross-links. Monodomain films with homogeneously aligned mesogens wereprepared by the two-step cross-linking method. Under irradiation with UVlight, the film contracted as much as 20% along the alignment direction ofmesogen with trans–cis isomerization of azobenzene moieties (Figure 1.9).After the UV light was turned off, the sample returned to the initial shapethrough thermal cis–trans isomerization. Ikeda and coworkers preparedcross-linked polyacrylates containing azobenzene moieties in all monomerunits [66–68]. The monodomain films showed three-dimensional bendingwhen irradiated with UV light (Figure 1.10a) [67, 68]. Due to the high con-centration of azobenzene in the chemistry employed by Ikeda, most of thephotons are absorbed near the surface of the sample, which localized themechanical response to the surface of the film, resulting in bending analogousto a bimetallic strip. Moreover, the polydomain orientation of the materials

�

� �

�

1.1 Introduction 15

UV

Dark

0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0

Time (min)

0.00

0.02

0.04

0.06

0.08

0.10

0.12

0.14

0.16

0.18

0.20

0.22

Contr

action fra

ction

0.0 50.0 100.0 150.0 200.0 250.00.00

0.02

0.04

0.06

0.08

0.10

0.12

0.14

0.16

0.18

0.20

0.22

Figure 1.9 Photoinduced contraction and extension of a CLCP film. (Finkelmann et al. [62].Reproduced with the permission of American Physical Society.)

allowed for the direction of the deflection of the films to be dictated by linearlypolarized UV light (Figure 1.10b) [66]. These two seminal works demonstratedfor the first time large-amplitude photomechanical effects, enabled by assimi-lating the photoresponsivity of azobenzene with the orientation and alignmentprovided by CLCPs.Although the photoinduced deformation of the glassy CLCPs prepared

by Ikeda was observed at temperature higher than Tg in the initial reports,it has been found that CLCPs show photomechanical effects even in glassymaterials [69]. The detailed mechanism of photomechanical effects is stillunder discussion [70]. The alignment change by LPL through trans–cis–transcycles has also been applied to photomechanical systems [71]. Under optimalconditions, LPL induces an alignment change of the azobenzene mesogens,which then results in anisotropic contraction or expansion of surface regionand subsequent deformation of the film depending on the polarization state oflight.Afforded by the unique properties of light, specifically the spatiotemporal

control of intensity, polarization, and wavelength, the mechanical responseof these materials can be all-optically triggered, adjusted, and erased. Fur-thermore, other sophisticated motions such as twisting [71], oscillation [72],

�

� �

�


Light

(a)

(b)

UV

Vis

Rubbing direction

Reaction

UV UV

Vis Vis

Substrate

Substrate

366 nm

366 nm

366 nm

366 nm

>540 nm

>540 nm

>540 nm

>540 nm

−90°

−135° −45°

0°

Figure 1.10 Photoinduced deformation of CLCP films. (a) Bending of a monodomain film.(b) Direction-selective bending of a polydomain film by linearly polarized light. (Yuet al. [66]. Reproduced with the permission of Nature Publishing Group. Ikeda et al. [67].Reproduced with the permission of John Wiley and Sons.)

�

� �

�

1.1 Introduction 17

rotation [73], and translation (motion) [74] have been observed in CLCPs(Figure 1.11) [96–103]. These manifestations of the photomechanical responseof these materials have been shown to strongly depend on the initial alignmentof mesogens and polarization state of incident light.As discussed, the initial alignment of the mesogens can be easily controlled

using glass cells coated with adequate alignment layers such as polyimides toprepare CLCPs by the in situ method. CLCP films functionalized with azoben-zene were prepared in two typical alignment modes, parallel (homogeneous)and normal (homeotropic) and, as expected, showed distinct mechanicalresponses to light irradiation [104]. Specifically, upon irradiation, a CLCP ina planar alignment bent toward the actinic light source while the CLCP filmprepared in a homeotropic alignment bent away from the light source. This isexplainable as the UV light causes the surface of the planarly aligned CLCPfilm to contract while a CLCP with homeotropic alignment expands, resultingin the opposite bending direction. CLCPs prepared with hybrid orientationsreferred to as splay (90∘ twist from a planar to a homeotropic orientation) ortwisted nematic (90∘ twist in planar-to-planar orientation) alignment were alsoinvestigated [105].These films bent toward an actinic light source if the surfaceof the film near the light source had a homogeneous alignment of mesogens.The bending motions of these films were faster and larger than those of thefilms with uniaxial planar alignment because the top and the bottom layers ofthe films cooperate such that the exposed surface undergoes a contraction,while the back surface expands due to the variation in the orientation of thematerial induced by the twist in the director profile.The performance of CLCPs can be enhanced and extended by generating

multimaterial laminates where photoresponsive materials are used to localizea mechanical response to incident light. Photoresponsive material systemswith good mechanical properties could be prepared through laminationof CLCP layers on flexible plastic sheets such as polyethylene [73, 76]. Alight-driven plastic motor was fabricated with a laminated film and two pulleys(Figure 1.11c) [73]. Simultaneous irradiation with UV and visible light ledto rotation of the belt and the pulleys. The bending of the UV irradiatedpart is supposed to produce torque on the small pulley, which results in therotation. Additionally, CLCP films can be laminated on multiple parts of thepolymer substrate (Figure 1.11e and f) [76, 77, 98]. This lamination enabledarbitrary motions similarly to a robotic arm. The connection between thephotoactive layer and the plastic sheet was reinforced with electron beamirradiation [106]. CLCP films have also been combined with functional layersshowing photon upconversion, which enables photoactuation by red andnear-IR light (Figure 1.11g) [78, 107]. Recently, the photoresponsive andmechanical properties of these material systems were further improved bypreparing interpenetrating polymer networks of CLCPs and amorphouspolymers [108].

�

� �

�


(a) (b) (c) (d)

(e) (f) (g) (h)

(j) (k) (l)

(n) (p)

(v)(t)

(i)

(m)

(q) (r)

(u) (w)

(x) (y) (z)

(s)

(o)

Figure 1.11 Various three-dimensional motions of CLCP systems induced by light.(a) Oscillation [75]. (b) Swimming [74]. (c) Light-driven plastic motor [73]. (d) Inchwormwalk [76]. (e) Robotic arm [76]. (f ) Manipulation of an object [77]. (g) Actuation throughtissues [78]. (h) Gripper [128]. (i) Crawling up [128]. (j) Adaptive liquid lens [80]. (k) Localizedactuation [81]. (l) Tactile device [82]. (m) Heliotropism [83]. (n) Microparticle [138].(o) Artificial cilia [85]. (p) Pillar array [86]. (q) Size-changeable pores [87]. (r) Fiber [88].(s) Micropump [89]. (t) Snap-through [90]. (u) Deformation into cone [114]. (v) Accordionfolding [115]. (w) Checkerboard pattern [115]. (x) Photoswitchable stripes [116]. (y) Dynamic3D finger print [79]. (z) Winding of spring [118]. (See color plate section for the colorrepresentation of this figure.)

�

� �

�

1.1 Introduction 19

A variety of novel CLCPmaterials have been created with the recently devel-oped methods to control alignment of mesogens in films with complex order.As described earlier, the motion of CLCPs can be programmed by controllingthe initial alignment of mesogens. Programming the local anisotropy in CLCPsleads to complex deformation. In particular, surface-mediated photoalignmenttechnique allows for precise control of molecular alignment without genera-tion of dust and static electricity. These benefits have motivated considerableinterest and research at their use as alignment layers in LC displays [109–111].In this alignment method, substrates coated with photoaligning materials areirradiated with LPL to define a preferential alignment direction. Importantly,local exposures using photomasks realize patterned alignment, which can beas small as 1 μm [111–113]. Broer and coworkers prepared CLCP films withcomplex order applying the surface-mediated alignment technique [114]. Thedirector in the alignment layer was controlled by irradiationwith linearly polar-ized UV light through a photomask while rotating a substrate and a polarizer.Deformation of CLCP films into cone and saddle shapes was observed depend-ing on the alignment patterns upon irradiationwith IR light (Figure 1.11u).Thisprocedure was applied to three-dimensional control of molecular alignmentas well (Figure 1.11v and w) [115]. Complicated patterns have also been pro-duced with chiral nematic LCs by electric field or self-assembly (Figure 1.11xand y) [79, 116]. White and coworkers demonstrated that precise control ofalignment at amicroscopic scale enables photoinduced changes in surfacemor-phology of CLCP films [117]. Furthermore, winding and unwindingmotions ofpolymer springs were achieved using CLCPs with twisted nematic alignment(Figure 1.11z) [118]. The ability to locally orient CLCP systems is a topic ofintense current research and will be detailed in later chapters [119, 120].

1.1.5 Photomechanical Effects in Polymeric Materials and CompositesSystems since 2000

Concurrently, researchers also developed and characterized materials andcomposite systems designed to generate photomechanical effects employingphotothermal processes. Most notably, carbon nanotubes (CNTs) havebeen extensively utilized as light absorbers. Photomechanical effects inpolymer/CNT composites were first reported in non-LC systems. Vaia andworkers dispersed multiwalled carbon nanotube (MWCNT) in thermo-plastic elastomers and irradiated the prestrained composite films with IRlight [121]. The photothermal conversion by CNTs led to the melting ofstrain-induced polymer crystallites and then macroscopic deformation of thefilms. Terentjev and coworkers investigated the photoactuation mechanismin polymer/CNT composite systems in detail. Specifically, these authorsreported that polymer/CNT composite films show mechanical effects evenwithout mesogens when CNTs are uniaxially aligned [122, 123]. Moreover,

�

� �

�


pristine CNT films without polymer matrices also show photoinduced stress[124]. The photomechanical effects in polymer/CNT composites were greatlyenhanced by the use of CLCPs as matrices after the development of methods todisperse CNTs homogeneously [84, 125, 126]. Chen and coworkers succeededin introducing single-walled carbon nanotube (SWCNT) in CLCP matricesusing a conjugated polymer as a dispersing agent [127, 128]. Upon irradiationwith IR light, the composite material system showed large deformation thatwas synchronized with a rise in temperature of the film. The heat generatedby photothermal conversion produced by light absorption of the CNTsalters the alignment of mesogens. The deformation can be either two- orthree-dimensional depending on the initial alignment of mesogens and tem-perature distribution in the sample. After the actinic light is switched off, thesample returns to the initial shape as its temperature decreases. The homoge-nous CNTs/CLCP composites were also obtained by functionalizing matrixpolymers with pyrene [129–131]. The preparation and photomechanicalresponse of composites systems are detailed in Chapter 6.Light as an actinic stimulus is an especially attractive approach for actuation

of micro-sized samples. The remote control has been employed to generatephotomechanical effects in microspheres of amphiphilic polymers in solution,where Wang and coworkers prepared colloidal spheres of amphiphilic poly-mers containing azobenzenemoieties and hydrophilic groups [132–134]. Uponirradiation with linearly polarized visible light, the microspheres deformedto anisotropic ellipsoids. As the type of chromophores and the irradiationcondition were similar to those of surface mass transport systems, the defor-mation is attributed to the diffusion of polymer chains in each microsphere.Zhao and coworkers prepared micelles of amphiphilic block copolymerscomposed of hydrophilic poly(acrylic acid) and hydrophobic polymethacrylatecontaining azobenzene moieties [135]. Upon irradiation with UV and visiblelight, reversible dissociation and formation of the micelles were observed.Thisbehavior is caused by the change in hydrophobicity of azobenzene moietieswith photoisomerization, which alters their aggregation states. Furthermore,Li and coworkers succeeded in photoinduced deformation of polymersomesof amphiphilic block copolymers [136]. The mechanism of large deformationin bilayer structures is essentially the same as that in monolayer systems. Inaddition, recent development of replica molding and microfluidic techniquesenabled the fabrication of micrometer-sized CLCP particles (Figure 1.11n)[42, 137, 138]. These photoresponsive microparticles could find utility inenabling targeted drug delivery in nanomedicine [134].The light-induced transformation of self-assembled structures has also been

observed in thin films of block copolymers. Seki and coworkers preparedthin films of a block copolymer composed of poly(butyl methacrylate) andpolymethacrylate containing azobenzene moieties [139]. The films showedcylinder structures derived by nanophase separation of block copolymers. The

�

� �

�

1.1 Introduction 21

alignment of the nanocylinders depended on film thickness: the homeotropicalignment was observed in films with thickness greater than 70 nm, whilethe planar alignment was observed in thinner films. When the homeotropicfilms were irradiated with interference light, a periodic topography appearedthrough surface mass transport. The cylinders in the irradiated area showedplanar orientation, which was induced with a decrease in thickness. Thenanostructures of block copolymer films composed of poly(ethylene oxide)and polymethacrylate containing azobenzene moieties have also been inves-tigated [140]. The films showed hexagonally packed nanocylinder structureswith periodicity around 10 nm, which align orthogonally to the substrate. Theorientation of the cylinder was controlled by irradiation with LPL. Thus, thephotoinduced deformation at a mesoscale has been achieved [141–143].The influence of light upon orientation is not limited to CLCPs. Pho-

tomechanical responses have been documented in single crystals preparedfrom photoresponsive materials [144–147]. In 2006, Bardeen and coworkersreported photoinduced deformation of single crystals of 9-tert-butylanthroatethrough [4+ 4] photodimerization [148]. The change in crystal structure isdirectly transferred to the deformation of nanorods (Figure 1.12a). Irie andcoworkers found that single crystals composed of diarylethenes deform withphotoisomerization [149]. In this case, deformation is optically reversible byirradiation with UV and visible light. A thin, rectangular plate-like crystaldeformed to a lozenge shape (Figure 1.12b), which is consistent with thecrystal structure observed by X-ray diffraction. Moreover, a rod-like crystalbent toward the UV light source upon exposure through the shrinkage ofthe illuminated area. Various 3D motions have been observed in crystal sys-tems. The performance has been enhanced by the modification of molecularstructures. The size and toughness of crystalline materials were improvedby preparing a material by cocrystallization of a diarylethene derivativewith perfluoronaphthalene (Figure 1.12c) [150]. Upon irradiation with UVlight, cantilevers prepared from this cocrystal were shown to lift metal balls,which were 200–600 times heavier than the cantilever. The strong couplingof molecular and macroscopic shapes in crystal systems is expected to bringabout high energy conversion efficiency.In a similar way, photomechanical effects in high-performance polymeric

materials have also been examined.White and coworkers prepared amorphousand semicrystalline polyimides containing azobenzene moieties in cross-linksor main chains [151, 152]. By irradiation with LPL in the blue-green region ofthe electromagnetic spectrum, the films showed bending behavior similarly toCLCPs. A drastic change in the alignment of azobenzenemoieties enables largemotions even in amorphous systems. Compared with classical amorphous sys-tems before the development of photoresponsive CLCPs, various factors suchas molecular structures, irradiation condition, and penetration depth of light

�

� �

�


UV

Vis

(b)

(c)

UVfromright

Visfromright

(a)

UV

S

F2

F2 F2

F2F2F2

S S SEt

Et

2

t-Bu

t-Bu

t-Bu

−O

−O

OO

OO−

Et

Et

10 μm

1 mm

10 μm

Figure 1.12 Photoinduced deformation of organic crystals. (a) Anthracene ester. (Al-Kaysiet al. [148]. Reproduced with the permission of American Chemical Society.) (b)Diarylethene. (Kobatake et al. [149]. Reproduced with the permission of Nature PublishingGroup.) (c) Cocrystal of diarylethene and perfluoronaphthalene. (Morimoto and Irie [150].Reproduced with the permission of American Chemical Society.)

�

� �

�

1.1 Introduction 23

(balance between concentration of chromophores and sample thickness) havebeen modified.Photomechanical effects in materials have also been employed to generate

shape memory in polymeric materials [91]. Shape memory materials canbe deformed and fixed into a temporary shape and recover their originalpermanent shape upon exposure to an external stimulus. Lendlein et al.prepared polymers containing covalent cross-links and cinnamic acid groups.A flat film (permanent shape) was stretched by mechanical force, and bothsides of the film were evenly irradiated with UV light (𝜆> 260 nm) to activatecross-linking through [2+ 2] cycloaddition reaction of cinnamic acid groupsto fix a temporary elongated shape. After the external stress was released, thefilm remained in the elongated form. Irradiation of the elongated form withUV light at 𝜆< 260 nm under ambient condition induced cleavage of newlyformed photosensitive cross-links, allowing the film to recover its permanentshape. Other temporary shapes such as a corkscrew spiral shape were alsoproduced. Moreover, similar photoinduced shape memory effect was alsoobserved in CLCPs through realignment of mesogens and polymer chainsunder external force and irradiation with visible light [92]. Light-activated andmanipulated shape memory in polymeric materials is detailed in Chapter 10.Some of the glasses composed of low-molecular-weight dyes have been found

to show photomechanical effects. Photochromic molecules containing bulkysubstructures form amorphous films by spin coating and vacuum deposition.Nakano et al. created low-molecular-weight azobenzene glasses, which formsurface relief gratings upon irradiationwith LPL [93, 94]. Furthermore, bendingand translational motions of the photochromic molecular glasses were demon-strated [95, 153].Recent advances in carbonmaterials are opening new fields of photomechan-

ical effects. Kobayashi and Abe showed that magnetically levitating graphiteplaced on NdFeB permanent magnets can be moved by photoirradiation [154].This behavior is attributed to photothermally induced changes in the magneticsusceptibility of the graphite. They observed the rotation of the graphite discwith a speed over 200 rpm upon irradiation with sunlight. Chen and cowork-ers reported photoinduced propulsion of a bulk graphene-basedmaterial [155].Exposure to visible light leads to the emission of energetic electrons, whichpush the sample in the propagation direction of the laser beam.

1.1.6 Classification

We summarize this chapter by classifying photomechanical effects fromseveral points of view. The first classification can be made based on theactuation mechanisms: photochemical, photothermal, and photoelectric pro-cesses. Photochemical responses in the preparation of materials from organicchromophores such as azobenzene, spiropyran, and diarylethene have been

�

� �

�


reported. Photothermal processes have long been observed in dye-containingsystems and recently enhanced in nanocomposites with nanoparticles andnanocarbons. Photochemical processes are often accompanied by photother-mal effects, which should be carefully observed. Photoelectrical processeshave been mainly observed in inorganic solid systems such as ceramics.As evident in this chapter, photomechanical effects in many classes of

materials have been reported (Table 1.1). Amorphous and semicrystallinepolymers containing dyes have been extensively investigated. Variationsinclude dye-doped (guest–host) systems or polymer matrices with pho-toresponsive units covalently bonded in the side chains, main chains, andcross-links. Gel is a representative soft material, which can include electrolytesto be deformed by Coulombic interaction. CLCPs show large photomechanicaleffects due to cooperative motion afforded by the order of the material system.Nanocomposites are typically composed of polymer matrix and nanoparticlesor nanocarbons. Organic crystals, low-molecular-weight glasses, and carbonmaterials are novel types showing photomechanical effects. Inorganic systemsshowing photoelectrical effects have long been investigated.The shape of the materials would be significant in practical use (Table 1.2).

Films are basic forms in studying the photomechanical effects, the thicknessof which is typically <200 μm. The films or sheets with rectangular shapes are

Table 1.1 Classification based on the types of materials.

Material Example of compounds References

Polymer/dyeAmorphous or semicrystallinepolymer

Polyimide containing azobenzenein the main chain

[151, 152]

Gel Polyacrylamide functionalizedwith triphenylmethane

[19, 20]

PNIPAM functionalized withchlorophyllin

[21]

CLCP Cross-linked polysiloxanefunctionalized with azobenzene

[62, 63]

Cross-linked polyacrylatefunctionalized with azobenzene

[65, 66, 69, 71]

Nanocomposite CLCP doped with CNT [127–131]Organic crystal Single crystal of diarylethene [149]Inorganic solid PLZT ceramics [16–18]Low-molecular-weight glass Azobenzene glass [93–95, 153]Carbon material CNT [124]

Graphite on magnet [154]

�

� �

�

References 25

Table 1.2 Classification based on shapes of the materials.

Shape Example of photomechanical effects References

Film Deformation of various types of materialsFiber Bending of CLCP containing azobenzene [88]Thin film Surface mass transport in azobenzene polymer film [56–58]

Photoalignment of nanostructure formed with blockcopolymer

[139–143]

Monolayer Contraction and expansion of polymer containingazobenzene

[25–31]

Microparticle Deformation of colloidal sphere of amphiphilic polymer [132–134]Deformation of CLCP microparticle [138]

also called cantilevers. Fibers are also typical forms of polymers with cylindri-cal symmetry.Thin films with thickness in the order of 1–100 nm on substratesshowing surface mass transport have been investigated. Monolayers typicallyform at the air/water interface and tend to show large photomechanical effectsdue to the restriction in two dimensions.Microparticles are an emerging geom-etry for photomechanical effects, leveraging the wireless controllability of light.

References

1 Bell, A.G. (1880) On the production and reproduction of sound by light.American Journal of Science, s3-20, 305–324.

2 Vacca, J.R. (2006) Optical Networking Best Practices Handbook, John Wiley& Sons, Inc., Hoboken.

3 Wang, L.V. and Hu, S. (2012) Photoacoustic tomography: in vivo imagingfrom organelles to organs. Science, 335, 1458–1462.

4 Kuhn, W., Hargitay, B., Katchalsky, A., and Eisenberg, H. (1950)Reversible dilation and contraction by changing the state of ionizationof high-polymer acid networks. Nature, 165, 514–516.

5 Urban, M.W. (ed.) (2011) Handbook of Stimuli-Responsive Materials,Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim.

6 Bandara, H.M.D. and Burdette, S.C. (2012) Photoisomerization in differentclasses of azobenzene. Chemical Society Reviews, 41, 1809–1825.

7 Hartley, G.S. (1937) The cis-form of azobenzene. Nature, 140, 281.8 Lovrien, R. (1967) The photoviscosity effect. PNAS, 57, 236–242.9 van der Veen, G. and Prins, W. (1971) Photomechanical energy conversionin a polymer membrane. Nature Physical Science, 230, 70–72.

�

� �

�


10 Merian, E. (1966) Steric factors influencing the dyeing of hydrophobicfibers. Textile Research Journal, 36, 612–618.

11 Agolini, F. and Gay, F.P. (1970) Synthesis and properties of azoaromaticpolymers. Macromolecules, 3, 349–351.

12 Smets, G. and de Blauwe, F. (1974) Chemical reactions in solid polymericsystems. Photomechanical phenomena. Pure and Applied Chemistry, 39,225–238.

13 Matejka, L., Dušek, K., and Ilavský, M. (1979) The thermal effect in thephotomechanical conversion of a photochromic polymer. Polymer Bulletin,1, 659–664.

14 Kundys, B. (2015) Photostrictive materials. Applied Physics Reviews, 2,011301.

15 Tatsuzaki, I., Itoh, K., Ueda, S., and Shindo, Y. (1966) Strain along c axis ofSbSI caused by illumination in DC electric field. Physical Review Letters,17, 198–200.

16 Brody, P.S. (1983) Optomechanical bimorph actuator. Ferroelectrics, 50,27–32.

17 Uchino, K. (1989) Micro walking machines using piezoelectric actuators.Journal of Robotics and Mechatronics, 1, 124–127.

18 Uchino, K. (2010) Photostrictive actuators using piezoelectric materials, inAdvanced Piezoelectric Materials (ed. K. Uchino), Woodhead Publishing,Ltd., Cambridge, pp. 599–627.

19 Irie, M. and Kungwatchakun, D. (1984) Photoresponsive polymers.Mechanochemistry of polyacrylamide gels having triphenylmethane leucoderivatives. Makromolekulare Chemie, Rapid Communications, 5, 829–832.

20 Irie, M. (1986) Photoresponsive polymers. Reversible bending ofrod-shaped acrylamide gels in an electric-field. Macromolecules, 19,2890–2892.

21 Suzuki, A. and Tanaka, T. (1990) Phase-transition in polymer gels inducedby visible-light. Nature, 346, 345–347.

22 Schild, H.G. (1992) Poly(N-isopropylacrylamide): experiment, theory andapplication. Progress in Polymer Science, 17, 163–249.

23 Eisenbach, C.D. (1980) Isomerization of aromatic azo chromophores inpoly(ethyl acrylate) networks and photomechanical effect. Polymer, 21,1175–1179.

24 Zimmermann, E. and Stille, J. (1985) Photoresponsive polyquinolines.Macromolecules, 18, 321–327.

25 Blair, H.S., Pague, H.I., and Riordan, J.E. (1980) Photoresponsive effects inazo polymers. Polymer, 21, 1195–1198.

26 Blair, H.S. and McArdle, C.B. (1984) Photoresponsive polymers: 2. Themonolayer behaviour of photochromic polymers containing aromaticazobenzene residues. Polymer, 25, 1347–1352.

�

� �

�

References 27

27 Malcolm, B.R. and Pieroni, O. (1990) The photoresponse of anazobenzene-containing poly(l-lysine) in the monolayer state. Biopolymers,29, 1121–1123.

28 Menzel, H., Weichart, B., and Hallensleben, M.L. (1992)Langmuir-Blodgett-films of photochromic polyglutamates. Polymer Bul-letin, 27, 637–644.

29 Seki, T., Sekizawa, H., Morino, S., and Ichimura, K. (1998) Inherent andcooperative photomechanical motions in monolayers of an azobenzenecontaining polymer at the air-water interface. Journal of Physical Chem-istry B, 102, 5313–5321.

30 Seki, T. (2004) Dynamic photoresponsive functions in organized layer sys-tems comprised of azobenzene-containing polymers. Polymer Journal, 36,435–454.

31 Seki, T. and Tamaki, T. (1993) Photomechanical effect in monolayers ofazobenzene side chain polymers. Chemistry Letters, 22, 1739–1742.

32 Warner, M. and Terentjev, E.M. (2003) Liquid Crystal Elastomers, OxfordUniversity Press, Oxford.

33 Geelhaar, T., Griesar, K., and Reckmann, B. (2013) 125 years of liquidcrystals — a scientific revolution in the home. Angewandte Chemie, Inter-national Edition, 52, 8798–8809.

34 Tschierske, C. (2013) Development of structural complexity byliquid-crystal self-assembly. Angewandte Chemie, International Edition,52, 8828–8878.

35 Bremer, M., Kirsch, P., Klasen-Memmer, M., and Tarumi, K. (2013) TheTV in your pocket: development of liquid-crystal materials for the newmillennium. Angewandte Chemie, International Edition, 52, 8880–8896.

36 de Gennes, P.G. (1975) Réflexions sur un type de polymères nématiques.Comptes Rendus de l’ Academie des Sciences B, 281, 101–103.

37 de Gennes, P.G., Hébert, M., and Kant, R. (1997) Artificial muscles basedon nematic gels. Macromolecular Symposium, 113, 39–49.

38 Finkelmann, H., Kock, H.-J., and Rehage, G. (1981) Investigations onliquid-crystalline polysiloxanes. 3. Liquid-crystalline elastomers — a newtype of liquid-crystalline material. Makromolekulare Chemie, Rapid Com-munications, 2, 317–322.

39 Finkelmann, H., Kock, H.-J., Gleim, W., and Rehage, G. (1984) Investiga-tions on liquid crystalline polysiloxanes, 5. Orientation of LC-elastomersby mechanical forces. Makromolekulare Chemie, Rapid Communications,5, 287–293.

40 Zentel, R. (1989) Liquid crystalline elastomers. Advanced Materials, 1,321–329.

41 Küpfer, J. and Finkelmann, H. (1991) Nematic liquid single-crystal elas-tomers. Makromolekulare Chemie, Rapid Communications, 12, 717–726.

�

� �

�


42 Ohm, C., Brehmer, M., and Zentel, R. (2010) Liquid crystalline elastomersas actuators and sensors. Advanced Materials, 22, 3366–3387.

43 Wermter, H. and Finkelmann, H. (2001) Liquid crystalline elastomers asartificial muscles. e-Polymers, 1, 111–123.

44 Zentel, R. and Reckert, G. (1986) Liquid crystalline elastomers basedon liquid crystalline side group, main chain and combined polymers.Makromolekulare Chemie, 187, 1915–1926.

45 Broer, D.J., Finkelmann, H., and Kondo, K. (1988) In-situ photopolymer-ization of an oriented liquid-crystalline acrylate. Makromolekulare Chemie,189, 185–194.

46 Broer, D.J., Boven, J., Mol, G.N., and Challa, G. (1989) In-situ photopoly-merization of oriented liquid-crystalline acrylates, 3. Oriented polymernetworks from a mesogenic diacrylate. Makromolekulare Chemie, 190,2255–2268.

47 Broer, D.J., Gossink, R.G., and Hikmet, R.A.M. (1990) Oriented polymernetworks obtained by photopolymerization of liquid-crystalline monomers.Angewandte Makromolekulare Chemie, 183, 45–66.

48 Ikeda, T. (2003) Photomodulation of liquid crystal orientations for pho-tonic applications. Journal of Materials Chemistry, 13, 2037–2057.

49 Tazuke, S., Kurihara, S., and Ikeda, T. (1987) Amplified image record-ing in liquid-crystal media by means of photochemically triggeredphase-transition. Chemistry Letters, 16, 911–914.

50 Ikeda, T., Horiuchi, S., Karanjit, D. et al. (1990) Photochemically inducedisothermal phase-transition in polymer liquid-crystals with mesogenicphenyl benzoate side-chains. 2. Photochemically induced isothermalphase-transition behaviors. Macromolecules, 23, 42–48.

51 Ikeda, T., Kurihara, S., Karanjit, D., and Tazuke, S. (1990) Photochemi-cally induced isothermal phase-transition in polymer liquid-crystals withmesogenic cyanobiphenyl side-chains. Macromolecules, 23, 3938–3943.

52 Ikeda, T. and Tsutsumi, O. (1995) Optical switching and image storage bymeans of azobenzene liquid-crystal films. Science, 268, 1873–1875.

53 Weigert, F. (1919) Über einen neuen Effekt der Strahlung in lichtempfind-lichen Schichten. Verhandlungen der Deutschen Physikalischen Gesellschaft,21, 479–491.

54 Todorov, T., Tomova, N., and Nikolova, L. (1983) High-sensitivity mate-rial with reversible photo-induced anisotropy. Optics Communication, 47,123–126.

55 Rochon, P., Gosselin, J., Natansohn, A., and Xie, S. (1992) Opticallyinduced and erased birefringence and dichroism in azoaromatic polymers.Applied Physics Letters, 60, 4–5.

56 Rochon, P., Batalla, E., and Natansohn, A. (1995) Optically induced sur-face gratings on azoaromatic polymer films. Applied Physics Letters, 66,136–138.

�

� �

�

References 29

57 Kim, D.Y., Tripathy, S.K., Li, L., and Kumar, J. (1995) Laser induced holo-graphic surface relief gratings on nonlinear optical polymer films. AppliedPhysics Letters, 66, 1166–1168.

58 Natansohn, A. and Rochon, P. (2002) Photoinduced motions inazo-containing polymers. Chemical Reviews, 102, 4139–4176.

59 Ichimura, K., Suzuki, Y., Seki, T. et al. (1988) Reversible change in align-ment mode of nematic liquid crystals regulated photochemically bycommand surfaces modified with an azobenzene monolayer. Langmuir,4, 1214–1216.

60 Ichimura, K. (2000) Photoalignment of liquid-crystal systems. ChemicalReviews, 100, 1847–1874.

61 Ichimura, K., Oh, S.-K., and Nakagawa, M. (2000) Light-driven motion ofliquids on a photoresponsive surface. Science, 288, 1624–1626.

62 Finkelmann, H., Nishikawa, E., Pereira, G.G., and Warner, M. (2001) Anew opto-mechanical effect in solids. Physical Review Letters, 87, 015501.

63 Hogan, P.M., Tajbakhsh, A.R., and Terentjev, E.M. (2002) UV manipulationof order and macroscopic shape in nematic elastomers. Physical Review E,65, 041720.

64 Cviklinski, J., Tajbakhsh, A.R., and Terentjev, E.M. (2002) UV isomerisa-tion in nematic elastomers as a route to photo-mechanical transducer. TheEuropean Physical Journal E, 9, 427–434.

65 Li, M.-H., Keller, P., Li, B. et al. (2003) Light-driven side-on nematicelastomer actuators. Advanced Materials, 15, 569–572.

66 Yu, Y., Nakano, M., and Ikeda, T. (2003) Directed bending of a polymerfilm by light. Nature, 425, 145.

67 Ikeda, T., Nakano, M., Yu, Y. et al. (2003) Anisotropic bending andunbending behavior of azobenzene liquid-crystalline gels by light exposure.Advanced Materials, 15, 201–205.

68 Yu, Y., Nakano, M., and Ikeda, T. (2004) Photoinduced bending andunbending behavior of liquid-crystalline gels and elastomers. Pure andApplied Chemistry, 76, 1467–1477.

69 Harris, K.D., Cuypers, R., Scheibe, P. et al. (2005) Large amplitudelight-induced motion in high elastic modulus polymer actuators. Journal ofMaterials Chemistry, 15, 5043–5048.

70 White, T.J. (2012) Light to work transduction and shape memory in glassy,photoresponsive macromolecular systems: Trends and opportunities. Jour-nal of Polymer Science, Part B: Polymer Physics, 50, 877–880.

71 Tabiryan, N., Serak, S., Dai, X.-M., and Bunning, T. (2005) Polymerfilm with optically controlled form and actuation. Optics Express, 13,7442–7448.

72 White, T.J., Tabiryan, N.V., Serak, S.V. et al. (2008) A high frequencyphotodriven polymer oscillator. Soft Matter, 4, 1796–1798.

�

� �

�


73 Yamada, M., Kondo, M., Mamiya, J. et al. (2008) Photomobile polymermaterials: towards light-driven plastic motors. Angewandte Chemie, Inter-national Edition, 47, 4986–4988.

74 Camacho-Lopez, M., Finkelmann, H., Palffy-Muhoray, P., and Shelley, M.(2004) Fast liquid-crystal elastomer swims into the dark. Nature Materials,3, 307–310.

75 Serak, S., Tabiryan, N., Vergara, R. et al. (2010) Liquid crystalline polymercantilever oscillators fueled by light. Soft Matter, 6, 779–783.

76 Yamada, M., Kondo, M., Miyasato, R. et al. (2009) Photomobile polymermaterials — Various three-dimensional movements. Journal of MaterialsChemistry, 19, 60–62.

77 Cheng, F., Yin, R., Zhang, Y. et al. (2010) Fully plastic microrobots whichmanipulate objects using only visible light. Soft Matter, 6, 3447–3449.

78 Jiang, Z., Xu, M., Li, F., and Yu, Y. (2013) Red-light-controllableliquid-crystal soft actuators via low-power excited upconversion basedon triplet–triplet annihilation. Journal of the American Chemical Society,135, 16446–16453.

79 Liu, D. and Broer, D.J. (2014) Self-assembled dynamic 3D fingerprintsin liquid-crystal coatings towards controllable friction and adhesion.Angewandte Chemie, International Edition, 53, 4542–4546.

80 Xu, S., Ren, H., Lin, Y.-J. et al. (2009) Adaptive liquid lens actuated byphoto-polymer. Optics Express, 17, 17590–17595.

81 Camargo, C.J., Campanella, H., Marshall, J.E. et al. (2011) Localised actu-ation in composites containing carbon nanotubes and liquid crystallineelastomers. Macromolecular Rapid Communications, 32, 1953–1959.

82 Torras, N., Zinoviev, K.E., Camargo, C.J. et al. (2014) Tactile device basedon opto-mechanical actuation of liquid crystal elastomers. Sensors andActuators A, 208, 104–112.

83 Li, C., Liu, Y., Huang, X., and Jiang, H. (2012) Direct sun-drivenartificial heliotropism for solar energy harvesting based on aphoto-thermomechanical liquid-crystal elastomer nanocomposite.Advanced Functional Materials, 22, 5166–5174.

84 Star, A., Stoddart, J.F., Steuerman, D. et al. (2001) Preparation and prop-erties of polymer-wrapped single-walled carbon nanotubes. AngewandteChemie, International Edition, 40, 1721–1725.

85 van Oosten, C.L., Bastiaansen, C.W.M., and Broer, D.J. (2009) Printed arti-ficial cilia from liquid-crystal network actuators modularly driven by light.Nature Materials, 8, 677–682.

86 Yan, Z., Ji, X., Wu, W. et al. (2012) Light-switchable behavior of a microar-ray of azobenzene liquid crystal polymer induced by photodeformation.Macromolecular Rapid Communications, 33, 1362–1367.

�

� �

�

References 31

87 Zhao, J., Liu, Y., and Yu, Y. (2014) Dual-responsive inverse opal films basedon a crosslinked liquid crystal polymer containing azobenzene. Journal ofMaterials Chemistry C, 2, 10262–10267.

88 Yoshino, T., Kondo, M., Mamiya, J. et al. (2010) Three-dimensional pho-tomobility of crosslinked azobenzene liquid-crystalline polymer fibers.Advanced Materials, 22, 1361–1363.

89 Chen, M., Xing, X., Liu, Z. et al. (2010) Photodeformable polymer mate-rial: towards light-driven micropump applications. Applied Physics A, 100,39–43.

90 Shankar, M.R., Smith, M.L., Tondiglia, V.P. et al. (2013) Contactless, pho-toinitiated snap-through in azobenzene-functionalized polymers. PNAS,110, 18792–18797.

91 Lendlein, A., Jiang, H., Jünger, O., and Langer, R. (2005) Light-inducedshape-memory polymers. Nature, 434, 879–882.

92 Lee, K.M., Koerner, H., Vaia, R.A. et al. (2011) Light-activated shapememory of glassy, azobenzene liquid crystalline polymer networks. SoftMatter, 7, 4318–4324.

93 Nakano, H., Takahashi, T., Kadota, T., and Shirota, Y. (2002) Formationof a surface relief grating using a novel azobenzene-based photochromicamorphous molecular material. Advanced Materials, 14, 1157–1160.

94 Nakano, H., Tanino, T., Takahashi, T. et al. (2008) Relationship betweenmolecular structure and photoinduced surface relief grating formationusing azobenzene-based photochromic amorphous molecular materials.Journal of Materials Chemistry, 18, 242–246.

95 Nakano, H. and Suzuki, M. (2012) Photoinduced mass flow of pho-tochromic molecular materials. Journal of Materials Chemistry, 22,3702–3704.

96 Ikeda, T., Mamiya, J., and Yu, Y. (2007) Photomechanics ofliquid-crystalline elastomers and other polymers. Angewandte Chemie,International Edition, 46, 506–528.

97 Ikeda, T. and Ube, T. (2011) Photomobile polymer materials: from nano tomacro. Materials Today, 14, 480–487.

98 Wei, J. and Yu, Y. (2012) Photodeformable polymer gels and crosslinkedliquid-crystalline polymers. Soft Matter, 8, 8050–8059.

99 Palffy-Muhoray, P. (2012) Liquid crystal elastomers and light, in Liq-uid Crystal Elastomers: Materials and Applications (ed. W.H. DeJeu),Springer-Verlag, Berlin Heidelberg, pp. 95–118.

100 Broer, D.J., Bastiaansen, C.M.W., Debije, M.G., and Schenning, A.P.H.J.(2012) Functional organic materials based on polymerized liquid-crystalmonomers: supramolecular hydrogen-bonded systems. AngewandteChemie, International Edition, 51, 7102–7109.

�

� �

�


101 Fleischmann, E.-K. and Zentel, R. (2013) Liquid-crystalline ordering as aconcept in materials science: from semiconductors to stimuli-responsivedevices. Angewandte Chemie, International Edition, 52, 8810–8827.

102 Jiang, H., Li, C., and Huang, X. (2013) Actuators based on liquid crys-talline elastomer materials. Nanoscale, 5, 5225–5240.

103 Ube, T. and Ikeda, T. (2014) Photomobile polymer materials withcrosslinked liquid-crystalline structures: molecular design, fabrication, andfunctions. Angewandte Chemie, International Edition, 53, 10290–10299.

104 Kondo, M., Yu, Y., and Ikeda, T. (2006) How does the initial alignment ofmesogens affect the photoinduced bending behavior of liquid-crystallineelastomers? Angewandte Chemie, International Edition, 45, 1378–1382.

105 van Oosten, C.L., Harris, K.D., Bastiaansen, C.W.M., and Broer, D.J. (2007)Glassy photomechanical liquid-crystal network actuators for microscaledevices. The European Physical Journal E, 23, 329–336.

106 Naka, Y., Mamiya, J., Shishido, A. et al. (2011) Direct fabrication of pho-tomobile polymer materials with an adhesive-free bilayer structure byelectron-beam irradiation. Journal of Materials Chemistry, 21, 1681–1683.

107 Wu, W., Yao, L., Yang, T. et al. (2011) NIR-light-induced deformation ofcross-linked liquid-crystal polymers using upconversion nanophosphors.Journal of the American Chemical Society, 133, 15810–15813.

108 Ube, T., Takado, K., and Ikeda, T. (2015) Photomobile materials withinterpenetrating polymer networks composed of liquid-crystalline andamorphous polymers. Journal of Materials Chemistry C, 3, 8006–8009.

109 Gibbons, W.M., Shannon, P.J., Sun, S.-T., and Swetlin, B.J. (1991)Surface-mediated alignment of nematic liquid crystals with polarizedlaser light. Nature, 351, 49–50.

110 Schadt, M., Schmitt, K., Kozinkov, V., and Chigrinov, V. (1992)Surface-induced parallel alignment of liquid crystals by linearly polymer-ized photopolymers. Japanese Journal of Applied Physics, 31, 2155–2164.

111 Yaroshchuk, O. and Reznikov, Y. (2012) Photoalignment of liquid crystals:basics and current trends. Journal of Materials Chemistry, 22, 286–300.

112 Chen, J., Johnson, D.L., Bos, P.J. et al. (1996) Radially and azimuthallyoriented liquid crystal alignment patterns fabricated by linearly polarizedultraviolet exposure process. Applied Physics Letters, 68, 885–887.

113 Lee, B.-W. and Clark, N.A. (2001) Alignment of liquid crystals with pat-terned isotropic surfaces. Science, 291, 2576–2580.

114 de Haan, L.T., Sánchez-Somolinos, C., Bastiaansen, C.M.W. et al. (2012)Engineering of complex order and the macroscopic deformation of liquidcrystal polymer networks. Angewandte Chemie, International Edition, 51,12469–12472.

115 de Haan, L.T., Gimenez-Pinto, V., Konya, A. et al. (2014) Accordion-likeactuators of multiple 3D patterned liquid crystal polymer films. AdvancedFunctional Materials, 24, 1251–1258.

�

� �

�

References 33

116 Liu, D., Bastiaansen, C.W.M., den Toonder, J.M.J., and Broer, D.J. (2012)Photo-switchable surface topologies in chiral nematic coatings. Ange-wandte Chemie, International Edition, 51, 892–896.

117 McConney, M.E., Martinez, A., Tondiglia, V.P. et al. (2013) Topographyfrom topology: photoinduced surface features generated in liquid crystalpolymer networks. Advanced Materials, 25, 5880–5885.

118 Iamsaard, S., Aßhoff, S.J., Matt, B. et al. (2014) Conversion of light intomacroscopic helical motion. Nature Chemistry, 6, 229–235.

119 Ware, T.H., McConney, M.E., Wie, J.J. et al. (2015) Voxelated liquid crystalelastomers. Science, 347, 982–984.

120 White, T.J. and Broer, D.J. (2015) Programmable and adaptive mechanicswith liquid crystal polymer networks and elastomers. Nature Materials,14, 1087–1098.

121 Koerner, H., Price, G., Pearce, N.A. et al. (2004) Remotely actuatedpolymer nanocomposites — stress-recovery of carbon-nanotube-filledthermoplastic elastomers. Nature Materials, 3, 115–120.

122 Ahir, S.V. and Terentjev, E.M. (2005) Photomechanical actuation inpolymer-nanotube composites. Nature Materials, 4, 491–495.

123 Ahir, S.V., Squires, A.M., Tajbakhsh, A.R., and Terentjev, E.M. (2006)Infrared actuation in aligned polymer-nanotube composites. PhysicalReview B, 73, 085420.

124 Ahir, S.V., Terentjev, E.M., Lu, S.X., and Panchapakesan, B. (2007) Thermalfluctuations, stress relaxation, and actuation in carbon nanotube networks.Physical Review B, 76, 165437.

125 Dalton, A.B., Stephan, C., Coleman, J.N. et al. (2000) Selective interactionof a semiconjugated organic polymer with single-wall nanotubes. Journalof Physical Chemistry B, 104, 10012–10016.

126 Chen, J., Liu, H., Weimer, W.A. et al. (2002) Noncovalent engineering ofcarbon nanotube surfaces by rigid, functional conjugated polymers. Journalof the American Chemical Society, 124, 9034–9035.

127 Yang, L., Setyowati, K., Li, A. et al. (2008) Reversible infrared actuation ofcarbon nanotube-liquid crystalline elastomer nanocomposites. AdvancedMaterials, 20, 2271–2275.

128 Kohlmeyer, R.R. and Chen, J. (2013) Wavelength-selective, IR light-drivenhinges based on liquid crystalline elastomer composites. AngewandteChemie, International Edition, 52, 9234–9237.

129 Ji, Y., Huang, Y.Y., Rungsawang, R., and Terentjev, E.M. (2010) Dispersionand alignment of carbon nanotubes in liquid crystalline polymers andelastomers. Advanced Materials, 22, 3436–3440.

130 Marshall, J.E., Ji, Y., Torras, N. et al. (2012) Carbon-nanotube sensitizednematic elastomer composites for IR-visible photo-actuation. Soft Matter,8, 1570–1574.

�

� �

�


131 Ji, Y., Marshall, J.E., and Terentjev, E.M. (2012) Nanoparticle-liquid crys-talline elastomer composites. Polymers, 4, 316–340.

132 Li, Y., He, Y., Tong, X., and Wang, X. (2005) Photoinduced deformationof amphiphilic azo polymer colloidal spheres. Journal of the AmericanChemical Society, 127, 2402–2403.

133 Li, Y., He, Y., Tong, X., and Wang, X. (2006) Stretching effect of linearlypolarized Ar+ laser single-beam on azo polymer colloidal spheres. Lang-muir, 22, 2288–2291.

134 Wang, D. and Wang, X. (2013) Amphiphilic azo polymers: molecularengineering, self- assembly and photoresponsive properties. Progress inPolymer Science, 38, 271–301.

135 Wang, G., Tong, X., and Zhao, Y. (2004) Preparation ofazobenzene-containing amphiphilic diblock copolymers forlight-responsive micellar aggregates. Macromolecules, 37, 8911–8917.

136 Mabrouk, E., Cuvelier, D., Brochard-Wyart, F. et al. (2009) Bursting of sen-sitive polymersomes induced by curling. PNAS, 106, 7294–7298.

137 Yang, H., Ye, G., Wang, X., and Keller, P. (2011) Micron-sized liquid crys-talline elastomer actuators. Soft Matter, 7, 815–823.

138 Sun, Y., Evans, J.S., Lee, T. et al. (2012) Optical manipulation ofshape-morphing elastomeric liquid crystal microparticles doped withgold nanocrystals. Applied Physics Letters, 100, 241901.

139 Morikawa, Y., Nagano, S., Watanabe, K. et al. (2006) Optical alignmentand patterning of nanoscale microdomains in a block copolymer thin film.Advanced Materials, 18, 883–886.

140 Yu, H., Iyoda, T., and Ikeda, T. (2006) Photoinduced alignment ofnanocylinders by supramolecular cooperative motions. Journal of theAmerican Chemical Society, 128, 11010–11011.

141 Seki, T., Nagano, S., and Hara, M. (2013) Versatility of photoalignmenttechniques: from nematics to a wide range of functional materials. Poly-mer, 54, 6053–6072.

142 Seki, T. (2014) New strategies and implications for the photoalignment ofliquid crystalline polymers. Polymer Journal, 46, 751–768.

143 Yu, H. (2014) Recent advances in photoresponsive liquid-crystallinepolymers containing azobenzene chromophores. Journal of MaterialsChemistry C, 2, 3047–3054.

144 Irie, M. (2008) Photochromism and molecular mechanical devices. Bulletinof the Chemical Society of Japan, 81, 917–926.

145 Irie, M., Fukaminato, T., Matsuda, K., and Kobatake, S. (2014) Pho-tochromism of diarylethene molecules and crystals: memories, switches,and actuators. Chemical Reviews, 114, 12174–12277.

146 Kim, T., Zhu, L., Al-Kaysi, R.O., and Bardeen, C.J. (2014) Organic pho-tomechanical materials. ChemPhysChem, 15, 400–414.

�

� �

�

References 35

147 Nath, N.K., Panda, M.K., Sahoo, S.C., and Naumov, P. (2014) Thermallyinduced and photoinduced mechanical effects in molecular single crystals— a revival. CrystEngComm, 16, 1850–1858.

148 Al-Kaysi, R.O., Müller, A.M., and Bardeen, C.J. (2006) Photochemicallydriven shape changes of crystalline organic nanorods. Journal of the Amer-ican Chemical Society, 128, 15938–15939.

149 Kobatake, S., Takami, S., Muto, H. et al. (2007) Rapid and reversible shapechanges of molecular crystals on photoirradiation. Nature, 446, 778–781.

150 Morimoto, M. and Irie, M. (2010) A diarylethene cocrystal that convertslight into mechanical work. Journal of the American Chemical Society, 132,14172–14178.

151 Wang, D.H., Lee, K.M., Yu, Z. et al. (2011) Photomechanical response ofglassy azobenzene polyimide networks. Macromolecules, 44, 3840–3846.

152 Lee, K.M., Wang, D.H., Koerner, H. et al. (2012) Enhancement of pho-togenerated mechanical force in azobenzene-functionalized polyimides.Angewandte Chemie, International Edition, 51, 4117–4121.

153 Nakano, H. (2010) Direction control of photomechanical bending ofa photochromic molecular fiber. Journal of Materials Chemistry, 20,2071–2074.

154 Kobayashi, M. and Abe, J. (2012) Optical motion control of maglevgraphite. Journal of the American Chemical Society, 134, 20593–20596.

155 Zhang, T., Chang, H., Wu, Y. et al. (2015) Macroscopic and direct lightpropulsion of bulk graphene material. Nature Photonics, 9, 471–476.

�

� �

�

37

2

Photochromism in the Solid StateOleksandr S. Bushuyev and Christopher J. Barrett

Department of Chemistry, McGill University, Montreal, Canada

2.1 Molecular Photoswitches in the Solid State

Photochromism is a reversible transformation of two chemical species betweentwo energetic states caused by absorption of visible light or, in a broad sense,any electromagnetic radiation. In principle, any double-well potential systemthat reversibly cycles between two energetic states by absorption of light canbe photochromic (a “photoswitch”). Practical photochromic systems, however,need to cleanly switch without the buildup of unwanted side products andpossess a sufficiently high potential barrier as to ensure the reasonable stabilityof both of the energetic states. Numerous classes of molecules that reversiblyisomerize between multiple structural configurations in response to externalstimuli fall within the realm of molecular switches. Common examples ofphotochromic molecular switches are illustrated in Figure 2.1. Some of themost extensively studied small-molecule motifs include (but are not limited to)azobenzenes, whose isomerization between E and Z conformations about adouble bond mimics flapping motions [1–3]; diarylethenes and spiropyrans, inwhich conformational changes are accompanied by ring-opening or -closingreactions [4–6]; anthracene and coumarin derivatives that reversibly dimerizewith one another [7, 8]; and overcrowded alkenes, hydrazones, and imines,which behave as rotors that can revolve about a rigid internal axis [9–12].Covalent modifications of the parent switch and rotor molecules providenear-infinite opportunities to design alternate isomerization pathways andto tune actuation stimuli. Additionally, noncovalent interactions betweenmolecules such as hydrogen bonding, hydrophobic effects, and 𝜋–𝜋 stackingenable the construction of dynamic host–guest systems, expanding the versa-tility of switches via supramolecular self-assembly of two or more components[13].These dynamic molecules represent the smallest building blocks available


�

� �

�

38 2 Photochromism in the Solid State

N

N

S SR1 R2

F2

F2

F2

R′ON

R

(a)

(b)

(c)

hv

N N

S SR1R2

F2

F2

F2

hv

hv ′

N+

R

O–

R′

hv

Δ, hv ′

Δ, hv′

Figure 2.1 Examples of common small-molecule photoswitch isomerization reactions:(a) azobenzenes; (b) diarylethenes; (c) spiropyrans.

for a bottom-up approach to synthesize functional mechanical devices thatexhibit motion. While being chemically diverse, all of the aforementionedswitch components operate based on similar principles [14], where absorptionof a photon manipulates the relative energy barrier between two states.Photoswitches are usuallymost extensively studied in solution, where each of

the isomerization reactions is most easily characterized and can be consideredto occur independently. The focus of this chapter, however, is the reversibletransformation of photochromic molecules in the solid state where they aresupported by a surface or incorporated into an amorphous polymer hostmatrix, liquid-crystalline polymer network (LCN), or liquid crystal elastomer(LCE) or grown as single crystals. Typically, chromophores are embeddedinto a solid matrix both for study and for application as real devices. As aresult, matrix effects are inescapable; the behavior of the chromophore isaltered due to the matrix, and in turn, the chromophore alters the matrix[15]. Although either could be viewed as a hindrance, both can in fact bequite useful: the chromophore can be used as a delicate probe of the matrix(free volume, polarizability, mobility, morphology, viscoelasticity, etc.), andwhen the matrix couples to chromophore motion, molecular motions can betranslated to larger length scales. An illustrative example of this translation isnanometer-thick “command surfaces” of azobenzene chromophores that candictate the alignment of micron-thick adjacent layers of otherwise inert liquidcrystals [15].

�

� �

�

2.2 Molecular and Macroscopic Motion of Azobenzene Chromophores 39

Since the two energetic states of photoswitching molecules exhibit dif-ferences in absorption, photochromism, in a literal sense, means a changeof color of a material upon absorption of electromagnetic radiation. At thesame time, chemical isomerization between the two energetic states canbring about motion and conformational change. In order for a moleculeto change conformation, movement of a part or of the whole molecule hasto occur. Such molecular motion may be carefully engineered and harnessedto produce a secondary function and enable solid-state photochromism todrive many useful phenomena. The history of photochromism owes much tothe azobenzene chromophore; therefore, we cover it in greater detail in thefollowing section.

2.2 Molecular and Macroscopic Motionof Azobenzene Chromophores

Photoinducedmotion in azobenzenes, or any other photoswitch, occurs due tothe geometric change that occurs upon absorption of light. In cis-azobenzene,the phenyl rings are twisted at 90∘ relative to the C—N=N—C plane [16, 17].Isomerization reduces the distance between the 4 and 4′ positions from0.99 nm in the trans state to 0.55 nm in the cis state [18–20]. This geometricchange increases the dipole moment: while the trans form has no dipolemoment, the cis form has a dipole moment of 3.1 D [21]. The free-volumerequirement of the cis form can be much larger than that for the trans form[22], and the free volume required to cycle between these two states is stilllarger. It has been estimated that the minimum free-volume pocket requiredto allow isomerization to proceed through a transition state via the inversionpathway is 0.12 nm3 [16, 23] and via the rotation pathway is approximately0.38 nm3 [24]. The effects of matrix free-volume constraints on photochemicalreactions, in general, have been considered [25]. The geometrical changes inazobenzene are very large by molecular standards, and it is thus no surprisethat isomerization can modify a wide range of material properties. Morerecent measurements via high-pressure spectroscopy (104–105 atm) on theforce applied and energy exerted through this isomerization suggest thatazobenzene is indeed an extremely powerful molecular unit, and employmentin actuators depends largely on clever engineering of themechanical advantagegained and is not inherently material limited.This molecular displacement generates a significant nanoscale force, which

has been measured in single-molecule force spectroscopy experiments [26, 27]and compared well to theory [28]. In these experiments, illumination causescontraction of an azobenzene polymer, showing that each chromophore canexert pN to nN molecular forces on demand. The ability to activate and powermolecular-level devices using light is of course attractive in many applications,

�

� �

�


since it circumvents the limitations inherent to diffusion or wiring and permitsa remote (or even quite distant) power supply. The fast response and absenceof side reactions in azo isomerization are also advantageous. Coupling thesemolecular-scale motions to useful work is of course the next challenging step.To realize this aim, a wide variety of molecular switches have been synthe-sized and examined. For example, an azo bond linking two porphyrin ringsenabled photocontrol of electron transfer [29], and in another example, dramat-ically different hydrogen-bonding networks (intermolecular and intramolec-ular) could be favored based on the isomeric state of the azo group linkingtwo cyclic peptides [30, 31]. Other recently reported examples include osmoticpressure pumps [32], created by the photocontrolled solubility of azobenzene,analytical columns that increase the effluent rate of developing solvents [33],reversible light-controlled conductance switching [34], photoresponsive goldnanoparticle solvation [35], and network formation [36].While it is important to study the nanometer-scale azobenzene molecular

conformational changes that give rise to macroscopic phenomena, by farthe most useful applications to actuation are the reversible changes thatcan result in changes to bulk phenomena or to macroscopic motion overthe micrometer- to centimeter-size scale. The first consideration is perhapswhether the host material can expand or contract to an appreciable extent.In floating monolayers at a liquid surface, it is well established that the largermolecular size of the cis isomer leads to the corresponding lateral expansionof many tens of percentages [37], which can modify other bulk properties. Forexample, this allows photomodulation of a monolayer’s water contact angle[38] or surface potential [39]. Using fluorinated azo polymers, good photo-control was demonstrated over photopatterning [40, 41], and wettability hasbeen demonstrated [42, 43]. A monolayer of azo-modified calixarene, whenirradiated with a light gradient, produced a surface energy gradient sufficientto move a macroscopic oil droplet [15]. In a more recent work, surfactants ofazobenzene were used to create a liquid–liquid interface between oleic aciddroplets in an aqueous solution [44]. Photoisomerization of the azobenzenesurfactant created a wavelength-dependent interfacial tension capable ofinducing interfacial flow, and this interfacial flow then generated large-scaledroplet motion in a direction opposite to the gradient. The photocontrolleddroplet motion was thus used to direct droplets into the trajectories of variousshapes and letters. It also suggests possible applications of the aforementionedmaterials to microfluidics. Modest photoinduced contact angle changes forthin polymer films have also been reported [45]. Recently, an azobenzenecopolymer assembled into polyelectrolyte multilayer showed a modest 2∘change in contact angle with UV (ultraviolet) light irradiation. However, whenthe same copolymer was assembled onto a patterned substrate, the change incontact angle upon irradiation was enhanced to 70∘ [46]. The fact that surfaceroughness plays a role in contact angle is well established and shows that many

�

� �

�

2.3 Photomechanical Effects 41

systems can be optimized to give rise to a large change in surface propertiesthrough clever amplification.Perhaps, the most visible demonstration of macroscopic motion induced by

the isomerization of azobenzene is the mechanical bending and unbendingof freestanding LCN and LCE films [47, 48]. Bending occurs in these rela-tively thick films because the free surface (which absorbs light) contracts,whereas the interior of the film (which is not irradiated owing to the strongabsorption of the upper part of the film) does not contract. Because thedirection of bending can be controlled with polarized light, the materialsenable full directional photomechanical control [49]. Other examples includeexpansion–bending of cantilevers coated with an amorphous azobenzene thinfilm [50] and macroscopic contraction–bending of fibers and cantilevers madeof azobenzene liquid-crystalline elastomers [51–57]. One can also invert thecoupling of mechanical and optical effects: by stretching an elastomeric azofilm containing a grating, one can affect its wavelength-selection propertiesand orient the chromophores [58].Depending on the nature of the function and whether it is performed on

a micro-, meso-, or macroscopic scale, solid-state macroscopic transforma-tions fall in the realm of photomechanical solids, surface mass transport,molecular machines, or structural changes in network architectures. Thereis no clear boundary between these classes, as molecular machines performphotomechanical motion, photomechanical materials could be seen as molec-ular machines scaled up in three dimensions, and surface mass transporteffects are not always easily distinguishable from photomechanical motion.Chromophore, as a structural element in different classes of networks, is afairly new motif that bridges the gaps between all three of these phenomena.We will attempt to categorize the effects, but the reader should also keepin mind the interrelation of these arbitrary classifications. We will mostlydiscuss molecular machines as surface-organized photochromic moleculesperforming work in the environment (matrix), while the photomechanicalsection will cover photochromes in scaled-up freestanding bulk materialssuch as polymer films and crystals. Mass transport usually occurs close tothe surface, while switching inside the networks is typically viewed as a bulkproperty change. To give the reader a sense of the historical development,we will discuss these effects in chronological order of their discovery anddevelopment.

2.3 Photomechanical Effects

The photomechanical effect is a change of shape of a material upon exposureto light (not due simply to heat expansion) and is a property of some materi-als incorporating light-switchable chromophores. In its essence, it refers to the

�

� �

�


direct transformation of light into mechanical motion. For ease of observationand with the goal of maximizing photomechanical efficiency, such materialsare generally fabricated into high-aspect-ratio films forming a thin actuator. Inthe most simple terms, actuators are the systems that convert a given type ofstimulus – in this case, energy of photons – into mechanical motion.If an actuator is defined as an energy transducer converting an input energy

into mechanical motion, then azobenzene-based systems are excellent can-didates for photomechanical actuation for many niche applications involvingsmall size, localized actuation, remoteness of the power source, and freedomfrom the encumbrance of batteries, electrons, and internal moving parts. Themost convincing demonstration of macroscopic motion due to azo isomeriza-tion is the mechanical bending and unbending of freestanding polymer thinfilms [47, 48]. As the direction of bending can be controlled via the polarizationof the light, the materials enable full directional photomechanical control[49] and have been used to drive macroscopic motion of a floating film [59].For a thin film floating on a water surface, a contraction in the direction ofpolarized light was seen for LC materials, whereas an expansion was seen foramorphous materials [60]. A related amplification of azo motion to macro-scopic motion is the photoinduced bending of a microcantilever coated withan azobenzene monolayer [50]. Other examples include macroscopic bendingand three-dimensional control of fibers made of azobenzene liquid-crystallineelastomers [51–53], light-driven micro valves [61], and all-plastic motors[1]. In this section, a survey summary of some of the manifestations of thephotomechanical effect leading to macroscale actuation with azobenzene andother photoswitches is provided.

2.3.1 Photomechanical Effects in Amorphous Azo Polymers

Perhaps, not surprisingly, photomechanical effects in organic materials werefirst noted for the azobenzene chromophore. Merian is often credited with thefirst observation of the photomechanical effect in azobenzenes, when, in 1966,he observed that an azobenzene-treated nylon filament shrank upon irradiationwith a Xe daylight lamp [62].The relatively unimpressive overall contraction ofunder 0.1% of the original length and complexity of the samplemeant that it wasnot until 20 years later that anyone took much notice of this curious report. Inthe 1980s, Eisenbach prepared poly(ethyl acrylate) networks cross-linked withazobenzene chromophore [63]. The samples exhibited photoinduced contrac-tion of around 0.2% upon irradiation with UV light and corresponding expan-sion after visible light irradiation. Upon further investigations, Matejka et al.reported that upon the increase of loading of azobenzene into the materialabove 5%, a photoinduced contraction of nearly 1% was achieved [64–66].Thin-film polymer actuators capable of responding to external stimuli

and deforming are the most desirable for practical applications, in either

�

� �

�


amorphous or organized form (such as liquid crystalline). Photoinducedreversible changes in elasticity of semi-interpenetrating network films bear-ing azobenzene moieties were achieved by UV and visible light irradiation[67]. These network films were prepared by cationic copolymerization ofazobenzene-containing vinyl ethers in a linear polycarbonate matrix. Thenetwork film showed reversible deformation by switching the UV light onand off, and the photomechanical effect was attributed to a reversible changebetween the highly aggregated and dissociated states of the azobenzenegroups [67–69]. In other studies, similar films of azobenzene-containingvinyl ethers with polycaprolactone have achieved rapid (0.1min) anisotropicdeformation and recovery.The films, placed under constant tensile stress, werestretched perpendicular and parallel to the tensile stress before irradiation.Photoisomerization of these films resulted in film contraction for stretchingparallel to the tensile stress and film elongation for stretching perpendicularto the tensile stress. The photomechanical response was observed to increasewith film stretching and was speculated to arise from anisotropic responsescaused by the isomerization-induced vibration of azobenzene molecules,which decreases the modulus of the deformed amorphous area [70]. Otherpolymer films that exhibit high bending intensity and large bending angles(90∘) have also been reported [71].

2.3.2 Actuation in Liquid-Crystalline Polymers

In amorphous polymers, photomechanical deformations mostly occur inan isotropic and uniform way, that is, there is no preferential direction fordeformation. Anisotropic materials, such as in liquid-crystalline materials,provide direction to the mechanical response. A particularly promising classof materials for efficient photoinduced actuation are LCNs and LCEs. LCEsare lightly cross-linked polymers in which the high alignment order of themesogens can be coupled with the motions of the highly elastic polymernetwork. This coupling gives rise to many characteristic properties of LCEs.Upon heating, the alignment order of the LCE films decreases, and when theLC–isotropic phase-transition temperature is exceeded, the films exhibit acontraction along the mesogen alignment direction. Such anisotropic defor-mation can be very large, and along with the versatile mechanical properties ofthe polymer network and the reversibility of the process (upon cooling, LCEfilms revert back to their original size), LCEs show great potential as artificialmuscles [72–77].The application potential of photocontrolled actuators was further promoted

when Ikeda et al. reported various photoinduced 3D motions (bending) ofazobenzene LC gels and elastomers [47, 48, 78]. The bending is driven by agradient in the isomerization-induced reduction in the LC alignment order:the majority of the incident UV irradiation is absorbed within a relatively thin

�

� �

�


surface layer of the film, which generates asymmetric strain and subsequentdeformation. The process is reversible: UV irradiation destroys the mesogenalignment through trans–cis isomerization and causes the sample to bend,whereas irradiation with visible green light restores the azobenzenes to thetrans form and the film regains its original unbent state. The nature of bendingstrongly depends on the details of the material system. Homogeneouslyaligned polymer systems bend in the mesogen alignment direction [78],whereas the bending direction of polydomain LCEs can be controlled bylinearly polarized light [47]. The latter serves as an example of repeatable andprecisely controlled photoinduced deformation along any chosen direction,enabling full photomechanical directional control. It is an important steptoward practical applicability of light-driven actuators. Another example ofdirection control is provided by artificial muscle-like photochromic fibers, thebending direction of which can be controlled by changing the location of theilluminating source [79, 80]. Conversely, homeotropically aligned cross-linkedLC polymer films were observed to exhibit a completely different bendingbehavior; upon exposure to UV light, they bent away from the light source,due to isotropic expansion of the sample surface upon trans–cis isomerization[81]. The initial chromophore alignment is not the only way to control thedirectionality of the photoinduced bending: Tabiryan et al. demonstrated thatthe bending direction can be controlled with the polarization direction ofthe excitation beam, which was attributed to light-induced reorientation ofthe azobenzene moieties [82]. More recently, Van Oosten et al. showed thatthe bending direction can be controlled by designing the material to bearinternal composition gradients within the LC polymer network [83], and as thelatest example, Shishido and coworkers showed that the bending direction canalso be dictated by the nature of bonding between the azobenzene moietiesand the cross-linked polymer network [84].With appropriate engineering, the photoinduced deformations (expan-

sion/contraction and bending) can be translated into “real-life” actuation, todesign proof-of-principle micromachines capable of producing applicablework. As the first example of such engineering, Ikeda, with Yamada, Barrett,and coworkers, translated the photoinduced deformations of a cross-linkedliquid-crystalline polymer (CLCP) film into rotational motion [1]. Theylaminated a film of azobenzene with a thin polyethylene sheet, joined two endsof the composite film to create a continuous ring, and mounted it onto a pulleysystem. Upon simultaneous irradiation with both UV and visible light on eachof the pulleys to “pull” and “push,” the belt was driven in counterclockwiserotational motion (Figure 2.2a). Other recent examples by the same researchcollaboration include an “inchworm” locomotion achieved by attaching asheet of azo-LCN on a flexible polyethylene (PE) substrate with asymmetricsliding friction [85]. In this application, the film undergoes photomechanicalcontraction, while the asymmetric end shapes on the PE films act as a ratchet,

�

� �

�


8 s 24 s

0 s 16 s

White spot

as a marker

(a)

(b) 0 s 4 s 7 s

CLCPs

UV Vis

UV

17 s 22 s

VisUV

11 s

Figure 2.2 (a) Series of photographs showing the rotation of the light-driven plastic motorwith the LCE laminated film induced by simultaneous irradiation with UV and visible light.(Yamada et al. [1]. Reproduced with the permission of Royal Society of Chemistry.) (b) Seriesof photographs of the flexible “robotic arm” motion of the azo-LCE laminated film inducedby irradiation with UV and visible light. Arrows indicate the direction of light irradiation.(Yamada et al. [85]. Reproduced with the permission of Royal Society of Chemistry.)

�

� �

�


directing the motion of the film. Robotic arm-like actuation of flexible PEsheets was also demonstrated using azo-polymer “hinges” (Figure 2.2b).Different sections of a flexible PE film were laminated with azo-CLCPs,which enabled specific optical control (expansion or contraction) at variousindividually addressable positions of the film. The sections containing theazo-CLCPs thus functioned as hinge joints, acting as “arms” with remotecontrol over “elbows” and “wrists.”The latest advancement addressed an important problem inherent to the

laminated azo-CLCP films: even if their mechanical strength is improvedby the flexible polymer substrate, the poor adhesion between the two layersprevents efficient deformation transfer from the photoactive layer to thepolymer substrate. This can be overcome by connecting the active and passivelayers by chemical bonding (using e-beam cross-linking) [86].The durability ofsuch adhesive-free bilayer structures was significantly improved as comparedto adhesive-containing laminated films, and they might provide a routetoward increasing the optical–mechanical energy conversion efficiency of thelight-driven motors.In the previous examples of photo-drivenmotions, the primary energy source

was the combination of UV and visible light sources, which gave rise to locallyaddressable photoinduced contraction/expansion of the photoactive polymerfilms. UV light is harmful to many living organisms, however, so it is impor-tant to develop photo-driven actuators powered by visible light, and ultimatelysunlight, for any bio-related applications, and in general, there is an advantageto avoiding UV light to lessen material degradation. The first sunlight-drivenphotomobile materials, employing photoresponsive azotolane moieties, weredeveloped by Yu and coworkers [87], who also fabricated visible-light-drivenmicrorobots capable of lifting up and moving an object weighing 10mg, 10times the weight of the robotic arm itself [57, 88]. This integrated system con-sisted of several azo-LCN films on PE substrates connected by joints to mimicthe arm, wrist, hand, and even fingers of the human arm.The robotic arm couldbe bent and manipulated to perform complex actions by individually address-ing the various photoactive sections, for instance, an object could be picked upor dropped by addressing the “fingers,” while the entire arm could be movedby addressing it at different “elbow” locations. Later on, White and cowork-ers demonstrated a clever photo-fueled catapult motion, capable of launchingan object at a rate of 0.3m/s using moderate-intensity blue-light irradiation[89]. Yu and coworkers also designed a similar composite material, in whichupconverting nanophosphors assisted in inducing the photoinduced deforma-tion using near-infrared (NIR) (980 nm) light [90].White and coworkers have reported the prospect of high-frequency

photo-driven oscillators [51–53]. They designed azo-LCN materials andintegrated into cantilevers capable of achieving oscillation frequencies as

�

� �

�


“Off” “Activated”

“On”

Ar+ off

Ar+ off

E⊥n

E⊥n

E||n

Figure 2.3 The optical protocol for activating the light-powered oscillation of a cantilever.The nematic director (n) is positioned parallel to the long axis of the polymer cantilever ofdimension 5 mm × 1 mm × 50 mm. When exposed to light polarized orthogonal to n (E⊥n ),bending occurs toward the laser source. Cycling the Ar+ laser from E⊥n to E||n can turnoscillation “on,” while blocking the Ar+or returning the polarization of the laser beam to E⊥nturns the oscillation “off.” (White et al. [51]. Reproduced with the permission of Royal Societyof Chemistry.) (See color plate section for the color representation of this figure.)

high as 270Hz and an energy conversion efficiency of 0.1% upon irradia-tion with a focused blue laser beam, with a range of motion close to 180∘(Figure 2.3). The cantilevers possessed a storage modulus ranging from 1.3to 1.7GPa and were shown to bend faster and attain larger bending angleswith monodomain orientation, increasing azobenzene concentration, andreduced thickness. The bending angle was also dependent on the polarizationof the incoming light as well as atmospheric pressure. These cantilevers alsooscillated under a focused beam of sunlight [53] and thus offer the potentialfor remotely triggered photoactuation, adaptive optics, and, most importantly,solar energy harvesting. Such high-frequency oscillators could power aminiaturized micro-opto-mechanical system as they contain both the forcegeneration component (azobenzene) and the kinematic structure (cantilever)in a single unit.The aforementioned studies employed azo-LCN entirely composed of

azobenzene mesogens. In such systems, practically all of the incident irradi-ation is absorbed within the near-surface region with a thickness of 1–2 μm.As typical film thicknesses used are in the order of 10–20 μm, the majority of

�

� �

�


the azobenzene moieties in the bulk of the film remain dark and unaffectedby incident light. As a result, the efficiency of the photomechanical effectgenerated in such plain-azobenzene actuators is far from optimal. This wasfirst addressed by Broer and coworkers, who designed densely cross-linkedhigh-elastic-modulus polymer actuators with relatively low azobenzene con-centration [91]. Indeed, as shown by Palffy-Muhoray and coworkers as earlyas in 2004, even nonphotoresponsive LCEs doped with low concentrations ofazobenzene dyes can exhibit remarkable and unprecedented photoinduceddeformation behavior [59]. In fact, it has been recently shown that the opti-mum photoinduced response (in terms of the stress generated) is achievedusing a moderate concentration of azobenzene moieties, supplemented withhigher concentrations of nonphotoactive mesogens [92]. The largest mechan-ical force generated by photoirradiation of the various films was measuredas 2.6MPa. Detailed studies have also been performed on the cross-linkerconcentration dependence of azo-LCN. The cross-linking density changes theelastic modulus and the thermomechanical properties of the material systemin a delicate manner, playing an important role in the mobility of the polymersegments, and in general, low cross-linker concentration is favorable foroptimizing the photoinduced/thermally induced deformation of cross-linkedLC polymers, whereas high cross-linker concentration (high modulus) ispreferable for high photoinduced stress generation [76, 93, 94]. A recent obser-vation by Shishido and coworkers suggests that the photoinduced bending ofazo-CLCPs is accompanied by a significant, 2.5-fold decrease in the Young’smodulus of the sample upon UV irradiation [94]. Such “photo-softening” wasobserved to be the most pronounced in a low-cross-linker-concentration (andlow-modulus) sample, which also exhibited the most efficient photoinducedbending. Upon increasing the cross-linker concentration (and the modulus),both photoinduced bending and the photo-softening effect became lessefficient, indicating that there might be a profound connection between thephoto-softening and the photomechanical properties of azo-LCNs. In order toprobe such connections further, the Broer group offered a route to maximizephoto-expansion and, consequently, photomechanical bending [95]. Theyfound that higher expansion was possible upon simultaneous irradiation of anazo-LCN with UV and visible light, stimulating dynamic trans↔cis conversionand enhancing the free volume of a system by a factor of 4. Surface protrusionsas high as 12% were achieved on these azo-LNC films compared to just a 3%expansion upon single wavelength irradiation.More recent notable advances in the field of photomechanical poly-

mers report on the development of more complex movement patterns,designs that respond to a broader range of stimuli, and better addressabilityor spatial precision [96, 97]. Katsonis and coworkers, for example, recentlydescribed impressive control over the helical motion of azobenzene-containingliquid-crystalline polymer springs, mimicking the extensile function of plant

�

� �

�


UV (365 nm)

(a)

(b)

(c)

Vis

Magnets

Magnets

dRdL

Figure 2.4 The central kink in the mixed helicity ribbon (a) incorporating azobenzenechromophore undergoes a piston-like shuttling motion upon irradiation with visible (b) andUV (c) light. A magnet is connected to the kink and is transported and transmits thepush–pull shuttling motion to a second magnet placed 10 mm below. (Iamsaard et al. [98].Reproduced with the permission of Nature Publishing Group.)

tendrils [98] (Figure 2.4). The helical deformations were preprogrammed byincluding chiral azobenzene dopants and control of the relative orientationof the aligned liquid crystals within each spring, respectively. Chiral dopantsinduce a left-handed and right-handed twist in the liquid-crystalline film.Depending on the direction, in which such a film is cut, it will curl, twist, or doboth upon irradiation with light. Complex extensile and contractile coiling andtwisting helical motions are possible, and mechanical energy can be exportedfrom the system by the use of a pair of magnets.

2.3.3 Photosalient, Photochromic, and Photomechanical Crystals

While polymeric materials offer ease of processing and fabrication, crystallinemotifs offer a path to near-perfect three-dimensional arrangements of dynamicmolecular building blocks as well as a facile means to monitor the motion ofthe crystals using nuclear magnetic resonance (NMR) and X-ray diffraction.

�

� �

�


Furthermore, energy transduction in crystals is faster, and actuating motionis generally of larger magnitude compared to polymeric assemblies becauserigidly packed molecules have to work as a cooperative assembly, whereas ina less-ordered medium, motion of the same molecular switches may be lessdirectionally biased and ultimately less efficient. Often derided as a “chemicalcemetery” after Leopold Ruzicka’s famous remark, crystal machines insteadhave witnessed a remarkable resurgence in recent years with discoveriesof controlled or spontaneous actuation, reversible or irreversible move-ments, and reactivity [99–101]. Entire crystals of photochromic moleculescan serve as actuators and convert the external stimulus such as light intomechanical energy. Depending on the photomechanical response, such crystalactuators can be categorized as photosalient or photomechanical crystals[75, 102, 103]. Photosalient crystals are an interesting subset of the moregeneral thermosalient material class, which exhibit spontaneous actuation(jumping) upon heating or irradiation. Thermosalient crystals developed byNaumov and coworkers harness the energy of polymorphic transformationsthat occur upon heating of crystals. The effect is driven by extremely rapidanisotropic expansion and contraction of the unit cell axes upon a phasetransition that was found to be 104 times faster than regular crystal-to-crystalphase transformations [104]. This class of crystalline compounds is comprisedof a diverse range of materials including brominated organic molecules,terephthalic acid, and organometallic complexes [105, 106]. Although thereexists little directional control or foresight into which compounds will exhibitthe effect, the explosive “popcorn” crystals nonetheless exhibit impressivecentimeter-scale jumping movements that greatly exceed the crystal dimen-sions. Light-activated chemical processes within crystals, such as changesin the coordination sites of small ligands or [2+ 2] cycloaddition reactions,have also been shown to result in “popping” under UV irradiation [102, 107].Similarly to thermosalient materials, little control is possible over “popping”crystals, and thus, their utilization as functional molecular machines is difficultto envisage. To circumvent this problem, Sahoo et al. developed smart hybridmaterials that incorporate thermosalient microcrystallites on flexible sodiumcaseinate films to impart directionality to the crystals’ movements [108].The material represents a successful marriage of crystals to biocompatiblepolymeric films in one system, combining the benefits of the plasticity of softpolymers and the efficient, fast actuation of leaping crystalline solids.Photomechanical materials can move, bend, twist, or curl when exposed to

light in a wider variety of motions and usually with greater control. Some ofthe most promising photomechanical crystalline systems for converting lightinto mechanical work have been proposed by Irie and coworkers and are basedon diarylethene photoswitches [109]. Light absorption triggers pericyclicring-opening and -closing reactions of this photoswitch throughout the crystaland is responsible for expansion and contraction, respectively, of the unit cell

�

� �

�


that consequently leads to photomechanical bending of the crystal. Structuralstudies on crystals of diarylethene derivatives link the initial speed of curvaturechange to crystal thickness [110]. The bending behavior is also dependent onthe crystallographic face that is irradiated, which is attributed to differencesin molecular packing. While the geometric change that occurs during theisomerization of a single diarylethene photoswitch is modest, the collectiveaction of arrays of molecules in the crystal lattice can produce macroscopicmotion. If smartly engineered, such actuating crystals can perform workpushing or lifting objects many times their weight [109, 111], rotating gears[112], or acting as an electrical circuit switch (Figure 2.5) [113].

“ON”

(a)

(b)

(c) (d)

“ON”

UV

UV

Vis

“OFF”

“OFF” (a)

1 mm

1.0 mm

100 μm 100 μm 100 μm 100 μm 100 μm 100 μm

1 mm

1 mm 1 mm

1 mmUV

Figure 2.5 Photomechanical systems based on diarylethene crystals that convert light intomechanical work. (a) A rod-like crystal pushes a gold microparticle that is 90 times heavierthan the crystal when irradiated with UV light. Bending of the crystal pushes themicroparticle up to 30 μm. (Kobatake et al. [109]. Reproduced with the permission of NaturePublishing Group.) (b) Rotation of gears facilitated by the reversible bending of a crystallineactuator upon UV and visible irradiation. (Terao et al. [112]. Reproduced with the permissionof John Wiley and Sons.) (c) Irradiation of gold-coated diarylethene crystals with UV andvisible light enables the ON/OFF photoreversible current switching of an electric circuit.(Kitagawa and Kobatake [113]. Reproduced with the permission of Royal Society ofChemistry.) (d) Superimposed photographs of a crystal cantilever lifting a lead ball with amass 275 times larger than that of the crystal upon irradiation with UV light from theunderside of the actuator. (Morimoto and Irie [111]. Reproduced with the permission ofAmerican Chemical Society.)

�

� �

�


Other commonly studied photomechanical crystalline architectures arecomposed of anthracenes or salicylidenephenylethylaminemolecules. Bardeenand coworkers investigated the [4+ 4] dimerization reaction of anthraceneswhere the photoreaction results in reversible or irreversible twisting ofcrystalline microribbons [114, 115]. The curling and twisting motion may beattributed to strain between spatially distinct reactant and product domains asa result of differential absorption by different regions of the crystal or intrinsicsolid-state reaction kinetics [116]. Another crystalline molecular machine wasproposed by Koshima et al. based on photomechanical action in plate-likecrystals of salicylidenephenylethylamine (Figure 2.6) [117].

H3CH3C

N

enol-(S)-1 keto-(S)-1

H OO

UV

1.0 mm

δ

H

Nhv

Δ, hv′

Figure 2.6 (a) Photoinduced proton transfer in the S enantiomers of chiralsalicylidenephenylethylamines upon keto-enol tautomerism. (b) Superimposedphotographs of a chiral enol-(S)-1 crystal before and after irradiating the top of the crystalactuator with ultraviolet light. The crystalline cantilever achieved 26 nJ of work by lifting a4.00 mg metal ring a height, 𝛿, of 0.65 mm. Various photomechanical lifting works wereachieved with different enantiomeric compositions within the crystal: the racemic crystal,enol-(rac)-1, achieved 59 nJ of work by lifting a weight with a mass 300 times larger thanthat of the crystal (not shown). (Koshima et al. [117]. Reproduced with the permission ofAmerican Chemical Society.)

�

� �

�


These actuators are capable of lifting weights up to 300 times the mass of thelever. The motion is linked to geometric changes in the molecules producedupon tautomeric transformation triggered by light absorption and consequentproton migration. Collective reorganization of the molecules within the latticeleads to small uniaxial cell expansion, which in turn results in bending of thecrystal.While, perhaps, being the workhorse molecule of most photomechanical

studies, the azobenzene chromophore has been extensively investigatedmainlyon surfaces, in polymers, and within liquid crystals, but it has been far lessfrequently studied by the crystalline photomechanical community [118–122].This is due, in part, to the sterically demanding trans-to-cis iso-

merization process, which may be impeded in a crystal. Surprisingly,Koshima et al. demonstrated reversible photomechanical bending of thintrans-4-(dimethylamino)azobenzene plates upon UV irradiation, concludingthat isomerization can still occur inside the crystal [123]. That report wasfollowed by a study of thin crystalline plates and needles of pseudostilbenes(azobenzenes with short-lived cis-states), which were capable of submil-lisecond bending and relaxation upon irradiation with visible light [124].Pseudostilbene bending crystals offer the fastest speed of bending–relaxationcycles as the whole event can take less than a second and is the only system thatcompletely circumvents the need for UV light to induce isomerization. Morerecent reports have focused on elucidating the mechanistic aspects of azoben-zene isomerization in crystals focusing on in situ X-ray diffraction studies ofirreversible cis–trans isomerization in crystals [125], and cocrystals [126], ofthe molecule (Figure 2.7).These reports demonstrated that cis→trans isomerization in crystals ismedi-

ated by a transient amorphous state. While the sterically demanding azoben-zene isomerization reaction requires considerable free volume to occur andis rarely possible in a single-crystal-to-single-crystal manner, an amorphiza-tion mechanism enables the photochemical reaction to proceed despite theconstraints of the crystal lattice. Amorphous intermediates in the crystal werecorroborated by the loss of diffraction spots upon irradiation with visible lightat a low temperature. Remarkably, isomerization within the crystals of azoben-zene represents a topotactic process in which the orientation of the resultanttrans crystal phase is dependent on the initial crystal orientation and effectivelyrepresents templated crystal growth directed by light.The growing number of investigations on dynamic molecular crystals that

can perform as actuators demands new theoretical models, as will be describedin detail inChapter 3. Variousmodels have been reported in the context of poly-mer films, rods, or plates, such as the analysis by Warner and Mahadevan onphotodeformation of nematic elastomers [127].However, careful structural andkinetic considerations based on the spatial density and uniform orientationsof photoactive molecules must be taken into account for these models to be

�

� �

�


Cocrystal-1

1.0 mm

hv

Amorphous

intermediateCocrystal-2

Figure 2.7 Amorphous phase-mediated azobenzene isomerization and photomechanicalbending in a cocrystal (1.2 mm × 90 μm × 20 μm). The cocrystal contains a 1:1 ratio of ahalogen bond acceptor (cis-1,2-bis(4-pyridyl)ethylene) and halogen bond donor species.Bromine or iodine atoms on perfluorinated azobenzenes act as halogen bond donors,having a linear interaction with the pyridine nitrogen atoms on the halogen bond acceptor.Irradiation of the cocrystals with a 532-nm laser facilitates cis-to-trans isomerization of thehalogen bond donors, via amorphous intermediates, determined by X-ray diffraction.(Bushuyev et al. [126]. Reproduced with the permission of Royal Society of Chemistry.)

fully applicable to crystalline systems. While the molecular motifs responsiblefor crystal motions are diverse, an attempt to unify photomechanical processeswas made by Nath et al., who proposed a mathematical treatment of photome-chanical crystal bending by accounting for the gradual profile of the productin the crystal, irradiation time, direction, and power using the azobenzene dye,Disperse Red 1 (DR1), as a model compound [128].The model is applicable forany photomechanical crystal system and should allow for an easier compari-son between the different platforms for efficiency, modulus, stress, and otherparameters critical for optimization of the process for practical use. Ultimately,with the help of theoretical frameworks and empirical data, the goal of futuredevelopment of photomechanical systems needs to be directed toward robustand fatigue-resistant designs capable of even faster and reliable actuation overthousands of cycles.

2.4 Solid-State Photochromic Molecular Machines

Inspired by the complexity and hierarchical organization of biologicalmachines, the design of artificial molecular machines that exhibit controlledmechanical motion and perform sophisticated tasks is an ultimate pursuitof molecular-scale engineering [129]. The design and synthesis of moleculesthat can undergo reversible structural changes with various stimuli havereceived considerable attention, but there remain far fewer reports of dynamicmolecular systems such as cleverly designed motors and pumps where the

�

� �

�

2.4 Solid-State Photochromic Molecular Machines 55

mechanistic action of their molecular components has been exploited to docontrolled work in their environment. The very definition of a machine hasbeen debated ever since biochemistry Professor Isaac Asimov began to layout the early laws of robotics over 75 years ago in his storytelling [130]. Inan effort to advance the field, a stringent language has been sought to differ-entiate simple molecular switches from motors that are capable of drivinga system away from equilibrium [131, 132]. While the working principles ofthese molecular devices cannot be easily compared to macroscopic analogs,a molecular machine has been defined as a multicomponent system withdefined energy input that is capable of performing a measurable and usefulsecondary function either at the nanoscale or, if amplified through collectiveaction, at the macroscale [133]. The system should ideally act in a reversiblemanner, with the capacity to complete repeated mechanical operations.Spatial and temporal control over this motion and work performed are furtherhallmarks of successful machines, helping in differentiating deliberate actuatedmechanics that leverage Brownian motion from undirected thermal effectsand distinguish useful machines from simple molecules that wiggle or diffuserandomly.The most exciting recent reports on molecules and assemblies exemplify

practical and advantageous attributes that can be applied to other systems andshow the most promise in successfully advancing the development of artificialmolecular machines. Ultimately, the precise and robust integration into higherdimensional architectures that take advantage of mechanical action by manycomponents is essential to bridge the gap between actuating molecular motionand performing microscopic, mesoscopic, and macroscopic work.

2.4.1 Nanostructure Functionalization

The integration of switchable molecular systems into inorganic nanostructuresand nanoparticle assemblies enables the manipulation of the hybrid material’soptical properties in situ. Two common methods to direct the optical prop-erties of nanostructured materials are the active tuning of refractive indicesat the surface of plasmonic nanoparticles functionalized with switchablemolecules [134–136] and physically modulating interparticle distances or ori-entations [137]. The intelligent functionalization of larger organic or inorganicnanostructures with small molecular switch and rotor components expandsthe versatility of these dynamic systems for applications such as drug deliveryand for tuning the chemical and physical properties of the hybrid materials[138, 139]. Exemplary demonstrations of artificial molecular machineriesthat employ functionalized nanostructures include mesoporous nanocrystalsmodified with molecular nanoimpellers, valves, or gates for the capture andrelease of cargo with external control [140].Numerous switch motifs have been utilized as “gatekeepers” to control

payload release from these versatile materials including coumarins [141, 142]

�

� �

�


and azobenzenes [143, 144]. Besides the necessity for biocompatibility, tissuespecificity, and high loading capability, to implement these systems in vivo forbiomedical applications such as controlled drug release, noninvasive actuationmechanisms are also required because some stimuli may be detrimental tobiological environments [145]. In a recent example, a nanoimpeller systemdeveloped by Croissant et al. utilizes azobenzene photoisomerization as adriving force to release the anticancer drug camptothecin from mesoporoussilica nanoparticles (Figure 2.8) [146].Contrary to classic designs in which azobenzene trans-to-cis isomerization

is triggered by UV light (which is harmful to living cells), this system is basedon two-photon excitation (TPE) of a fluorophore with NIR light. The use ofNIR light facilitates isomerization of the azobenzene moieties through Försterresonance energy transfer (FRET) from a nearby fluorophore. Isomerization ofthe azobenzene nanoimpellers subsequently kicks out the camptothecin cargo,leading to cancer cell death in vitro. Using TPE with NIR light to trigger drugrelease from mesoporous nanoparticles has not yet been extensively exploredbut offers the benefits of deeper tissue penetration in the biological spectralwindow (700–1000 nm) and lower scattering loss [147, 148]. The TPE-baseddesigns illustrate how the actuated mechanics of photoswitches can be tai-lored by their immediate surroundings by coupling simple switches or rotors totheir nanostructured environment, provided that the fluorophores have largetwo-photon absorption cross sections and sufficient emission quantum yields(>0.5) for FRET.Azobenzene chromophores hold much promise for application as artificial

muscles through linear polymerization, placing the azo photoswitch directlyinto the polymer backbone instead of just the side groups. By incorporatingazobenzene within the main chain of a linear assembly, the culmination ofmodest dimensional changes of merely a few Ångström for each chromophorecan amplify in concert and result in dramatic changes in the contour lengthof the polymer. Utilizing this strategy, Gaub and coworkers demonstrated thecapability of individual polyazobenzene peptides to perform mechanical workby tethering one end of the chain to a substrate and the other to a flexiblecantilever to measure the force exerted by the contracting polymer uponphotoisomerization [26]. The extent of polymer deformation, and thus theusefulness of the molecules for optomechanical applications, depends on boththe conformational rigidity of the backbone and minimization of electroniccoupling between azobenzene moieties [149]. The synthesis of rigid-rodpolymers that include azobenzene within a poly(para-phenylene) backboneis one strategy to maximize photodeformation, enabling accordion-likecompression and extension of chains upon cycling with UV and visible light(Figure 2.9a) [150]. Lee and coworkers demonstrated that these single-chainpolymeric assemblies may even exhibit crawling movements when depositedonto an octadecylamine-modified graphite surface and imaged with scanning

�

� �

�


1.0

0.8

0.6

0.4

Absorb

ance

0.2

0.0300

OO

OO

NH

N N HN

OO

SOOO

O O

Two-photon fluorophore

O

OO

NNH

SiOO

O

C8H17O

OC8H17

NN O

O

AA FH

NN

NN

2 hv(760 nm)

100 nmMAFAF

NPs

FRET

NN

NN

NN

NN

DRUG

(e) (f)

(d)

(b)

(a)

(c)400 500 600

Absorption

Azobenzene

Emission

fluorophore

Wavelength (nm)

Azobenzene nanoimpeller

Figure 2.8 Two-photon excitation (TPE) of a fluorophore to facilitate Förster resonanceenergy transfer (FRET) to photoisomerize azobenzene nanoimpellers on mesoporous silicananocrystals and subsequent cargo release. (a) Overlap of the emission spectrum of thefluorophore and absorption spectrum of the azobenzene nanoimpeller enables FRET.(b) The chemical structure of the fluorophore. (c) Structure of the two-photon fluorophore.(d) Photoisomerization of azobenzene using two-photon (760 nm) excitation of thefluorophore. (e) Schematic of the mesoporous silica nanocrystal. (f ) Transmission electronmicroscopy image of a single nanocrystal. Light-activated nanovalves that utilizenear-infrared irradiation such as this TPE-based mechanism show promise for targeted drugdelivery applications and should be further explored to extend their scope. (Croissant et al.[146]. Reproduced with the permission of John Wiley & Sons.)

�

� �

�


357 nm

Extended

(a)

(b)

> 400 nm

Compressed

n

P1

(86%-Z)

P1

(100%-E)

n > 30

UV

9 min

UV

19 min

20 nm

X

XX

20 nm 20 nm

R = C12H25

NN

NN

R

R

R

R

Φ

Φ

III

III

i ii iii

Figure 2.9 (a) Schematic of a main-chain azobenzene-containing polymer (P1; R=C12H25)with a poly(para-phenylene) backbone. Irradiation with ultraviolet (UV) or visible lightfacilitates photoisomerization of azobenzene and conversion to the compressed andextended conformations, respectively. (Bleger et al. [150]. Reproduced with the permissionof John Wiley & Sons.) (b) Scanning force microscopy images of P1 deposited on a modifiedgraphite surface. The polymer crawls along the surface as it contracts upon UV irradiation.Demonstrating control over movement direction and the functionalization or tethering ofthe polymer strands to scaffolds may enable the macromolecules to perform work by liftingweights or transporting cargo. (Lee et al. [151]. Reproduced with the permission of ACSNano.)

�

� �

�


force microscopy (Figure 2.9b) [151]. Chemically or physically cross-linkedsupramolecular assemblies of these linear photomechanical polymers maybe envisioned to behave as actuators, to lift weights, and to perform othertypes of work with greater resistance to deformation fatigue compared toindividual strands [152]. For example, Fang et al. reported on using a simplemelt spinning method to fabricate hydrogen-bonded cross-linked fibersof azobenzene-containing main-chain polymers that were prepared via aMichael addition reaction (Figure 2.10) [153]. The authors also investigatedthe photoinduced mechanical properties of the fibers, reporting a maximumstress generated by a single fiber of 240 kPa upon UV irradiation at 35 ∘C.Thisforce is similar to the maximal tension forces of some chemically cross-linkedazobenzene-containing polymer fibers and even human striated muscles(ca. 300 kPa) [80].

2.4.2 Two-Dimensional Assemblies and Surface Functionalization

To leverage the mechanical motion of ensembles of molecules, directionalityis mandatory to overcome the chaotic (isotropic) generation and application offorce. Similarly to Archimedes’ need for a place to stand tomove the Earth witha lever, a surface may be utilized to instill directionality to harness the power oflarge numbers of photochromic molecular machines. Two-dimensional cover-age by molecular switches and rotors on planar surfaces provides advantages

O

NN

O

O

Vis, 60 s

from right

Vis, 60 s

from left

UV, 15 s

from right

UV, 15 s

from leftNH x

MP-m (m = 2, 6, 10)

O(CH2)mO

O

NN

O

O NH3+–

OOCCF3M-m (m = 2,6, 10)

TEA

(a)

(b)

(c)Michael addition

O(CH2)m O

Figure 2.10 (a) Synthetic route and chemical structures of acrylate-type azobenzenemonomers. (b) Supramolecular hydrogen-bonding interactions between main-chainpolymers to facilitate physical cross-linking. (c) Photographs of a polymeric fiber fabricatedby simple melt spinning. The fiber reversibly bends upon irradiation with ultraviolet andvisible light. The fibers demonstrate robust photodeformation fatigue resistance and highthermal stability and show promise for applications as photomechanical actuators. (Fanget al. [153]. Reproduced with the permission of American Chemical Society.)

�

� �

�


over isolated molecules or functionalized nanoparticles by facilitating themanipulation of physical and chemical properties of a material at the micro-,meso-, and macroscales. For example, through the amplification of collectivemolecular mechanical motion, the integration of small-molecule switches androtors into ordered arrays has resulted in dynamic control over work function,refractive index, and surface wettability [41, 154–157]. Molecular pumpsbased on host–guest interactions composed of cyclodextrin and azobenzenedesigned by Sen and coworkers perform such a function by external stim-ulation with light [158]. These hybrid systems are organized within gels oradsorbed directly on glass substrates (Figure 2.11a and b). Upon UV lightabsorption, azobenzene molecules isomerize and leave their cyclodextrinhosts. The created cavity is then promptly filled with water molecules. Theamplified and collective actions of the multitude of neighboring pumps createa steady flow of fluid around the surface at a rate of about 2 μm/s (Figure 2.11cand d). The pump can also be activated by chemical stimuli and rechargedby visible light irradiation. Despite these impressive examples, there existfew reports on the integration of molecular switches and rotors in planarassemblies because of the challenging design rules that accompany surfacefunctionalization, as described as follows. Self-assembled monolayers (SAMs),Langmuir–Blodgett (LB) films, and layer-by-layer (LBL) assemblies are allrelatively well-understood organic thin-film technologies that can be employedto fabricate nanoscale functional surfaces [159]. Within SAMs, intermoleculardistances, molecular orientation, and substrate–molecule interactions stronglyinfluence whether assembled switches and rotors retain their functionality dueto the varied chemistries of their interfaces [160]. Physically and electronicallydecoupling these functional moieties from surfaces or from neighboringmolecules is often necessary to avoid steric constraints or quenching of excitedstates [138, 161, 162]. For example, molecular rotors can be tethered such thatthe axis of rotation is aligned parallel or perpendicular to the surface, in eitheraltitudinal or azimuthal orientations, respectively. Feringa and coworkersreported the tunable and reversible wettability of gold surfaces modified withSAMs of altitudinal rotors based on light-driven overcrowded alkenes bearingperfluorinated alkyl chains [163].Taking advantage of unhindered rotation enabled by the superior altitudi-

nal orientation of the rotor units dramatically modified the surface energy withresulting water contact angle changes of as much as 8–22∘ owing to differencesin the orientation of the hydrophobic perfluorobutyl group. The photoconver-sion efficiency and rotation speed of these surface-bound rotors are still gen-erally lower than those for free molecules in solution, highlighting how properspatial arrangement and sufficient room to rotate are necessary parameters tooptimize these dynamic molecular motifs to retain their large-scale function-ality [164].

�

� �

�


Physical stimuli

Fluid chamber

(PEG) NN

H2O

H2O

H2O

(PEG) H2O

Tracer particles

O

GelGel

0 min

(c)

(b)

(a)

(d)

60 min

O

OO

OSi

O

OO

O

OO

OSi

GlassGlass

O

i-Pr

x

y

OH2Otrans-Azo

Flu

id flo

w

cis-Azo

O

β-CD gel

α-CD

Top layer

Bottom layer

O

O

OO

OSi

Glass

O

i-Pr

x

y4-Acryloyl-4′-dimethyl

aminoazobenzene

N-i-Propyl acrylamide UV

Vis

β-CD gel

Chemical stimuli

Figure 2.11 (a) Schematic of a dual-responsive micropump on a glass surface. Light orchemical stimuli may be used to induce fluid flow by a β-cyclodextrin–polyethylene glycol(β-CD-PEG) gel upon isomerization of the azobenzene moiety. (b) Schematic of the directfunctionalization of glass surfaces by covalently tethering azobenzene-containingmolecules. Reversible formation or disassociation of the host/guest complex withα-cyclodextrin results in fluid pumping. (c) Optical microscopy image of tracer particles insolution above a β-CD-PEG gel on a glass surface before irradiation. (d) Optical microscopyimage of tracer particles accumulating at the edge of a β-CD-PEG gel after irradiation withultraviolet light for 1 h. Scale bars, 50 μm. The reversibility of the host/guest interactionmakes the design particularly appealing for rechargeable microdevices. (Patra et al. [158].Reproduced with the permission of American Chemical Society.)

�

� �

�


2.5 Surface Mass Transport and Phase Change Effects

In 1995, an unexpected and unprecedented optical effect was discoveredin polymer thin films containing the azo chromophore DR1. The Natan-sohn/Rochon research team [165] and the Tripathy/Kumar collaboration[166] simultaneously and independently discovered a large-scale surface masstransport when the films were irradiated with a light interference pattern.In this experiment, two coherent laser beams with a wavelength in the azoabsorption band are intersected at the sample surface to interfere. The sampleusually consists of a thin spin-cast film (10–1000 nm) of an amorphous azopolymer on a transparent substrate. The sinusoidal light interference patternat the sample surface leads to a sinusoidal surface topology patterning, thatis, a relief grating often referred to in the literature as a surface relief grating(SRG), though the effect is not limited to just gratings, and might more accu-rately and generally be called photopatterning, phototransport, or all-opticalpatterning (Figure 2.12). These gratings were found to be significantly large,up to hundreds of nanometers, as confirmed by AFM, which means that thelight induced the motion of many hundreds of nanometers of the polymerchains to “walk” across the substrate surface. The SRGs diffract very cleanlyand efficiently, and in retrospect, it is clear that many early reports on the largediffraction efficiency prior to 1995, attributed then to birefringence, were in

600

300

0

4

3

2

1

12

34

5y(μm)

z(nm)

x(μm)

Figure 2.12 AFM image of a typical surface relief grating (SRG) optically inscribed into anazo polymer film. Grating amplitudes of hundreds of nanometers, on the order of theoriginal film thickness, are easily obtained. In this image, the approximate location of thefilm–substrate interface has been set to z = 0, based on the knowledge of the film thickness.(Mahimwalla et al. [2]. Reproduced with the permission of Springer.)

�

� �

�

2.5 Surface Mass Transport and Phase Change Effects 63

fact probably due instead to surface gratings unbeknown to the experimenters.The process occurs readily at room temperature (well below the Tg of theamorphous polymers used) with moderate irradiation (1–100mW/cm2) overseconds to minutes. The phenomenon is a reversible mass transport, notirreversible material ablation, since a flat film with the original thickness isrecovered upon heating above Tg. Critically, it requires the presence andisomerization of azobenzene chromophores, as other absorbing but noniso-merizing chromophores do not produce SRGs.Many other systems can exhibitoptical surface patterning [167], but the amplitude of the modification is muchsmaller, does not involve mass transport, and usually requires additionalprocessing steps. The all-optical patterning unique to azobenzenes has beenstudied intensively since its discovery, and many reviews of the remarkablebody of experimental results are available [168–171].In a typical inscription experiment, a sinusoidally varying light pattern is

generated at the sample surface, and what results is a sinusoidal surface pro-file: an SRG. This is the pattern most often reported in the literature, becauseit is most conveniently generated (by intersecting two coherent beams) andmost easily monitored (by recording the diffraction intensity at a nonabsorb-ing wavelength, usually using a HeNe laser at 633 nm). However, it must beemphasized that the azo surfacemass transport can produce arbitrary patterns.Essentially, the film encodes the impinging light pattern as a topography pattern(as a Fourier transform), holographically encoding both the spatial intensityand the polarization patterns of the incident light. What appears to be essen-tial is a gradient in the intensity and/or polarization of the incident light field.For instance, a single focused Gaussian laser spot will lead to a localized pitdepression, and a Gaussian line will lead to an elongated trench [172]. In prin-ciple, any arbitrary pattern could be generated through an appropriate mask,interference/holographic setup, or laser rastering [170].Concomitant with the inscription of a surface relief is a photo-orientation

of the azo chromophores, which depends on the polarization of the incidentbeam(s). The orientation of chromophores in SRG experiments has been mea-sured using polarized Raman confocal microspectrometry [173–175], and thestrong surface orientation has been confirmed by photoelectron spectroscopy[176]. What is found is that the chromophores orient perpendicular to thelocal polarization vector of the impinging interference pattern. Thus, for a(+45∘, −45∘) two-beam interference, in the valleys (x= 0), the electric fieldis aligned in the y-direction, so the chromophores orient in the x-direction;in the peaks (x=Λ/2), the chromophores orient in the y-direction; and inthe slope regions (x=Λ/4), the electric field is circularly polarized and thusthe chromophores are nearly isotropic. For a (p, p) two-beam interference, itis observed that the chromophores are primarily oriented in the y-directioneverywhere, since the impinging light pattern is always linearly polarized in

�

� �

�


the x-direction. Mass transport may lead to perturbations in the orientationaldistribution, but photo-orientation remains the dominant effect.The anisotropic volume grating that is submerged below a SRG apparently

also leads to the formation of a density grating under appropriate conditions.It was found that upon thermal annealing (heating just beyond Tg) of an SRG,which erases the surface grating and restores a flat film surface, a densitygrating began growing beneath the surface (and into the film bulk) [177, 178].This density grating only develops where the SRG was originally inscribed, andit appears that the photo-orientation and mass transport lead to the nucleationof liquid-crystalline “seeding aggregates” that are thermally grown into largerscale density variations. The thermal erasure of the SRG, with concomitantgrowth of the density grating, has been both measured [179] and modeled[180]. The diffraction of a visible-light laser primarily probes the surfacerelief, whereas a simultaneous X-ray diffraction experiment probes the densitygrating. The formation of a density grating is similar to, and consistent with,the production of surface topography [181] and surface density patterns [182],as observed by tapping-mode AFM on an azo film exposed to an optical nearfield. In these experiments, it was found that volume is not strictly conservedduring surface deformation [183], consistent with changes in density.Mass transport effects are not limited to the polymeric azo materials but can

also occur in crystals of photochromic molecules. Primarily, such observationsrepresent directional melting and crystal growth as well as self-propulsion in amedium, which is related to the previously discussed photosalient crystals. Afirst entry into this field was provided by Milam et al. with the observation ofswimming, sinking, and stationary azobenzene crystals in a triacrylate solution[184]. The motility was rationalized by the creation of concentration/surfacetension gradients around the crystal/liquid interface upon exclusion of tri-acrylate solvent from the growing crystal front. More recently, Hoshino et al.demonstrated how irradiation-induced trans–cis conversion in the crystals ofazobenzenes can lead to directed melting of crystals [185]. Simultaneously,Norikane et al., by careful choice of the position and identity of a substituent ofon the azobenzene core, have identified the structures amenable to melting byUV light (Figure 2.13) and utilized this phase transition to selectively patterna copper surface [186]. These observations once again proved the feasibilityof reversible trans–cis-azobenzene isomerization in carefully tailored singlecrystals.The observation of light-mediated melting also led to a completely new phe-

nomenon – apparent directional “crawling” of single crystals on a glass surfacedriven by the melting transition. Upon simultaneous visible and UV irradi-ation, single crystals of trans-3,3′-dimethylazobenzene “crawl” along the flatglass surface (and even vertically) away from the UV light source (Figure 2.14)[187]. The motion is driven by melting and crystallization of the crystals at thefront and rear edge, andwhile the shape of the crystal continuously changes, the

�

� �

�

2.6 Photochromic Reactions in Framework Architectures 65

RO N RO RON N

N NOR OR OR

3a: R = C12H25

3b: R = C6H13

2a: R = C12H25

2b: R = C6H13

2c: R = C10H21

1a: R = C12H25

1b: R = C6H13

(a)

(b) (c)

N

Figure 2.13 (a) Structures and (b) photographs of the crystalline powders of azobenzenesutilized in the study. (c) The same powders after irradiation with 365 nm light for 30 min at100 mW/cm2. (Norikane et al. [186]. Reproduced with the permission of American ChemicalSociety.) (See color plate section for the color representation of this figure.)

optical axis remains constant. Melting-driven motility of crystals is an impor-tant step in the development of self-propelled objects [188, 189] and enhancesthe understanding of crawling phenomenon already observed in photochromicazobenzene-containing glasses and polymers [190, 191]. Such directed surfacetransport of azobenzene materials is a complimentary (and inverted) observa-tion of the liquid mass transport on the surface of azobenzene-functionalizedsurfaces pioneered by Ichimura et al. [192–194]

2.6 Photochromic Reactions in FrameworkArchitectures

A new avenue of research in solid-state photochromic reactions was openedwith the development of metal–organic frameworks (MOFs) and similarnetwork-type crystalline systems [195]. Specifically, upon realization thatMOFs can survive in relatively harsh environments and are capable of post-functionalization while retaining very large surface area [196, 197], effortswere made to prepare and study photoswitchable MOFs with the goal ofcarbon dioxide absorption (Figure 2.15) [198, 199]. Following the idea thatazobenzene molecules can only isomerize when used as pendant groups on

�

� �

�


(a) (c)

10 m

10 m

50 m 50 m 50 m

50 m 50 m 50 m

365 nm 465 nm

Sample

θUV

θVIS

φ

Microscope

(b)

(d)

(g) (h)

(e) (f)

(i)

Figure 2.14 (a and b) Motion of single crystals of trans-3,3′-dimethylazobenzene on a glasssurface. (c) Schematic representation of the irradiation setup; (d–i) microscope images oftranslational motion of trans-3,3′-dimethylazobenzene after irradiation time, t, min 0 (d), 3(b), 6 (f ), 10 (g), 15 (h), 20 (b). Dashed white and dark gray lines represent the initial positionsof crystals and droplets, respectively. (Uchida et al. [187]. Reproduced with the permission ofNature Publishing Group.)

linkers, photoactive MOFs were prepared and successfully tested to regulatemethane absorption [200]. The azobenzene chromophore usually does notact as a truly bistable switch as the lifetime of the cis form is usually short.Recently, however, the Hecht group has reported o-fluorinated azobenzenesthat have lifetimes of a cis form of over 2 years in solution [201]. Castellanoset al. utilized such fluorinated azobenzenes to prepare a MOF, which isaddressable by green and blue light and has potential as a bistable gas-storingmaterial [202].A simple and elegant approach to photoswitchable gas absorption in MOFs

was proposed by Lyndon et al. Instead of covalent modification of an MOF ora linker unit, they opted for postsynthetic treatment of the surface of a MOF

�

� �

�

2.6 Photochromic Reactions in Framework Architectures 67

HO

(a)

(b)

HO

O

O

NN UV

HO

HO

O

O

NN

Δ

Figure 2.15 (a) Isomerization of the azobenzene ligand within an MOF referred to asPCN-123. (b) Schematic illustration of CO2 uptake in the parent MOF-5 structure andPCN-123 network in trans and cis states. (Park et al. [198]. Reproduced with the permissionof American Chemical Society.)

by the azobenzene dye, methyl red. The dye coated the surface and preventedabsorption of CO2 inside the MOF [203]. However, upon irradiation, the poreswould open and the MOF absorbed CO2. In the same vein, photochromism ofazobenzene molecules inside the pores of an MOF was shown by the Kitagawagroup to direct structural changes in the network and, as a consequence,regulate gas sorption [204]. Incorporation of azobenzene chromophore asa guest molecule into the network led to a phase change of the networkfrom tetragonal to orthorhombic crystal system upon UV irradiation, whichresulted in drastically different uptake profiles. While most of the researchis performed under the assumption that switching of azobenzene is onlypossible in an MOF when chromophores are used as pendant groups [199], arecent report by Baroncini et al. may warrant a closer inspection of this idea.Tetrameric star-shaped azobenzene molecules assemble a porous networkand then undergo reversible isomerization aided by partial amorphizationof the sample [205]. The isomerization in its turn changes the porosity andconsequently the gas uptake of the network.While most of the effort in photoresponsive MOFs was directed toward

the study of azobenzene-type chromophores, Walton et al. produced pho-tochromic architectures employing diarylethene chromophore [206, 207].Irradiation of a crystal they refer to as UBMOF-1 with UV light would turn the

�

� �

�


crystal red, indicating a successful ring-closure reaction of the diarylethenechromophore. However, unlike in the work by Irie and coworkers on pho-tochromic crystals [109], when diarylethene photochromes are introducedinto the MOF scaffold, the reverse reaction can only take place upon digestionof the MOF into its constituents by a strong acid. The reversible switchingof diarylethene unit inside a network was since achieved by Luo et al., whichallowed light-triggered desorption of up to 75% of CO2 upon sequential UVand visible irradiation [208].

2.7 Summary and Outlook

As discussed in this chapter, reversible photoisomerization of photochromicmolecules in the solid state can be leveraged to control larger scale materialproperties in response to light. While azobenzene is the most studied of thechromophores, various other photoswitches are increasingly being utilized.Light is an efficient power source for many of these applications, offering adirect conversion of photonic energy into mechanical motion without require-ments for energy converters, amplification, or other subsystems. Light is alsoan ideal triggering mechanism, since it can be localized (in time and space), isselective, nondamaging, and allows remote activation and remote delivery ofenergy to a system. Thus, for sensing, actuation, and motion, photoresponsivematerials are of great interest. Photochromic materials have demonstrated awide variety of switching behaviors, from altering optical properties, to surfaceenergy changes, to even eliciting bulk material phase changes. Azobenzene isa leader among the small class of photoreversible molecules, and azo crystals,polymers, and other supramolecular azo materials are promising candidatesfor enabling the potential applications of these systems discussed in this bookbecause of their ease of incorporation and efficient and robust photochemistry.At the same time, for nonpolymeric materials, diarylethenes have shown greatpromise despite their overall lower fatigue resistance. This chapter describedthe light-induced effects observed in thin films, crystals, amorphous polymers,and LCNs and LCEs containing various photochromes. The effects rangefrom full macroscopic light-driven actuation to matter transport across thesurfaces, phase changes, and modification of gas sorption and storage capacity.The unifying limitation, however, is that the mechanical forces produced thusfar and the efficiency for light energy conversion are still far from optimal.LCEs in particular are promising materials for artificial muscles and motorsdriven by light, and in these systems, not only two-dimensional but alsothree-dimensional motions have now been achieved, which are competitiveand promising for many applications as soft actuators. However, many prob-lems still remain unsolved, such as fatigue resistance and biocompatibility ofthese materials, which need further intensive investigation.

�

� �

�

References 69

References

1 Yamada, M., Kondo, M., Mamiya, J.I. et al. (2008) Angewandte Chemie,International Edition, 47, 4986–4988.

2 Mahimwalla, Z., Yager, K.G., Mamiya, J. et al. (2012) Polymer Bulletin, 69,967–1006.

3 Bandara, H.M.D. and Burdette, S.C. (2012) Chemical Society Reviews, 41,1809–1825.

4 Wirth, J., Hatter, N., Drost, R. et al. (2015) Journal of Physical ChemistryC, 119, 4874–4883.

5 Irie, M., Fukaminato, T., Matsuda, K., and Kobatake, S. (2014) ChemicalReviews, 114, 12174–12277.

6 Berkovic, G., Krongauz, V., and Weiss, V. (2000) Chemical Reviews, 100,1741–1753.

7 Trenor, S.R., Shultz, A.R., Love, B.J., and Long, T.E. (2004) ChemicalReviews, 104, 3059–3077.

8 Whalley, A.C., Steigerwald, M.L., Guo, X., and Nuckolls, C. (2007) Journalof the American Chemical Society, 129, 12590–12591.

9 Bauer, J., Hou, L.L., Kistemaker, J.C.M., and Feringa, B.L. (2014) Journal ofOrganic Chemistry, 79, 4446–4455.

10 Su, X., Voskian, S., Hughes, R.P., and Aprahamian, I. (2013) AngewandteChemie International Edition, 52, 10734–10739.

11 Greb, L. and Lehn, J.M. (2014) Journal of the American Chemical Society,136, 13114–13117.

12 Tatum, L.A., Su, X., and Aprahamian, I. (2014) Accounts of ChemicalResearch, 47, 2141–2149.

13 Qu, D.-H., Wang, Q.-C., Zhang, Q.-W. et al. (2015) Chemical Reviews, 115(15), 7543–7588.

14 Astumian, R.D. (2012) Nature Nanotechnology, 7, 684–688.15 Ichimura, K. (2000) Chemical Reviews, 100, 1847–1873.16 Naito, T., Horie, K., and Mita, I. (1991) Macromolecules, 24, 2907–2911.17 Uznanski, P., Kryszewski, M., and Thulstrup, E.W. (1991) European Poly-

mer Journal, 27, 41–43.18 de Lange, J.J., Robertson, J.M., and Woodward, I. (1939) Proceedings of the

Royal Society of London A, A171, 398–410.19 Hampson, G.C. and Robertson, J.M. (1941) Journal of the Chemical Soci-

ety, 409–413.20 Brown, C.J. (1966) Acta Cryst, 21, 146–152.21 Hartley, G.S. (1937) Nature, 140, 281.22 Naito, T., Horie, K., and Mita, I. (1993) Polymer, 34, 4140–4145.23 Paik, C.S. and Morawetz, H. (1972) Macromolecules, 5, 171–177.24 Lamarre, L. and Sung, C.S.P. (1983) Macromolecules, 16, 1729–1736.

�

� �

�


25 Weiss, R.G., Ramamurthy, V., and Hammond, G.S. (1993) Accounts ofChemical Research, 25, 530–536.

26 Hugel, T., Holland, N.B., Cattani, A. et al. (2002) Science, 296, 1103–1106.27 Holland, N.B., Hugel, T., Neuert, G. et al. (2003) Macromolecules, 36,

2015–2023.28 Neuert, G., Hugel, T., Netz, R.R., and Gaub, H.E. (2005) Macromolecules,

39, 789–797.29 Tsuchiya, S. (1999) Journal of American Chemical Society, 121, 48–53.30 Steinem, C., Janshoff, A., Vollmer, M.S., and Ghadiri, M.R. (1999) Lang-

muir, 15, 3956–3964.31 Vollmer, M.S., Clark, T.D., Steinem, C., and Ghadiri, M.R. (1999) Ange-

wandte Chemie, International Edition, 38, 1598–1601.32 Masiero, S., Lena, S., Pieraccini, S., and Spada, G.P. (2008) Angewandte

Chemie, International Edition, 47, 3184–3187.33 Fujiwara, M., Akiyama, M., Hata, M. et al. (2008) ACS Nano, 2,

1671–1681.34 Pakula, C., Zaporojtchenko, V., Strunskus, T. et al. (2010) Nanotechnology,

21 (46), 465201.35 Raimondo, C., Reinders, F., Soydaner, U. et al. (2010) Chemical Communi-

cations, 46, 1147–1149.36 Kimoto, A., Iwasaki, K., and Abe, J. (2010) Photochemical and Photobiolog-

ical Sciences, 9, 152–156.37 Higuchi, M., Minoura, N., and Kinoshita, T. (1995) Colloid and Polymer

Science, 273, 1022–1027.38 Siewierski, L.M., Brittain, W.J., Petrash, S., and Foster, M.D. (1996) Lang-

muir, 12, 5838–5844.39 Stiller, B., Knochenhauer, G., Markava, E. et al. (1999) Materials Science

and Engineering: C , 8–9, 385–389.40 Moller, G., Harke, M., Motschmann, H., and Prescher, D. (1998) Lang-

muir, 14, 4955–4957.41 Feng, C.L., Zhang, Y.J., Jin, J. et al. (2001) Langmuir, 17, 4593–4597.42 Chen, T., Xu, S., Zhang, F. et al. (2009) Chemical Engineering Science, 64,

4350–4357.43 Delorme, N., Bardeau, J.-F., Bulou, A., and Poncin-Epaillard, F. (2005)

Langmuir, 21, 12278–12282.44 Diguet, A., Guillermic, R.-M., Magome, N. et al. (2009) Angewandte

Chemie, International Edition, 48, 9281–9284.45 Sarkar, N., Sarkar, A., and Sivaram, S. (2001) Journal of Applied Polymer

Science, 81, 2923–2928.46 Jiang, W.H., Wang, G.J., He, Y.N. et al. (2005) Chemical Communications,

3550–3552.47 Yu, Y., Nakano, M., and Ikeda, T. (2003) Nature, 425, 145.

�

� �

�

References 71

48 Ikeda, T., Nakano, M., Yu, Y. et al. (2003) Advanced Materials, 15,201–205.

49 Yu, Y.L., Nakano, M., Maeda, T. et al. (2005) Molecular Crystals andLiquid Crystals, 436, 1235–1244.

50 Ji, H.F., Feng, Y., Xu, X.H. et al. (2004) Chemical Communications,2532–2533.

51 White, T.J., Tabiryan, N.V., Serak, S.V. et al. (2008) Soft Matter, 4,1796–1798.

52 White, T.J., Serak, S.V., Tabiryan, N.V. et al. (2009) Journal of MaterialsChemistry, 19, 1080–1085.

53 Serak, S., Tabiryan, N., Vergara, R. et al. (2010) Soft Matter, 6, 779–783.54 Deng, W., Li, M.H., Wang, X.G., and Keller, P. (2009) Liquid Crystals, 36,

1023–1029.55 Hrozhyk, U., Serak, S., Tabiryan, N. et al. (2009) Optics Express, 17,

716–722.56 Kondo, M., Matsuda, T., Fukae, R., and Kawatsuki, N. (2010) Chemistry

Letters, 39, 234–235.57 Cheng, F.T., Zhang, Y.Y., Yin, R.Y., and Yu, Y.L. (2010) Journal of Materials

Chemistry, 20, 4888–4896.58 Bai, S. and Zhao, Y. (2001) Macromolecules, 34, 9032–9038.59 Camacho-Lopez, M., Finkelmann, H., Palffy-Muhoray, P., and Shelley, M.

(2004) Nature Materials, 3, 307–310.60 Bublitz, D., Helgert, M., Fleck, B. et al. (2000) Applied Physics B, 70,

863–865.61 Chen, X., Tao, J., Zou, G. et al. (2010) Applied Physics A: Materials Science

& Processing, 100, 223–230.62 Merian, E. (1966) Textile Research Journal, 36, 612–618.63 Eisenbach, C.D. (1980) Polymer, 21, 1175–1179.64 Matejka, L., Dusek, K., and Ilavsky, M. (1979) Polymer Bulletin, 1,

659–664.65 Matejka, L. and Dusek, K. (1981) Makromolekulare

Chemie-Macromolecular Chemistry and Physics, 182, 3223–3236.66 Matejka, L., Ilavsky, M., Dusek, K., and Wichterle, O. (1981) Polymer, 22,

1511–1515.67 Kim, H.K., Wang, X.S., Fujita, Y. et al. (2005) Macromolecular Rapid Com-

munications, 26, 1032–1036.68 Kim, H.K., Wang, X.S., Fujita, Y. et al. (2005) Polymer, 46, 5879–5883.69 Kim, H.K., Wang, X.S., Fujita, Y. et al. (2005) Macromolecular Chemistry

and Physics, 206, 2106–2111.70 Tanaka, S., Kim, H.K., Sudo, A. et al. (2008) Macromolecular Chemistry

and Physics, 209, 2071–2077.71 Zhang, C., Zhao, X., Chao, D. et al. (2009) Journal of Applied Polymer Sci-

ence, 113, 1330–1334.

�

� �

�


72 Küpfer, J., Nishikawa, E., and Finkelmann, H. (1994) Polymers forAdvanced Technologies, 5, 110–115.

73 Wermter, H. and Finkelmann, H. (2001) e-Polymers, 1, 111–123.74 Gennes, P.G.D., Hebert, M., and Kant, R. (1997) Macromolecular Sym-

posia, 113, 39–49.75 Ikeda, T., Mamiya, J., and Yu, Y.L. (2007) Angewandte Chemie, Interna-

tional Edition, 46, 506–528.76 Elias, A.L., Harris, K.D., Bastiaansen, C.W.M. et al. (2006) Journal of Mate-

rials Chemistry, 16, 2903–2912.77 Yang, H., Ye, G., Wang, X., and Keller, P. (2011) Soft Matter, 7, 815–823.78 Yu, Y., Nakano, M., and Ikeda, T. (2004) Pure and Applied Chemistry, 76,

1467–1477.79 Nakano, H. (2010) Journal of Materials Chemistry, 20, 2071–2074.80 Yoshino, T., Kondo, M., Mamiya, J. et al. (2010) Advanced Materials, 22,

1361–1363.81 Kondo, M., Yu, Y., and Ikeda, T. (2006) Angewandte Chemie, International

Edition, 45, 1378–1382.82 Tabiryan, N., Serak, S., Dai, X.M., and Bunning, T. (2005) Optics Express,

13, 7442–7448.83 Van Oosten, C.L., Corbett, D., Davies, D. et al. (2008) Macromolecules, 41,

8592–8596.84 Priimagi, A., Shimamura, A., Kondo, M. et al. (2011) ACS Macro Letters,

1, 96–99.85 Yamada, M., Kondo, M., Miyasato, R. et al. (2009) Journal of Materials

Chemistry, 19, 60–62.86 Naka, Y., Mamiya, J.-I., Shishido, A. et al. (2011) Journal of Materials

Chemistry, 21, 1681–1683.87 Yin, R.Y., Xu, W.X., Kondo, M. et al. (2009) Journal of Materials Chem-

istry, 19, 3141–3143.88 Cheng, F., Yin, R., Zhang, Y. et al. (2010) Soft Matter, 6, 3447–3449.89 Lee, K.M., Koerner, H., Vaia, R.A. et al. (2011) Soft Matter, 7, 4318–4324.90 Wu, W., Yao, L., Yang, T. et al. (2011) Journal of American Chemical

Society, 133, 15810–15813.91 Harris, K.D., Cuypers, R., Scheibe, P. et al. (2005) Journal of Materials

Chemistry, 15, 5043–5048.92 Kondo, M., Sugimoto, M., Yamada, M. et al. (2010) Journal of Materials

Chemistry, 20, 117–122.93 Lee, K.M., Koerner, H., Vaia, R.A. et al. (2010) Macromolecules, 43,

8185–8190.94 Shimamura, A., Priimagi, A., Mamiya, J.-I. et al. (2011) Journal of Nonlin-

ear Optical Physics and Materials, 20, 405–413.95 Liu, D. and Broer, D.J. (2015) Nature Communications, 6, Article number:

8334.

�

� �

�

References 73

96 Ionov, L. (2015) Langmuir, 31, 5015–5024.97 Ware, T.H., McConney, M.E., Wie, J.J. et al. (2015) Science, 347, 982–984.98 Iamsaard, S., Aßhoff, S.J., Matt, B. et al. (2014) Nature Chemistry, 6,

229–235.99 Dunitz, J.D., Schomaker, V., and Trueblood, K.N. (1988) Journal of Physical

Chemistry, 92, 856–867.100 Nath, N.K., Panda, M.K., Sahoo, S.C., and Naumov, P. (2014) CrystEng-

Comm, 16, 1850–1858.101 Lange, C.W., Foldeaki, M., Nevodchikov, V.I. et al. (1992) Journal of the

American Chemical Society, 114, 4220–4222.102 Naumov, P., Sahoo, S.C., Zakharov, B.A., and Boldyreva, E.V. (2013) Ange-

wandte Chemie International Edition, 52, 9990–9995.103 Kim, T., Zhu, L., Al-Kaysi, R.O., and Bardeen, C.J. (2014) ChemPhysChem,

15, 400–414.104 Panda, M.K., Runcevski, T., Sahoo, S.C. et al. (2014) Nature Communica-

tions, 5, 4811.105 Sahoo, S.C., Panda, M.K., Nath, N.K., and Naumov, P. (2013) Journal of

American Chemical Society, 135, 12241–12251.106 Sahoo, S.C., Sinha, S.B., Kiran, M.S.R.N. et al. (2013) Journal of American

Chemical Society, 135, 13843–13850.107 Medishetty, R., Husain, A., Bai, Z.Z. et al. (2014) Angewandte Chemie

International Edition, 53, 5907–5911.108 Sahoo, S.C., Nath, N.K., Zhang, L.D. et al. (2014) RSC Advances, 4,

7640–7647.109 Kobatake, S., Takami, S., Muto, H. et al. (2007) Nature, 446, 778–781.110 Kitagawa, D. and Kobatake, S. (2014) Photochemical & Photobiological Sci-

ences, 13, 764–769.111 Morimoto, M. and Irie, M. (2010) Journal of American Chemical Society,

132, 14172–14178.112 Terao, F., Morimoto, M., and Irie, M. (2012) Angewandte Chemie Interna-

tional Edition, 51, 901–904.113 Kitagawa, D. and Kobatake, S. (2015) Chemical Communications, 51,

4421–4424.114 Zhu, L.Y., Al-Kaysi, R.O., and Bardeen, C.J. (2011) Journal of American

Chemical Society, 133, 12569–12575.115 Kim, T., Al-Muhanna, M.K., Al-Suwaidan, S.D. et al. (2013) Angewandte

Chemie International Edition, 52, 6889–6893.116 Kim, T., Zhu, L.C., Mueller, L.J., and Bardeen, C.J. (2014) Journal of Amer-

ican Chemical Society, 136, 6617–6625.117 Koshima, H., Matsuo, R., Matsudomi, M. et al. (2013) Crystal Growth &

Design, 13, 4330–4337.118 Kumar, A.S., Ye, T., Takami, T. et al. (2008) Nano Letters, 8, 1644–1648.

�

� �

�


119 Zheng, Y.B., Payton, J.L., Chung, C.H. et al. (2011) Nano Letters, 11,3447–3452.

120 Vapaavuori, J., Goulet-Hanssens, A., Heikkinen, I.T.S. et al. (2014) Chem-istry of Materials, 26, 5089–5096.

121 Goulet-Hanssens, A. and Barrett, C.J. (2013) Journal of Polymer SciencePart A: Polymer Chemistry, 51, 3058–3070.

122 Ikeda, T. and Tsutsumi, O. (1995) Science, 268, 1873–1875.123 Koshima, H., Ojima, N., and Uchimoto, H. (2009) Journal of American

Chemical Society, 131, 6890–6891.124 Bushuyev, O.S., Singleton, T.A., and Barrett, C.J. (2013) Advanced Materi-

als, 25, 1796–1800.125 Bushuyev, O.S., Tomberg, A., Friscic, T., and Barrett, C.J. (2013) Journal of

American Chemical Society, 135, 12556–12559.126 Bushuyev, O.S., Corkery, T.C., Barrett, C.J., and Friscic, T. (2014) Chemical

Science, 5, 3158–3164.127 Warner, M. and Mahadevan, L. (2004) Physical Review Letters, 92, 134302.128 Nath, N.K., Pejov, L., Nichols, S.M. et al. (2014) Journal of American

Chemical Society, 136, 2757–2766.129 Mahadevan, L. and Matsudaira, P. (2000) Science, 288, 95–99.130 Asimov, I. (1942) Astounding Science-Fiction, 1, 94.131 Kay, E.R., Leigh, D.A., and Zerbetto, F. (2007) Angewandte Chemie Inter-

national Edition, 46, 72–191.132 Coskun, A., Banaszak, M., Astumian, R.D. et al. (2012) Chemical Society

Reviews, 41, 19–30.133 Abendroth, J.M., Bushuyev, O.S., Weiss, P.S., and Barrett, C.J. (2015) ACS

Nano, 9, 7746–7768.134 Baudrion, A.L., Perron, A., Veltri, A. et al. (2013) Nano Letters, 13,

282–286.135 Zheng, Y.B., Kiraly, B., Cheunkar, S. et al. (2011) Nano Letters, 11,

2061–2065.136 Joshi, G.K., Blodgett, K.N., Muhoberac, B.B. et al. (2014) Nano Letters, 14,

532–540.137 Lee, J.-W. and Klajn, R. (2015) Chemical Communications, 51, 2036–2039.138 Klajn, R., Stoddart, J.F., and Grzybowski, B.A. (2010) Chemical Society

Reviews, 39, 2203–2237.139 Swaminathan, S., Garcia-Amorós, J., Fraix, A. et al. (2014) Chemical

Society Reviews, 43, 4167–4178.140 Song, N. and Yang, Y.-W. (2015) Chemical Society Reviews, 44, 3474–3504.141 Mal, N.K., Fujiwara, M., and Tanaka, Y. (2003) Nature, 421, 350–353.142 Guardado-Alvarez, T.M., Devi, L.S., Russell, M.M. et al. (2013) Journal of

the American Chemical Society, 135, 14000–14003.143 Mei, X., Yang, S., Chen, D.Y. et al. (2012) Chemical Communications, 48,

10010–10012.

�

� �

�

References 75

144 Li, T., Lohmann, F., and Famulok, M. (2014) Nature Communications, 5,Article number: 4279.

145 Yang, Y.W. (2011) MedChemComm, 2, 1033–1049.146 Croissant, J., Maynadier, M., Gallud, A. et al. (2013) Angewandte Chemie

International Edition, 52, 13813–13817.147 Croissant, J., Chaix, A., Mongin, O. et al. (2014) Small, 10, 1752–1755.148 Gary-Bobo, M., Mir, Y., Rouxel, C. et al. (2011) Angewandte Chemie Inter-

national Edition, 50, 11425–11429.149 Bléger, D., Dokic, J., Peters, M.V. et al. (2011) The Journal of Physical

Chemistry B, 115, 9930–9940.150 Bleger, D., Liebig, T., Thiermann, R. et al. (2011) Angewandte Chemie

International Edition, 50, 12559–12563.151 Lee, C.L., Liebig, T., Hecht, S. et al. (2014) ACS Nano, 8, 11987–11993.152 Wie, J.J., Chatterjee, S., Wang, D.H. et al. (2014) Polymer, 55, 5915–5923.153 Fang, L.J., Zhang, H.T., Li, Z.D. et al. (2013) Macromolecules, 46,

7650–7660.154 Crivillers, N., Osella, S., Van Dyck, C. et al. (2013) Advanced Materials,

25, 432–436.155 Zheng, Y.B., Yang, Y.W., Jensen, L. et al. (2009) Nano Letters, 9, 819–825.156 Chen, M. and Besenbacher, F. (2011) ACS Nano, 5, 1549–1555.157 Ahmad, N.M., Lu, X.Y., and Barrett, C.J. (2010) Journal of Materials Chem-

istry, 20, 244–247.158 Patra, D., Zhang, H., Sengupta, S., and Sen, A. (2013) ACS Nano, 7,

7674–7679.159 Zasadzinski, J.A., Viswanathan, R., Madsen, L. et al. (1994) Science, 263,

1726–1733.160 Pathem, B.K., Claridge, S.A., Zheng, Y.B., and Weiss, P.S. (2013) Annual

Review of Physical Chemistry, 64, 605–630.161 Tegeder, P. (2012) Journal of Physics: Condensed Matter, 24, 394001.162 Pathem, B.K., Zheng, Y.B., Payton, J.L. et al. (2012) Journal of Physical

Chemistry Letters, 3, 2388–2394.163 Chen, K.Y., Ivashenko, O., Carroll, G.T. et al. (2014) Journal of American

Chemical Society, 136, 3219–3224.164 London, G., Chen, K.Y., Carroll, G.T., and Feringa, B.L. (2013)

Chemistry--A European Journal, 19, 10690–10697.165 Rochon, P., Batalla, E., and Natansohn, A. (1995) Applied Physics Letters,

66, 136–138.166 Kim, D.Y., Tripathy, S.K., Li, L., and Kumar, J. (1995) Applied Physics

Letters, 66, 1166–1168.167 Yamaki, S., Nakagawa, M., Morino, S., and Ichimura, K. (2000) Applied

Physics Letters, 76, 2520–2522.168 Delaire, J.A. and Nakatani, K. (2000) Chemical Reviews, 100, 1817–1846.

�

� �

�


169 Viswanathan, N.K., Kim, D.Y., Bian, S. et al. (1999) Journal of MaterialsChemistry, 9, 1941–1955.

170 Natansohn, A. and Rochon, P. (2002) Chemical Reviews, 102, 4139–4176.171 Yager, K.G. and Barrett, C.J. (2001) Current Opinion in Solid State &

Materials Science, 5, 487–494.172 Bian, S., Li, L., Kumar, J. et al. (1998) Applied Physics Letters, 73,

1817–1819.173 Labarthet, F.L., Bruneel, J.L., Buffeteau, T., and Sourisseau, C. (2004)

Journal of Physical Chemistry B, 108, 6949–6960.174 Lagugne-Labarthet, F., Bruneel, J.L., Rodriguez, V., and Sourisseau, C.

(2004) Journal of Physical Chemistry B, 108, 1267–1278.175 Labarthet, F.L., Bruneel, J.L., Buffeteau, T. et al. (2000) Physical Chemistry

Chemical Physics, 2, 5154–5167.176 Henneberg, O., Geue, T., Pietsch, U. et al. (2004) Applied Physics Letters,

84, 1561–1563.177 Pietsch, U., Rochon, P., and Natansohn, A. (2000) Advanced Materials, 12,

1129–1132.178 Geue, T., Henneberg, O., Grenzer, J. et al. (2002) Colloids and Surfaces A,

198-200, 31–36.179 Geue, T.M., Saphiannikova, M.G., Henneberg, O. et al. (2003) Journal of

Applied Physics, 93, 3161–3166.180 Pietsch, U. (2002) Physical Review B, 66, 155430.181 Watanabe, O., Ikawa, T., Hasegawa, M. et al. (2000) Molecular Crystals

and Liquid Crystals, 345, 629–634.182 Ikawa, T., Mitsuoka, T., Hasegawa, M. et al. (2000) Journal of Physical

Chemistry B, 104, 9055–9058.183 Keum, C.D., Ikawa, T., Tsuchimori, M., and Watanabe, O. (2003) Macro-

molecules, 36, 4916–4923.184 Milam, K., O’Malley, G., Kim, N. et al. (2010) Journal of Physical Chem-

istry B, 114, 7791–7796.185 Hoshino, M., Uchida, E., Norikane, Y. et al. (2014) Journal of the American

Chemical Society, 136, 9158–9164.186 Norikane, Y., Uchida, E., Tanaka, S. et al. (2014) Organic Letters, 16,

5012–5015.187 Uchida, E., Azumi, R., and Norikane, Y. (2015) Nature Communications, 6,

Article number: 7310.188 Nakata, S., Hata, M., Ikura, Y.S. et al. (2013) Journal of Physical Chemistry

C, 117, 24490–24495.189 Nagai, K.H., Sumino, Y., Montagne, R. et al. (2015) Physical Review Letters,

114, 168001.190 Suzuki, M. and Nakano, H. (2012) Journal of Photopolymer Science and

Technology, 25, 159–160.

�

� �

�

References 77

191 Nakano, H. and Suzuki, M. (2012) Journal of Materials Chemistry, 22,3702–3704.

192 Ichimura, K., Oh, S.K., and Nakagawa, M. (2000) Science, 288, 1624–1626.193 Oh, S.-K., Nakagawa, M., and Ichimura, K. (2002) Journal of Materials

Chemistry, 12, 2262–2269.194 Kausar, A., Nagano, H., Ogata, T. et al. (2009) Angewandte Chemie, Inter-

national Edition, 48, 2144–2147.195 Yaghi, O.M. and Li, H. (1995) Journal of American Chemical Society, 117,

10401–10402.196 Li, H., Eddaoudi, M., O’Keeffe, M., and Yaghi, O.M. (1999) Nature, 402,

276–279.197 Chui, S.S.-Y., Lo, S.M.-F., Charmant, J.P.H. et al. (1999) Science, 283,

1148–1150.198 Park, J., Yuan, D., Pham, K.T. et al. (2012) Journal of American Chemical

Society, 134, 99–102.199 Modrow, A., Zargarani, D., Herges, R., and Stock, N. (2011) Dalton Trans-

actions, 40, 4217–4222.200 Modrow, A., Zargarani, D., Herges, R., and Stock, N. (2012) Dalton Trans-

actions, 41, 8690–8696.201 Bléger, D., Schwarz, J., Brouwer, A.M., and Hecht, S. (2012) Journal of the

American Chemical Society, 134, 20597–20600.202 Castellanos, S., Goulet-Hanssens, A., Zhao, F. et al. (2016) Chemistry – A

European Journal, 22, 746–752.203 Lyndon, R., Konstas, K., Thornton, A.W. et al. (2015) Chemistry of Materi-

als, 27, 7882–7888.204 Yanai, N., Uemura, T., Inoue, M. et al. (2012) Journal of American Chemi-

cal Society, 134, 4501–4504.205 Baroncini, M., d’Agostino, S., Bergamini, G. et al. (2015) Nature Chemistry,

7, 634–640.206 Walton, I.M., Cox, J.M., Coppin, J.A. et al. (2013) Chemical Communica-

tions, 49, 8012–8014.207 Patel, D.G., Walton, I.M., Cox, J.M. et al. (2014) Chemical Communica-

tions, 50, 2653–2656.208 Luo, F., Fan, C.B., Luo, M.B. et al. (2014) Angewandte Chemie, Interna-

tional Edition, 126, 9452–9455.

�

� �

�

79

3

Photomechanics: Bend, Curl, Topography, and TopologyDaniel Corbett1, Carl D. Modes2, and Mark Warner3

1School of Chemical Engineering and Analytical Science, The University of Manchester, Manchester, UK2Center for Studies in Physics and Biology, The Rockefeller University, New York, NY, USA3Cavendish Laboratory, Department of Physics, University of Cambridge, Cambridge, UK

The thread running through this book is the conversion of light into moleculartransformations (on the nanoscale) and thence into mechanical distortions onthe macroscopic scale. One is aiming for optoactuation, for the conversion oflight energy into alternative forms, or for optically driven elements for morecomplex devices. The molecules being transformed by light should be in theirsolid phases in order for light energy to be transduced. The solids investigatedare crystals, amorphous (glassy, polydomain, or plastic) solids, orientationallyordered glasses and elastomers, and even classic elastomers. The common ele-ment is the photochromic entity that suffers a shape change upon excitation byphoton absorption–as discussed by Barrett in Chapter 2.Molecular shape change, most typically from rod-like to bent, intuitively sug-

gests a reduction in packing efficiency and hence a dilation of the solid. Eitherthe photochromics are the single species present, or they are present as guestsin the solid matrix. Either way, if the solid (e.g., a molecular crystal) has thephotochromics directionally ordered, then the contribution to the creation offree volume along and perpendicular to the preferred direction will be differentand so, on top of a background level of isotropic swelling, there will be direc-tionally dependent distortion. For solids such as crystals, glasses, and com-posites, the picture seems apposite and the responses are large (2–10% strain)compared with conventional thermal response in solids, but small comparedwith those of photoresponsive elastomers (20–400% strain). To our knowledge,there are no quantitative, mechanistic pictures for the response of the formersolids to light. Elastomers, on the other hand, are liquid-like locally, and thereare well-developed, quantitative models of how they respond to order changeinduced by the deformations of photochromic guests. Free-volume creationplays a minor role in photoelastomers.


�

� �

�

80 3 Photomechanics: Bend, Curl, Topography, and Topology

The directionally ordered amorphous solids that have been studied areeither liquid-crystalline (LC) glasses or elastomers. They are the subjectof this chapter, and in particular, the mechanism of elastomer responsewill be explored. Most commonly studied are the nematic solids, with onlydirectional order, and also cholesteric solids (Chapter 9), where the directional(i.e., nematic) order rotates about a perpendicular axis. Furthermore, highlyinteresting are the smectic forms of solids, which are additionally layered. If,further, there is a tilt of the nematic direction with respect to the layer normal,then there can be ferroelectricity, which can then be optically controlled ifthere are photochromic guests.These extraordinary optomechanical effects can be further enriched by the

variation of optical polarization–something that is easy to control and is uniqueto these solids in that their mechanics follow such optical adjustments. Therestoration of the ground, and hence the undistorted, state can be via a thermaldecay for which one simply has to wait or by a decay to the ground state stim-ulated by light of a different color. It seems that the light of the second colordoes not simply accelerate recovery but can also make a cycle of free-volumecreation and recovery that induces rotation of the anisotropy of the solid andthus gives rise to entirely new phenomena (see Chapter 9).Thus, the new phenomena afforded by light-responsive mechanics are much

richer than those in conventional mechanics. More subtle control arises and,as we see in the remainder of this chapter, bend, twist, intrinsic curvature, andtopology change are possible.A theme encountered throughout this volume is bend and twist. These both

stem from a response that differs in depth through the thickness of the sheet orcantilever. There are two ways of achieving this differential response:

1. If the light is absorbed by the photochromic species present, then itbecomes less intense with depth and thereby causes less contraction.Accordingly, there must be a bend toward the direction from which thelight impinges. The type of absorption profile is then vital. Is it Beer-like,that is exponential, which is obtained when the number of absorbingspecies is slightly depleted by the light? Or, is it non-Beer because asignificant depletion of the absorbing species takes place (i.e., a significantfraction of photochromics are excited and no longer absorb the incidentlight)? The difference is easily tested, and the two types of absorptionlead to profoundly different response. It is equally clear that the degreeof loading of a sample with photochromic guests is also significantlyimportant. If, for instance, the conversion of 5% of the species present isalready sufficient to saturate the mechanical response, having more guestsis pointless since they absorb light but cause no additional mechanicalresponse.

�

� �

�

3.1 The Photomechanics of Liquid-Crystalline Solids 81

2. Differential response with depth can arise because the ordering directioncan vary with depth.This possibility arises more directly in LC solids, whichwe discuss in the following.

To go beyond bend and twist to achieve intrinsic curvature of the space ofthe initial solid, with attendant stretch and thus strong actuation, one requiresdirector variation in the plane of the sample, rather thanwith depth.We believethis to be the strangest and most distant frontier of photomechanics and is thesubject of the second part of this chapter.We now concentrate on the photomechanical response in nematic LC glasses

and elastomers.

3.1 The Photomechanics of Liquid-Crystalline Solids

An LC solid combines solidity with liquid crystallinity, that is, with directionalorder of rods without their having positional order. Such noncrystalline solidsare necessarily glasses or elastomers. Their photoresponse arises because theirmechanical state depends on their LC order, and this can be changed optically(as well as by heat, solvent, and other stimuli). It is interesting to explore thefollowing:

1. How photoresponse occurs. The underlying mechanisms are best exploredwhen the response is for a solid with a uniform director, n, and where allparts of the sample are equally illuminated.

2. How decay with depth of the stimulating beam of light gives a depth-dependent response and hence bend.

3. How nonuniform directors give a wonderfully subtle and complex response:(a) Bend and curl occur when the director varies through the solid thick-

ness, and hence, the extent of the response in a given direction also varieswith depth. This response, in some sense, mimics the stimulus vary-ing with depth in a uniform system, for example, when light is stronglyabsorbed.

(b) Topography and Gaussian curvature can change if the variation ofthe director is in the plane of a sheet of solid, rather than through thethickness.

(c) Topology changes if conflicting mechanical responses within a field ofn(r) can only be resolved by opening slits and so on.

This chapter first addresses the mechanisms of solid-body photoresponse andthen explores the response extended to bend and curl. Then, it moves on, inSection 3.4, to the issues of curvature, topography, and topology–what onemight term “metricmechanics” since the natural lengths and angles are inducedto change, as space becomes intrinsically curved.

�

� �

�


3.2 Photomechanics and Its Mechanisms

The directional (LC) order of densely packed rods can be reduced if guestdye rods can be made to bend. Packing is then less efficient, and translationalentropy is no longer simply maximized by rods being parallel (the Onsagersteric ordering mechanism). This picture describes how liquids of rods (classicliquid crystals and molten polymer liquid crystals) have their order changedoptically because suitable dye molecules that absorb a photon then bend(photoisomerize).Elastomers are essentially liquids that cannot flow in any macroscopic sense

but have the extreme molecular mobility at the local scale that liquids have.Their LC form consists of polymers with pendent or in-chain LC-forming rods,and the classic mechanism for photoinduced order reduction applies to themas well. We discuss LC glasses separately. The LC polymers in elastomers arelightly cross-linked, which prevents their large-scale motion and makes theelastomers (marginally) solid-like. Dense cross-linking leads to high-modulusglassy materials without high molecular mobility. Typically, a concentration of3–5% guest dye rods in a network of nematic polymers is sufficient, when bent,to destroy their hosts’ nematic order. More guests would absorb more light butnot cause any further effect, the order already being destroyed, and therefore,higher guest concentrations are pointless for converting light to mechanicalresponse.As described in Chapter 2 by Barrett, one requires rods with a photosensitive

core or central bond that, upon absorbing a photon and being promoted fromthe straight, trans ground state to the bent, cis excited state, causes the entirerod to bend. The classic example of such a photoisomerizing central unit isazobenzene; see Figure 3.1(a).The bent guests decrease or destroy the nematic order of their rod-like host

molecules. With reduced nematic order, the network of chains that were previ-ously elongated along the ordering direction n are now, on average, isotropicin shape; see Figure 3.1(b). The solid that the chains comprise is less elon-gated along the director; it has contracted (as shown in the illustration, by afactor of 1∕𝜆 where 𝜆 > 1). Conversely, a block of elastomer in the isotropicstate elongates by a factor of 𝜆 when it is no longer illuminated, the photoac-tive dye guests thus recover to the trans state, and the hosts’ nematic orderis restored. The destruction of order by induction of molecular bend, with itsconsequent mechanical effect, is entirely analogous to that of order reductionby heating and its mechanical consequence. The mapping of these two pro-cesses onto each other was the core of the first theoretical and experimentalinvestigation of nematic photoelastomer response [1].The connection between the distribution of chain shapes and mechanics

for elastomers is well understood and can be easily extended to nematic elas-tomers. Since chains are fluidic and explore a myriad of configurations, they

�

� �

�

3.2 Photomechanics and Its Mechanisms 83

Photon

trans (straight) cis (bent)

Azobenzene

Nematic Isotropic

Recovery

1

Assume spans deform aswhole solid:

Block of rubber

(a)

(b)

Photon

Recovery

R0

n

λ R

λ1

R = λ . R0=

N

N

N

N

UV

365nm

465nm

T° or UV

Figure 3.1 (a) (Top line) A nematic liquid of rods with dye guests, shown with theirazobenzene cores as open (trans) or filled (cis) dots. The two isomers of azobenzene. (b) Ablock of rubber, of unit dimensions, composed of polymer chains linked together(cross-links shown as dots). The nematic rods driving the chains to anisotropy are notshown. When the chains extend (right block, in the dark), the block suffers an extension by afactor of 𝜆 along the director (presaged in the isotropic state (left). One can assume thatspans R between reticulation points deform in proportion to the body, that is, according tothe deformation gradient 𝛌.

are easily averaged over (even when aligned) to free energies and mean-squaresizes of chains. Their fluidity means that (i) they are also capable of largeextensions and (ii) their moduli are low (in the range of 105 to 5 × 106 Pa). Theratio of the mean-square polymer chain size along and perpendicular to thedirector is denoted by r. The ratio r can be determined by light or neutronscattering. Within a freely jointed chain model, it can be calculated that

r = (1 + 2Q)∕(1 − Q), (3.1)

whereQ is the nematic order parameter;Q = 0 in the isotropic state; and hence,r = 1 as expected–the chains are, on average, isotropic. In the unphysical limitof perfect order, Q = 1, the chains are perfectly stretched out and, becausetheir thickness is not taken into account, r → ∞, which is equally unphysicalor rather ungeometrical–a shortcoming that is easily remedied. The model isextremely good and makes powerful predictions: one can show [2] that, upon

�

� �

�


increasing their order from Q = 0 to Q > 0, and hence r increases from r = 1to r > 1, elastomers suffer an elongation by a factor of 𝜆 along the director nand a contraction perpendicular to n of 1∕

√𝜆, where

𝜆 = r1∕3 > 1. (3.2)

The spontaneous distortion has sometimes been denoted by 𝜆m to distinguishit from imposed deformations, but in this review, we are entirely concernedwith spontaneous deformations and will mostly drop the subscript m. Coolingfrom the isotropic state (or first illuminating to create isotropy and then placinga sample in the dark to recover relative to the illuminated state) gives r = 1 →r ∼ 60 in an extreme case, and hence, 𝜆 ∼ 4; see Ref. 3 and also Figure 3.2.A length doubling (dark) or halving (light) is not unusual for photomechan-

ical response in elastomers. Conversely, using the nematic state as a referencestate, then, heating or illumination leads to a contraction by a factor of 𝜆 ≡1∕𝜆m < 1. The power of the model to incorporate nematic order is seen exper-imentally through the connection between order and elongation, that is, Q and𝜆, which follows from Equation 3.1, that is,

𝜆3 = (1 + 2Q)∕(1 − Q). (3.3)

See Ref. 4 (Figure 2b), [5], and [2] (Section 6.1) for a discussion of this connec-tion between nematic order, chain elongation, and spontaneous distortion.Imposed deformations of elastomers are at constant volume, since volume

change is expensive compared with shape change. The bulk modulus is≳109 Pa, that is, about 103 − 104 × 𝜇, the shear modulus. Photoinduced shapechanges in elastomers are also at constant volume since local molecularrearrangements are easy. The 1∕

√𝜆 response in the perpendicular directions

Isotropy OrderedHeating/light - Cooling/dark

1

1.5

2

2.5

3

3.5

4λ

20 40 60 80 100 120

0 g

5 g10 g15 g

T

Figure 3.2 A nematic elastomer in a measuring cylinder supporting a mass. It is initially hot(first frame). Cold air is blown down the cylinder, the nematic order is restored, and theelastomer grows in length by more than a factor of 3. The process is quickly reversible uponsubsequent heating. The fractional length change 𝜆 of the same elastomer supportingdifferent masses responding to temperature, T . (Tajbakhsh and Terentjev [3]. Reproducedwith the permission of Springer.)

�

� �

�


seen earlier preserves the volume, since the product of these two-dimensionalchanges with that of 𝜆 along the director is the fractional volume change andis unity.To introduce glasses, one can summarize elastomers; shape change factors 𝜆

are large, materials are weak (soft), and, related to the latter, deformations are ata constant volume. Glasses, by contrast, are densely cross-linked, for instance,epoxies, with largely immobile component rods (their ordering direction needsto be set during formation). Heating or illumination has a weaker effect on rodalignment. Even above a conventional polymer glass transition, glasses onlypartially soften since mobility is suppressed by the dense cross-linking. Theyare the opposite of elastomers in nearly every aspect: their mechanisms arenot well understood, their deformations are small, their moduli are high, andthey are not volume-preserving under deformation (whether spontaneous orimposed). Thus, for glasses,

𝜆 ∼ 1.04 (1∕𝜆 ∼ 0.94), (3.4)𝜇 ∼ 109 Pa (hard, strong), (3.5)

𝜆⊥= 𝜆−𝜈

(12<∼𝜈<∼3

)

, (3.6)

where the perpendicular response, denoted by 𝜆⊥, is characterized by the expo-

nent 𝜈, which one could call the optothermal Poisson ratio by analogy to thechange of perpendicular size 𝜆−𝜈P on an imposed deformation 𝜆, where 𝜈P isthe conventional Poisson ratio of mechanics (𝜈P ≃ 1∕3 for steel, 𝜈P ≃ 1∕2 as wehave seen earlier for elastomers, and 𝜈P ≃ 0 for cork).Thus, glasses suffer a relative volume change by a factor of 𝜆 × 𝜆−𝜈 × 𝜆−𝜈 =

𝜆1−2𝜈 , which is the product of the relative length changes of the sides of the

body.1 If the nematic state is irradiated, there is a contraction 𝜆 < 1 along thedirector, where V∕V0 = 𝜆

1−2𝜈> 1 for 𝜈 > 1∕2. Nematic glasses dilate upon

irradiation, as one might expect. They have a molecular environment that isnot mobile–a bend guest is disruptive of the unadapting hosts, forcing themapart. The values of 𝜈 will be vital later in our “metric mechanics” since italso determines whether the area increases or decreases upon irradiation anddetermines whether emergent shells can be anchored or not. The change ofdensity upon irradiation has beenmeasured using a clever floatation techniqueby Liu et al. [6], who were also interested in the value of 𝜈 in their method ofachieving surface relief by irradiating an LC glass layer stuck on to a substrate.Being constrained by its attachment, volume change can only be accommo-dated by expansions into the third dimension, hence generating topographicalchanges where light is allowed to fall–an aim of the second part of thischapter.

1 Thus, 𝜆, 𝜆−𝜈 , and 𝜆−𝜈 are the three principal elements of the deformation gradient tensor, 𝛌,that we have met before and will use again in Equation 3.24.

�

� �

�


3.2.1 Absorption, Photomechanics, and Bend Actuation

A variety of experiments have investigated the effects of illumination onfilms/cantilevers of nematic elastomers incorporating photoactive dyemolecules [7–10]. We consider the situation shown in Figure 3.3; light ofan appropriate frequency is incident on the surface of a film of photoactivenematic elastomer. The director of the nematic material is taken to be alongthe long axis of the film. Dye molecules are linear (trans) in their ground stateand bent (cis) when excited by photon absorption. We denote the fraction ofcis(trans)-molecules by nc (nt) with nc = 1 − nt . The number fraction of cisincreases by illumination with light I(x) and decreases by thermal recoveryat a rate 1∕𝜏 , where 𝜏 is the thermal recovery time and I(x) is the intensity(Poynting flux) at depth x into the film or cantilever. Thus,

𝜕nc

𝜕t= ΓI(x)nt −

nc

𝜏, (3.7)

where Γ is a constant, which contains an absorption cross section per dyemolecule and a quantum efficiency. It is convenient to scale time by thethermal recovery time, thus t = t∕𝜏 . We can then identify a material intensityIm = 1∕Γ𝜏 . We define a dimensionless intensity (x) = I(x)∕I0 and a dimen-sionless characteristic intensity 𝛼 = I0∕Im, where I0 is the intensity of light atthe surface. In terms of these reduced variables, we have

nc = −[𝛼(x) + 1]nc + 𝛼(x). (3.8)

Here, we have neglected cis absorption, background absorption, and scattering.The parameter 𝛼 measures the ratio of the forward and backward rates for

the generation of cis molecules. A large 𝛼 value suggests a large deviation fromthe equilibrium nc = 0, while a small 𝛼 value corresponds to the Lambert–Beerlimit and has nc ≃ 0. The photostationary state, given by nc = 0, is

nc(x) =𝛼(x)

1 + 𝛼(x) . (3.9)

Intensity is reduced with depth x by photon absorption by trans nematogens,which is described by a modified Lambert–Beer law:

𝜕𝜕x

= −nt(x)

d, (3.10)

W

d

R

I(X)

X Figure 3.3 A beam of light, travelingin the x-direction is incident on thesurface of film of a photoactivenematic elastomer, absorption leadsto bend. (Corbett and Warner [11].Reproduced with the permission ofAmerican Physical Society.)

�

� �

�


where the Beer length d subsumes cross sections, number densities, andso on. We observe that the intensity (x) depends on nt(x), which itselfdepends on (x) via either Equation 3.8 (dynamic) or 3.9 (static). Creation ofcis isomers lowers the nematic order and leads to a photocontraction alongthe director. The simplest expression for this photostrain is 𝜖p = Pnc(x, t),where P is the dimensionless photoresponsivity. For 𝜖p ∼ −0.04 and nc ∼ 0.8,then P ∼ −1∕20. If the photostrain varies with depth, a mean strain K and acurvature 1

Rwill result as the solid aims to reduce the cost of deviating from its

new, local natural length. The effective strain is

𝜖(x) = xR+ K − 𝜖p(x). (3.11)

The stress corresponding to this strain is simply 𝜎(x) = E𝜖(x), where E is anappropriate Young’s modulus, assumed as constant. Integrating the stress andthe moment of the stress through the thickness w of the film to get the forceand the torque and setting these to zero give

0 = ∫w

0E[ x

R+ K − Pnc(x, t)

]

dx

= ∫w

0E[ x

R+ K − Pnc(x, t)

]

x dx. (3.12)

Solving these two equations, we obtain the (scaled) curvature w∕R:

wR

= −12Pw2 ∫

w

0

(w2− x

)

nc(x)dx. (3.13)

These equations certainly hold for the steady-state response, and also, thedynamic response provided inertia is unimportant.

3.2.1.1 Photostationary Dye Populations andMechanical ResponseUsing the steady-state population for the trans population nt(x) =1∕(1 + 𝛼(x)) (see Equation 3.9) in Equation 3.10 gives

d 𝜕𝜕x

= − 1 + 𝛼 (3.14)

Integrating with (0) = 1, we obtain:

log [(x)] + 𝛼((x) − 1) = −x∕d (3.15)

A formal solution to this equation is (x) = 1𝛼

WL[𝛼 exp(𝛼 − x∕d)], whereWL(…) is the LambertW function. In Figure 3.4(a), we show the intensity plot-ted as a function of x∕d for several values of 𝛼. We can see two extremal typesof behavior for the intensity: (i) for 𝛼 ≪ 1, we obtain typical Lambert–Beerbehavior (x) = exp(−x∕d), while (ii) for 𝛼 ≫ 1, we have (x) = 1 − x∕(𝛼d)for distances up to x ≈ 𝛼d followed by a return to exponential decay. Forlarge 𝛼 ≫ 1, light penetrates much farther into the film than the suggested

�

� �

�


1.0 0.16

0.14

0.12

0.1

0.08

0.06

0.04

0.02

0 42

0.8

0.6

0.4

0.2

42 6 8

(a) (b)

10 12 14

X/d

I

0.1

0.1

4.5

20.5

d/R

0.5 15

2α = 10

α = 9

6 8 10 12 14 16 18 20

w/d

Figure 3.4 (a) The decay in reduced light intensity with depth for various reduced incidentintensities 𝛼. (b) The reduced curvature d∕R as a function of reduced beam thickness w∕dfor several values of 𝛼. (Corbett and Warner [11]. Reproduced with the permission ofAmerican Physical Society.)

standard Lambert–Beer decay length d. The result that light can penetratemuch deeper than expected might well explain the unusual phenomenonthat samples with high dye loading, and thus, short Lambert–Beer lengths drelative to the cantilever thickness w still show appreciable bending. Extendingthe aforementioned analysis to include the effects of an optically stimulatedcis-to-trans (nc → nt) reaction is possible [12] and is essential to explain someoptomechanical effects [13].Taking the photostationary solution for (x) and inserting it into

Equation 3.13 and performing a change of variables for the integrationfrom x to via Equation 3.10, we arrive at an expression for the curvature interms of the reduced intensity

𝜔at the back surface:

dR= 12𝛼

(dw

)3 [wdw − (1 − w)

(

1 − w2d

)

− 𝛼

2(1 − 2

w)]

, (3.16)

Aswe can see, the curvature depends onw∕d and 𝛼, both directly and indirectlythrough the back surface intensityw = (w∕d, 𝛼). Plots of the scaled curvatured∕R as a function of w∕d are shown in Figure 3.4(b) for several values of 𝛼.For a given w∕d, the curvature initially increases with increasing 𝛼 as opticalpenetration through the sample becomes greater and then reduces as gradientsof strain are reduced.Thus, appreciable curvature arises experimentally even incantilevers for which w ≫ d. For example, as discussed in Ref. 7, curvature isinduced in samples for which w ≈ 100d.

3.2.1.2 Dynamical Intensity and Dye PopulationsIntroducing the absorbance (x, t) = − log allows us to combineEquations 3.8 and 3.10 into a single partial differential equation for the

�

� �

�


spatiotemporal development of the absorbance:

𝜏𝜕𝜕t

= xd− + 𝛼(e− − 1), (3.17)

wherewe havemade use of the condition(0, t) = 0. Using the initial condition(x, 0) = x∕d allows us to complete the quadrature to determine(x, t):

t𝜏= ∫

(x,t)

x∕d

d′

x∕d − 𝛼 −′ + 𝛼e−′ . (3.18)

Investigating this integral, we see that the limit t∕𝜏 → ∞ corresponds to thevanishing of the denominator in the integrand; this produces the Lambert Wfunction solution for the steady-state absorbance. The spatial profiles for theintensity ( ≡ e−) obtained by solving this integral are shown for several val-ues of t∕𝜏 and 𝛼 in Figure 3.5. For t = 0, the profile (x, 0) is an exponentialdecay, while at long times (t = 5𝜏), the profile is essentially linear out to x ∼𝛼d = 10d, followed by a region with exponential decay. At intermediate times,the profile first saturates at small x, that is, the intensity for small x approachesthe equilibrium solution while the intensity for larger x remains relatively low.The number of photoactive units in the surface regions is reduced; thus, morelight penetrates the surface regions and the profile for larger x also approachesthe Lambert W form. For 𝛼 = 30, the ultimate penetration is deeper and theapproach to the saturated state is sharper. Essentially, a saturation wavefront

1.0

I

0.8

0.6

0.4

0.2

0 5 10 15 20 25 30x/d

0

0.1

0.5

2

0.5

0.75

1 5

1 t/τ = 5

Figure 3.5 Intensity versus reduced depth for 𝛼 = 10 (light gray) and 𝛼 = 30 (dark gray) atreduced times t∕𝜏 as marked. The Lambert–Beer law holds for any 𝛼 at t = 0. (Corbettet al. [14]. Reproduced with the permission of American Physical Society.)

�

� �

�


passes through the sample. For both values of 𝛼, the emergent light at the backsurface is far in excess of any Lambert–Beer prediction.Van Oosten et al. [15] demonstrated some particularly interesting results of

bending for a glassy photoactive film. When their sample is illuminated fromone side, the cantilever bends toward the light source, while when illuminatedfrom the reverse side, the sample initially bends toward the light source but ulti-mately, in the steady state, bends away from the light source.This was explainedin terms of a compositional gradient, which leads to a positionally varying pho-toresponsivity, that is, 𝜖p(x, t) = P(x)nc(x, t). Assuming a linear form for P(x) =a1 + a2x, one obtains the following expression for the curvature:

dR= 12

(dw

)3 [

− w2d

P(w) log w − a2d12

(wd

)3

+∫w

0log {

a1 + a2d(

2 xd− w

2d

)}

dx]

. (3.19)

Reversing the side of illumination is equivalent to changing the sign of a2 and,for suitable values of a2∕a1, can lead to backbend.

3.2.1.3 Polydomain PhotosolidsA polydomain nematic solid has its director pointing from region to regionin random directions. It is not necessarily divided into discrete regions with adifferent director but is possibly subject rather to a random wandering of thedirector–mostly the length scale of director variation is sufficiently fine that itis hard to discern what type of random structure exists in the director field. Inpolydomain photoelastomers, these competing directions of order have nuga-tory effects on heating and cooling–elongations and contractions are in randomdirections, and volume is conserved, leading overall to no gross mechanicalresponse. However, light offers a subtle response and a route to controllingstrains by tuning light polarization.Yu et al. [7] found that the direction of curl of a nematic polydomain glass fol-

lowed the direction of polarization of the light that was normally incident withrespect to the initial plane of the solid sheet. As the direction of polarizationwas rotated, the direction of curl responds. Harvey and Terentjev [16] insteadfixed the length of a polydomain nematic elastomer and followed the buildupof stress depending on the direction of the polarization of light with respect tothe clamping direction.Theoretical descriptions of the response have been advanced in Refs 17

and 18. This latter work deals with the mechanisms by how different regionsin a nematic polydomain elastomer have differing new natural dimensionscorresponding to how their local director is aligned to the incident light’s polar-ization. The overall response is calculated as a best response to the ensembleof differing natural length changes, allowing partial director rotation, where

�

� �

�


that is advantageous. The approach is approximate in that it assumes the samemechanical response locally and globally–a kind of mechanical mean fieldknown as the Taylor limit in such problems.As well as determining the direction and magnitude of the mechanical

response, it is also described [18] how, as the intensity increases and the orderis largely lost, the magnitude of the response must then diminish. Imagine thetotal loss of order due to optical effects: the response would just be equivalentto heating to the isotropic state, and then, there is no overall advantage toany particular macroscopic deformation and the sample remains unchanged.Thus, the response must be nonmonotonic with intensity, even for polarizedlight. The case of incident unpolarized light is also treated. Now there is onlyone and not two preferred directions, namely the direction of propagation,and the effect differs from that of polarized light.

3.2.1.4 Photomechanics versus Thermal Mechanics upon IlluminatingPhotosolidsIn the context of polydomains, one can revisit the issue of the mechanism forphoto-response: the fact that such mechanical response is polarization depen-dent has been taken as a demonstration that the effects we are concerned within this entire chapter are essentially optical, that is, due to photoisomerizationrather than just that of heat being delivered by light, the conversion from lightto heat being achieved by dye molecules. If one assumes that optical effectsare simply due to heat being preferentially delivered to domains aligned withthe polarization, then difficulties arise: any unique direction associated withthe incident polarized light is also lost if one assumes that heat, released on theback-decay from the cis excited to trans ground state in regions alignedwith thepolarization direction, is then transferred quickly to other regions.The assump-tion of short times is reasonable: Hon et al. [19] obtained D ≈ 1.5 × 10−7 m2/sfor the heat diffusion coefficient of a side-chain nematic elastomer, while Broer-man et al. [20] obtained D ≈ 1.1 × 10−7 m2/s for an isotropic, silicone-basedelastomer. Assuming director correlation over l <∼ 1 μm, we estimate the char-acteristic time for heat to diffuse to another region to be ∼ l2∕D = 10 μs, muchshorter than the observed mechanical response times.Another view is that absorption is Beer-like, and the characteristic lengths

are shorter than the characteristic sample thicknesses. Generally, Beer lengthsare short, and absorption confined to a Beer layer would not activate a sufficientvolume of the sample to have a significant mechanical effect. As explained inSection 3.2.1, there is, however,much evidence that photodepletion of the transisomer does take place and that penetration is deep and nonexponential, thatis, non-Beer. Such population changes also give rise to characteristic dynamics,for instance, in photomechanics and also in photoinduced changes in ferro-electricity in liquid crystals. The original experiments by Yu et al. [7] have avery pronounced dynamics, indeed leading to a bend that overshoots and even

�

� �

�


eclipses the incoming light (a mysterious phenomenon that has been analyzedin Ref. 21). Ignoring non-Beer effects, one can argue that heat is generated ina Beer surface layer, too thin to influence the mechanics. The heat then dif-fuses into the bulk and generates a heat-mediated order parameter reductionand thus photocontractions [22, 23]. Difficulties arise in polydomain samplesif one assumes that absorption in a thin layer transfers heat to the bulk. Sincepolydomain elastomers deform at constant volume, with no preferred directionselected by the delivered heat there can be no mechanical response.On the other hand, experiments performed on the mechanical response

to light that has had the UV component filtered out also show effects(P. Palffy-Muhoray, private communication). It is not easy to reconcile thesetwo frameworks - perhaps, it is possible that conversion, though not strongaway from the main UV line, still takes place and bent monomers contributeas we have argued earlier. Using a nonisomerizing dye, which neverthelessabsorbs light preferentially according to polarization and produces heat indomains of a particular orientation, should present a decisive test (see [22]where disperse orange 11 dye was used), and results were indeed similar tothose from isomerizing dyes).

3.3 A Sketch of Macroscopic Mechanical Response inLC Rubbers and Glasses

A simple route to beyond planar contraction and elongation is to have n, thelocal direction of response, to vary through the thickness (a) by twisting fromone in-plane direction at the top to another on the bottom of the sheet or beamof material or (b) by n splay–bending between being in-plane on one face andbeing normal (homeotropic) to the other face, Figure 3.6(a) and (b).The strategyof director variation to obtain bend was introduced by Broer et al. in thermal[24] and in optical [25] cases, an example of the latter being Figure 3.6(c). Curl-ing also results [8] when the phase of the twist is adjusted. See also theory,simulation, and experiment [26–28]. A bi-glass, a nematic cantilever bondedto a nonnematic one, see Figure 3.6(d), is analogous to a bimetallic strip and iseasily analyzed.As we have described, a gradient of photoresponse leads to the photoinduc-

tion of bend for systems with simple, homogeneous director fields; see Refs30–32, for examples of actuation via a gradient of stimulus. Similarly, solventconcentration in the material resulting from exposure to vapor at one face, ora gradient in temperature through the cantilever, will cause bend. All thesegradient-of-stimulus methods will work for uniform director conformations.More subtly, considerable additional mechanical control can be achieved byusing a polydomain rather than uniform director nematic solid and exploitingthe specificity of response to the polarization of light [7, 33].

�

� �

�

3.3 A Sketch of Macroscopic Mechanical Response in LC Rubbers and Glasses 93

x xy y

y

z

2h

z

z

(a) (b)

(c) (d)

(e) R

θ ϕ

Figure 3.6 (a) Splay–bend director conformation with n varying in the yz-plane upontraversing the cantilever. (b) Twist director conformation with n twisting in the xy-planeupon traversing the cantilever. (c) Nematic glass photocantilever before and 0.04, 0.2, and0.5 s after illumination with UV light [25]. (d) A bi-rubber undergoing heating (figures from[29], Prof. EM Terentjev). (e) A bending cantilever, with thickness 2h and radius of curvatureR, the central plane being denoted by a light line, now curved, from which materialpositions z are measured through the thickness. (figures from [29], reproduced with thepermission of The Royal Society.)

�

� �

�


Splay–bend of the director field allows the induction of the easiest mechan-ical response to envisage, giving rise to pure bend of the material–fromFigure 3.6(a), one sees that illumination induces contraction along the bottomface and elongation along the upper face (directions perpendicular to thedirector). Bending toward the contracting face goes some way to satisfyingthesemechanical requirements, the resulting radius of curvature being the bestcompromise: the bent state has a linear variation of in-plane strain throughthe thickness z (see Figure 3.6e for the geometry) and would only be stress-freeif at every depth, the photostrain 𝜖syy = 𝜆yy − 1 were also linear in z, that is, if𝜖syy = 𝜖

s0z∕R + 𝜖s, where 𝜖s0 is some scale of strain and 𝜖s is a mean photostrain

associated with 𝜖syy(z).2 Then the photoinduced strains match those arising

purely from geometry, see Figure 3.6(d), when the beam conforms locallyto a sector of a circle. The compromise between the actual, linearly varyingstrains imposed by geometry and those ideal for the physical response to heator light, given the director variation, is the controller of actual response. SeeRef. 29 for a detailed discussion and the resultant curvature. The directorrotates in the yz-plane from parallel to the bottom boundary (z = −h) throughto normal on the top (z = h). The angle 𝜃(z) the director makes with they-axis is

𝜃(z) = 𝜋z4h

+ 𝜋

4. (3.20)

The zeroth and first moments of the photostrain are needed tomatch the meanstrain and varying strain discussed earlier and give

1∕R = − 6h𝜋2 (𝜖∥ − 𝜖⊥) e = 1

2(𝜖∥ + 𝜖⊥), (3.21)

where the small-strain thermal or optical distortions for a uniform system,along and perpendicular to the director, 𝜖∥ and 𝜖

⊥, are, 𝜆 − 1 and 𝜆

−𝜈 − 1,respectively, in the limit of 𝜆 ∼ 1. In this limit, 𝜖

⊥= −𝜈𝜖∥ in analogy to

conventional Poisson effects. [We refer the reader to [29, 30] for details ofwhere a gradient of stimulation (namely a light beam attenuating as it passesthrough the sample thickness) causes curvature.]This case of the response of a solid with a splay–bend director field is also

very simple, in that the transverse response 𝜖xx is not a function of depth, andthere is no need to accommodate bend in the x-direction as there was in the y.Bend in both directions at the same time would yield Gaussian curvature andhence stretch as well–a very high energy deformation (the avoidance of whichis the preoccupation of the last part of this review, which is concerned with thephotogeneration of stretch-free Gaussian curvature from initially flat sheets).

2 We deal with small strains here since in bending, although the effects are large, the distortionsonly need to be small.

�

� �

�

3.3 A Sketch of Macroscopic Mechanical Response in LC Rubbers and Glasses 95

2h

x

y

z

(a) (b) (c)

(d)

Figure 3.7 A cantilever of thickness 2h (a) before and (b) after imposed or spontaneousdistortion into a saddle shape, that is, with curvatures in two directions of opposite signs.(c) A nematic solid “swimmer” supported on a pin and forming a saddle in response toillumination from above (figure from Prof. P Palffy-Muhoray). Figures (a)–(c) from Ref. [29].(d) A classic beam (light color, seen along its length) being forced to bend. The straight edge(black) placed transversely across it reveals the transverse curvature (From [29], reproducedwith the permission of The Royal Society.)

Twist fields, by contrast, give mechanical responses where one must confrontthe issue of Gaussian curvature:Returning to Figure 3.6(b) showing twist, one can see the imperative to bend

as earlier, but simultaneously and with opposite curvature in two directions. Atthe bottom surface, the director being along y, there is elongation along x andat the top, there is contraction along x because there the director is along x. Butat the same time, at the bottom, there is contraction along y and at the top, thereis elongation along y. Thus, in the zx-plane, the surface bends upward, whilein the zy-plane, it bends downward. The desired state of bends, according tospontaneous distortions, is a saddle. See Figure 3.7(b), which is doubly curved,having evolved from (a). Figure 3.7(c) shows an experimental saddle.The twist configuration has been explored thermally [24] and optically [25]

(an optical response of a twisted nematic photocantilever being shown inFigure 3.6(c)) by the Broer group. The angle 𝜙(z) the director makes with they-axis is given as for 𝜃 in Equation 3.20. It determines the spontaneous strain,which now has off-diagonal components:

𝜖s =

⎛⎜⎜⎜⎝

𝜖∥sin2𝜙 + 𝜖

⊥cos2𝜙 (𝜖∥ − 𝜖⊥) sin𝜙 cos𝜙 0

(𝜖∥ − 𝜖⊥) sin𝜙 cos𝜙 𝜖∥cos2𝜙 + 𝜖⊥sin2

𝜙 0

0 0 𝜖⊥

⎞⎟⎟⎟⎠

. (3.22)

�

� �

�


The 𝜖sxx = 𝜖⊥+ (𝜖∥ − 𝜖⊥)sin2

𝜙(z), and the corresponding 𝜖syy, components con-tribute to the weak curvatures, mid-plane strains, and some amount of meanshear.The zeroth and first moments of 𝜖s determine the curvatures 1∕R and the

mean strains to give [29]

1∕Royy = − 6

h𝜋2 (𝜖∥ − 𝜖⊥) eyy =12(𝜖∥ + 𝜖⊥)

1∕Roxx =

6h𝜋2 (𝜖∥ − 𝜖⊥) exx =

12(𝜖∥ + 𝜖⊥)

(3.23)

and there is an exy that does not influence bend. For this geometry, there ismax-imal asymmetry also of the xx- and yy thermal/optical strains, providing equaland opposite drives to bend in the xz-plane as in the yz-plane, that is, 1∕Ro

xx =−1∕Ro

yy, in contrast to what happens in the splay–bend case. Indeed, the x andy directions are equivalent, and the maximal saddle (anticlastic) response iscreated.Curvature in more than one direction simultaneously will be the subject

of the rest of this review. We have seen it in small-strain response in twistednematic cantilevers, Figure 3.7(b), but such curvature is seen in classic systemsas well due to classic Poisson effects where a strain imposed in one directiongives rise to a strain of opposite sense in the other directions–the so-calledanticlasticity. Figure 3.7(b) could be the profile of a classic solid with, forinstance, bend in the longitudinal sense–see also Figure 3.7(d). But as the bendincreases, then the transverse bend in response shifts the material further awayfrom the neutral plane of the original bend and induces ever more stretch,which is very expensive. A good classic example is the builder’s metal tapemeasure, which is curved in one direction (transverse) naturally. Attempts tobend it longitudinally cause stretch, and the tape resists until the cost is toohigh, and it snaps to a localized bend of the opposite sense, while flatteningin this bend region so that it is only bent in one sense at a time. As thespontaneous strains increase in twisted nematic cantilevers, the simultaneousdouble bends lead to too much stretch, and bend in one of the two directions(the transverse one) is suppressed, corresponding to the experiment at largestrains, see Figure 3.6(c). The issue of suppression of anticlasticity in nematiccantilevers is dealt with in Ref. 28. The issue of transverse extent, and hencethe ultimate deviation from the neutral plane and thus stretch cost, is decisivewith regard to the suppression of anticlasticity. It is also vital in the selectionof the type of curled structure [26–28], either cylindrical (spiral) or ribbonhelices.The interplay between bend and stretch, and routes to its induction with

nematic director distributions in photosolids, is particularly significant. It canlead to topographical and topological changes, to which we now turn.

�

� �

�

3.5 Continuous Director Variation, Part 1 97

3.4 Photo- and Heat-Induced Topographicaland Topological Changes

As we have seen, liquid crystal order in solids allows for a mechanical responseto light, heat, and other stimuli that are aligned to the liquid crystal director.Remarkable elongations and contractions arise for uniform director fields,and for directors varying through the thickness of a beam, one can alsoachieve extreme bends. However, for effective actuation, one desires a “strong”response, that is, one that involves stretch. Blocking a photoinduced bendgenerates little force, as does blocking the elongation of a sheet or beam(that would suffer an Euler buckling instability). Blocking a contraction of astrip leads to its effective stretch and that is strong, but such a mode is notalways convenient or practical for a device application. In general, one wantsto develop stretch by heating or illumination, even for an extensional-typedistortion of the natural dimensions of a body, which normally leads to weakresponse.The key to circumventing these practical objections rests in the curvature–if

a system that develops a photoinduced curvature were to be blocked, thenstretches would arise, along with their large forces. A system that curves intoa cylinder is still flat in the sense that it lacks Gaussian curvature: a sheetof paper can be wrapped around a cylinder without stretch or wrinkling.By contrast, a sphere has Gaussian curvature 1∕R2 everywhere, where R isits radius. A flat sheet cannot wrap it without extreme distortions, the verydistortions of circumferences and in-plane radii that are the map-maker’sproblem.A route to switchable curvature, and hence to new actuation mechanisms, is

to have the director field n(r) varying in-plane in a nematic solid sheet, ratherthan through its thickness. We shall explore two types of director variation:(i) continuous, either with 2D topological defects in the in-plane field (discli-nations), or nondefective director variation and (ii) discontinuous jumps inotherwise uniform director fields that lead to an origami-like response but ofa radically new form. Both routes lead to Gaussian curvature of the initiallyflat space, and hence to compelling new opportunities for “strong” stretchactuation.

3.5 Continuous Director Variation, Part 1

A simple example serves to illustrate the induction of Gaussian curvature inphotoresponsive, nonuniform director nematic solids. Consider concentric cir-cles of n(r) [34, 35], a topological defect (disclination) in 2D with topologicalcharge m = 1, or such a disclination in 3D but where escape by the director

�

� �

�


Circle in plane

R

r ′sinϕ

r ′r

n

Rθ

Rsinθ

R

πR/2

3R

(a)

(b)

θ

P′ = 2π r′sinϕ

P = 2π

ϕ

λ < 1

θ = π/2

Figure 3.8 (a) A circular disc on a flat sheet, with circular director lines, contracts by 𝜆 alongthe director. The ratio of perimeter to in-material radius changes from 2𝜋, forcing thesurface to become conical with opening angle 𝜙. A sphere has ratios of perimeters of circles(lines of latitude) to in-material radii that also differ from 2𝜋, but by amounts that depend onthe size of the circles. (b) A circular disc with, for instance, radial director lines and whenheated contracts by 𝜆 along the director and elongates by 𝜆−𝜈 along the circumferences.The ratio of perimeter to in-material radius increases beyond 2𝜋, forcing the surface tobecome an “anticone” to dispose of the surplus perimeter. This can be seen by the trajectoryof the disc’s perimeter on the unit sphere or in the anticone with, here, three wavelengths inthe azimuthal sense. For very large amplitudes, the shape is a ruff. (Modes and Warner [36].Reproduced with the permission of American Physical Society.)

into the third dimension has been suppressed. See the chapters by Broer et al.and of White et al. that describe how these structures are made. A contrac-tion by a factor of 𝜆 < 1 along n and an elongation of 𝜆−𝜈 in the perpendiculardirections mean that the circumference P transforms to P′ = P𝜆 = 2𝜋r𝜆, whilethe radius becomes r′ = 𝜆

−𝜈r. Then the usual ratio P∕r = 2𝜋 characteristic ofcircles in a plane becomes instead P′∕r′ = 2𝜋𝜆1+𝜈 < 2𝜋 for the ratio of the newcircumference to the new in-material radius; see Figure 3.8(a).Clearly, the new surface cannot be flat, but in this case, it is a cone where the

ratio P′∕r′ is independent of r. The opening angle 𝜙 is given by sin𝜙 = 𝜆1+𝜈 .

Cones are flat in some sense–their generators are straight lines (the in-planeradii), so where is the Gaussian curvature that is upsetting the relationship

�

� �

�


between circumference and radius in the two-dimensional space? It is apositive Gaussian curvature, concentrated on the tip, and of magnitudeK = 2𝜋(1 − sin𝜙) [34, 35]. Elastomers, which have large deviations of 𝜆 from1, yield extreme cones experimentally–see Ref. 37–while glasses also yieldcones [38].We are familiar with these signatures of curvature of a surface from

spheres–the connection between circumferences and in-plane radii is also not2𝜋: take the hemisphere shown in Figure 3.8(a) where the length from pole toequator is 𝜋R∕2, giving a circumference (at the equator) to in-material radiusof 2𝜋R∕(𝜋R∕2) = 4, which is< 2𝜋.The difference now is that this ratio dependson the area of the surface enclosed by the circumference, since the amountof Gaussian curvature captured increases as the surface does, in this case ofconstant Gaussian curvature. One can see this by considering in Figure 3.8(a)a circle at angle 𝜃 down from the pole, which has a perimeter 2𝜋R sin 𝜃 and anin-material radius R𝜃. The ratio 2𝜋 sin(𝜃)∕𝜃 tends to 2𝜋 as 𝜃 → 0, suggestingthat the limiting circle is effectively in flat space or, equivalently, that theamount of Gaussian curvature captured tends to zero as its area does–quitereasonable on a sphere where the Gaussian curvature is evenly distributed.Negative curvature and saddle-like surfaces are achieved in such circular

director systems by having 𝜆 > 1 (cooling or darkness) and thus having anexcess of circumferential distance relative to what would be expected in flatspace given their in-plane radii, see Figure 3.8(b); they can be called anticones[34, 35] and are well characterized experimentally [38]. When they have a largeamplitude [36], they are ruffs, similar to the collars worn in the sixteenth andseventeenth centuries. These cones and anticones are the simplest possibleshell systems. All other continuous, nonuniform fields lead to delocalizedcurvature, to which we will return.Conceptually similar problems of emergent topography arise in the growth

of plant leaves, which become crinkled if they grow proportionately morein circumferential directions than in radial directions. The effect is neitherreversible nor switchable, but the aforementioned ideas were presaged inthe work of Amar et al. [39, 40]. We note that Gaussian curvature can beachieved by other related means where the growth is not anisotropic, butrather just a swelling that is spatially varying [41, 42], but in this review, weconcentrate on anisotropic systems since they respond to light. Another routeto topographical change is to form solid films on substrates and vary theirthickness by optical or other means, for instance, shining spots of light onnematic photoelastomer films where there is a localized mechanical response[30]. Yet another method is to have a textured film, for instance, a cholestericwith its axis in or perpendicular to the plane of the film, where the effect oflight or heat is to make the film selectively thicker or thinner, depending onwhere it is acting [6].

�

� �

�


We have considered azimuthal n fields in heating and cooling (equivalent toillumination and darkness). Radial fields, now, respectively, in cooling or heat-ing, achieve the same geometric effect. The perpendicular nematic field, the“orthogonal dual,” has this general property–the effect of heating becomes, inthe dual, that of cooling, and vice versa. Essentially, 𝜆 → 1∕𝜆 (actually → 𝜆

−𝜈)in passing between an n field and its dual.

3.6 Mechanico-Geometric Effects, Part 1

Consider loading a cone at its tip with the aim of pressing it back down to theplane. To do so, avoiding any symmetry-breaking snap-through for the pur-poses of illustration, is to stretch the circumferences and to compress the radii,both “strong” deformations. From everyday experience, one can bend the han-dle of a spoon but not distort its cup (which is curved), since to do so would beto create stretch and compression.The same is true with fragments of egg shell,which resist flattening by identical mechanisms.Ware et al. [43] have exploited this principle, using arrays of rising pyramids

in thin sheets, to lift loads highly exceeding their ownmass. Mechanical advan-tage is thus one motivation for the creation of curved shells. Closely relatedwould be creating pumps, sucking up liquid as the shell rises above a microflu-idic channel, valves in similar applications, responsive lenses, and even switch-able surface topography in, for example, aerodynamics, haptics, and optics.Changes in topography are one response to activating a solid with nonuni-

form director fields. An alternative response is instead topological–a flat sheetcould open up holes instead, for instance, to make a switchable microsieve orfilter [44]; see later Section 3.10 on origami.


Defects (disclinations) in the nematic in-plane field with topological chargem ≠ 1 can also be obtained [37], for instance, m = 6 and even much highercharges. They lead to delocalized curvature rather than just concentrated at apoint (the tip of a cone for m = 1).One can think more generally what is happening to the lengths (and angles)

in the 2-space that is deforming. The deformation gradient tensor 𝛌, the localspecification of the spatial rate of change of separation of neighboring points xin the target space with the separation of points r in the reference space, 𝜕x∕𝜕r,is

𝝀(r) = (𝜆 − 𝜆−𝜈)n n + 𝜆−𝜈𝜹. (3.24)

�

� �

�


It just encodes what we had before, namely there is (locally) elongation or con-traction by a factor 𝜆 along n(r) and contraction or elongation by 𝜆−𝜈 in thetwo directions perpendicular to n (one of which is in-plane, the other along theplane normal).The metric tensor for the space after deformation is

g = 𝛌† ⋅ 𝛌 = (𝜆2 − 𝜆−2𝜈)n n + 𝜆−2𝜈𝜹. (3.25)

It is a 2 × 2 symmetric object that measures distances between the points in thespace. When it varies spatially, the space has Gaussian curvature.Disclinations of charge m have an angle 𝛼 = (m − 1)𝜃 for the director, with

respect to the radius vector, as a function of the azimuthal location 𝜃. Fordefects more complicated than the circular/radial +1s so far considered,consult Figure 3.9.One can show [45, 46] that the Gaussian curvature is distributed as

K(r, 𝜃) = m(m − 1)(𝜆2𝜈 − 𝜆−2)r2

cos[2(m − 1)𝜃]. (3.26)

The angular dependence is cos(2𝛼(𝜃)), and since 𝛼 is independent of r, the onlyr-dependence must be explicitly 1∕r2 in order to get the dimensions of K cor-rect. Note that 𝜆→ 1∕𝜆 (i.e., light to dark) is equivalent to 𝛼 → 𝛼 + 𝜋∕2, bothactions reversing the sign of curvature–see the aforementioned remarks aboutorthogonal dual distributions. Furthermore, for defects withm ≠ 1, any consis-tent local rotation Δ𝜙 is in effect with a global rotation rescaled by the chargeon the defect, that is, a rotation of Δ𝜙∕(m − 1) [45].The shapes induced by this change of Gaussian curvature are difficult, or

perhaps impossible, to be determined uniquely from the curvature and otherdegrees of freedom in the spatially varying metric: a structure with a given cur-vature could be induced to bend, avoiding stretch and hence Gaussian curva-ture change, and other shapes are induced, albeit with different bend energies,while they share the same curvature. A high charge system is shown to be veryslightly deformed in Figure 3.9(c) and to be highly deformed after a tempera-ture rise in (d). They have a complex structure, but the symmetry demanded inEquation 3.26 for this m = 6 example is very clear.Such solids had already been obtained for advanced optics by Sanchez-

Somolinos and Broer et al. (see also the review [47]), unbeknownst to mechan-ics theory before the prediction of the coupling of complex director fields insolids to shape evolution under light or heat. A simple example of where theemerging shape is obvious is a quasi-continuous form of an m = 1∕2 defect,the hemi-stadium; Figure 3.9(e) shows two hemi-stadia joined together to forma full stadium, that is, two +1∕2 defects. Trivially, one can see that the straightsection just contracts (and elongates perpendicularly) and that the ends formhalves of cones. The central section, being connected with the half cones,is required to bend, without stretch as in a cylinder, along the ridge of the

�

� �

�


Heating

m = +6 defect (c)

(a)

(d)

(f)(e)

(b)

n

½ –½θ

βα

Figure 3.9 (a) The coordinates for a disclination showing at an angle 𝜃 the director makingan angle 𝛽(𝜃) = m𝜃 with the reference axis. The angle of the director to the radial direction,𝛼, is hence 𝛼 = 𝛽 − 𝜃 = (m − 1)𝜃. (b) Defects of order m = 1

2and m = − 1

2. (c, d) An m = +6

defect before and after heating, respectively. Slight crinkling in the almost flat state indicatesthe presence of the m = 6 defect pattern in the sheet. Arrows show the 10-fold symmetryexpected from a curvature variation given by Equation 3.26, that is, as cos[2(m − 1)𝜃] withm = 6. (e) Two hemi-stadia director patterns joined together by a region of uniform director.Under actuation, a tent, with half-cone ends, is formed. (f ) A paper model, folded to form thejunction line to the uniform section and having a conical end, is decorated with referencearrows to indicate the points of the flat material in (c) corresponding to those in the risenshell. (Ware et al. [37]. Reproduced with the permission of American Chemical Society.)

resultant tent that forms. The ease of envisaging what the resultant structuresare when the director field has such simple component parts motivates ourorigami approach discussed in Section 3.10. Indeed, it is easy to fold a paperto illustrate the half-conical ends and how they might be joined to a uniformsection, see Figure 3.9(f ). But before turning to a radically new origami, wefirst deal with the mechanical and geometric consequences of nontopologicaldefect distributions of the director field.

�

� �

�



Sharon et al. [46] and Mostajeran [48] show that both 1D Cartesian and circu-larly symmetric n(r) distributions can yield interesting curvature distributionsand hence interesting shapes. The Cartesian system has a director at an angle𝜓(x2) to the x1-axis that varies only in the x2-direction, being simply translatedin x1, that is,

n(x2) = cos(𝜓(x2))x1 + sin(𝜓(x2))x2. (3.27)

For sufficiently small 1 − 𝜆, the Gaussian curvature is

K = −12(𝜆−2 − 𝜆2𝜈)(𝜓 ′′ sin(2𝜓) + 2𝜓 ′2 cos(2𝜓)). (3.28)

Given a choice of the curvature K , one can solve this differential equation forthe director distribution 𝜓(x2) that generates the K . These resultant patternsthen lead, for instance, to spherical spindles (“pointy” or “thorny” spheres, seeRef. 49), when K is a positive constant, or to hyperbolic spheres when K isconstant and negative. See representative director patterns and their experi-mentally resulting shells in Figure 3.10.Director spirals offer routes to more general distributions of curvature and

varieties of surface revolution [46, 48, 50]: See Figure 3.10(b) for a logarithmicspiral, first treated in Refs 34, 35, that is also an m = 1 defect with which wehave dealt earlier, but where now the angle 𝛼 takes a more general, constant,value between 𝛼 = 0 (radial) and 𝛼 = 𝜋∕2 (azimuthal).All values of 𝛼, since they do not vary with r, yield cones or anticones, with

the Gaussian curvature still localized at the tip. For 𝜆 < 1, contraction along thedirector, varying through log spirals from 𝛼 = 𝜋∕2 to 𝛼 = 0 (anti-cones), thereis an intermediate state, which is flat for a director angle of 𝛼c given by

cot2(𝛼c(𝜆)) = 𝜆1+𝜈 . (3.29)

For contractions (illumination, heating) along the director, 𝜆 < 1, the criticalangle always increases above 𝜋∕4, and vice versa for elongations (darkness,cooling).Figure 3.10(c) shows a qualitatively different example where the angle 𝛼 of

the director to the radial direction varies with r but is independent of 𝜃–theexample is of 𝛼(r) = r∕L, with L a constant setting the length scale–and thisvariation induces delocalized curvature. The curvature is explicitly [50]

K = 12(𝜆−2 − 𝜆2𝜈)

((

𝛼′′ + 3

r𝛼′)

sin(2𝛼) + 2𝛼′2 cos(2𝛼))

. (3.30)

Clearly, the scale of K is 1∕L2.Figure 3.10(d) shows the resultant curvatureK(r) for this choice of𝛼(r), which

will then drive the resultant surface. The determination of the correspondingsurfaces is difficult and in general a insolvable issue. Further information from

�

� �

�


K(r)

r/L

(a)

(b) (c) (d)

(i) (ii) (iii) (iv) (v)

2 4 6 8 10

Figure 3.10 Cartesian and circular director patterns generating Gaussian curvature. (a)Cartesian: From left to right: (i) The initially flat configuration of a circular glassy film 15 μmin thickness and 7.1 mm in diameter. (ii) The positive Gaussian curvature pattern. Thedashed circle indicates the boundaries of the circular film. (iii) The formation of positiveGaussian curvature in the actuated state from two distinct viewing angles. (iv) The negativecurvature pattern obtained as the orthogonal dual director field. (v) The formation ofnegative Gaussian curvature in the actuated state from two viewing angles. Circular: (b) Alogarithmic spiral nematic pattern with 𝛼 = 𝜋

4. (c) The director field defined by 𝛼(r) = r∕L,

and (d) the resulting Gaussian curvature distribution K = K(r) (From Mostajeran et al. [50].)

the three independent contributors to the metric tensor g is required. This is aproblem under active theoretical investigation. Important cases are surfaces ofconstant curvature:Spherical spindles, and spherical caps as a special case, emerge from solving

Equation 3.30 with K = const. > 0 to give the solution [50]:

cos(2𝛼(r)) = c1 −12

C(K)r2 + c2∕r2, (3.31)

where C(K) = K∕(𝜆−2 − 𝜆2𝜈) and where c1 and c2 are constants; see Figure 3.11.Setting c2 = 0 for compact discs (without a central region around r = 0 being

excised), the pattern extends to a radius r∕L =√

2(1+c1)C(K)

, whereupon the patterncan be extended to the outer annuli; see Figure 9 of [50]. Spindles form if c1 ≤1 − 2∕(1 + 𝜆1+𝜈), with spherical caps in the case of equality; see Figure 3.11(c).

�

� �

�


(a) (b) (c)

(d) (e) (f)

Figure 3.11 Spherical spindles: (a) The top half of a spindle of constant Gaussian curvature

K > 0 and parameter 𝜌 ∈ (0, 1∕√

K). The spindle arises as a surface of revolution. (b) Thedirector field defined by Equation 3.31 with K > 0 and c1 = 1 − 2∕(1 + 𝜆1+𝜈). The solid curveindicates the circle of radius r0 whose length is unchanged by the pattern. (c) The directorfield on a disc of radius r0 and the spherical cap of fixed boundary that is expected to form.Hyperbolic spheres: (d) A hyperbolic sphere. (e) The director field defined by Equation 3.31with K < 0 and c1 = −1. The solid circle indicates the circle of radius r0 whose length isunchanged by the stimulation of the pattern and is shown in (f ). (From Ref. [50].)

Hyperbolic cones for K = const. < 0 arise for the same range of c1 as earlier;see Figure 3.11(d) that illustrates a typical hyperbolic cone arising from a nega-tive curvature spiral, Figure 3.11(e) and (f ) for the pattern up to the radius thatdoes not change with stimulation.One can also ask these questions regarding light-induced curvature change

not simply in an initially flat sheet decorated by a director field but also in analready initially curved structure such as a sphere [49]. Now the director fieldmust have defects adding up to charge+2, most simply a+1 defect at each pole(the “hairy ball theorem”), or perhaps, 4 defects of charge+1/2 distributed overthe spherical surface. In the former case, the spherical shell becomes pointy (or“thorny”); see Ref. 49 for further details. Ohm et al. [51] have obtained solid andshell-like nematic structures, using very clever microfluidics, which do indeedbecome pointy as their nematic order is changed.

�

� �

�


3.9 Mechanico-Geometric Effects, Part 2

Spindles are pointy spherical shells and have Gaussian curvature concentratedat their poles as well as distributed on their surface. The pointy character maywell be of use in sensors and in precision actuation. In all these surfaces, withpositive or negative Gaussian curvature, transverse loads applied at their tipsalong their axis cause in-material stretches and hence their actuation is strong.To be effective mechanical devices, these shells should mostly be anchored

at their perimeters. To avoid strain mismatch between the shell edge and thesurrounding, passive substrate, one needs an unchanging perimeter [50]. Theradius r0 in the reference state of the disc that is invariant after a contraction 𝜆along n is (see Figure 3.11)

r0 = 1𝜆

√2K√1 + c1 − 𝜆2𝜈(2 + (c1 − 1)𝜆2) (3.32)

→2

√K

√𝜆𝜈−1 − 𝜆2(𝜈−1),

the latter case being for spherical caps when c1 = 1 − 2∕(1 + 𝜆1+𝜈). Thus,boundary matching and hence stress-free anchoring can only be achieved ata particular 𝜆. A system anchored at r = r0 will only be stress-free when 𝜆achieves the particular value that gives the said r0. It is not yet known whatthe states of intervening 𝜆 are–is there a sudden jump to a raised shell as theappropriate 𝜆 is approached? It is a question that might have considerabletechnological ramifications if, for instance, the sudden operation of a particulareffect is required.A shell that rises out of the plane, but which is anchored at an unchang-

ing circumference, and hence unchanging in-space radius, necessarily has anincreasing area as 𝜆 < 1. The condition of increasing area is

A∕A0 = 𝜆 × 𝜆−𝜈 > 1, (3.33)

that is, the product of the fractional in-plane length changes exceeds unity. For𝜆 < 1, then 𝜈 > 1 is required. Rubber has 𝜈 = 1∕2, and some nematic glasseshave 𝜈 <∼ 1 for thermal contractions. However, optically induced length changesseem to be accompanied by much greater photo-Poisson effects, and 𝜈 > 1 isthe likely norm.Hyperbolic spheres are also pointy and can also be, in principle, neutrally

anchored at a given radius r0:

r0 =1𝜆

√2K√𝜆2𝜈(2 + (c1 − 1)𝜆2) − (1 + c1), (3.34)

where r0 has to be less than the maximum disc size; see Figure 3.11 for anexample.

�

� �

�

3.10 Director Fields with Discontinuities–Advanced Origami 107

We now focus on director fields with discrete, rather than continuous,changes of direction.

3.10 Director Fields with Discontinuities–AdvancedOrigami!

Directors that are piecewise uniform and joined by a strict “grammar” also gen-erate aGaussian curved response upon excitation, but the curvature is localizedto the corners of a fold structure. It is tempting to identify this response withorigami since the surfaces are often folded along the lines of junction of patchesof uniform director.Many groups have attempted to emulate origami by actuat-ing folds in a flat sheet, but the “origami” we develop here is of a fundamentallydifferent and unique nature.Consider Figure 3.12(a); a circular sheet is folded several times. The sheet is

grasped at X on its perimeter, and the radius OX is placed on top of OY byfolding down along OZ by 𝜋∕2 so that the shaded sector OXZ sits on top ofsector OZY ; see Figure 3.12(b) and then finally (c). Essentially, these two sec-tors are “lost” (tucked under), and a circumference around O has effectivelybecome shorter–reminiscent of how we formed cones. In classic origami tech-nique, we have created a 3D structure with a “corner,” which, when the seamsand tucked-under sections of the sheet are ignored, has localized Gaussian cur-vature. This effective Gaussian curvature arises due to the “removal” of an areaof the sheet through folding.By contrast, consider the square director pattern of Figure 3.12(d). Diago-

nals have been dotted to focus the eye on where regions of uniform directorare joined to each other. It is clear that contraction along the director, andelongation perpendicular, upon illumination can only be resolved by forminga square pyramid–a 3D surface with Gaussian curvature localized at the apex,see Figure 3.12(e). Actually, the bend energy cost localized at the folds (edges)is, by convexity, reduced by relaxing the square pyramid to a cone (Figure 3.12f )where the bend is evenly distributed. Later, we see, for practical devices, thatthis relaxation can be disturbed by taking an array of these squares or otherconcurrent sources of Gaussian curvature.The “vocabulary” of such sectors, for instance, within the dotted lines of

Figure 3.12(d), and the “grammar” of how they must be combined, was set outin Refs 45, 52. A good working vocabulary is seen in Figure 3.13.Upon activation, that is, contraction by 𝜆< 1 along n and elongation 𝜆−𝜈 >1

perpendicular to n, the angle characterizing each sector changes. Consider suit-able right angle triangles with n along a side and a perpendicular to n alonganother side, or for the circular arc sectors, consider radii and arcs changing by𝜆 or 𝜆−𝜈 ; an example triangle is given in gray in Figure 3.13(c), the sides of whichare marked by the ratios 𝜆−𝜈 and 𝜆 by which they change upon activation–in

�

� �

�


’

A

A′

C

C′

O

O′

O

B

B′

(d)

(e)(f)

(a) (b) (c)

X

Y

OZ

ϕ

Figure 3.12 (a, b) folding under the area to be “lost” around an apex, which then effectivelydevelops Gaussian curvature at its tip and forms a faceted surface (c). This is still isometricorigami. (d) A sheet with a concentric square pattern of director suffers contraction alongthe director to become the square pyramid (e). This pyramid releases high bend energyalong its edges by relaxing into a cone (f ).

this case, their ratio clearly gives the tangent of the new angle 𝜃′∕2 to which theoriginal triangle’s apex semiangle 𝜃∕2 changes. The angle transformations are,for the circular sectors (a) and (b), respectively [52],

𝜃′ = 𝜆

1+𝜈𝜃; 𝜃

′ = 𝜆−1−𝜈

𝜃. (3.35)

The triangular elements (c) and (d) change, respectively, as

𝜃′ = 2 tan−1(𝜆−1−𝜈 tan(𝜃∕2)); 𝜃

′ = 2 tan−1(𝜆1+𝜈 tan(𝜃∕2)). (3.36)

For (e) and (f ), respectively, the changes are

𝜃′ = 2 tan−1(𝜆1+𝜈 tan(𝜃∕2)); 𝜃

′ = 2 tan−1(𝜆−1−𝜈 tan(𝜃∕2)). (3.37)

The grammar of placing sectors together dictates that (a) n is continuousacross a boundary, or (b) the director may discontinuously change directionacross the boundary, but only if the director makes the same angle with theboundary on both sides. An example of (a) is illustrated in Figure 3.9 when the

�

� �

�

3.10 Director Fields with Discontinuities–Advanced Origami 109

θ/2

θ/2

(a)

(b)

(c)

(c)

(d)

(e)

(f)

θ

θ

θ

θ

λ

λ–ν

Figure 3.13 A vocabulary of patches of director that lead to angle changes in discretesections of a shell as the distortion 𝜆 changes.

semicircular form of a + 12defect, which led to a hemiconical end to a tent, was

joined to the remaining hemi-stadium, which formed the ridged body of theresulting tent; see Figure 3.9. Rule (b) arises since any deformation gradient 𝛌should lead to components of contraction or elongation along the junction thatare the same on each side. Otherwise, the solid would tear itself apart uponactivation due to the finite accruedmismatch between the opposite sides of theboundary. This requirement of “rank 1 (R1) connectedness” is fundamental toall deformations of textured solids and is used, for instance, in the deforma-tions and transformations of shape memory alloys, such as Martensite [53]. Itcan be seen, for example, that the four triangular sectors in the square patternof Figure 3.12(d) are combined in this way.The origami arising from responsive nematic solids has recently been termed

“nonisometric” (non-length-preserving) origami in contrast to conventionalorigami where folds cause area (and length) to disappear [54]. But on the facesthat do survive in normal origami, lengths and angles are preserved–a trian-gle drawn in Figure 3.12(a) that does not cross the shaded sectors retains itsstraight sides, and the sums of its internal angles remains 𝜋. If the triangle ispunctured by what will become the apex, and the triangle’s sides traverse theshaded sectors in such a way that a connected, three-sided shape results afterfolding, then the sum of its internal angles now exceeds 𝜋–a sign of enclosed

�

� �

�


Gaussian curvature–but the sides remain straight. By contrast, a triangle drawnin Figure 3.12(d) transforms, in this new origami, into a three-sided shape, thesides of which are no longer even geodesics–there has been an essential noni-sometric transformation.Such a transformation leads to the stretch and strong actuation that we

explored before in the continuous case.

3.11 Mechanico-Geometric Consequencesof Nonisometric Origami

Defects in arrays can create very subtle, beautiful, and potentially useful effects,far more so than they can in isolation. The consequence of combinations intoarrays can be both topographical, as before, and topological:The relaxation of square pyramids to smooth cones is not possible when they

are formed from an array of activated adjacent squares, since adjacent pyramidsshare sides of their bases–they cannot become circular simultaneously abouttwo centers, and they remain square; see Figure 3.14(a) and (b). Square pyra-mids, or even isolated, smooth circular cones, when loaded at the tip along aperpendicular to the base are induced to stretch around their flanks and strongactuation results–seeWare et al.’s exploitation of such arrays that lift hundredsof times their own weight [43].A linear chain of alternating± 1

2defects can act as a grain boundary separating

two regions of a uniform director, see Figure 3.14(c). Now photodeformationcauses the region that becomes too long in the transverse direction to disposeof the additional length by forming ridges, Figure 3.14(d), while the region thatis relatively too short remains planar. Eventually, the bend cost of the ridgesand furrows in the intensely folded region becomes too high, and the bend isdisposed of in both regions, which then are faceted, Figure 3.14(e).An array of textural pieces constructed from the vocabulary of wedges is

shown in Figure 3.14(f ); combined with slits or cuts through the sheet at themarked locations, the sheet exhibits topological changes as the slits openinto fully realized holes (Figure 3.14g) instead of displaying the previoustopographical transformations. This response has been used to create light- orheat-controllable sieves [44].

3.12 Conclusions

Nematic solids have a close relation between their mechanical state, forinstance, elongation or contraction along the director, and their state ofnematic order. As well as the classic route of changing order by heat, one has

�

� �

�

3.12 Conclusions 111

+1

–1

(a)(b)

(e)

(f)(g)

(d)

(c)

Figure 3.14 Arrays and networks of defects for topography and topology. (a) An array ofpyramids rises from a flat sheet decorated with a director field consisting of concentricsquares. These are discrete forms of m = +1 defects, but now the square pyramids intowhich they rise (b) cannot relax to cones. The ±1 defects are identified by two labelledexamples in red. Rising pyramids, similar to these, were used to lift heavy loads [43]. (c) Tworegions of uniform director can be welded together by a grain boundary of (discrete) ±1∕2defects (circled in blue and red); after [52]. (d) Upon weak deformation, a transverselyshrunk, still planar region in contact with a ridged region remains. (e) Upon strongeractuation, both regions are turned into parts of a faceted bottle in order to reduce theoverall bend energy along the ridges. (f ) An array of + 1

2and −1 defects, where neighboring

+ 1

2s’ cores are connected by slits (heavy lines) that are as yet unopened. (g) Contraction

along the directors around the defects leads to the opening of slits while remainingplanar–a topological rather than topographical change; taken from Ref. [44] whereexperimental realizations are shown. (Ware et al. [43]. Reproduced with the permission ofThe American Association for the Advancement of Science.) (See color plate section for thecolor representation of this figure.)

�

� �

�


the more subtle and more tunable tool of light to change the order and hencethe mechanical state. We have concentrated on the latter. The changes indimension with order parameter change are large for nematic elastomers, andstill very significant, on the scale of conventional solids, in nematic glasses.Theunderlying contractions and elongations that we initially explain and exploretheoretically and experimentally are rendered still more unusual by havingdirector variation, which is a route to spatial variation in mechanical response.Director variation through the thickness of a material leads to bend and curl,

both potentially complex because of the bend occurring inmore than one direc-tion at a time.We initially deal with curved response either in the weak limit ormore strongly, where we discuss the suppression of Gaussian curvature.In the second part of the review, we have suggested how director variation

in plane, both continuous and discrete, can lead to new mechanical possibili-ties utterly inaccessible to solids without nematic order–namely the creationof Gaussian curvature with the stringent consequences of stretch and anglechanges it brings, but without the energetic cost of stretch. Indeed, it is theavoidance of elastic stretch that would arise, where these deformations wereblocked and were not free to develop, that is proposed to be at the heart ofnew, strong forms of actuation. Continuous variation leads to classic curvedshells, while discrete variation leads to faceted surfaces. Discrete variation ofthe director has a superficial similarity with classic origami, but in reality, it isradically different in that it is nonisometric–one does not fold away the surfacebut rather relies on the intrinsic length changes, something one could, in boththe continuous and discrete cases, termed as “metric mechanics.”

References

1 Finkelmann, H., Nishikawa, E., Pereira, G.G., and Warner, M. (2001) A newopto-mechanical effect in solids. Physical Review Letters, 87, 015 501.

2 Warner, M. and Terentjev, E.M. (2007) Liquid Crystal Elastomers, Oxford,revised, paperback edition Aufl.

3 Tajbakhsh, A. and Terentjev, E. (2001) Spontaneous thermal expansion ofnematic elastomers. European Physical Journal E, 6, 181–188.

4 Clarke, S., Tajbakhsh, A., Terentjev, E., and Warner, M. (2001) Anomalousviscoelastic response of nematic elastomers. Physical Review Letters, 86,4044–4047.

5 Finkelmann, H., Greve, A., and Warner, M. (2001) The elastic anisotropy ofnematic elastomers. European Physical Journal E, 5, 281–293.

6 Liu, D., Bastiaansen, C.M.W., de Toonder, J.M.J., and Broer, D.J. (2012)Light-induced formation of dynamic and permanent surface topologies inchiral-nematic polymer networks. Macromolecules, 45, 8005–8012, doi:10.1021/ma301628h.

�

� �

�

References 113

7 Yu, Y., Nakano, M., and Ikeda, T. (2003) Photomechanics: directed bendingof a polymer film by light. Nature, 425, 145.

8 Harris, K.D., Cuypers, R., Scheibe, P., van Oosten, C.L., Bastiaansen,C.W.M., Lub, J., and Broer, D.J. (2005) Large amplitude light-inducedmotion in high elastic modulus polymer actuators. Journal of MaterialsChemistry, 15, 5043.

9 Tabiryan, N., Serak, S., Dai, X.M., and Bunning, T. (2005) Polymer film withoptically controlled form and actuation. Optics Express, 13 (19), 7442–7448,doi: 10.1364/OPEX.13.007442.

10 Kondo, M., Yu, Y., and Ikeda, T. (2006) How does the initial alignment ofmesogens affect the photoinduced bending behavior of liquid-crystallineelastomers? Angewandte Chemie International Edition, 45 (9), 1378–1382,doi: 10.1002/anie.200503684.

11 Corbett, D. and Warner, M. (2007) Linear and nonlinear photoinduceddeformations of cantilevers. Physical Review Letters, 99, 174 302.

12 Corbett, D. and Warner, M. (2008) Bleaching and stimulated recovery ofdyes and of photocantilevers. Physical Review E, 77 (5), 051 710.

13 Kneževic, M., Warner, M., Copic, M., and Sánchez-Ferrer, A. (2013) Pho-todynamics of stress in clamped nematic elastomers. Physical Review E, 87,062 503, doi: 10.1103/PhysRevE.87.062503.

14 Corbett, D., van Oosten, C.L., and Warner, M. (2008) Nonlinear dynamicsof optical absorption of intense beams. Physical Review A, 78, 013 823.

15 van Oosten, C.L., Corbett, D., Davies, D., Warner, M., Bastiaansen, C.W.,and Broer, D.J. (2008) Bending dynamics and directionality reversal inliquid crystal network photoactuators. Macromolecules, 41 (22), 8592–8596.

16 Harvey, C.L.M. and Terentjev, E.M. (2007) Role of polarization and align-ment in photoactuation of nematic elastomers. European Physical Journal E,23, 185–189.

17 Dunn, M.L. (2007) Photomechanics of mono- and polydomain liquid crystalelastomer films. Journal of Applied Physics, 102, 013 506.

18 Corbett, D. and Warner, M. (2008) Polarization dependence of opticallydriven polydomain elastomer mechanics. Physical Review E, 78, 061 701.

19 Hon, K.K., Corbett, D., and Terentjev, E.M. (2008) Thermal diffusion andbending kinetics in nematic elastomer cantilever. European Physical JournalE, 25 (1), 83–89.

20 Broerman, A.W., Venerus, D.C., and Schieber, J.D. (1999) Evidence for thestress-thermal rule in an elastomer subjected to simple elongation. Journalof Chemical Physics, 111, 6965–6969.

21 Corbett, D., Xuan, C., and Warner, M. (2015) Deep optical penetra-tion dynamics in photo-bending. Physical Review E, 92, 013 206, doi:10.1103/PhysRevE.92.013206.

22 Dawson, N.J., Kuzyk, M.G., Neal, J., Luchette, P., and Palffy-Muhoray, P.(2011) Experimental studies of the mechanisms of photomechanical effects

�

� �

�


in a nematic liquid crystal elastomer. Journal of the Optical Society ofAmerica B, 28 (8), 1916.

23 Dawson, N.J., Kuzyk, M.G., Neal, J., Luchette, P., and Palffy-Muhoray, P.(2011) Modeling the mechanisms of the photomechanical response of anematic liquid crystal elastomer. Journal of the Optical Society of AmericaB: Optical Physics, 28 (9), 2134.

24 Mol, G.N., Harris, K.D., Bastiaansen, C.W.M., and Broer, D.J. (2005)Thermo-mechanical responses of liquid-crystal networks with a splayedmolecular organization. Advanced Functional Materials, 15, 1155–1159.

25 van Oosten, C.L., Harris, K.D., Bastiaansen, C.W.M., and Broer, D.J. (2007)Glassy photomechanical liquid-crystal network actuators for microscaledevices. European Physical Journal E, 23, 329–336.

26 Sawa, Y., Ye, F., Urayama, K., Takigawa, T., Gimenez-Pinto, V., Selinger,R.L.B., and Selinger, J.V. (2011) Shape selection of twist-nematic-elastomerribbons. Proceedings of the National academy of Sciences of the UnitedStates of America, 108 (16), 6364–6368, doi: 10.1073/pnas.1017658108.

27 Sawa, Y., Urayama, K., Takigawa, T., Gimenez-Pinto, V., Mbanga, B.L., Ye,F., Selinger, R.L.B., and Selinger, J.V. (2013) Shape and chirality transitionsin off-axis twist nematic elastomer ribbons. Physical Review E, 88, 022502-1-8, doi: 10.1103/PhysRevE.88.022502.

28 Warner, M., Modes, C.D., and Corbett, D. (2011) Suppression of curvaturein nematic elastica. Proceedings of the Royal Society of London Series A, 466,3561–3578.

29 Warner, M., Modes, C.D., and Corbett, D. (2011) Curvature in nematicelastica responding to light and heat. Proceedings of the Royal Society ofLondon Series A, 466, 2975–2989.

30 Warner, M. and Mahadevan, L. (2004) Photoinduced deformationsof beams, plates, and films. Physical Review Letters, 92, 134 302, doi:10.1103/PhysRevLett.92.134302.

31 Jin, L., Jiang, X., and Huo, Y. (2006) Light-induced nonhomogeneity andgradient bending in photochromic liquid crystal elastomers. Science inChina Series G: Physics, Mechanics and Astronomy, 49, 553–563, doi:10.1007/s11433-006-0553-x.

32 Jin, L., Yan, Y., and Huo, Y. (2007) Light-Induced Bending and Smart Con-trol of Photochromic Liquid Crystal Elastomers. Proceedings of SPIE 6423,International Conference on Smart Materials and Nanotechnology in Engi-neering, November 14, 2007, vol. 64230W, doi: 10.1117/12.779349.

33 Corbett, D. and Warner, M. (2006) Nonlinear photoresponse of disorderedelastomers. Physical Review Letters, 96, 237 802.

34 Modes, C.D., Bhattacharya, K., and Warner, M. (2010)Disclination-mediated thermo-optical response in nematic glass sheets.Physical Review E, 81, R060 701.

�

� �

�

References 115

35 Modes, C.D., Bhattacharya, K., and Warner, M. (2011) Gaussian curvaturefrom flat elastica sheets. Proceedings of the Royal Society of London Series A,467 (2128), 1121–1140, doi: 10.1098/rspa.2010.0352.

36 Modes, C.D. and Warner, M. (2015) Negative Gaussian curvature frominduced metric changes. Physical Review E, 92, 010 401, doi: 10.1103/Phys-RevE.92.010401.

37 Ware, T.H., Perry, Z.P., Middleton, C.M., Iacono, S.T., and White, T.J.(2015) Programmable liquid crystal elastomers prepared by thiol-enephotopolymerization. ACS Macro Letters, 4, 942–946, doi: 10.1021/acs-macrolett.5b00511.

38 de Haan, L.T., Sánchez-Somolinos, C., Bastiaansen, C.M.W., Schenning,A.P.H.J., and Broer, D.J. (2012) A theory for programming complex shape inthin nematic elastomer and glass sheets. Angewandte Chemie InternationalEdition, 51 (50), 12 469–12 472, doi: 10.1002/anie.201205964.

39 Dervaux, J. and Amar, M.B. (2008) Morphogenesis of growing softtissues. Physical Review Letters, 101 (6), 068 101, doi: 10.1103/Phys-RevLett.101.068101.

40 Müller, M.M. and Amar, M.B. (2008) Conical defects in growingsheets. Physical Review Letters, 101 (6), 156 104, doi: 10.1103/Phys-RevLett.101.156104.

41 Klein, Y., Efrati, E., and Sharon, E. (2007) Shaping of elastic sheets by pre-scription of non-Euclidean metrics. Science, 315 (5815), 1116–1120, doi:10.1126/science.1135994.

42 Kim, J., Hanna, J.A., Byun, M., Santangelo, C.D., and Hayward, R.C. (2012)Designing responsive buckled surfaces by halftone gel lithography. Science,335 (6073), 1201–1205, doi: 10.1126/science.1215309.

43 Ware, T.H., McConney, M.E., Wie, J.J., Tondiglia, V.P., and White, T.J.(2015) Voxelated liquid crystal elastomers. Science, 347 (6225), 982–984,doi: 10.1126/science.1261019.

44 Modes, C.D., Warner, M., Sánchez-Somolinos, C., de Haan, L.T., and Broer,D. (2013) Angular deficits in flat space: remotely controllable apertures innematic solid sheets. Proceedings of the Royal Society of London Series A,469 (6225), 20120 631, doi: 10.1098/rspa.2012.0631.

45 Modes, C.D. and Warner, M. (2011) Blueprinting nematic glass: systemat-ically constructing and combining active points of curvature for emergentmorphology. Physical Review E, 84 (2), 021 711, doi: 10.1103/Phys-RevE.84.021711.

46 Aharoni, H., Sharon, E., and Kupferman, R. (2014) Geometry of thinnematic elastomer sheets. Physical Review Letters, 113 (25), 257 801, doi:10.1103/PhysRevLett.113.257801.

47 Tabiryan, N.V., Nersisyan, S.R., Steeves, D.M., and Kimball, B.R. (2010) Thepromise of diffractive waveplates. Optics & Photonics News, 21 (3), 40–45,doi: 10.1364/OPN.21.3.000040.

�

� �

�


48 Mostajeran, C. (2015) Curvature generation in nematic surfaces. PhysicalReview E, 91, 062 405, doi: 10.1103/PhysRevE.91.062405.

49 Modes, C.D. and Warner, M. (2012) Responsive nematic solid shells: topol-ogy, compatibility, and shape. EPL (Europhysics Letters), 97 (3), 36 007, doi:10.1209/0295-5075/97/36007.

50 Mostajeran, C., Warner, M., Ware, T.H., and White, T.J. (2016) Encod-ing Gaussian curvature in glassy and elastomeric liquid crystal solids.Proceedings of the Royal Society of London Series A, 472, 20160 112, doi:10.1098/rspa.2016.0112.

51 Ohm, C., Fleischmann, E.K., Kraus, I., Serra, C., and Zentel, R. (2010) Con-trol of the properties of micrometer-sized actuators from liquid crystallineelastomers prepared in a microfluidic setup. Advanced Functional Materials,20, 4314–4322, doi: 10.1002/adfm.201001178.

52 Modes, C.D. and Warner, M. (2012) The Activated Morphology of GrainBoundaries in Nematic Solid Sheets. Proceedings of SPIE 8279, Emerg-ing Liquid Crystal Technologies VII, February 9, 2012, vol. 82790Q, doi:10.1117/12.916788.

53 Kaushik Bhattacharya’s Group (2007) Microstructure of Martensite, OxfordUniversity Press, Oxford.

54 Plucinsky, P., Lemm, M., and Bhattacharya, K. (2016) A theory for program-ming complex shapes in thin nematic elastomer and glass sheets. PhysicalReview E, 94 (1), 010701(R), doi: 10.1103/physreve.94.010701.

�

� �

�

117

4

Photomechanical Effects in Amorphousand Semicrystalline PolymersJeong Jae Wie

Department of Polymer Science and Engineering, Inha University, Incheon, South Korea

4.1 Introduction

This chapter focuses on photomechanical responses observed in photorespon-sive polymeric materials that employ photochromic constituents to transducelight into macroscopic mechanical responses. As discussed, azobenzene hasand continues to be widely used to convert light into a conformational changeat molecular level in these materials. This chromophore was first observed toundergo photoisomerization in 1937 [1]. In polymers, the photoisomerizationof azobenzene was shown to affect the conformation of polymeric segmentby Lovrien in 1967. Here, irradiation with UV light significantly reducedthe solution viscosity of a mixture of low-molecular-weight chrysophenine(CHP) chromophore and polymethacrylic acid (PMAA). Lovrien referredto the response as the photoviscosity effect [2]. Upon UV irradiation,trans-azobenzenemolecules (9Å) in the CHP absorb light and photoisomerizeinto the cis form (5.5Å), which affects both the CHP solubility in aqueous solu-tion and the binding force to PMMA, resulting in a photoinduced reduction inthe viscosity.Shortly afterward, Agolini andGay demonstrated reversible photo- and ther-

mal contraction of a solid polymeric material functionalized with azobenzene[3]. The linear and semicrystalline polyimide contracted upon UV light irra-diation resulted in trans–cis isomerization of the azobenzene moieties. Subse-quent examination by Paik and Morawetz focused on the kinetics of trans–cisphotoisomerization of azobenzene in polymeric systems, noting that the iso-merization is considerably reduced in glassy polymeric systems in comparisonwith those in solution [4]. Smets reported similar suppression in the kinetics


�

� �

�

118 4 Photomechanical Effects in Amorphous and Semicrystalline Polymers

for spirobenzopyran in a system with a polyester backbone, below the glasstransition temperature of the polymer [5].These results focused on fundamen-tal explorations of photoisomerization in polymeric materials, which are thefirst signs of the strong influence of the polymer physical properties on theresponse of the materials.In 1980, Eisenbach reported UV- and visible-light-induced reversible dimen-

sional change from amorphous cross-linked polymers once again containingthe azobenzene chromophore [6]. The expansion and contraction ranges werevery limited, 0.15–0.25%. Despite this small strain, this report was perhapsthe first to show convincingly that light through photochemistry results inphotomechanical transduction.After the work of Eisenbach, the focus of the field shifted. Related to the

overall theme of this chapter, in 1995, the formation of surface relief gratings(SRGs) was reported in thin, azobenzene-functionalized polymeric films(10 nm to 1 μm) when exposed to spatially polarized light [7, 8]. The amplitudeof the gratings reached several hundred nanometers measured by atomicforce microscopy (AFM). In many of these examinations, the deformationsare surprisingly large compared to the thickness of films. As touched on inChapter 3, the SRG is formed by mass transport of material. Inherently, themobility of linear polymers allows for greater movement and larger amplitudesurface reliefs. Surprisingly, photoinduced surface reliefs are larger in glassymaterials. While this optically reversible photomechanical deformation pro-cess can occur withmild light intensity (1–100mW/cm2) at room temperature,it requires photoisomerization of azobenzene chromophores.In 2001, Finkelmann, Warner, and coworkers reported a large magnitude of

reversible uniaxial shape change from monodomain (MD) liquid crystallineelastomers (LCEs) upon UV irradiation [9]. The LCEs are prepared withazobenzene chromophores cross-linked with polysiloxane main chains.Optically induced deformation of as much as 20% strain was experimentallydemonstrated. Subsequent to this study, reversible macroscopic photome-chanical actuation in any arbitrary direction was reported by Ikeda andcoworkers in 2003 [10]. Azobenzene-containing liquid-crystalline monomersare covalently cross-linked using acrylate chemistry, and the molecular pho-tomechanical effects are translated through cross-linked junctions, resultingin macroscopic-scale motion. Here, the bending direction of the polymericmonolith is precisely controlled in any direction by the orientation of lin-early polarized UV (𝜆= 366 nm) light. Irradiation with higher wavelengthlight triggers the reverse photoisomerization, from cis-azobenzene intotrans-azobenzene. This results in the recovery of the initial flat geometry forthe bent polymeric film. One of the main advantages of liquid-crystallinepolymers is facile programmability of molecular alignment and their controlby linearly polarized light. Chapters 5 and 9 deal with photomechanicaleffects of liquid-crystalline polymer networks. In this chapter, we focus on

�

� �

�

4.3 The Amorphous Polymer State 119

photomechanical effects in amorphous and semicrystalline polymers that arepolymers in the bulk solid states.

4.2 Polymeric Materials

In the numerous literature reports to date, photomechanical effects in amor-phous and semicrystalline polymeric materials have shown strong dependenceon the morphology, free volume, and other variables. Accordingly, I begin thisdiscussion by detailing the fundamental properties and terms used to describethese materials.As touched on in Chapter 2 and detailed in Chapter 7, photomechanical

responses can in fact be observed in small molecules that form crystallinesolids. Here, we are concerned with polymeric materials. To define the scope ofthis chapter, it is known that in the case of small organic molecules, an increasein molecular weight dramatically changes their material properties. Alkanehydrocarbons exist as gases at room temperature with carbon chain lengthsup to 4 (methane, ethane, propane, and butane gases) [11]. For hydrocarbonlengths between 5 and 11, these materials are liquid. Between 9 and 25 carbonatoms, the viscosity of the liquid increases, and eventually, crystalline solidscan be obtained with 25–50 carbon atoms. These materials, despite theircomparatively large molecular weights, are not the focus of this chapter. Anymolecules having molecular weights greater than 10 kDa are categorized asmacromolecules. Although macromolecule is sometimes used as a synonymof polymer, macromolecules do not necessarily contain repeating units withinthe molecular structures. The term “polymer” is derived from the two Greekwords meaning “many” (poly) and “units” (mer). Accordingly, polymers arematerials with numerous repeating units. By increasing the repeating unitfrom 1 through 3, the molecules are called monomer, dimer, and trimer,respectively. Molecules consisting of larger number of repeating units arereferred as oligomers if their molecular weights do not exceed 10 kDa thresh-old to be macromolecules. Unlike macromolecules, the minimum molecularweight for polymers does not have this clear-cut criteria, and it is still a subjectof debate. However, commonly, polymer scientists often utilize 25 kDa as arough number [11].

4.3 The Amorphous Polymer State

In the bulk solid states, polymers are classified as amorphous and semicrys-talline polymers. In the amorphous state, polymers have neither a first-ordermelting transition nor X-ray diffraction patterns due to the absence of regularcrystalline regions. During cooling and heating processes, amorphous poly-mers only go through a second-order glass–rubber transition. This transition

�

� �

�


temperature is called glass transition temperature (Tg). Above the Tg, glassypolymer chain segments become rubbery by obtaining long-range cooperativemolecular mobility at several Kuhn length scale [11]. Here, the Kuhn length, Lk,is defined by [12–14]

Lk = limL→∞

⟨R𝜃

2 (L)⟩L

, (4.1)

where ⟨R𝜃

2(L)⟩is the mean-square end-to-end distance under Flory 𝜃 condi-tions and L the contour (curvilinear) length.The Kuhn length is the characteristic length scale of chains being noncorre-

lated with the neighboring segments so that the Kuhn segments can be consid-ered as if they are freely jointed chains (FJC, the simplest mathematical model).This correlation has an exponential decay function, and the rate of decay isscaled by the persistence length, Lp, defined by [12–14]:

⟨ai ⋅ aj⟩

Lb2 = exp

(

− LLp

)

, (4.2)

where ai = ri − rj, ri being the position of the ith monomer in space, andLb = |ai|. In the case of ideal polymers at equilibrium, the Kuhn segment isequal to Lb.Persistence length is roughly half of the Kuhn length (Lk ∼ 2Lp) and is a mea-

sure of the polymer rigidity defined by [15]

Lp =Bs

kBT, (4.3)

where Bs is the bending stiffness, kB is the Boltzmann’s constant, and T is theabsolute temperature. Hence, stiffer molecules have longer persistence length.The molecular rigidity of polymer can be quantified using the ratio of Kuhn

length (or persistence length) to the chain diameter. Stiffness of polymersis strongly influenced by chemical structures of main-chain molecules andside groups, intermolecular forces, and steric hindrance. This is discussed inSection 4.6.Several Kuhn-length-scale coordinated molecular motions result in vis-

coelasticity and the Tg of polymeric materials that are distinctively differentfrom small organicmolecules.While typically only 1–4 chain atoms participatein coordinated molecular motion below polymer Tg, 10–50 chain atoms areinvolved in this collective motion in the polymer Tg region [11]. Here, theintrinsic requirement for the motion is physical space for segments due toexcluded volume of polymers originating from Pauli’s exclusion principle.Therefore, free volume is an inevitable correlated parameter in order toconsider the segmental mobility of polymers. In photomechanical effects inpolymers, light is the input source to generate sufficient energy for mechanicaloutput, which is translated through polymer segments. Accordingly, it is

�

� �

�

4.4 The Semicrystalline Polymer State 121

expected and has been reported that many of these factors can stronglyinfluence the resulting photomechanical response.The influence as well as the dependence of photoisomerization in the solid

state on free volume has been extensively detailed. Naito et al. reported thata free-volume fluctuation model successfully fit experimental photoisomer-ization reactions in amorphous polymer solids by considering free-volumedistribution and its local fluctuation. For amorphous polymers with smallerfree-volume fluctuation, reduced trans–cis photoisomerization occurred inazobenzene molecules [16].Factors affecting the free volume of polymers include temperature and

pressure as well as the intermolecular packing of polymer chain segments.For example, the steady state cis isomer concentration for the trans–cisphotoisomerization of azobenzene decreases with lowering temperaturein poly(methylmethacrylate) (PMMA) due to reduced local free-volumefluctuations and the rate constant [16]. Regarding the intermolecular packingof polymer chains, crystalline segments have markedly different molecularpacking from amorphous region.

4.4 The Semicrystalline Polymer State

Based on entropic considerations, polymers prefer random coil conformationwithout long-range molecular ordering rather than aligned lamellar structuresby folding and stacking, tending to form crystals. A number of factors areknown to inhibit or prevent crystallization in polymers including high molec-ular weight, irregular backbone, and functionalization with bulky groups.Hence, relatively short chains with regular molecular structure promotemolecular packing of polymers and overcome entropic costs, resulting in thedevelopment of crystallinity. The degree of crystallinity of even highly orderedpolymers, however, is prohibited from reaching 100% with the lowest freeenergy due to the polymer entanglements, defects from chain ends and chainfolding, and the polydisperse nature of polymer molecular weight [17]. Hence,these materials are commonly and accurately referred to as semicrystallinepolymers. Accordingly, they are distinguished from truly crystalline materialsdetailed in Chapter 7.As shown in Figure 4.1, totally amorphous polymers have additional free vol-

ume in comparison with semicrystalline polymers due to structural disorder.It is well known that free volume and the segmental mobility are controllingparameters of trans–cis isomerization processes. It should not be surprisingthat the Williams–Landel–Ferry (WLF) theory [19] well describes thermalback isomerization processes of azobenzene in the rubbery state [20].TheWLFtheory is an empirical equation associated with Boltzmann time–temperaturesuperposition (TTS) to build a master curve. White and coworkers recently

�

� �

�


Semicrystalline(a)

(b)

Amorphous

Melt

Solid

Glass

Semicrystalline solid

Crystalline solid

Temperature

Liquid

Tg Tm

Sp

ecifi

c v

olu

me

Liquid crystal

Figure 4.1 (a) Schematics of the molecular structures in both melt and solid states forsemicrystalline, amorphous, and liquid crystal polymers. (b) Specific volume for totallyamorphous, semicrystalline, and crystalline polymers against temperature, upon coolingfrom the liquid melt. As crystallinity of polymers cannot be 100%, the curve for crystallinesolid illustrates the extreme limit by a totally crystalline solid. (Callister and Rethwisch [18].Reproduced with the permission of John Wiley & Sons.)

�

� �

�

4.4 The Semicrystalline Polymer State 123

103(a)

(b)

E′ (

MP

a)

102

101

10

1.2

1.0

0.8

0.6

0.4

0.2

0.0–60 –50 –40 –30 –20 –10

T – Tg (°C)

α/α R

T

0 10 20 30 40 50 60

35 60 85

Temperature (°C)

Eq

uili

bri

um

be

nd

ing

an

gle

(°)

110 135 160 185 210

35

30

25

20

15

10

5

Figure 4.2 (a) Temperature-resolved storage modulus (O, left axis) and equilibrium bendingangle (Δ, right axis) for cantilever bending of photoresponsive azobenzene-functionalizedpolymer. (b) Photomechanical master curve for photoresponsive azobenzene-functionalizedpolymer with various cross-link densities. The normalized bending angle is plotted as afunction of normalized temperature (T − T g). (Lee et al. [21]. Reproduced with thepermission of American Chemical Society.)

illustrated that a master curve of the photomechanical response of a glassy,liquid-crystalline material (visualized as cantilever bending) can be preparedvia WLF theory as evident in Figure 4.2 [21]. This graph illustrates a clearand strong relationship between polymer physics and photomechanicalbehaviors.In addition to the glass–rubber transition of amorphous polymers, semicrys-

talline polymers have an additional thermal transition temperature – melting

�

� �

�


temperature (Tm). The Tg and Tm of semicrystalline polymers have awell-known relation via Boyer–Beaman rule [22, 23]. This empirical relationpredicts the ratio of Tg to Tm, both in Kelvin, to be 1/2 for symmetric polymers(i.e., polyethylene, polyvinylidene fluoride) and Tg/Tm = 2/3 for asymmetricpolymers (i.e., polystyrene, polyisoprene). While the fundamental basis for theBoyer–Beaman rule is not yet clear, it is understood that this rule of thumb isa result of the relationship of both thermal transitions to common parameterssuch as specific volume, molecular stiffness, polarity of the polymer, moleculargeometry, and cross-link density. In a similar way, recent studies of photome-chanical effects in semicrystalline materials have been difficult to attribute toa specific variable, due to the strong intercorrelation of these factors.

4.5 Absorption Processes

As detailed in Chapter 2, the transduction of light energy into mechanicaldeformation is typically driven by the absorption of light by a photochromicmoiety. Without the inclusion of chromophores (i.e., azobenzenes) or othermeans of conjugation, most polymers are insulators having large energyband gap (Eg > 5 eV) [24]. Trans-azobenzene when embedded into polymericmaterials can absorb visible light (n→ 𝜋*) in addition to strong absorption byUV (𝜋→𝜋*). The absorption of blue and violet light results in appearance oforange color of azobenzene compounds in both small molecule and polymericforms.The absorption of light is not uniform through the bulk ofmostmaterials, due

to the strong absorption coefficient of azobenzene and other chromophoresused in these examinations. Accordingly, more light is absorbed at the filmsurface and exponentially attenuates at a constant rate through the materialthickness. This famous empirical relation is the Beer–Lambert law (also calledthe Beer–Lambert–Bouguer law or simply Beer’s law) [25–27], and in a sim-ple form, transmitted light intensity at any arbitrary depth, x, can be calculatedusing the following equation:

I(x) = I0 exp(

−xl

)

, (4.4)

where I(x) is the transmitted intensity at a depth x, I0 the incident intensity, andl the penetration length.When x is equal to l, the ratio of transmitted intensity to incident intensity

becomes I(x)/I0 = 1/e= 0.37. Hence, the penetration depth is defined as thedepth where the light intensity falls to 37% of original intensity. The inverseof the penetration depth 1/x is the absorption coefficient, 𝛼, with the dimen-sions of inverse length. The absorption coefficient is often used as an indicatorof how strongly absorbing a material is at a given wavelength.

�

� �

�

4.5 Absorption Processes 125

While the Beer–Lambert law is a powerful analytical tool, Lambertian propa-gation of light is only valid at low chromophore concentration and light intensi-ties. Serra andTerentjev pointed out that the nonlinear regime is very accessibleand therefore common inmany of thematerial systems inwhich photomechan-ical effects have been examined and reported [28]. Irradiation of actinic lightat the wavelength of the trans-azobenzene absorption results in trans–cis pho-toisomerization. This decrease in trans isomer concentration leads to reducedabsorption at the absorption of the trans isomer, similarly to photobleaching.The change in concentration of cis isomers can be estimated by [29]

[cis]t =1 −ODt∕OD0

1 − 𝛼cis∕𝛼trans[trans]0, (4.5)

where [cis]t is the concentration of cis isomer at time t, [OD]t/[OD]0 the ratio ofoptical density at time t and zero, [trans]0 the initial concentration of trans iso-mer, and 𝛼cis/𝛼trans the ratio of absorption coefficient (also known as extinctioncoefficient) of cis and trans isomers.After calculating the change in cis isomer concentration, the change in con-

centration of trans isomer can easily be obtained by

[trans]t = [trans]0 − [cis]t , (4.6)

where [trans]t is the concentration of trans isomer at time t.By taking the effective trans–cis photobleaching effects originating from the

photoisomerization into account, Serra andTerentjevmodified these equationsto predict nonlinear absorption behaviors as follows [28].At small x/l limit (x/l ≪ 1),

ln(

I(x, t)I0

)

∼ −xl

[

1 − A1 + A

(1 − e−𝛾(1+A)t)]

, (4.7)

where nondimensional parameter A= I0kTC/𝛾 , which represents the balancebetween photo and thermal isomerization at a given incident intensity. Theterm kTC is the trans–cis constants of photoisomerization, and 𝛾 is the rate ofspontaneous thermal cis–trans isomerization.In the photostationary state at small x/l [28],

ln(

I(x)I0

)

∼ −x∕l1 + A

. (4.8)

Serra and Terentjev indicate that the nonlinear absorption kinetics arenot limited to azobenzene molecules but are essentially applicable to allchromophores.Absorption kinetics of azobenzene moieties in polymeric materials are influ-

enced by molecular structures. Semicrystalline polymers possess lower freevolume and slower chain relaxation compared to an amorphous counterpart.The structural differences lead to different absorption behaviors of azobenzene

�

� �

�


moieties in polymeric materials. Specifically, Wang and Weiss have reportedthat crystalline segments are subject cis isomers to considerable stress, due tothe constrainedmolecular structures [30]. Accordingly, the cis isomers in crys-talline structures have a lower activation energy and faster kinetics for cis–transisomerization compared to amorphous polymers.When azobenzene chromophores are subject to irradiation of linearly

polarized blue light (i.e., 𝜆= 442 or 445 nm), the dichroic absorption of theazobenzene molecules and rotational freedom of the azo bond result in thestatistical reorientation perpendicular to light polarization through what isknown as the trans–cis–trans reorientation (Weigert’s effect) process [31].Azobenzene chromophores oriented along the light polarization absorb thelargest quantity of light and experience trans–cis isomerization, followedby subsequent cis–trans back isomerization. Since both isomeric forms aredichroic, molecules aligned parallel to the linear polarization of the incidentlight source are more likely to absorb light, and accordingly, the moleculesin this orientation are depleted. On the other extreme, molecules that areoriented normal to the incident polarization are least likely to absorb aphoton, and this orientation is populated. The result is azobenzene moleculesstatistically rotated perpendicular to the incident light polarization.

4.6 Photomechanical Effects in Amorphous andSemicrystalline Azobenzene-Functionalized Polymers

Researchers at the Air Force Research Laboratory have recently completeda series of systematic investigations focused on elucidating the relationshipof photomechanical effects in amorphous and semicrystalline polymericmaterials to a variety of factors including crystallinity, segmental mobility, andcross-linking. Due to the facile nature of the chemistry as well as numerousreadily available precursors, these studies on polyimide materials were under-taken. Notably, the experiments in all cases were conducted 150–375K belowthe glass transition temperatures of the materials. Despite this, appreciablephotomechanical responses were observed.

4.6.1 Influence of Crystallinity on Photomechanical Responseof Polyimides

White and coworkers investigated the effect of crystallinity on the pho-toisomerization and photomechanical deflection of linear copolymersof amorphous and semicrystalline polyimides [32]. The polyimides weresynthesized by polymerizing 4-4′-diaminoazobenzene (50mol%) with eitherrigid rod-like pyromellitic dianhydride (PMDA) or the flexible dianhydride1,1,1,3,3,3-hexafluoro-2,2-bis(4-phthalic anhydride) propane (6FDA). Thechemical structures of PMDA and 6FDA are presented in Figure 4.3(a). The

�

� �

�

4.6 Photomechanical Effects in Amorphous and Semicrystalline 127

80

PMDA(a)

(b)

O

O O

O

NN

O O

O

N NN

1–xxn

O

N

F3C CF3

N N

6FDA

70

60

50

40

30

Bendin

g a

ngle

(°)

Cry

sta

llinity (

%)

20

10

0

105

100

95

90

85

800 2 4 6 8

Time (h)

Amorphous azo-PI-6FDA

Semicrystalline azo-PI-PMDA

Norm

aliz

ed a

bsorb

ance (

%)

10 12 14 16 18

0 10 20 30 40

PMDA concentration (mol%)

50 60 70 80 90 100

40

30

20

10

0

Figure 4.3 (a) Photomechanical response (bending angle, left axis) and crystallinity (rightaxis) as a function of semicrystalline PMDA concentration for a series ofazobenzene-functionalized linear polyimides. Cantilevers (5 mm × 1 mm × 0.02 mm) in theinset images are obtained after 1 h of continuous irradiation at 100 mW/cm2 intensity with𝜆= 442 nm light polarized parallel to the long axis of the cantilever. (b) Normalized peakabsorbance for the semicrystalline azo-PI-PMDA and amorphous azo-PI-6FDA against timeof continuous light (𝜆= 442 nm) irradiation at 100 mW/cm2. (Lee et al. [31]. Reproduced withthe permission of John Wiley and Sons.)

�

� �

�


employment of the bulky and flexible 6FDA dianhydride perturbed molecularpacking and resulted in a completely amorphous polyimide while the corre-sponding material prepared with PMDA exhibited about 35% crystallinity. Byintroducing increasing concentration of PMDA into the 6FDA material, thedegree of crystallinity (right axis in Figure 4.3a) increased from 0% to 35%.Upon exposure to 𝜆= 442 nm light, which was linearly polarized parallel

to the main axis of the cantilever for 1 h, azobenzene-functionalized poly-imides bent toward the light. Use of light in this wavelength regime resultsin the trans–cis–trans reorientation mechanism described earlier [31]. Ascrystallinity increases at increased PMDA concentration, the observablephotomechanical response decreased. The photomechanical strain is visuallyillustrated by the deflection of a cantilever, which is measurable by the angulardisplacement between the tip and the mounting point of the cantilever. It isimportant to note that the molar concentration of azobenzene chromophoreis identical at 50mol% in the series of polyimides. The reason for the inverselyproportional relation between the photomechanical response and crystallinityis attributable to the influence of free volume on the ability for an azobenzenechromphore to isomerize. This is evident in the absorption spectra in whichthe amorphous 6FDA polyimide absorbed had substantially larger photoiso-merization compared to the semicrystalline PMDA polyimide after 16 h oflight irradiation, as shown in Figure 4.3(b).Interestingly, the large photomechanical response of the amorphous

polyimide is observed despite a concurrent increase in Tg. In addition,accompanying the variation in crystalline content was a variation in modulus.The 6FDA polyimide exhibited a modulus of 3.80GPa, while the rigid PMDAshowed 6.12GPa modulus. Another study without azobenzene chromophoresalso reported similar crystal structures and mechanical properties for PMDAand 6FDA polyimides. Rigid PMDA had various sharp crystal peaks asobserved by X-ray with as much as a 12.2GPa modulus. Conversely, flexible6FDA had only 3.8GPa of modulus and was completely amorphous phasesevident by structureless X-ray patterns [33]. Since the dense molecularpacking of ordered crystalline regions has higher molecular stiffness aswell as bulk density, crystallinity and modulus effects are intrinsically verydifficult parameters to deconvolute in the case of semicrystalline polymers.Therefore, separation of the effect of modulus on photomechanical effectsfrom crystallinity should be undertaken in amorphous polymers with minimalcrystallinity.

4.6.2 Backbone Rigidity

An Ashby plot, a way to illustrate the trade-offs of particular types of actuatingmaterials, commonly plots Young’s modulus against the density of the materialsystem [34]. Materials usually have a positive slope (relationship) between

�

� �

�


Young’s modulus and density. As mentioned in the previous section, orderedcrystalline regions have a denser molecular packing, leading to high bulk den-sity of semicrystalline polymers versus amorphous polymers. In addition tothe simple density (molecular packing) effect, the proximity between polymersegments of crystalline region promotes stronger intermolecular secondarybonding, due to the reduction in intermolecular distance. Thus, amorphouspolymers have much smaller secondary bonding between adjacent chain seg-ments with chain misalignment. As a consequence, the degree of crystallinitystrongly affects the mechanical properties of semicrystalline polymers. Forexample, the crystallinity index increase from 0.3 to 0.6 in polyethylene causesa concurrent increase in Young’s modulus of an order of magnitude [18].Thus, amorphous polymer systems are more suitable to study the influenceof molecular rigidity on the photomechanical response of the polyimides wehave examined.In 1991, Naito et al. investigated the effect of the rigidity of the polymer main

chain on the photoisomerization of azobenzene chromophore [29]. When4-dimethylamino-4′-nitroazobenzene (DANAB) chromophore is molecularlydispersed in polyetherimide (PEI) film, trans–cis photoisomerization ofDANAB is significantly suppressed when compared to that occurring inmethylcyclohexane solution (Figure 4.4). This dependence can be observedin other polymers. At a constant DANAB concentration, smaller final cisconcentrations are measured from more rigid polyimide system than in moreflexible polycarbonate (PC) and PEI environment. This is attributed to thereduced free-volume fluctuation within rigid polymer segments than in moreflexible PC and PEI systems. While this study provides insight into the effectof molecular rigidity on photoisomerization of azobenzene chromophore, theimpact on photomechanical response had not been (or was not?) examined.Recently, the influence of the rigidity of the polymer backbone on the pho-

tomechanical response of polyimides was examined [35]. Here, the azobenzenemoiety was covalently bonded into the main chain of the linear polyimides asshown in Figure 4.5. The properties of the polyimides can be controlled byadjusting the dianhydride and/or diamine precursors. Here, the compositionof dianhydride (DA) is varied from the flexible oxy-4,4′-di(phthalic anhydride)(OPDA) to the more rigid 6FDA, 3,3′,4,4′-benzophenonetetracarboxylicdianhydride (BTDA), 3,3′,4,4′-biphenyltetracarboxylic dianhydride (BPDA),and PMDA. PMDA, BPDA, and BTDA are composed of skeletal sp2-carbons.Accordingly, the rigidity of the repeat unit can be qualitatively assessedbased on the number of single C—C bonds (sp3-carbons) between the twophenylene rings. The OPDA is more flexible than the material prepared with6FDA due to the rotational freedom of oxygen linkage in OPDA and thesteric limitations of 6FDA imparted by the sp3-swivel moiety between thephenylene rings. Thus, the backbone rigidity of the polyimides is ranked asPMDA>BPDA>BTDA> 6FDA>OPDA. Kuhn length or persistence length

�

� �

�


2

(a)

(b)

1

0

ε/10

4ε/

10

4

300 400 500

Wavelength (nm)

Wavelength (nm)

0

0

02N N

N N(CH3)2

15304560

120600

Irradiation time (s)


10203060

120180300600

400

5

4

3

2

1

0500 600 700 800

600 700

Figure 4.4 Temporal observation of UV–vis absorption spectra of 4-dimehtylamino-4′-nitroazobenzene (DANAB) in methylcyclohexane solution (a) and in polyetherimide film (b)at 230 K. (Naito et al. [29]. Reproduced with the permission of Nature Publishing Group.)

is a parameter often used in theoretical calculations to describe the conforma-tional rigidity of polymers [36, 37], where larger Kuhn/persistence length is theindicative of a more rigid polymer. Prior literature supported the correlation ofdianhydrides and molecular rigidity [33]. In addition, the molecular rigidity istranslated into bulk modulus of the polyimide filmsmeasured in tensionmode:1.34GPa, 1.37GPa, 1.66GPa, 1.99GPa, and 2.51GPa moduli for polymersprepared with OPDA, 6FDA, BTDA, BPDA, and PMDA, respectively.In this report, X-ray examination confirmed the amorphous nature of

OPDA-, 6FDA-, and BTDA-based polyimides. The polyimides with BPDAand PMDA had less than 1.5% of crystallinity, which is a negligible amount inlight of the previous study [32]. Thus, any deleterious influence of crystallinity

�

� �

�


3

(b) (c)

(a)

(i) (ii) (iii) (iv) (v)

(i′) (ii′) (iii′) (iv′) (v′)

Ab

so

rba

nce

Ab

so

rba

nce

3

2

1

0

2

1

0300 350 400 450

Wavelength (nm) Wavelength (nm)

OPDA

More flexible backbone

PMDA

PMDA

Ar

O

O

N N

O

Ar

O

O NN

O O

NN

n

OC

CF3

CF3

=

BPDABTDA 6FDA OPDA

O

500 550 600 300 350 400 450 500 550 600

Figure 4.5 (a) Polymer backbone rigidity effects on photoinduced bending response oflinear azobenzene-containing polyimides upon exposure to linearly polarized 445 nm light(E||x) at 120 mW/cm2 for 1 h. The backbone rigidity is systematically varied by usingdifferent diamines (DA) including (i) PMDA, (ii) BPDA, (iii) BTDA, (iv) 6FDA, and (v) OPDA. Theeffect of backbone rigidity on retention or relaxation of the photomechanical response isobserved by storing the cantilevers in dark for 10 days after irradiation (i′) PMDA, (ii′) BPDA,(iii′) BTDA, (iv′) 6FDA, and (v′) OPDA. Polymer backbone also significantly affected lightabsorption behaviors as evident in UV–vis absorption spectra of (b) PMDA and (c) OPDAsamples upon the irradiation of linear polarized (E||x) 445 nm light at 60 mW/cm2 for 1 h.(Wang et al. [35]. Reproduced with the permission of American Chemical Society.)

�

� �

�


on photoisomerization and the resulting photomechanical responses canbe effectively ignored. The photomechanical response was monitored in thecantilever geometry with continuous irradiation of 𝜆= 445 nm light linearlypolarized parallel to the primary axis of the cantilever (E||x) at 120mW/cm2

for 1 h. From rigid PMDA and BPDA series, small magnitude of photoinducedstrain and partial photofixing is observed. Conversely, polymers preparedfrom the flexible dianhydrides exhibited large magnitude of bending angle andcomplete recovery of the original mechanical state after dark relaxation.This study also examined the force-generating capability of the materials, by

measuring photogenerated stress. Interestingly but not surprisingly, the role ofbackbone rigidity in creating photogenerated stress is opposite of that observedin the strain response.The comparatively rigid polymers prepared with PMDAand BPDA chemistries resulted in greater photogenerated stress.Figure 4.5(b) and (c) contrast the UV–vis absorption spectra of the polymeric

materials prepared from the rigid PMDA and the flexible OPDA, respectively.Similarly to the report by Naito, the backbone rigidity strongly governs thephotoisomerization process and resultant photomechanical responses. FlexibleOPDApolyimides demonstrated larger conversion of trans-azobenzene into cisisomer.

4.7 Molecular Alignment

Entropically, polymers prefer isotropic random coil conformation. In the caseof liquid crystalline polymers, anisotropic steric repulsions (excluded volumeinteractions) of mesogenic units provide the molecular driving force fororganization [38]. Maier and Saupe proposed that the orientational directionwas imparted by the anisotropic nature of the polarizability in mesogens [39,40]. The Onsager model successfully describes lyotropic disorder–order phasetransitions, and theMaier–Saupemodel explains thermotropic disorder–orderphase transitions [41]. Nonmesogenic polymeric materials also have certainamount of anisotropy since polymeric random coil is not perfectly spherical.The anisotropy and molecular rigidity of polymer increase at lower tempera-ture, which can be represented as increased Kuhn length. If the Kuhn segmentbecomes so large during cooling process of polymer melts, spontaneousmolecular ordering by crystallization process occurs [42].When the molecular driving force for crystallization is not sufficient, poly-

mer molecules remain amorphous with random orientation. However, appli-cation of external energy can enable molecular ordering of polymers by kineticeffects, for example, subjecting to uniaxial tension. External mechanical forceabove the critical point (yield stress) of the polymer causes irreversible plas-tic deformation and neck extension, corresponding to drawing for a polymericspecimen. Semicrystalline regions break apart into smaller sized crystallites,

�

� �

�

4.7 Molecular Alignment 133

and further drawing finally generates molecular orientation for semicrystallinepolymers. Due to the unique viscoelasticity of polymers, application of stress ina timescale shorter than the polymer relaxation timescale can result inmechan-ical failure. When polymer chains become rubbery at or above the Tg of thepolymer, the chains can now reorient along the stress field and dissipate loadunder uniaxial tension. Subsequent quenching locks-in the alignment of poly-meric segments.This prestrain process is called hot drawing, detailed since the1950s [43, 44].In daily life, pulling chewing gum is a good example of the hot-drawing poly-

mer processing technique utilizing the viscoelastic properties. Often, gum baseutilizes polyvinylacetate (PVAc), with a Tg of about 30 ∘C. In the mouth, gumis subjected to heating above its Tg. Pulling gum out of the mouth with thefingers is a process of applying uniaxial tension above the polymer Tg and sub-sequent quenching to below Tg outside the mouth. This process is analogousto hot drawing of polymer, converting random coil conformation of polymerinto aligned rod-like polymer conformations. In the case of liquid-crystallinepolymers, this is somewhat similar to what is referred to as the MD orienta-tion. Along the aligned direction, the polymer chain has the largest mechanicalstiffness. We aimed to ascertain whether alignment would enable better coop-eration and result in a larger magnitude photomechanical response.Various analytical tools can be employed to measure the molecular align-

ment of polymeric materials. The 2D patterns of wide-angle X-ray diffraction(WAXD) are useful to visualize the alignment of crystalline materials. As theX-ray diffraction signal is from periodic structures, regular lattice of crystallinestructures is suitable to be analyzed by WAXD experiments. Unorientedsemicrystalline polymers have isotropic ring patterns. Conversely, X-raypatterns of oriented polymers form arcs, and the degree of orientation canbe quantified from the information of azimuthal width of the arcs using thefollowing equation [45].

S =3⟨cos2𝜙⟩ − 1

2, (4.9)

where S is Hermans’ orientational order parameter and 𝜙 the azimuthal angle.The average cosine square of 𝜙 can be calculated by performing integration ofthe I(𝜙) data versus 𝜙 using the following equation [45].

⟨cos2𝜙⟩ =∫ 2π0 I(𝜙)cos2(𝜙) sin(𝜙)d𝜙

∫ 2π0 I(𝜙) sin(𝜙)d𝜙

. (4.10)

This value is indicative of the crystallinity of a given material. Various spec-troscopy techniques can be employed to further assess the orientation of thematerial. UV–vis measurement can be employed to assess the orientation ofchromophores by using linearly polarized input probe light. From themeasuredUV–vis absorbance spectra data of a given material at different polarization,

�

� �

�


polar plot can be established, visually illustrating the molecular orientation ofphotoresponsive moiety [46–48].

S =A∥ − A⟂

A∥ + A⟂. (4.11)

Here,A|| is the absorption parallel to alignment direction andA⟂ the absorptionperpendicular to alignment direction.This anisotropic light absorption is calleddichroism, and the dichroic ratio R is equal to the ratio of A|| to A⟂ [49]. Hence,the order parameter S can be rewritten as [49, 50]

S = (R − 1)(R + 2)

. (4.12)

The orientation of certain functional groups can also be probed with Ramanor Fourier transform infrared (FT-IR) spectroscopy. For example, the cyanobond at around 2230 cm−1 has been extensively examined in liquid-crystallinesystems [51, 52]. Equation 4.12 can be adapted to calculate the order parameterby FT-IR [53]. Employing Raman, FT-IR, X-ray, or other techniques can allowthe deconvolution of each aspect to better elucidate the underlying orientationand interactions at the molecular and macromolecular levels [52].An FT-IR polar plot is shown in Figure 4.6 [51]. Here, a cyano group

is included in a uniaxially aligned polymer. Initially, the film has a brightcolor at 45∘ offset from crossed polarizers due to the orientation. In situFT-IR measurement with mechanical stretching demonstrates the molecularreorientation along the axis of applied stress. From the draw ratio (L/L0),𝜆= 1.3, necked regions developed higher local strain compared to the non-necked region. Hence, a location-specific reorientation rate is expected dueto localized strain. As the necked regions propagate, complete and uniformreorientation is achieved at 𝜆= 2.Necking-induced location-specific results are summarized in Figure 4.6(b),

by plotting lateral deformation in y-direction (𝜆y) as a function of drawratio (𝜆x) for different locations on the film. The mechanical instability inthe necked regions resulted in faster molecular reorientation compared tothe nonnecked region. The arrows indicate the point where the completedirector-reorientation occurred. Regardless of the location of the polymericfilm, strong alignment is completed at about 70% strain from original gaugelength.Programming the molecular alignment allows anisotropic mechanical prop-

erties and actuation rather than three-dimensional expansion or contraction.While alignment of liquid-crystalline materials can be manipulated easily viaself-assembly, following alignment direction of command surface, this tech-nique is not applicable for most common polymers. Instead, aforementionedhot-drawing process can be employed to prestrain molecular structures toachieve directed stimuli-responsive actuation.

�

� �

�


(b)

(a)

Natural light

Neck formation

Initial director

270°

0

90°

180°

270°

0

90°

180°

270°

0

90°

180°

270°

0

90°

180°

1 mm

Crossed polarizersA P

λ = 1

λ = 1.3

λ = 1.5

λ = 2

Stretching direction

A

1

1

0.9

λ y

0.8

0.7

0.6

0.51.2 1.4 1.6

λx

1.8 2 2.2

A

X

MNE-86

CE

B C D E

Figure 4.6 (a) Optical microphotographs of hot-drawn polymers at various draw ratiosunder crossed polarizer (upper) and natural light (bottom) conditions. Polar plots ofabsorbance regarding the stretching vibration of the terminal cyano group in polymers atvarious draw ratios. (b) Lateral deformation in y-direction (𝜆y) at positions A, C, and E as afunction of overall stretch (𝜆x). The arrows indicate the point where the director-rotationcompletes. The point for the completion of director rotation for polymers is also shown.(Higaki et al. [51]. Reproduced with the permission of American Chemical Society.)

�

� �

�


In Figure 4.7(a), a UV–vis polar plot is constructed for cross-linked amor-phous polyimides composed of 20mol% azobenzene cross-linker using theabsorbance information at 𝜆= 355 nm (trans-azobenzene peak) as a functionof the polarization of the probe light. The as-prepared polyimide film effec-tively has a uniform absorbance at any direction before prestraining. Thispolyimide film is stretched by 70% from its initial length at 275 ∘C, whichis 50 ∘C higher than its Tg. At 275 ∘C, the sample is equilibrated for 5minand manually hot-drawn by tweezers followed by subsequent quenching toambient condition by air cooling. After the hot-drawing process, a dichroicabsorbance of the azobenzene chromophores was observed due to anisotropicmolecular alignment along the prestrained axis.The influence of drawing-induced alignment of the polyimide (and the

azobenzene chromophores) on the photomechanical response of the materialswas examined by bending observed in cantilevers with the dimensions of6mm× 1mm× 20 μm for length, width, and thickness, respectively. Thesamples were exposed with 𝜆= 442 nm linearly polarized light over 1 h at80mW/cm2. Photoinduced bending to 442 nm light polarized either paral-lel (E ∥ x) or perpendicular (E ⟂ x) to the primary axis of cantilevers wasexamined in the materials subjected to various prestrain values (0–70%).Up to 25% of prestrain, the polyimide films exhibit polarization-controlledbidirectional bending evident in the inset of Figure 4.7(b). The bidirectionalbending is indicative that the sign of strain can be optically controlled bylight, from contractile to expansive. With prestrain values exceeding 25%, onlyunidirectional bending was observed.Themolecular alignment by hot drawingnot only enhanced the strain response upon light irradiation but also increasedthe photogenerated stress response from 0.6 to 1.2MPa when the material wassubjected to 70% prestrain. The concurrent enhancement of photogeneratedstress and strain response by hot drawing is beneficial to improve actuatorperformance.

Figure 4.7 (a) Chemical structure of cross-linked amorphous polyimides composed of20 mol% azobenzene cross-linker. The polar plot is prepared for the polyimide sample withthe UV–vis spectrometer absorbance information at 𝜆= 355 nm (trans-azobenzene peak) asa function of the polarization of the probe light. Before prestraining, the as-preparedpolyimide film has uniform absorbance (•). After hot drawing to 70% prestrain (⚬), a dichroicabsorbance of the azobenzene chromophores was measured due to anisotropic molecularalignment along the strained axis. (b) Photoinduced bending angle ofazobenzene-containing polyimides measured in cantilever geometry (6 mm × 1 mm ×20 μm for length, width, and thickness, respectively) after 1 h of irradiation to 80 mW/cm2 oflinearly polarized 𝜆= 442 nm irradiation polarized both parallel (E||x) and perpendicular(E ⟂ x) to the primary axis of cantilevers subjected to various prestrain values (0–70%). (Leeet al. [32]. Reproduced with the permission of John Wiley and Sons.)

�

� �

�


O

N

N N N

O

O O

O O

O3

F3C CF3

n,m,l

150

120

90

60

30

0

330

300

270

240

210

180600

600

600

600

500

500

500

500

400

400

400

400

0 10 20 30 40

Strain (%)

50 60 70

100

80

60

40

20

0

−20

−40

−60

Bendin

g a

ngle

(°) E//x

E⊥x

(a)

(b)

�

� �

�


4.8 Annealing and Aging

The photochromic response to light irradiation in polymeric materials can bestrongly inhibited by free volume. As discussed earlier, the free volume of amaterial is a function of temperature. For example, the temperature depen-dence of the thermal back reaction in the rubbery state can be described bythe WLF theory [20].Segmental mobility of polymers is also strongly influenced by temperature.

For example, the diffusion coefficient of polymers is defined by

D =kBT

6π𝜂[3(MW)4𝜋N𝜌

]1∕3 , (4.13)

where D is the diffusion coefficient, kB Boltzmann’s constant, T the tempera-ture in kelvin, 𝜂 the viscosity of sustaining fluid, MW the molecular weight ofpolymer, N Avogadro’s number, and 𝜌 the density.This equation is derived from the Stokes–Einstein relation by substituting

hydrodynamic radius with the term related to the molecular weight of poly-mers. As evident from the equation, the diffusivity is not a strong function ofpolymer molecular weight. Instead, temperature has a much greater impacton molecular motion of polymers. Naito et al. demonstrated that the photoi-somerization process of azobenzene moiety dispersed in polymer is reducedat lower temperature due to the suppression of local mobility [29]. Whilethe polymer mobility is strongly related to polymer Tg, the correlation ofsegmental mobility with absolute temperature implies that Brownian motionof polymer is not perfectly restricted unless the temperature is maintained at 0K. Instead, supercooled glasses are subjected to slow but gradual physical agingfor extended periods of time even below the polymer Tg because amorphouspolymers are not in their equilibrium states [54]. Physical aging includesvarious mechanisms such as annealing, enthalpy relaxation, and volumetricrelaxation. Through physical aging, polymeric materials can be prepared withreduced free volume and decreased molecular configurational energy. Thisstructural modification inherently changes the energy landscape within whichany photoisomerization were to occur and thereafter affect photomechanicalresponses [55].Physical aging reduces free volume, and the polymer becomes stiffer. Both

parameters are already discussed in the earlier sections, and decreased pho-toisomerization and photomechanical actuation are expected. Vaia, White,and their coworkers investigated the effects of physical aging on photoisomer-ization and photomechanical response of glassy, azobenzene-functionalizedpolyimides [55]. The polyimide is again cross-linked amorphous polyimide

�

� �

�

4.8 Annealing and Aging 139

composed of 20mol% of trifunctional azobenzene cross-linker with identicalchemical structure shown in Figure 4.7. Initially, the polyimide films areheated to 270 ∘C (Tg + 50 ∘C) for 30min under vacuum environment in orderto erase the previous thermal history. Then, the polyimides were subjectedto two distinct physical aging processes – rapid quenching (RQ) and slowquenching (SQ) – to control the thermal history of glass. The RQ samplewas prepared by immediately immersing the material into liquid nitrogen(−196 ∘C) at the elevated temperature of 270 ∘C. This process is expected tokinetically trap the material in thermodynamic equilibrium attained aboveTg at this elevated temperature. The result is that the constrained polymerchains now have loosened molecular packing and increased free volume asshown in Figure 4.8(a). On the other hand, the SQ sample is slowly cooled inair at a cooling rate of 1 ∘C/min. The SQ procedure allows longer aging timeof polymers to approach the equilibrium expected for a glassy material. Theenergy landscape of an SQ sample is denser molecular packing and reducedfree volume.To confirm the influence of thermal history (physical aging) on the photome-

chanical responses, the samematerial subjected to these two distinct processeswas examined.The effect of the thermal processing on the azobenzene absorp-tion in the materials was investigated by UV–vis spectroscopy before and afterlight illumination as shown in Figure 4.8(b). The azobenzene-functionalizedpolyimide films were subjected to 100mW/cm2 of linearly polarized 442 nmgenerated by a helium cadmium (HeCd) laser. In the polar plot, the 0∘ axiscorresponds to the long direction of the polyimide cantilever. Before thelight irradiation, isotropic absorption behaviors are observed from boththe “as-prepared” and the RQ and SQ samples. These randomly orientedisotropic samples can be reoriented to the perpendicular direction of the lightpolarization via trans–cis–trans reorientations. Obvious anisotropy isdeveloped after the light irradiation for both RQ and SQ cases. However,significantly reduced amplitude of anisotropic absorption is measured fromSQ samples, which is indicative of reduced probability of photoisomerizationdue to dense local molecular packing. Conversely, larger free volume of looselypacked RQ samples results in a larger distortion of the population of reorientedazobenzene evident in the greater dichroism from the UV–vis polar plot data.The translation of the nascent photochemical response into photomechanicalresponses is demonstrated in Figure 4.8(c).After light illumination and removal, thematerial undergoes a relaxationwell

above 0K, where random thermal Brownianmotion takes place. If this randomthermal process is sufficient enough, dark relaxation can drive full recovery ofinitial flat cantilever shape.The environment for the azobenzene chromophorein the SQ sample is closer to perfect glass. Accordingly, the SQ sample has a

�

� �

�


Rapidly quenched (RQ) Slowly quenched (SQ)

Po

ten

tia

l e

ne

rgy

Configuration space

150

180

210

240

270

300

330

0

30

60

90

1201.0

1.0

1.0

1.0

0.9

0.80.8

0.8

0.9

0.9

0.9

150

180

210

240

270

300

330

0

30

60

90

1201.0

1.0

1.0

1.0

0.9

0.80.8

0.8

0.9

0.9

0.9

RQ SQ

(a)

(b)

Figure 4.8 (a) Schematic description of the potential energy landscape for glasses withdifferent configuration spaces. The left image describes the rapidly quenched (RQ) samplehaving larger free volume. The right image illustrates the slowly quenched (SQ) sample withdense environment. (b) Polar plots of the normalized absorption value at 𝜆= 355 nm for thephysically aged azobenzene-containing polyimides (left, RQ; right, SQ) at different lightirradiation conditions: (•) Before irradiation, ( ) after irradiation with linearly polarized442 nm light polarized along the y-direction (90−270∘ axis), or ( ) along the x-direction(0−180∘ axis), and ( ) 4 days after irradiation with linearly polarized 442 nm light polarizedalong the x-direction (0−180∘ axis). (c) Photomechanical bending is monitored with acantilever geometry (5 mm × 1 mm × 20 μm) upon 100 mW/cm2 intensity of 𝜆= 442 nmlight linearly polarized along the x-direction. Effect of different physical aging conditions iscontrasted by monitoring RQ (i−iii) and SQ (iv−vi) samples. The (i, iv) inset images showcantilevers before light irradiation and after 2 h of irradiation with polarized 442 nm light(parallel to the long axis of the cantilever (E||x)) shown in (ii, v). The (iii, vi) images arecaptured after 72 h of dark relaxation after the light irradiation. (d) Summarizedphotomechanical bending response of azobenzene-functionalized polyimide cantilevers forRQ ( ) and SQ ( ) during 2 h of continuous irradiation 100 mW/cm2 intensity of 𝜆= 442 nmlight linearly polarized to x-axis followed by subsequent dark relaxation. (Lee et al. [55].Reproduced with the permission of American Chemical Society.) (See color plate section forthe color representation of this figure.)

�

� �

�

4.8 Annealing and Aging 141

SQ

RQ

00

5

10

15

20

25

30

35

1 2 3 4 5 110 120

Time (h)On Off

Be

nd

ing

an

gle

(°)

RQ

SQ

E||x

E||x OFF

OFF

72 h

72 h

(iv) (v) (vi)

(i) (ii) (iii)

(c)

(d)

Figure 4.8 (Continued)

reduced amount of azobenzene chromophores reoriented and, because of theincreased likelihood of molecular interactions due to the constrained space,achieved full recovery of isotropic absorption behaviors as well as flatteningof the deformed cantilever over time. Unlike the SQ sample, the photoactivechromophore in the RQ sample maintained the photoinduced changes inabsorptive behavior even after removal of the light illumination.This is evidentin the measured dichroism from the UV–vis polar plot (Figure 4.8b) as well asthe retention of the deformed cantilever (Figure 4.8c and d). It is clear fromthis illustration that the process history is critical not only to produce reliableresults but also as a method to enhance the photomechanical output observedin these materials by enabling greater photoresponsivity at the molecularlevel.

�

� �

�


4.9 Sub-Tg Segmental Mobility

The glassy-to-rubbery transition temperature, Tg, is associated with 3–5 Kuhnsegments of the global main chain α-relaxation process and also called T

𝛼. In

addition to the α-relaxation process, polymeric materials can also exhibit anumber of comparatively less-studied sub-α transitions including β-, γ-, andδ-relaxations.The sub-α transitions are associated with local segmental mobil-ity, and the β-transition typically originates from whole side chain, stereoiso-mer chain, or local mode relaxation of a polymer on a length scale comparableto the Kuhn segment. The γ-transition in glassy polymers is conditioned by alocal bond (i.e., methyl group relaxation) or side-chain mobility (bending andstretching) at length scale smaller than Kuhn segment, and δ-transition is fromeven smaller local relaxation mode. Due to the generic cooperative relationsbetween α- and β-transitions, the T

𝛽(K)/T

𝛼(K) ratio for polymers ranges from

0.6 to 0.9 as well known from Boyer’s rule, and approximately, it is often scaledas T

𝛽(K) ∼ 0.75 T

𝛼(K) for many polymers unless large kinetic effects play a

role [56].Photochromic process dramatically varies at transition temperatures includ-

ing sub-Tg transitions. At each T𝛼and sub-Tg transition, photochromic

processes have breaks of Arrhenius plots, indicating the importance ofsub-Tg transitions in photochemistry [57]. In the temperature range ofT𝛽<T <T

𝛼, nonexponential decay of photoluminescence was reported for

benzophenone chromophore dispersed in various polymer matrices such asPMMA, poly(isopropyl methacrylate) (PIPMA), and poly(methyl acrylate)(PMA) [58]. Here, the sub-Tg segmental mobility is based on the rotationof the chromophore molecules at a few monomer unit scales. These studiesimply the possible impact of sub-Tg relaxation behavior of polymers onphotomechanical processes.In order to more directly elucidate the role of local segmental mobility in

photomechanical responses realizable in these materials, Wie, White, andcoworkers prepared a model system. To select this system, we consideredthat densely packed crystalline structures have lower free volume and lightabsorbance compared to amorphous polymers. As it is very difficult to haveexactly the same degree of crystallinity in different materials, clearly, amor-phousmaterials allow formore direct comparison. However, even in this subsetof materials, as we have discussed earlier, a number of factors can influencethe observed response including the rigidity of the backbone. Hence, ideally astudy would examine materials with identical chemical compositions but widevariation in the β-transition. Toward this end, Wie, White, and coworkersemployed a series of positional isomers to prepare azobenzene-functionalizedamorphous polyimides [59]. To synthesize amorphous polyimides, 6FDA-typechemistry is adopted, and various other parameters are controlled includingazobenzene concentrations, film geometry, and light intensity.

�

� �

�

4.9 Sub-Tg Segmental Mobility 143

Positional isomers have the identical chemical formula and carbon skeleton,but the locations of functional groups are different. Within the report, posi-tional isomers of phenyl ethers are included with para and meta connectivity(see Figure 4.9a) within the main chain of both linear and cross-linked poly-imides. The para phenyl ether is capable of rotational motion at low torsionalpotential, providing substantial β-transition due to the intrasegmental mobil-ity below T

𝛼[60]. By computer simulation, Toshchevikov and co-workers

reported that chemically attached azobenzene chromophores result in largermagnitude of photomechanical deformation in comparison with azobenzenechromophores dispersed in structures due to mechanical coupling betweenchemical network and chromophores [61]. The crosslink density effect willbe discussed in the next section. Conversely, the intrasegmental mobility byrotational motion is hindered in the meta phenyl ether due to large rotationalenergy barriers and allows negligible β-transition. As shown in Figure 4.9(b),a large amount of sub-Tg β-transition for freely rotating para phenyl ether isconfirmed by dynamic mechanical analyzer (DMA) from the peak of loss mod-ulus below the α-transition. In the case of meta isomer, significantly smallermagnitude in the β-transition is measured, due to the restricted rotationalfreedom from the meta isomer. Despite the contrasted segmental mobility, theUV–vis absorption spectra of para and meta isomers are not distinguishablefor both as-prepared and light-irradiated samples. Although the photoisomer-ization capabilities of para and meta isomers are very similar, a considerablylarger photomechanical response is observed in the material composed withthe para phenyl ether moiety (Figure 4.9c). This unique experimental evidencedemonstrates that larger photomechanical deformation is not necessarily amere result of larger photoisomerization process. Instead, photomechanicalactuation is often substantially affected by the local dynamics of the physicalproperties of the material, in this case, the segmental mobility.As discussed earlier, most variables are successfully controlled by using

positional isomers. However, it is noteworthy that the backbone of polyimidescomposed with para isomers are slightly stiffer compared to meta isomers(2.38GPa vs 2.27GPa in linear polyimides and 2.77GPa vs 2.40GPa forcross-linked polyimides) determined from storage modulus value at 25 ∘C.This feature allows interesting advantage as an actuator. In general, actuatorstrade-off force and displacement. For example, stiffer actuators generate largestress (force), but strain (displacement) response is limited. Similarly, softactuators have large strain output but suffer from restricted force. The paraisomer of the polyimides can achieve concomitant enhancement of photo-generated stress and strain originated from larger stiffness and β-transition,respectively. This study suggests the critical role of sub-Tg segmental mobil-ity and importance of molecular engineering to design photomechanicalactuators.

�

� �

�


(a)

175

Freely rotate No rotational freedom

Large β-transition

Small β-transition

Temperature (°C)

150

125

100

E′ (

MP

a)

Bendin

g a

ngle

(°)

75

50

25

0

0

0 15 30 45 60 75

Time (min)

90 105 120 4200 4400

15

30

45

60

75

90

100–100 200 300 4000

(c)

(b)

O OO O

OFF

(ii)

(i)

(i′)

(ii′)

�

� �

�

4.10 Cross-Link Density 145

4.10 Cross-Link Density

Polymerization reactions with difunctional monomers produce linear poly-mers. Linear polymers are reprocessible by dissolving them in solvents or bythermal processing techniques. In contrast, materials with n-functional groups(when n > 2) produces cross-linked polymers with networked structures.The cross-linking makes these polymers insoluble in solvents. However,cross-linked polymers do swell when exposed to solvents, and the swellingratio can be utilized to estimate the cross-link density of polymers. Cross-linkdensity of polymers can also be calculated using Flory’s rubber elasticity theoryas follows [12],

𝜈e =E′high

(3RThigh), (4.14)

where 𝜈e is the cross-link density, E′high is the storage modulus at Thigh. Thigh is

the temperature in the rubbery plateau region.In the case of cross-linked polymers, the cross-linking process during

polymerization develops shrinkage stress that can leave residual stress withinthe polymer. Polymerization-induced shrinkage occurs by reducing the inter-molecular distance between the monomers and converts into covalent bondsin the polymer state. When shrinkage stress causes dimensional stability issuesfor applications, ring-opening polymerization can be employed in order tominimize the volumetric shrinkage during the cross-linking. Before and afterthe ring-opening polymerization, the number and types of chemical bondsare identical. Recently, thiol–ene or thiol–yne polymerizations have also beenproposed toward the lower shrinkage stress by delaying the gelation process.Thiol–ene is suitable only for low cross-link density polymers with low Tgand modulus. When enhanced thermomechanical properties are desirable,thiolvinyl–yne polymerizations can be introduced [62, 63].In consideration of the influence of cross-link density on other parameters,

the increase in cross-linking density can enhance Tg by decrease of confor-mational entropy. It also causes increase in mechanical stiffness and restrictsthe thermal expansion of polymers. As a result, cross-linking limits the

Figure 4.9 (a) Chemical structure of freely rotating para isomer (left) and nonrotating metaisomer (right) of phenyl ether linkages within azobenzene-functionalized polyimides.(b) Loss modulus plotted against temperature for para (—) and meta (---) isomers.(c) Time-resolved monitoring of photomechanical bending angle of azobenzene-functiona-lized polyimide cantilever consisted of para (•) and meta (⚬) isomers. Linearly polarized445 nm light is irradiated for 60 min followed by 72 h of dark relaxation. Inset imagesindicate bending angle at 60 min time mark for continuous light irradiation on (i) meta and(ii) para isomers. Images marked with (′) are captured after 72 h relaxation at dark. (Wie et al.[59]. Reproduced with the permission of American Chemical Society)

�

� �

�


thermomechanical strain response at the same thermal energy input [52]. Inphotomechanical material systems, the increase in cross-link density results inmaterials with enhanced rigidity. As has been discussed, this reduces the overallmechanical response that is realized [21]. Toshchevikov and coworkers pro-vided theoretical calculation using a regular cubic network model constructedfrom freely jointed rod-like segments (Kuhn segments). Photomechanicaldeformation is turned out to be very sensitive to chemical structures, and tightcross-linking tends to decrease photomechanical strain response at fixed lightintensity [37].One notable means to hybridize these materials systems to maintain the

mechanical integrity associated with cross-linked polymers but the pho-tomechanical response of less-rigid systems has been proposed by Ikeda.An “interpenetrating polymer network” (IPN) is a material system thathas two or more cross-linked polymers. IPNs are commonly composed ofelastomers and hydrogels in order to obtain desirable mechanical propertiesbut retain the stimuli-responsive nature of these material classes. Recently,Ikeda and coworkers reported on the synthesis of IPNs composed of poly(alkylmethacrylate)s (PAMAs) and azobenzene liquid-crystalline polymers [64].Thefirst network with azobenzene moiety is formed by photopolymerization of amonofunctional monomer (A6AB6) and a bifunctional cross-linker (DA6AB)in the presence of LC solvent (1BZ6) as shown in Figure 4.10(a). Subsequently,LC solvent is removed, giving porous film of the first polymer network. Thisporous film is immersed in a mixture of alkyl methacrylate-based monomers(MMA/BMA/DDMA) and a cross-linker (EGDMA). The second polymernetwork is again obtained via photopolymerization where the carbon numberin alkyl methacrylate-based monomers varied from 1 to 4 and 12.The smallestaliphatic carbon number, MMA, is expected to have the largest cross-linkdensity. As expected, the MMA sample resulted in the largest mechanicalproperties (Figure 4.10b) and the smallest light absorption due to reduced freevolume. With the conjunction of increased stiffness and reduced light absorp-tion behavior, the smallest photomechanical strain response (Figure 4.10c) isobserved from theMMA sample, again demonstrating the influence of variouspolymer properties on the underlying principles of photoisomerization andphotomechanical processes.

4.11 Concluding Remarks

As discussed within this chapter, various parameters of polymer physics areintercorrelated and are responsible for photoisomerization and photomechani-cal energy transduction processes.The total isolation of interplay from all thosefactors is extremely difficult, and often, convolution of the various factors isinevitable to some extent. However, the consideration of this issue is critical toprecisely analyze the photomechanical behaviors, and the effort to build up a

�

� �

�

4.11 Concluding Remarks 147

(a) O

A6AB6

O O

ON

N44

(b)

(c)

OO

O

OO

O

OO

5

O

DA6AB

1BZ6

EGDMA

O O

ON

N44

O

O

MMA (R = CH3)

BMA (R = C4H9)

DDMA (R = C12H25)

0

0

10

20

30

Str

ess (

MP

a)

40

50

0.05 0.10

Strain

120

90

60

30

00 60 120 180


Be

nd

ing

an

gle

(°)

240 300

PAzo: Control (No IPN)PAzoTP: Porous PAzo (No IPN)IPNs with:

PAzo/PMMA

PAzo/PBMA

PAzo/PDDMA

PAzo

PAzoTP

O

O

MMA (R = CH3)

BMA (R = C4H9)

DDMA (R = C12H25)

R

O

OR

PAzo/PMMA

PAzo/PBMA

PAzo/PDDMA

PAzo

PAzoTP bendingangle

Figure 4.10 (a) Chemical structures of compounds used in this study. (b) Stress–straincurves of 20-μm-thick films upon stretching along the director axis of azobenzene moieties.(c) Photomechanical bending behavior of films (3 mm × 1 mm × 16 μm) upon irradiationwith UV (10 mW/cm2) and visible (40 mW/cm2) light. (Ube et al. [64]. Reproduced with thepermission of Royal Society of Chemistry.)

�

� �

�


model system is very important to establish solid basis to understand funda-mentals of the structure–property relationships governing the photomechani-cal behaviors of polymeric materials.

References

1 Hartley, G.S. (1937) The cis-form of azobenzene. Nature, 140 (3537), 281.2 Lovrien, R. (1967) The photoviscosity effect. Proceedings of the National

academy of Sciences of the United States of America, 57, 236–242.3 Agolini, F. and Gay, F.P. (1970) Synthesis and properties of azoaromaticpolymers. Macromolecules, 3 (3), 349–351.

4 Paik, C.S. and Morawetz, H. (1972) Photochemical and thermal isomer-ization of azoaromatic residues in the side chains and the backbone ofpolymers in Bulk132. Macromolecules, 5 (2), 171–177.

5 Smets, G. (1972) Photochromic behaviour of polymeric systems and relatedphenomena. Pure and Applied Chemistry, 30 (1), 1–24.

6 Eisenbach, C.D. (1980) Isomerization of aromatic azo chromophores inpoly(ethyl acrylate) networks and photomechanical effect. Polymer (Guildf ),21, 1175–1179.

7 Kim, D.Y., Tripathy, S.K., Li, L., and Kumar, J. (1995) Laser-induced holo-graphic surface relief gratings on nonlinear optical polymer films. AppliedPhysics Letters, 66 (10), 1166.

8 Rochon, P., Batalla, E., and Natansohn, A. (1995) Optically induced surfacegratings on azoaromatic polymer films. Applied Physics Letters, 66 (2), 136.

9 Finkelmann, H., Nishikawa, E., Pereira, G.G., and Warner, M. (2001) A newopto-mechanical effect in solids. Physical Review Letters, 87 (1), 15501.

10 Yu, Y., Nakano, M., and Ikeda, T. (2003) Photomechanics: directed bendingof a polymer film by light. Nature, 425 (6954), 145.

11 Sperling, L.H. (2006) Introduction to Physical Polymer Science, 4th edn, JohnWiley & Sons, Inc, Hoboken.

12 Flory, P.J. (1953) Principles of Polymer Chemistry, Cornell University Press,Ithaca, NY.

13 Flory, P.J. (1989) Statistical Mechanics of Chain Molecules, Hanser, NewYork.

14 Rubinstein, M. and Colby, R.H. (2003) Polymer Physics, Oxford UniversityPress, Oxford.

15 Mofrad, M.R.K. and Kamm, R.D. (2011) Cytoskeletal Mechanics Models andMeasurements in Cell Mechanics, Cambridge University Press, Cambridge.

16 Naito, T., Horie, K., and Mita, I. (1993) Photochemistry in polymer solids:12. Effects of main-chain structures and formation of hydrogen bondson photoisomerization of azobenzene in various polymer films. Polymer(Guildf ), 34 (19), 4140–4145.

�

� �

�

References 149

17 Callister, W.D. Jr., and Rethwisch, D.G. (2010) Materials Science and Engi-neering: An Introduction, 8th edn, John Wiley & Sons, Inc, Hoboken.

18 Callister, W.D. Jr., and Rethwisch, D.G. (2012) Fundamentals of MaterialsScience and Engineering: An Integrated Approach, John Wiley & Sons, Inc,Hoboken.

19 Williams, M.L., Landel, R.F., and Ferry, J.D. (1955) The temperaturedependence of relaxation mechanisms in amorphous polymers and otherglass-forming liquids. Journal of the American Chemical Society, 77 (14),3701–3707.

20 Eisenbach, C.D. (1979) Cis-trans isomerization of aromatic azo compoundsbuilt in the polyester segment of poly(ester urethanes). Polymer Bulletin, 1,517–522.

21 Lee, K.M., Koerner, H., Vaia, R.A. et al. (2010) Relationship betweenthe photomechanical response and the thermomechanical properties ofazobenzene liquid crystalline polymer networks. Macromolecules, 43 (19),8185–8190.

22 Beaman, R.G. (1952) Relation between (apparent) second-order transitiontemperature and melting point. Journal of Polymer Science, 9 (5), 470–472.

23 Boyer, R.F. (1954) Relationship of first- to second-order transition tempera-tures for crystalline high polymers. Journal of Applied Physics, 25 (7), 825.

24 Kao, K. (2004) Dielectric Phenomena in Solids, Academic Press, San Diego.25 Bouguer, P. (1729) Essai d’optique sur la gradation de la lumière, Claude

Jombert, Paris, pp. 16–22.26 Lambert, J.H. (1760) Photometria sive de mensura et gradibus luminis, colo-

rum et umbrae, Eberhardt Klett, Germany, p. 391.27 Beer, A. (1852) Bestimmung der Absorption des rothen Lichts in farbigen

Flüssigkeiten. Annalen der Physik und Chemie, 86, 78–88.28 Serra, F. and Terentjev, E.M. (2008) Nonlinear dynamics of absorption and

photobleaching of dyes. Journal of Chemical Physics, 128 (22), 224510.29 Naito, T., Horie, K., and Mita, I. (1991) The effect of polymer rigidity

on photoisomerization of 4-dimethylamino-4′-nitroazobenzene. PolymerJournal, 23 (6), 809–813.

30 Wang, C. and Weiss, R.G. (2003) Thermal cis → trans isomerization ofcovalently attached azobenzene groups in undrawn and drawn polyethylenefilms. Characterization and comparisons of occupied sites. Macromolecules,36 (11), 3833–3840.

31 Lee, K.M., Tabiryan, N.V., Bunning, T.J., and White, T.J. (2012) Pho-tomechanical mechanism and structure–property considerations in thegeneration of photomechanical work in glassy, azobenzene liquid crystalpolymer networks. Journal of Materials Chemistry, 22 (2), 691–698.

32 Lee, K.M., Wang, D.H., Koerner, H. et al. (2012) Enhancement of pho-togenerated mechanical force in azobenzene-functionalized polyimides.Angewandte Chemie International Edition, 51 (17), 4117–4121.

�

� �

�


33 Ree, M., Kim, K., Woo, S.H., and Chang, H. (1997) Structure, chain orienta-tion, and properties in thin films of aromatic polyimides with various chainrigidities. Journal of Applied Physics, 81 (2), 698.

34 Ashby, M.F. (2010) Materials Selection in Mechanical Design, 4th edn,Butterworth-Heinemann, Oxford.

35 Wang, D.H., Wie, J.J., Lee, K.M. et al. (2014) Impact of backbone rigid-ity on the photomechanical response of glassy, azobenzene-functionalizedpolyimides. Macromolecules, 47 (2), 659–667.

36 Toshchevikov, V.P., Saphiannikova, M., and Heinrich, G. (2012) Theory oflight-induced deformation of azobenzene elastomers: influence of networkstructure. Journal of Chemical Physics, 137 (2), 24903.

37 Toshchevikov, V., Saphiannikova, M., and Heinrich, G. (2012) Light-induceddeformation of azobenzene elastomers: a regular cubic network model.Journal of Physical Chemistry B, 116 (3), 913–924.

38 Onsager, L. (1949) The effects of shape on the interaction of colloidal parti-cles. Annals of the New York Academy of Sciences, 51 (4), 627–659.

39 Maier, W. and Saupe, A. (1959) Eine einfache molekular-statistische Theorieder nematischen kristallinflüssigen Phase. Teil l1. Zeitschrift für Natur-forschung A, 14 (10), 882–889.

40 Maier, W. and Saupe, A. (1960) Eine einfache molekular-statistische Theorieder nematischen kristallinflüssigen Phase. Teil II. Zeitschrift für Natur-forschung A, 15 (4), 287–292.

41 Hu, W. and Frenkel, D. (2005) Polymer crystallization driven by anisotropicinteractions, in Interphases and Mesophases in Polymer Crystallization III(ed. A. Giuseppe), Springer-Verlag, Berlin/Heidelberg, pp. 1–35.

42 Flory, P.J. (1956) Statistical thermodynamics of semi-flexible chainmolecules. Proceedings of the Royal Society A: Mathematical, Physicaland Engineering Science, 234 (1196), 60–73.

43 Cleereman, K.J., Karam, H.J., and Williams, J.L. (1953) Modern Plastic, Bre-skin Publications Inc., New York, NY.

44 Picot, C., Duplessix, R., Decker, D. et al. (1977) Neutron scattering by uni-axially hot stretched polystyrene samples. Macromolecules, 10 (2), 436–442.

45 Koerner, H., Luo, Y., Li, X. et al. (2003) Structural studies ofextension-induced mesophase formation in poly(diethylsiloxane) elastomers:in situ synchrotron WAXS and SAXS. Macromolecules, 36, 1975–1981.

46 Park, M.-K. and Advincula, R.C. (2002) In-plane photoalignment of liq-uid crystals by azobenzene−polyelectrolyte layer-by-layer ultrathin films.Langmuir, 18 (11), 4532–4535.

47 Bobrovsky, A., Boiko, N., Shibaev, V., and Stumpe, J. (2004) Com-parative study of photoorientation phenomena in photosensitiveazobenzene-containing homopolymers and copolymers. Journal of Pho-tochemistry and Photobiology A: Chemistry, 163 (3), 347–358.

�

� �

�

References 151

48 Bobrovsky, A., Shibaev, V., Cigl, M. et al. (2014) Photochromic LC–polymercomposites containing azobenzene chromophores with thermally stableZ-isomers. Journal of Materials Chemistry C, 2 (22), 4482.

49 Neff, V.D. (1974) Liquid Crystals and Plastic Crystals, Ellis Horwood,Chichester.

50 Noel, C., Laupretre, F., Friedrich, C. et al. (1984) Synthesis and meso-morphic properties of a new thermotropic liquid-crystalline “backbone”copolyester. Polymer (Guildf ), 25 (6), 808–814.

51 Higaki, H., Takigawa, T., and Urayama, K. (2013) Nonuniform and uniformdeformations of stretched nematic elastomers. Macromolecules, 46 (13),5223–5231.

52 Wie, J.J., Lee, K.M., Ware, T.H., and White, T.J. (2015) Twists and turnsin glassy, liquid crystalline polymer networks. Macromolecules, 48 (4),1087–1092.

53 Wu, Y., Demachi, Y., Tsutsumi, O. et al. (1998) Photoinduced alignment ofpolymer liquid crystals containing azobenzene moieties in the side chain.1. Effect of light intensity on alignment behavior. Macromolecules, 31 (2),349–354.

54 Kovacs, A.J. (1964) Transition vitreuse dans les polymères amorphes. Etudephénoménologique. Fortschr Hochpolym-Forsch, 3, 394.

55 Lee, K.M., Koerner, H., Wang, D.H. et al. (2012) Tailoring the photome-chanical response of glassy, azobenzene-functionalized polyimides byphysical aging. Macromolecules, 45, 7527–7534.

56 Boyer, R. (1977) Encyclopedia of Polymer Science and Technology, vol. II,Wiley-VCH Verlag GmbH & Co. KGaA, Hoboken.

57 Horie, K., Tsukamoto, M., and Mita, I. (1985) Photochemistry in poly-mer solids-7. Photochromic reaction of spiropyran in polycarbonate film.European Polymer Journal, 21 (9), 805–810.

58 Horie, K., Morishita, K., and Mita, I. (1984) Photochemistry in polymersolids. 3. Kinetics for nonexponential decay of benzophenone phospho-rescence in acrylic and methacrylic polymers. Macromolecules, 17 (9),1746–1750.

59 Wie, J.J., Wang, D.H., Lee, K.M. et al. (2014) Molecular engineering ofazobenzene-functionalized polyimides to enhance both photomechanicalwork and motion. Chemistry of Materials, 26, 5223–5230.

60 Mi, Y., Stern, S.A., and Trohalakib, S. (1993) Dependence of the gas perme-ability of some polyimide isomers on their intrasegmental mobility. Journalof Membrane Science, 17, 41–48.

61 Petrova, T., Toshchevikov, V., Saphiannikova, M. et al. (2015) Light-induceddeformation of polymer networks containing azobenzene chromophoresand liquid crystalline mesogens. Soft Matter, 11 (17), 3412–3423.

62 Lee, T.Y., Carioscia, J., Smith, Z., and Bowman, C.N. (2007) Thiol-allylether-methacrylate ternary systems. Evolution mechanism of

�

� �

�


polymerization-induced shrinkage stress and mechanical properties. Macro-molecules, 40 (5), 1473–1479.

63 Lowe, A.B. (2014) Thiol-yne “click”/coupling chemistry and recent appli-cations in polymer and materials synthesis and modification. Polymer(Guildf ), 55, 5517–5549.

64 Ube, T., Takado, K., Ikeda, T. et al. (2015) Photomobile materials withinterpenetrating polymer networks composed of liquid-crystalline andamorphous polymers. Journal of Materials Chemistry C, 3 (31), 8006–8009.

�

� �

�

153

5

Photomechanical Effects in Liquid-Crystalline PolymerNetworks and ElastomersTimothy J. White

Dayton, OH, USA

5.1 Introduction

Liquid-crystalline materials are synonymous with their use in the displays thatsurround us in our daily lives. The widespread application of liquid crystals indisplay applications is enabled by their unique assimilation of properties: thelong-range fluidity of liquids combined with the anisotropic optical propertiesof crystalline solids. In displays, the relationship of orientation of the liquidcrystal director and birefringence has been leveraged by concurrent advancesin thin-film transistors to locally control the transmission and spectral output(color) of light visualized in a pixel. The display community, in response toseemingly insatiable consumer demand, is currently focused on improvementsin resolution and speed as well as realization of flexible form factors.Liquid crystallinity can also be observed and retained in polymers. Polymeric

materials maintaining liquid crystallinity are also anisotropic not only in theiroptical properties but also in their mechanical properties. Employment of thesensitivity of these materials to thermal, electrical, and optical stimuli to trans-duce these energy inputs into mechanical response (deflection, deformation,or motion) has been recently subject to a general review [1]. Here, I detail anexhaustive overview focused exclusively on the preparation of photoresponsiveliquid-crystalline networks and elastomers and the mechanical responses theyexhibit.Before focusing on the specific nuances of these fascinating materials, let

us first discuss the semantics used to describe them and the general syntheticmethods by which they are prepared.

5.1.1 What Is a Liquid Crystal Polymer, Polymer Network, or Elastomer?

Many terms have been used to describe polymeric materials maintainingliquid crystallinity including polymeric liquid crystals (PLCs), liquid crystal


�

� �

�

154 5 Photomechanical Effects in Liquid-Crystalline Polymer Networks and Elastomers

polymers (LCPs), cross-linked liquid crystal polymers (CLCPs), liquid crystalpolymer networks (LCNs), and liquid crystal elastomers (LCEs). Many of theseterms have specific connotations to practicing researchers that unfortunatelyare often not appreciated by new entrants to this field nor by the broaderscientific community. For clarity, Figure 5.1 illustrates the difference in thematerials and the semantic convention used in this book and most commonin the current literature.The term liquid-crystalline polymer is often used as a general descriptor,

but within the practicing research community, this term describeshigh-performance yet linear polymeric materials such as Vectra or Kevlar. Thechemical structure of Kevlar is given as an inset in Figure 5.1(a). These poly-mers self-organize to form liquid-crystalline phases. The local order of thesematerials dominates their properties, in particular resulting in exceptionallylow coefficients of thermal expansion as well as high modulus (stiffness)despite their linear nature. A number of common, commercially availableLCPs are widely known. Kevlar is the basis of bulletproof vests, fishing line,and even drumheads.The distinguishing feature of liquid-crystalline polymer networks (referred to

by some groups as cross-linked liquid-crystalline polymers, such as in Chapters1 and 2) from LCPs is that the materials are cross-linked [2]. Liquid-crystallinepolymer networks are almost exclusively prepared from monomer precursorsthat are liquid crystalline. Materials within this subclass can be prepared tobe glassy or elastomeric. Elastomeric liquid-crystalline polymer networksare most often referred to as LCEs. Due to the differences in the preparationprocedures, historically, glassy and elastomeric liquid-crystalline polymernetworks have been viewed as distinct subclasses. However, as recentlyillustrated [3], the mechanical response of LCNs is a continuum dictated bythe extent of cross-linking. In general terms, glassy liquid-crystalline polymernetworks tend to have glass transition temperatures ranging from 60 to 100 ∘Cwith moduli parallel to the nematic director in the range of 1–2GPa. LCEshave glass transition temperatures below 25 ∘C with moduli on the order of100MPa or lower. This chapter separately discusses the photomechanicalresponse of liquid-crystalline polymer networks and elastomers.

5.1.2 How Are Liquid-Crystalline Polymer Networks and ElastomersPrepared?

5.1.2.1 Polysiloxane ChemistriesThe chemistry synonymous with Heino Finkelmann and the preparationof LCEs are reported in a series of works titled “Investigations on LiquidCrystalline Polysiloxanes” [4–6].The generalized reaction scheme is illustratedin Figure 5.2(a) [7]. The approach employs a hydrosilylation reaction, in whichthe Si—H bonds react via addition of unsaturated functional groups, such as

�

�

(a) LC polymers(“polymeric liquid crystals”) (b) “Glassy” LC polymers Networks

Tg > 200 °C

For example For example

E ∼ >2 GPa

ΔS ∼ 0%

Tg ∼ 40–100 °C

... Intramolecular interactions

E ∼ 0.8–2 GPa

ΔS ∼ 5%

Tg ∼ < 20 °C

E ∼ 0.1 MPa

ΔS ∼ 90%

(3) LC Elasfomers

n

n

n

O

O O

OO

H3C CH2 C2H5

Side-chain mesogen

Crosslinker

O

C O6

Si

H3C 2CH CH3CH211Si Si

(H-bonding, etc.) Crosslinks Mesogen Crosslinks Mesogen

Figure 5.1 Notional properties and molecular configurations of (a) liquid crystal polymers, (b) liquid crystal polymer networks, and (c) elastomers.

�

�

(c)

(b)(a)

W2

W1

1010

x

10

O

σe)(σi

End on

Side on

Tred

1.21.11.00.90.8

1

2

I/l0

3

4

5 Mainchain

Mechanical

field σe

Second

cross-linking

First

cross-linking

Toluene

Pt catalyst

60 °C, 24 h

OO

O

O

O

O OMe +

H60

SiO O SiSi

+

Figure 5.2 (a) General illustration of the preparation of liquid crystal elastomers by hydrosilylation reaction (used with permission from Ref. [7]).(b) Alignment of liquid crystal elastomers prepared with this chemistry is induced mechanically, through the “Finkelmann method” (used withpermission from Ref. [8]). (c) The strain (l/l0) of liquid crystal elastomers is known to be strongly dependent on the mesogen connectivity. (usedwith permission from Ref. [9].) (White and Broer [1]. Reproduced with the permission of Nature Publishing Group.)

�

� �

�

5.1 Introduction 157

vinyl monomers. The reaction is catalyzed by platinum (Pt). In the originalefforts, the vinyl groups react with and functionalize the polysiloxane mainchain as side-chain mesogens. In the capstone study of the initial examinations[4], Finkelmann, for the first time, prepared cross-linked materials to formelastomeric polymer networks that retained the nematic, cholesteric, andsmectic phases.The monomer and macromer precursors employed in subsequent studies

were varied to elucidate the structure–property–performance relationships ofthe materials chemistry to the order of the polymer networks and, ultimately,to the mechanical response. While the materials were shown to maintainliquid-crystalline character at themicroscale, thematerials were polydomain atthe macroscale. To realize the “artificial-muscle”-like properties first predictedby De Gennes [10], the mesogens should be aligned to cooperatively sum thestrain at the molecular level into a macroscopic deformation. Toward, this end,in 1991, Finkelmann demonstrated a two-step preparation method illustratedin Figure 5.2(b) to prepare single-crystal or “monodomain” LCEs [8, 11]. Thereaction proceeds in two steps – an initial reaction to form a material with suf-ficient mechanical integrity to be handled. At this point, the reaction is haltedand the material is aligned by mechanical stretching (similar to the ‘training’of shape memory alloys). Under sustained mechanical load, the reaction iscompleted to realize an aligned and fully reacted liquid-crystalline elastomer.Subsequent examinations have reported the preparation of LCEs with side-on[12] and main-chain [13] mesogenic units. As illustrated in Figure 5.2(c) [9],the mechanical strains generated from thermotropic phase transitions observ-able in LCE prepared with side-chain, side-on, and main-chain mesogenicprecursors vary widely [14]. Main-chain LCEs produce the largest deformationto stimulus.

5.1.2.2 Free Radical or Cationic PhotopolymerizationConcurrently, other groups focused on synthetic methods to prepare bothglassy and elastomer liquid crystal networks with free-radical or cationicpolymerization starting from low-molar-mass liquid-crystalline precursors.By far the most common method to prepare liquid-crystalline polymernetworks is and has been the free-radical polymerization of (meth)acrylateliquid-crystalline monomers (sometimes called reactive mesogens). Someexamples of commonly employed liquid-crystalline monomers are illustratedin Figure 5.3. Comparatively, alignment of these materials is simplified by theliquid-crystalline character of the precursors that makes them conducive tosurface alignment techniques widely employed in the fabrication of displays.The polymerization of these monomers was originally thermally initiated.

Photoinitiated polymerization (photopolymerization) is advantageous as itdecouples the reaction from the thermotropic nature of the starting materials[15]. A wide range of liquid-crystalline monomers can be mixed with one

�

�

ON

O

O

O O

OOO

O

O

∗ ∗O

N

O O O O

O

O

OO O O

O

NO

ON

N

OO O

O

“1azo”, “DA6AB”

“C6M”, “RM82” “C3M”, “RM257”

“RM23” “SLO4151”

“2azo”, “A6AB2”

O

O

O

O

O

O O

OO

O

O

O

Figure 5.3 Common liquid crystal monomers employed in the preparation of liquid-crystalline polymer networks and elastomers. Theazobenzene monomers “1azo” and “2azo” are frequently used in many of the examinations reported in this chapter.

�

� �

�

5.2 Optically Responsive Liquid Crystal Polymer Networks 159

another and even small concentrations of non-liquid-crystalline cross-linkersto prepare materials with mixed side- and main-chain mesogen connectivity[15–31]. Liquid-crystalline polymer networks with side-on mesogens havealso been reported [32–34]. Preparation of liquid crystal networks withmain-chain units requires the employment of a difunctional monomer, such asthe common diacrylate monomer RM82. If homopolymerized, this monomerwill yield a glassy liquid-crystalline polymer network with Tg around 80 ∘C anda modulus on the order of 1.5–2.0GPa (parallel to the nematic director) [26].The Tg and modulus can be reduced by including monoacrylate monomers,as pursued by Zentel and coworkers [31], Broer et al. [20], and Fridrikh andTerentjev [35]. However, at monoacrylate concentrations necessary to yieldliquid-crystalline polymer networks that are elastomers at room temperature,alignment/orientation of the materials is dominated by the side-chain orside-on mesogens.Recently, we have reported two methods to increase the molecular weight

between cross-links while maintaining the main-chain character of theresulting LCEs. The first simply subjects diacrylate liquid crystal monomerssuch as RM82 to an amine-catalyzed Michael addition reaction [36]. Byregulating the ratio of amine to acrylate functional groups, the glass transitiontemperature can be reduced from the glassy state of homopolymerized RM82(Tg of 80 ∘C) to elastomers with Tg as low as 7 ∘C [3]. The reaction occurs inone pot but must be slow so that the material can retain the surface alignmentas the molecular weight and viscosity increase. The second method employsthiol–ene chemistry with acrylate cross-linker (such as RM82) to prepare LCEsvia a simple and rapid photopolymerization [37]. A number of other groupshave reported on a variety of reaction chemistries in the recent literature[38–40]. Many recent papers also report on chemistries with reconfigurablecovalent and adaptive networks and employ them to postprocess the materialafter preparation [41–46].

5.2 Optically Responsive Liquid Crystal PolymerNetworks

5.2.1 Historical Overview

Light is a pervasive resource and has been employed since ancient times forpractical purposes – ranging from telling time by sundial to capturing solarradiation and transducing it into electrical power. With the advances in thegeneration of synthetic light with ever-increasing variety of wavelengths, itshould not be a surprise that scientists and engineers have pursued transducingthis input energy into a mechanical output. Light has a number of potentialadvantages as an actinic stimulus: it can be contactless, it can traverse long

�

� �

�


distances (remote separation of source and function), it can be easily turned onand off (temporal control), it can be easily patterned (intensity, polarization,and phase), and it comes in numerous wavelengths.Themethod and paradigmfor the transduction of light energy into mechanical work were first articulatedby Lovrien, where he referred to “light-energy transducers” nearly 50 years ago[47].The pursuit of large-scale and efficient transduction of light into work has

been subject to intense research from an impressively wide array of disciplines,primarily emanating from chemists, physicists, and engineers. Given theburgeoning growth, this topic has been reviewed regularly in the recentliterature [48–53]. As detailed hereto, the conversion of light into large-scalemechanical output (typically measured as strain) was unable to exceed 1% inphotoresponsive amorphous or semicrystalline polymeric materials. A collab-orative effort among Finkelmann, Warner, and coworkers realized large-scaleoptically reversible strain of 20% (Figure 5.4), by preparing and employing anazobenzene-functionalized LCE for the first time [54]. This paper instigated arenaissance in the research of materials and methods to transduce light energy

↑ Temperature from 298 to 313 K0.22

0.20

0.18

0.16

0.14

0.12

Contr

action f

raction

0.10

0.08

0.06

0.04

0.02

0.000.0 10.0 20.0 30.0 40.0

0.0

0.220.200.180.160.140.120.100.08

0.060.040.020.00

50.0 100.0 150.0 200.0 250.0

Time (min)

50.0 60.0 70.0 80.0 90.0

Figure 5.4 Photoinduced contraction as a function of time for a azobenzene-functionalizedliquid crystal elastomers upon UV irradiation. (Inset) Relaxation of contraction in theabsence of light. The symbols represent data collected at 298 K (*), 303 K (circle), 308 K(triangle), and 313 K (square). (Finkelmann et al. [54]. Reproduced with the permission ofAmerican Physical Society.)

�

� �

�


into mechanical work. The work inspired peers within the liquid-crystallinecommunity as well as brought new entrants from related communities inmechanics, photochemistry, and optics to explore the possibilities affordedby photoresponsive liquid-crystalline elastomers and polymer networks. Asof writing, this paper has been subject to nearly 500 citations from across theinternational research community.Shortly following the report of Finkelmann and coworkers was a paper

by Ikeda and coworkers in the journal Nature [55]. The authors prepareda unique LCN (heated above the glass transition temperature) entirelycomposed of azobenzene mesogens [55]. Irradiation of this material withlinearly polarized UV light (Figure 5.5) showcased, for the first time, theability to control the directionality of the deflection of a photoresponsive,

366 nm >540 nm

>540 nm

>540 nm

>540 nm366 nm

366 nm

366 nm

–45°

–90°

–135°

0°

Figure 5.5 All-optical control of bending direction and flattening in anazobenzene-functionalized liquid-crystalline polymer network, in the polydomainorientation. The film in this experiment was heated above the glass transition temperature.Irradiation with linearly polarized UV light in the orientations inset into the images dictatesthe direction of the bending. Irradiation with light >540 nm restores the films to the flatcondition. (Yu et al. [55]. Reproduced with the permission of Nature Publishing Group.) (Seecolor plate section for the color representation of this figure.)

�

� �

�


azobenzene-functionalized material.Thematerial maintained an exceptionallylarge absorption coefficient, which upon irradiation limited the penetrationdepth of UV light into the film. Accordingly, the photogenerated strain waslimited to the irradiated surface, inducing bending similar to a bimetallic strip.The work of Ikeda and coworkers further enticed the research community,edifying the association of photomechanical effects in materials and liquidcrystallinity. Subsequent to this highly visible report, a considerable volume ofresearch by these authors and other groups theoretically and experimentallyexplored the opportunity space to control the photomechanical response ofthese materials with light intensity, polarization, and wavelength.

5.2.2 Photochromic and Liquid Crystalline

Azobenzene has been and remains by far the most common means fortransducing light into mechanical deformation in polymeric materials. Thefundamentals of the photochemistry of azobenzene in the solid state werediscussed in Chapter 2. Included in Figure 5.3 are examples of two commonlyused azobenzene-based liquid-crystalline monomers. Azobenzene in thethermodynamically stable trans configurations maintains a strong absorbanceof UV light. Upon absorption of a photon, azobenzene can isomerize into thecis isomer. As first reported in 1987, the photoinduced conformational changeof azobenzene from rod-like (trans) to bent (cis) is highly disruptive to liquidcrystallinity and the source of phototropic, order-decreasing phase transitions[56, 57].Embedding azobenzene into polymeric materials either as a guest molecule

[58, 59] or through covalent attachment reduces the quantum efficiencyof conversion from the trans to cis isomer, as extensively reported in thephotochemistry literature [60, 61]. In LCEs, such as the work of Finkel-mann [54], irradiation of the azobenzene-functionalized LCE induces aphototropic order–disorder transition [62] much as in conventional, fluidic,azobenzene liquid crystal mixtures [56, 57]. The photoinduced loss of orderin azobenzene-functionalized LCEs that accompanies UV irradiation (e.g.,trans–cis isomerization) results in a considerable volume change as theanisotropic structure of the network collapses, resulting in the large increase inthe magnitude of the photogenerated strain to 20% or greater. Importantly, theirradiation of azobenzene-functionalized LCNs with UV light (e.g., trans–cisisomerization) does not induce phase transition nor large-scale changes in theorder parameter of cross-linked networks in the glassy state, yet still leads toanisotropic deformation [63]. The photochemistry in elastomeric and glassyliquid crystal networks does relax over time and has been recently shown to bestrongly and unexpectedly dependent on the mobility of the polymer networks[64, 65]. Cis–trans isomerization within these materials can be expedited withirradiation with light of higher wavelength, exceeding 530 nm [55].

�

� �

�


A related photochemical mechanism has also been employed to inducephotomechanical effects in azo-LCE and azo-LCN materials, referred toas trans–cis–trans reorientation (the Weigert effect) [66, 67]. Of crucialimportance to this mechanism is the use of light with a wavelength that isnearly equivalently absorbed by both the trans and cis isomers of azobenzene.Under these conditions, absorption of a photon by a trans or cis isomer isstatistically equivalent and thus can induce both trans–cis and cis–transisomerization. Due to the rotational freedom of the azo bond as well as thedichroic absorbance of azobenzene, irradiation with linearly polarized light inthis wavelength regime (typically the blue-green, 440–488 nm) can generatea statistical buildup of trans- or cis-azobenzene oriented in a distributionthat is orthogonal to the electric field vector of the linearly polarized light.Accordingly, this can allow the same wavelength of light to alter the orientationof a portion of the azobenzene chromophores such that both contractile andexpansive strains (or shear) can be realized. This mechanism has been widelyemployed in the generation of surface relief gratings in glassy, amorphousazobenzene-functionalized polymers – under experimental conditions asmuch as 300 ∘C below the glass transition temperature, detailed extensivelyin Chapter 4 [68–74]. Recent reports have shown that the photoinducedreorientation in glassy azobenzene-functionalized liquid-crystalline polymernetworks can be retained for substantial periods of time and can be a mech-anism for shape memory (Figure 5.6) [75]. As will be discussed in Chapter9, Liu and Broer report that employing the use of two light sources cantailor the effectiveness of this mechanism enabling improvement of thephotomechanical responses [65].

(a) (c)(b) (d)

(e) (f)

Figure 5.6 All-optical control of shape memory in an azobenzene-functionalizedliquid-crystalline polymer network. The originally flat film (a) was mechanically deformed(b) and irradiated (c) with linearly polarized 442 nm light. After the film was removed fromthe light (d), the deformed shape was retained (shape memory). The material retained thedeformed state indefinitely. After irradiation with circularly polarized light (e), the film wasrestored to the original flat shape (f ). (Lee et al. [75]. Reproduced with the permission ofRoyal Society of Chemistry.)

�

� �

�


5.2.3 Photomechanics

The deformation of a given material under certain irradiation conditions isgoverned by the language of mechanics. Since the time of Archimedes andEuclid, mechanics have fascinated scientists and engineers.The photomechan-ical effects discussed here are complex due to the additional considerationsimposed by optics, photochemistry, and polymer science. While the commu-nity is currently populatedwith chemists and physicists, ultimately, if andwhenthesematerials are employed asmechanically active elements in a larger assem-bly, it is important that thematerials and research ongoingwithin this area con-textualize the results in the appropriate language and framework of mechanics.To date, two primary geometries have been employed as a means to visually

depict and characterize the photomechanical response of azo-LCN materials:the film and cantilever geometry. In the parlance employed here, the filmgeometry is in the limit where light is nearly equally absorbed across thethickness of the film. Accordingly, the film will exhibit near-uniform strainthrough the thickness, which will manifest itself in a change in length in onedimension and an equal but oppositely signed change in another dimension.However, due to the limited strain generated in azo-LCN materials as wellas the large azobenzene concentrations that are typically examined, the filmgeometry is very rarely used.More commonly, researchers have employed the cantilever geometry to

visualize the photomechanical output of a material. The cantilever geometry isone inwhich the length is larger than thewidth and considerably larger than thethickness. In this regime, the bend is caused by nonuniform strain generationthrough the sample thickness that is reticent to a bimetallic strip. The straingradient through the sample thickness is dictated by the strong attenuation oflight due to the absorbance of azobenzene (or other materials) that localizesthe strain on the exposed surface. The directionality of the strain dictates themagnitude of the bend or twist. If the strain is contractile along the length of thecantilever (long axis), the cantilever can deflect toward the light source. If thestrain is expansive along the length of the cantilever (long axis), the cantileverwill deflect away from the light source. If the strain is offset to the principalaxes of the cantilever, the cantilever can twist. In all cases, the magnitude ofthe deflection is strongly dependent on the thickness, aspect ratio, irradiationconditions (intensity, polarization, wavelength), and properties of the material(azobenzene concentration, domain orientation of the LCN, thermomechan-ical properties). Warner and coworkers have examined the underlying physicsand mechanics of the photomechanical responses of azo-LCE and azo-LCNsystems in a series of papers [76–83] as well as a recent review [49]. Corbett,Modes, and Warner detailed the mechanics of these materials in Chapter 3,specifically focusing on the integration of complex topological director profilesto induce a variety of intricate and potentially useful shape transformations.

�

� �

�

5.3 Literature Survey 165

5.3 Literature Survey

By employing the basic principles introduced hereto, azobenzene-functionalized liquid-crystalline polymer networks and elastomers havebeen shown to generate a number of remarkable photomechanical outputs. Avariety of materials have been synthesized to date, many employing mixturesof the monomers presented in Figure 5.3. In this section, I review the synthesisand characterization of photomechanical responses observed in materialsprepared with hydrosilylation chemistry as well as poly(meth)acrylates.

5.3.1 Photomechanical Effects in PolysiloxaneMaterials and Analogs

As discussed, Finkelmann et al. synthetized LCEs from the hydrosilylationreaction of the siloxane precursor PHMS and three side-chain monomersin addition to two cross-linkers – a conventional divinyl monomer and 20%(by mol) vinyl methacrylate monomer with an azobenzene linker [54]. Theauthors employed the two-stage alignment technique, at times referred toas the “Finkelmann method,” [11] to prepare the material, with the azoben-zene monomer acting as the second-stage cross-linker due to the use of themethacrylate group.The authors detail the thermotropic as well as phototropicresponse to UV light irradiation. The material deforms by about 24% to heatand upward of 17% to light (at room temperature).Subsequent studies of similar materials by Terentjev and coworkers detail

the underlying structure–property–performance relationships. In the first[84], a variety of materials were prepared and examined to ascertain the roleof azobenzene connectivity (side chain vs main chain). The five samples hada large variation in the Tni and concurrent variation in azobenzene concen-tration. The variations in the thermomechanical response of the materialscomplicate the implications, but the authors conclude that the largest elonga-tion attributable to photomechanical deformation is observed in a sample withonly 18mol% side-chain azobenzene. Subsequent studies have clarified that,indeed, main-chain azobenzene particularly in cross-linking sites enables themost efficient translation of light into mechanical deformation [85]. Perhapsmost notably, this work wonderfully and directly confirms the relationshipbetween light and the disruption of order in these materials. A subsequentreport details the performance of a related material in generating stress (force)and relationship to temperature [62]. Sanchez-Ferrer et al. [85] andVelasco andcoworkers [86] have built upon this prior literature to clarify the association ofazobenzene connectivity (side chain or main chain/cross-linking) as well as toexplore variations in the azobenzene chromophore.Palffy-Muhoray and coworkers prepared an azobenzene-dye-doped

LCE, and examined extremely rapid deformations to pulsed laser irradi-ation [58]. Further investigations focused on the underlying connection

�

� �

�


of the photochemistry and mechanical response, by comparing the sen-sitivity of a variety of materials (both monodomain and polydomain)to polarized light of different wavelengths [87]. This contribution alsohighlighted that similar and, in fact, faster photomechanical responsescould be observed in LCEs prepared with guest additives of azobenzene.A recent study further edifies this conclusion, in a comparison of the photome-chanical response measured as actuation stress of LCEs prepared with guestadditives (Disperse Red 1 – azobenzene; Disperse Blue 14 – anthraquinone)and comparison of the response of LCEs prepared with a side-chain push–pullazobenzene group with similar structure to the Disperse Red dye [59]. Thedata shows that comparatively larger magnitude stress is realized in elastomersprepared with guest additives and that, comparatively, the differences in thephotochemistry (azobenzene can isomerize, anthraquinone cannot and thusradiates absorbed light as heat) are not distinguishable.

5.3.2 Photomechanical Effects in Poly(meth)acrylate Materialsand Analogs

Liquid-crystalline networks and elastomers prepared from free-radical,cationic, or related polymerization techniques have been long pursued.Perhaps, due to the comparative ease of preparing these materials as wellas the more straightforward alignment procedures, since the 2003 reportby Ikeda and coworkers [55], a greater number of papers have focused onmaterials prepared primarily through (meth)acrylate photopolymerization. Inthe original case of Ikeda and coworkers, the material was prepared from thecopolymerization of a monoacrylate azobenzene monomer with a diacrylateazobenzene monomer. As has been discussed, the authors elegantly andsuccinctly report polarization-directed bending and unbending of a glassyliquid-crystalline polymer network (in this case, heated above Tg) entirelycomposed of azobenzene mesogenic monomers [55]. A key detail to fullyappreciate this work is that the 8-μm-thick filmwas polydomain in orientation.Bending is observed despite the small thickness of the filmbecause of the strongattenuation of light near the exposed surface. Subsequent exposure to higherwavelength light (in this case,>530 nm) initiated cis–trans isomerization of theazobenzene chromophores and restored the film to approximately the flat state(e.g., nearly 100% trans isomer concentration). Subsequent exposure to lightoriented in a different axis was able to reconfigure the deflection of the film.Ikeda and coworkers populated the literature with a number of subsequent

examinations [88–91]. Notable recent efforts include the preparation andphotomechanical characterization of fibers [92], the inclusion of upconvertingnanoparticles to allow response to be triggered with higher wavelength (IR)irradiation [93], and systematic examination of the role of cross-linker in thegeneration of strain [94–97]. Concurrently, Keller and coworkers focused

�

� �

�

5.3 Literature Survey 167

their efforts on preparing liquid-crystalline networks with azobenzene unitsconnected as “side-on” mesogens and characterizing the mechanical responses[32, 98].Researchers at the Air Force Research Laboratory (AFRL) explored related

materials but focused on their response to irradiation with blue-green light.Building upon the work of Tabiryan et al. [99], polarization-controlled andbidirectional deflection was reported [100]. The reorientation of azobenzenein glassy polymeric materials – amorphous, semicrystalline, or liquid crys-talline – remains a mystery to many, and the response of azobenzene to lightin these wavelength regime continues to be a topic of fundamental study [101].Building upon prior examinations focused on preparing surface relief grat-ings in glassy, azobenzene polymer-based films [66, 102], AFRL researchersreported on using this wavelength to realize all-optical shape memory in thesematerials [75]. The materials were shown to retain the deformed state uponirradiation with linearly polarized blue-green light, which was subsequentlyunlocked by irradiation with circularly polarized blue-green light. Recentefforts by Liu report similar behavior employing two light sources to furtherenhance the effect [103].Dynamic photomechanical responses have been examined in these materials

as well. A collaboration between AFRL and a small company (BEAM Co.)reported on photo-driven oscillations in glassy azobenzene-functionalizedLCNs (Figure 5.7). Notably, these materials were of the identical chemistry

1.0

0.8

0.6

0.4

0.2

0.00 50

(a) (i) (ii)

(c)

(b)

100 150 200

Time (ms)

Inte

nsi

ty (

a.u

.)

250 300 350 400

Figure 5.7 Photo-driven oscillation of an azobenzene-functionalized liquid-crystallinepolymer network. (a) The oscillation induced with the multiline irradiation from an Argonion laser (457, 488, 514 nm) is shown in (i) and (ii) (laser light filtered). (b) The oscillationfrequency can be measured optically and strongly dependent on the film geometry,reaching as much as 270 Hz. (c) Oscillation can be induced with 442 nm light as well. (Whiteet al. [104]. Reproduced with the permission of Royal Society of Chemistry.)

�

� �

�


[104], as discussed in the initial work by Ikeda and coworkers [55], butmonodomain rather than polydomain. A subsequent report authored by Seraket al. clarified that the frequency of the observed oscillation matches theexpected resonant frequency of the cantilever [105]. Key to the generationof oscillatory responses is the employment of focused irradiation, which canallow the front and back surfaces of the cantilever to deflect into and out of thelight. These initial reports allude to the contribution of photothermal heatingto the effect, which have been further clarified in subsequent studies usingthermal imaging [106].In addition to conventional planar deflections, the flexural–torsional

(e.g., bending and twisting) responses have also been examined. The initialwork of Ikeda and coworkers [55] documents a flexural–torsional responsein the cases where the films bend along the diagonal of the cantilever.Polarization-controlled twisting has been reported by Tabiryan et al. [99]and recently at AFRL, under conditions that induce both static [96, 106]and oscillatory [107] deflections. While being interesting, the magnitude ofthe twisting is somewhat limited in these conventional domain orientations(monodomain or polydomain).One of the substantial advantages of LCN materials in comparison to other

stimuli-responsive polymeric materials is the ability to spatially or hierar-chically manipulate the orientation of anisotropy to generate desired effects.The splay and twisted nematic geometry can be retained in LCNs to generatenew and distinctive properties. Broer and coworkers have reported on a fiveorders of magnitude enhancement in work generation in planar deflectionsemploying these hierarchical structures (e.g., splay or twisted nematic). Inaddition to enhancing the magnitude of planar deflections, hierarchical LCNstructures also enhance the magnitude of flexural–torsional responses. Theseauthors had reported on photoinduced coiling of a TN film [108]. Buildingupon this work, Urayama and Sellinger reported on an experimental and the-oretical examination of ribbon formation (spiral and helicoidal) in a thermallyresponsive LCN material as discussed earlier [109]. AFRL researchers haverecently reported on photoinduced twisting and shape formation in twistednematic films with spatially patterned domain orientation (alternating withmonodomain), as shown in Figure 5.8 [110]. Other groups have reportedphotoinduced twisting subsequent to these prior efforts [111].In nearly all of the examples of photomechanical deformation in

poly(meth)acrylate materials, the deformations have been observed andreported in the glassy state. Extending upon the employment of theaza-Michael addition reaction to prepare high-molecular-weight oligomerswithmain-chainmesogenic groups, recently, an effort fromAFRL has reportedon the photoinduced deformation of cones in the so-called blueprinted elas-tomers prepared with azobenzene-functionalized polymer networks [64]. Asillustrated in Figure 5.9, light irradiation generates a large deformation in

�

� �

�

5.4 Outlook and Conclusion 169

(a) (b) (c)

Figure 5.8 Thermal (a) and photoinduced twisting of azobenzene-functionalizedliquid-crystalline polymer network (b and c). ((a) Sawa et al. [109]. Reproduced with thepermission of PNAS publication. (b) Wie et al. [110]. Reproduced with the permission of RSCpublication. (c) Iamsaard et al. [111]. Reproduced with the permission of Nature PublishingGroup.)

these materials that is highly dependent on the cross-link density as well as theazobenzene concentration. The deformation of these materials has increasedby at least a factor of 5 when compared to prior examination of photomechani-cal deformation of+1 defects imprinted into glassy azobenzene-functionalizedliquid-crystalline polymer networks [112].

5.4 Outlook and Conclusion

As has been discussed hereto in this chapter and those preceding, employinglight as a “smart” energy stimulus in concert with “smart”materials is a ripe areaof exploratory research.As this area grows andmatures, it is important for theseefforts to more strongly connect and baseline the material responses to priorliterature within this community as well as make connections to the broaderarea of stimuli-responsive materials. In this way, the uncertainty surroundingthe potential utility of these materials can be further unveiled.Contextualizing the large array of results discussed in this chapter is not

an easy task, as the results in some cases are self-contradictory. What canbe unequivocally stated is that liquid crystallinity can further improve andenhance the photomechanical response that can occur in more conventionalpolymeric materials, but it should be emphasized that liquid crystallinity isnot a requirement to realize photomechanical effects in polymeric materials.Further, it does seem to be consistent that the most effective way to transducelight into photomechanical deformation (stress or strain) is to embed azoben-zene in the cross-linking sites of the materials. With this said, considerableprogress must be made in the efficient conversion of energy, which has beencalculated to be extremely small [113].Throughout this book as well as in the literature cited throughout this

chapter, it has been continuously mentioned that the potential outlet of thesematerials is in actuation. Gauging from the literature, it is clear that different

�

� �

�


(a)

(c) 1 2 3

AP

5 mm

4 5

(d)

(e)

(f)

(b)

Figure 5.9 Azo-LCE compositions were prepared by aza-Michael addition reactionsincluding 2azo. Samples were prepared with +1 azimuthal defects subsumed in the centerof square films of 5 × 5 mm with 50 μ thickness. (a) Illustration of the director profiledescribed by a +1 azimuthal defect. (b) A representative photograph of a +1 azimuthaldefect within a azo-LCE taken between cross-polarizers. (c–f ) The five azo-LCE films wereplaced on a white surface and subjected to 365 nm irradiation of 100 mW/cm2 for 15 min.Photographs were taken to measure the relative deflection of the materials (c) duringexposure, (d) 5 s after exposure, (e) 2 min after exposure, and (f ) after 532 nm exposure(∼50 mW/cm2 for 10 min). (Ahn et al. [64]. Reproduced with the permission of John Wileyand Sons.) (See color plate section for the color representation of this figure.)

�

� �

�

References 171

authors mean entirely different things when they use the term actuation.Some define actuation simply as motion. Others think of an actuator as asystem in which the stimuli-responsive element is but a small part of a largersystem composed of amplifying elements and other mechanisms. What doesseem to be general is that the term actuation implies purpose. The purposeand employment of these materials will not be championed by the materialschemistry and physics community, but rather the mechanics community.A challenge facing the stimuli-responsive materials community at large is toidentify the key experiments and performance attributes required to securefurther interest by the pure mechanics community. This may identify potentialentry points to niche application demonstrations that would spur initialinvestigation of viable applications.One unique attribute of liquid-crystalline materials is the ease with which

locally patterned anisotropic materials can be prepared [36, 37]. Repeatedlydemonstrated in the recent literature, these patterns have been imprinted intothe topology of the director profile of the materials and have produced definedand repeatable shape formation.Thepreparation of “functional” (e.g., designed)monolithicmaterials is not simple to emulate in other stimuli-responsivemate-rials classes. As discussed in Chapter 9, the generation of shape-changing ordynamic topographical features has potential use in microfluidics [114], flowcontrol, solar energy harvesting [105, 115–117], and haptic displays [118–120].Finally, an exciting new area of functionality has been introduced in both

polysiloxane and poly(meth)acrylate liquid-crystalline polymer networks andelastomers. Concurrent to the examinations detailed within this chapter, aparallel community focused on the so-called covalent adaptive networks hasdeveloped [121]. These materials are also cross-linked and have functionalgroups that allow for bond breaking and reformation. Recently, the twocommunities have begun to merge, which is evident in a number of recentreports [41–46].

References

1 White, T.J. and Broer, D.J. (2015) Nature Materials, 14 (11), 1087–1098.2 Broer, D.J., Crawford, G.P., and Zumer, S. (2011) Cross-Linked Liquid

Crystalline Systems: From Rigid Polymer Networks To Elastomers, CRCPress, Boca Raton, FL, 605 pp.

3 Ware, T. and White, T. (2015) Polymer Chemistry, 6 (26), 4835–4844.4 Finkelmann, H., Kock, H.-J., and Rehage, G. (1981) Macromolecular Rapid

Communications, 2, 317–322.5 Finkelmann, H. and Rehage, G. (1980) Die Makromolekulare Chemie.

Rapid Communications, 1 (12), 733–740.

�

� �

�


6 Finkelmann, H. and Rehage, G. (1980) Die Makromolekulare Chemie.Rapid Communications, 1 (1), 31–34.

7 Krause, S., Zander, F., Bergmann, G. et al. (2009) Comptes Rendus Chimie,12 (1–2), 85–104.

8 Kuepfer, J. and Finkelmann, H. (1994) Macromolecular Chemistry andPhysics, 195 (4), 1353–1367.

9 Ohm, C., Brehmer, M., and Zentel, R. (2010) Advanced Materials, 22 (31),3366–3387.

10 De Gennes, P.G. (1975) Comptes Rendus de l’Académie des Sciences, B281,101.

11 Küpfer, J. and Finkelmann, H. (1991) Die Makromolekulare Chemie. RapidCommunications, 12 (12), 717–726.

12 Thomsen, D.L. III, Keller, P., Naciri, J. et al. (2001) Macromolecules, 34(17), 5868–5875.

13 Bergmann, G.H., Finkelmann, H., Percec, V., and Zhao, M. (1997) Macro-molecular Rapid Communications, 18 (5), 353–360.

14 Wermter, H. and Finkelmann, H. (2001) e-Polymers, Online.15 Broer, D.J. and Mol, G.N. (1987) ACS Symposium Series, 346, (Polymers

for High Technology: Electronics and Photonics), 417–423.16 Broer, D.J., Finkelmann, H., and Kondo, K. (1988) Makromolekulare

Chemie, 189 (1), 185–194.17 Bouten, P.C.P., Broer, D.J., Jochem, G.M.G., and Meeuwsen, T.P.M. (1989)

Polymer Engineering and Science, 29 (17), 1172–1176.18 Broer, D.J., Boven, J., Mol, G.N., and Challa, G. (1989) Makromolekulare

Chemie, 190 (9), 2255–2268.19 Broer, D.J., Hikmet, R.A.M., and Challa, G. (1989) Makromolekulare

Chemie, 190 (12), 3201–3215.20 Broer, D.J., Mol, G.N., and Challa, G. (1989) Makromolekulare Chemie,

190 (1), 19–30.21 Broer, D.J., Gossink, R.G., and Hikmet, R.A.M. (1990) Angewandte Makro-

molekulare Chemie, 183, 45–66.22 Broer, D.J. and Heynderickx, I. (1990) Macromolecules, 23 (9), 2474–2477.23 Broer, D.J. and Mol, G.N. (1991) Polymer Engineering and Science, 31 (9),

625–631.24 Broer, D.J., Mol, G.N., and Challa, G. (1991) Polymer, 32 (4), 690–695.25 Broer, D.J., Mol, G.N., and Challa, G. (1991) Makromolekulare Chemie,

192 (1), 59–74.26 Hikmet, R.A.M. and Broer, D.J. (1991) Polymer, 32 (9), 1627–1632.27 Hikmet, R.A.M., Lub, J., and Broer, D.J. (1991) Advanced Materials, 3

(7–8), 392–394.28 Hikmet, R.A.M., Zwerver, B.H., and Broer, D.J. (1992) Polymer, 33 (1),

89–95.

�

� �

�

References 173

29 Broer, D.J., Lub, J., and Mol, G.N. (1993) Macromolecules, 26 (6),1244–1247.

30 Lub, J., Broer, D.J., Hikmet, R.A.M., and Nierop, K.G.J. (1995) Liquid Crys-tals, 18 (2), 319–326.

31 Portugall, M., Ringsdorf, H., and Zentel, R. (1982) MakromolekulareChemie, 183 (10), 2311–2321.

32 Li, M.-H., Keller, P., Li, B. et al. (2003) Advanced Materials, 15 (7–8),569–572.

33 Buguin, A., Li, M.-H., Silberzan, P. et al. (2006) Journal of the AmericanChemical Society, 128 (4), 1088–1089.

34 Li, M.-H. and Keller, P. (2006) Philosophical Transactions of the Royal Soci-ety of London, Series A, 364 (1847), 2763–2777.

35 Fridrikh, S. and Terentjev, E. (1999) Physical Review E, 60 (2), 1847.36 Ware, T.H., McConney, M.E., Wie, J.J. et al. (2015) Science, 347 (6225),

982–984.37 Ware, T.H., Perry, Z.P., Middleton, C.M. et al. (2015) ACS Macro Letters, 4

(9), 942–946.38 Gelebart, A.H., McBride, M., Schenning, A.P.H.J. et al. (2016) Advanced

Functional Materials, 26 (29), 5322–5327.39 Martella, D., Parmeggiani, C., Wiersma, D.S. et al. (2015) Journal of Mate-

rials Chemistry C, 3 (34), 9003–9010.40 Yakacki, C.M., Saed, M., Nair, D.P. et al. (2015) RSC Advances, 5 (25),

18997–19001.41 Pei, Z., Yang, Y., Chen, Q. et al. (2014) Nature Materials, 13 (1), 36–41.42 Meng, F., Pritchard, R.H., and Terentjev, E.M. (2016) Macromolecules

(Washington, DC, United States), 49 (7), 2843–2852.43 Pritchard, R.H., Redmann, A.-L., Pei, Z. et al. (2016) Polymer, 95, 45–51.44 Li, Y., Rios, O., Keum, J.K. et al. (2016) ACS Applied Materials & Inter-

faces, 8 (24), 15750–15757.45 Yang, Y., Pei, Z., Li, Z. et al. (2016) Journal of the American Chemical

Society, 138 (7), 2118–2121.46 Ube, T., Kawasaki, K., and Ikeda, T. (2016) Advanced Materials

(Weinheim, Germany), 28 (37), 8212–8217.47 Lovrien, R. (1967) Proceedings of the National Academy of Sciences of the

United States of America, 57 (2), 236–242.48 Ikeda, T. and Zhao, Y. (2009) Smart Light-Responsive Materials:

Azobenzene-Containing Polymers and Liquid Crystals, John Wiley & Sons,Inc, Hoboken, NJ, p. 514.

49 Corbett, D. and Warner, M. (2009) Liquid Crystals, 36 (10–11),1263–1280.

50 Ikeda, T., Mamiya, J.-I., and Yu, Y. (2007) Angewandte Chemie, Interna-tional Edition, 46 (4), 506–528.

51 Ikeda, T. and Ube, T. (2011) Materials Today, 14 (10), 480–487.

�

� �

�


52 Koerner, H., White, T.J., Tabiryan, N.V. et al. (2008) Materials Today, 11(7–8), 34–42.

53 White, T.J. (2012) Journal of Polymer Science Part B: Polymer Physics, 50(13), 877–880.

54 Finkelmann, H., Nishikawa, E., Pereira, G.G., and Warner, M. (2001) Physi-cal Review Letters, 87 (1), 015501/1–015501/4.

55 Yu, Y., Nakano, M., and Ikeda, T. (2003) Nature, 425, 145.56 Tazuke, S., Kurihara, S., and Ikeda, T. (1987) Chemistry Letters, 5,

911–914.57 Ikeda, T. and Tsutsumi, O. (1995) Science, 268 (5219), 1873–1875.58 Camacho-Lopez, M., Finkelmann, H., Palffy-Muhoray, P., and Shelley, M.

(2004) Nature Materials, 3, 307–310.59 Marshall, J.E. and Terentjev, E.M. (2013) Soft Matter, 9 (35), 8547–8551.60 Paik, C.S. and Morawetz, H. (1972) Macromolecules, 5 (2), 171–177.61 Serra, F. and Terentjev, E.M. (2008) Macromolecules, 41 (3), 981–986.62 Cviklinski, J., Tajbakhsh, A.R., and Terentjev, E.M. (2002) European

Physical Journal E: Soft Matter, 9 (5), 427–434.63 Harris, K.D., Bastiaansen, C.W.M., Lub, J., and Broer, D.J. (2005) Nano

Letters, 5 (9), 1857–1860.64 Ahn, S.-k., Ware, T.H., Lee, K.M. et al. (2016) Advanced Functional Mate-

rials, 26, 5819–5826.65 Liu, D. and Broer, D.J. (2015) Nature Communications, 6, Article number:

8334.66 Viswanathan, N.K., Kim, D.U., Bian, S. et al. (1999) Journal of Materials

Chemistry, 9, 1941–1955.67 Cojocariu, C. and Rochon, P. (2004) Pure and Applied Chemistry, 76

(7–8), 1479–1497.68 Sekkat, Z., Wood, J., Knoll, W. et al. (1997) Journal of the Optical Society

of America B, 14 (4), 829–833.69 Wie, J.J., Wang, D.H., Lee, K.M. et al. (2014) Chemistry of Materials, 26

(18), 5223–5230.70 Lee, K.M., Wang, D.H., Koerner, H. et al. (2013) Macromolecular Chem-

istry and Physics, 214 (11), 1189–1194.71 Wang, D.H., Lee, K.M., Koerner, H. et al. (2012) Macromolecular Materi-

als and Engineering, 297 (12), 1167–1174.72 Lee, K.M., Wang, D.H., Koerner, H. et al. (2012) Angewandte Chemie,

International Edition, 51 (17), 4117–4121, S4117/1–S4117/2.73 Lee, K.M., Koerner, H., Wang, D.H. et al. (2012) Macromolecules, 45 (18),

7527–7534.74 Wang, D.H., Lee, K.M., Yu, Z. et al. (2011) Macromolecules, 44,

3840–3846.75 Lee, K.M., Koerner, H., Vaia, R.A. et al. (2011) Soft Matter, 7, 4318–4324.

�

� �

�

References 175

76 Warner, M. and Terentjev, E. (2003) Macromolecular Symposia, 200,81–92.

77 Warner, M. and Mahadevan, L. (2004) Physical Review Letters, 92 (13),134302/1–134302/4.

78 Corbett, D. and Warner, M. (2006) Physical Review Letters, 96 (23),237802/1–237802/4.

79 Corbett, D. and Warner, M. (2007) Physical Review Letters, 99 (17),174302/1–174302/4.

80 Corbett, D. and Warner, M. (2008) Physical Review E, 78 (6-1),061701/1–061701/13.

81 van Oosten, C.L., Corbett, D., Davies, D. et al. (2008) Macromolecules, 41(22), 8592–8596.

82 Warner, M., Modes, C.D., and Corbett, D. (2010) Proceedings of the RoyalSociety A: Mathematical, Physical and Engineering Science.

83 Warner, M., Modes, C.D., and Corbett, D. (2010) Proceedings of the RoyalSociety A: Mathematical, Physical and Engineering Science, 466 (2122),2975–2989.

84 Hogan, P.M., Tajbakhsh, A.R., and Terentjev, E.M. (2002) Physical ReviewE, 65 (4-1), 041720/1–041720/10.

85 Sanchez-Ferrer, A., Merekalov, A., and Finkelmann, H. (2011) Macro-molecular Rapid Communications, 32 (8), 671–678.

86 Garcia-Amorós, J., Martínez, M., Finkelmann, H., and Velasco, D. (2014)Physical Chemistry Chemical Physics, 16 (18), 8448–8454.

87 Harvey, C.L.M. and Terentjev, E.M. (2007) European Physical Journal E:Soft Matter, 23 (2), 185–189.

88 Yu, Y., Nakano, M., Shishido, A. et al. (2004) Chemistry of Materials, 16(9), 1637–1643.

89 Kondo, M., Maeda, T., Shishido, A. et al. (2005) Molecular Crystals andLiquid Crystals, 441, 297.

90 Kondo, M., Yu, Y., and Ikeda, T. (2006) Angewandte Chemie, InternationalEdition, 45 (9), 1378–1382.

91 Okano, K., Shishido, A., and Ikeda, T. (2006) Advanced Materials, 18 (4),523–527.

92 Yoshino, T., Kondo, M., Mamiya, J.-i. et al. (2010) Advanced Materials, 22(12), 1361–1363.

93 Wu, W., Yao, L., Yang, T. et al. (2011) Journal of the American ChemicalSociety, 133 (40), 15810–15813.

94 Mamiya, J.-i., Yoshitake, A., Kondo, M. et al. (2008) Journal of MaterialsChemistry, 18 (1), 63–65.

95 Kondo, M., Sugimoto, M., Yamada, M. et al. (2010) Journal of MaterialsChemistry, 20 (1), 117–122.

96 Lee, K.M., Tabiryan, N.V., Bunning, T.J., and White, T.J. (2012) Journal ofMaterials Chemistry, 22 (2), 691–698.

�

� �

�


97 Lee, K.M., Koerner, H., Vaia, R.A. et al. (2010) Macromolecules, 43 (19),8185–8190.

98 Deng, W., Li, M.-H., Wang, X., and Keller, P. (2009) Liquid Crystals, 36(10–11), 1023–1029.

99 Tabiryan, N., Serak, S., Dai, X.-M., and Bunning, T. (2005) Optics Express,13, 7442–7448.

100 White, T.J., Serak, S.V., Tabiryan, N.V. et al. (2009) Journal of MaterialsChemistry, 19 (8), 1080–1085.

101 Fang, G.J., Maclennan, J.E., Yi, Y. et al. (2013) Nature Communications, 4,2483.

102 Natansohn, A. and Rochon, P. (2002) Chemical Reviews, 102 (11),4139–4175.

103 Kumar, K., Schenning, A.P., Broer, D.J., and Liu, D. (2016) Soft Matter, 12(13), 3196–3201.

104 White, T.J., Tabiryan, N., Tondiglia, V.P. et al. (2008) Soft Matter, 4,1796–1798.

105 Serak, S., Tabiryan, N., White, T.J. et al. (2010) Soft Matter, 6, 779–783.106 Lee, K.M. and White, T.J. (2012) Macromolecules, 45 (17), 7163–7170.107 Lee, K.M., Smith, M.L., Koerner, H. et al. (2011) Advanced Functional

Materials, 15, 2913.108 Harris, K.D., Cuypers, R., Scheibe, P. et al. (2005) Journal of Materials

Chemistry, 15, 5043–5048.109 Sawa, Y., Ye, F., Urayama, K. et al. (2011) Proceedings of the National

Academy of Sciences of the United States of America, 108 (16), 6364–6368.110 Wie, J.J., Lee, K.M., Ware, T.H., and White, T.J. (2015) Macromolecules, 48

(4), 1087–1092.111 Iamsaard, S., Aßhoff, S.J., Matt, B. et al. (2014) Nature Chemistry, 6 (3),

229–235.112 McConney, M.E., Martinez, A., Tondiglia, V.P. et al. (2013) Advanced

Materials, 25 (41), 5880–5885.113 Oates, W.S. and BIN, J. (2015) Shape Memory Polymers for Aerospace

Applications: Novel Synthesis, Modeling, Characterization and Design,DEStech Publications, Lancaster, PA, p. 183.

114 van Oosten, C.L., Bastiaansen, C.W.M., and Broer, D.J. (2009) NatureMaterials, 8 (8), 677–682.

115 Li, C., Liu, Y., Huang, X., and Jiang, H. (2012) Advanced Functional Mate-rials, 22 (24), 5166–5174.

116 Hiscock, T., Warner, M., and Palffy-Muhoray, P. (2011) Journal of AppliedPhysics, 109 (10), 104506/1–104506/9.

117 Yin, R., Xu, W., Kondo, M. et al. (2009) Journal of Materials Chemistry, 19(20), 3141–3143.

118 Torras, N., Zinoviev, K.E., Camargo, C.J. et al. (2014) Sensors and Actua-tors, A, 208, 104–112.

�

� �

�

References 177

119 Campo, E.M., Roig, J., Roeder, B. et al. (2011) Proceedings of SPIE, 8107,8107, (Nano-Opto-Mechanical Systems), 81070H-1–81070H-10.

120 Camargo, C.J., Torras, N., Campanella, H. et al. (2011) Proceedings of SPIE,8107, 8107, (Nano-Opto-Mechanical Systems), 810709-1–810709-7.

121 Chatani, S., Kloxin, C.J., and Bowman, C.N. (2014) Polymer Chemistry, 5(7), 2187–2201.

�

� �

�

179

6

Photomechanical Effects in Polymer NanocompositesBalaji Panchapakesan1, Farhad Khosravi1, James Loomis2, andEugene M. Terentjev3

1Small Systems Laboratory, Department of Mechanical Engineering, Worcester Polytechnic Institute,Worcester, MA, USA2Department of Mechanical Engineering, University of Auckland, Auckland, New Zealand3Cavendish Laboratory, Department of Physics, University of Cambridge, Cambridge, UK

6.1 Introduction

The ability to convert light into other forms of energy is useful for manyapplications, from actuators to solar cells. In some systems, external stimulican trigger large changes in the internal state of the material, leading toa mechanical response much larger than the initial input [1]. The abilityto unlock this internal work is pivotal for many potential applications. Inthis chapter, we report photon-induced mechanical actuation observedin a polymer–nanotube/graphene composite when exposed to infrared(IR) radiation [1–3]. The polymer composites made of carbon nanotube(CNT)/graphene undergo photomechanical actuation with prestrains. Atsmall strains, the samples exhibit expansion and at large prestrains, thesamples exhibit contraction, when stimulated by near-IR photons. Theamplitudes of expansion and contraction are several orders of magnitudegreater than those for pristine polymer. The behavior is further modeledas a function of orientational ordering of nanotubes, which is induced bythe uniaxial extension. It is believed that no other materials can exhibit thiscontinuously reversing response, especially for such large magnitude, makingnanocomposites suitable for actuator applications.Actuatormaterials change their dimensions upon application of a given stim-

ulus, such as heat, electric voltage, or light, which makes them such attractivesystems to study.The industry has now widely adopted actuators with differingcharacteristics to fill a variety of technological requirements [4]. Some actua-tors have a one-way response,while others based on equilibriumhave reversible


�

� �

�

180 6 Photomechanical Effects in Polymer Nanocomposites

response to the given stimulus. Some (smart) materials display a latent abil-ity to actuate under specific conditions, such as shape memory alloys [5] orliquid crystal (LC) elastomers [6]. Other systems require the blending of two(or more) distinctly different materials to have a physical response leading tothe actuation process. The work presented here focuses on the second class ofmechanical actuation in equilibrium,which employs the use of CNTs/grapheneembedded in a polymer matrix.

6.2 Photomechanical Actuation in Polymer–NanotubeComposites

This section presents a simple polymer composite system (comparing withnematic elastomer composites of [7]) in which using MWNTs as fillers inpolymers produces a large mechanical response to the application of IRirradiation. Remarkably, we obtain both compressive and expansive responsemodes, depending on the external uniaxial strain that is applied to the com-posite sample. We are confident that this behavior depends on the nanotubeorientation within a homogenous polymer matrix.Multiwalled nanotubes (Nanostructured &AmorphousMaterials, Inc.) were

uniformly dispersed in polydimethylsiloxane (PDMS) at concentrations of 0.02,0.5, 1, 4, and 7wt%. A schematic of the apparatus is presented in Figure 6.1(a).For characterization purposes, the mechanical response of our nanocompositewas tested for different loadings of nanotubes in the cross-linked PDMSmatrix.The rubbery network becomes stiffer, and the Young modulus increases by afactor of 2, as the concentration of MWNTs increases from 0 to 4wt% loading.This is expected and compares well with the literature findings [8, 9].The subtlevariations in the measured moduli are perhaps due to the polymer–nanotubeinterface and relaxation of local stresses in the composites.It is imperative to characterize nanotube alignment quantitatively.

Wide-angle X-ray diffraction was used to determine the average nan-otube orientation as a function of the applied uniaxial strain. Figure 6.1(b)shows the characteristic features of the diffraction halos. This example is fora 7wt% sample, initially nonaligned, that is stretched by 𝜀= 0.33 (33%). TheBragg peak around 3.40Å corresponds to the (0 0 2) scattering plane, whichdescribes the intershell spacing periodicity within the multiwall tubes [10].The bright scattering ring corresponding to the length scale of ∼7.5Å is an

interesting feature and is similar to the pristine PDMS rubber prepared in thesame batch. In the pristine PDMS network, with no solvent, the only X-raycontrast arises due to the differences in the cross-link distribution. A clearscattering length is an indication of cross-link density fluctuations (in otherterminology called clustering). As the extensive theory of this phenomenonsuggests [11], at the given chain length and cross-linking density, the PDMS

�

� �

�

6.2 Photomechanical Actuation in Polymer–Nanotube Composites 181

180 270Azimuthal angle

(b) (c)(a)

Inte

nsity (

a.u

.)

3609007.5 Å

3.4 Å

High accuracy

balance

Test strip

Weighted

lower clamp

NIR laser

source

Upper

clamp

Linear

actuator

Manual positioning

stage

Figure 6.1 (a) Schematic of the photomechanical testing apparatus. (b) The X-ray imageshowing key scattering reflexes; the outer ring corresponding to 3.4 Å is the signal fromnanotubes. The inner ring is indicative of the correlation length of mesh size, see text. Thearrow shows the direction of the aligning strain. (c) The typical azimuthal intensity variation,I(𝛽), at a scattering angle corresponding to the outer (MWCNT, 3.4 Å) ring. (Ahir andTerentjev [1]. Reproduced with the permission of Nature Publishing Group.)

network is well below the “cross-link saturation threshold,” and the clusteringcorrelation length should be of the order of the mesh size. The length scaleof ∼7.5Å accurately represents this size, and accordingly, we believe thatthis scattering is a result of small-scale cross-link density fluctuations (forcomparison, similar conditions of scattering from a non-cross-linked PDMSmelt did not show such reflection). These fluctuations should not affect themacroscopic properties or even the local MWNT embedding properties. Asthe applied uniaxial strain increases, the 3.4Å (MWNT) ring develops anincreasing azimuthal bias [I(𝛽) in Figure 6.1c], indicating the orientationalordering of tubes. For instance, at the prestrain value of 𝜀 = 0.6 (=60%), thisinduced order, Q, reaches as high as 0.29. As a comparison, samples that wereaccidentally prestressed during preparation are shown to have a low orienta-tional order parameter, Q≤ 0.005. Hence, composites with no significant initialalignment reached much higher values of induced orientational order uponsubsequent stretching. We assume that this is due to a more rigid networkaround the tubes that attempt to deform affinely, thus imposing significantorientational bias than a loosely cross-linked gel under similar deformation.Furthermore, the change in orientation upon stretching is reversible, that is,equilibrium.The intriguing response of our nanocomposite samples to IR radiation

is shown in Figure 6.2(a). It presents the measured stress for the initiallynonaligned 1wt% samples. At the start of the experiment, a 2% prestrain(𝜀 = 0.02) was applied to the sample and then was allowed to mechanicallyequilibrate. The first data set represents the 2% prestrain line (the lowest curvein Figure 6.2). The plot shows the raw data of measured stress as a functionof time. The initial stress reading is simply the measure of Young’s modulus

�

� �

�


0.5

0.4

0.3

0.2

0.1

00 5 10 15

(a) (b)

(c)

20

0.1

0

IR off

IR on

0 10 20 30 40 50

Applied prestrain (%)

0

–20

–10

0

10

20

30

40

50

Exe

rted s

tress (

kP

a)

10 20 30 40

1 wt%

0.5 wt%

0.02 wt%

50

+2%

Change in

natu

ral le

ngth

+1%

0%

–3%

–2%

–1%

–4%


0.2

0.3

0.4

0.5

2%4%6%8%10%12%14%20%25%30%35%40%

Str

ess (

MP

a)

Time (min)

Figure 6.2 (a) Response to IR radiation at different values of prestrain. (b) Stress versuspre-strain data from Figure 6.2. The circle represents crossover from expansion tocompression. (c) The magnitude (in kilopascal) of exerted actuation stress, as a function ofprestrain, for samples with increasing MWCNT loading. The right y-axis shows thecorresponding actuation stroke: the change in natural length L0 upon IR irradiation. (Ahirand Terentjev [1]. Reproduced with the permission of Nature Publishing Group.)

E ∼ 1.15MPa, presented with open symbols on the plot in Figure 6.2(b). TheIR light source is switched on at a certain time, and the stress reading changes.In the case of the 2% prestrain sample, the change is downward, meaning thatthe sample initial length has expanded upon actuation. The new (IR-on) stressreading is presented on the plot with filled symbol in Figure 6.2(b). After aperiod of constant irradiation, during which the stress reading remains stable,the light source is switched off, and the stress reading returns to its originalvalue. This experiment is then repeated for the same sample at differentvalues of prestraining, up to 40%, as shown in Figure 6.2(a). The stress–strainpoints, with and without IR stimulations, are shown in Figure 6.2(b). Wehave deliberately conducted these experiments in a random sequence ofprestrain/equilibration cycles in order to investigate degradation. The clearlyconsistent trend proves the reversibility of the sample state.

�

� �

�


Figure 6.2(c) summarizes the actuation effect by presenting the magnitudesof stress step in the IR-on state, at different levels of prestrains and for sampleswith increasing MWNT loadings. Although this is not explicitly measured inour (isostrain) experiment, we can directly calculate the change of the origi-nal length of the samples upon actuation. This is shown on the same plot, byexplicitly illustrating the regions of expansion and contraction. Remarkably, allsampleswith different nanotube loadings appear to have a crossover at the samepoint of around 10% prestrain. An increase in the amplitude of this effect withloading is to be expected.There is no significant change in the stress value, as shown in Figure 6.2(a),

with time after the IR source is switched on. This means that the heat transferfrom the bulk face of the irradiated sample plays a marginal role in the mech-anism of mechanical actuation. The effect is highly reproducible over manycycles of irradiation, which clearly shows that no degradation occurs due tononradiative photon decay in the nanocomposite samples. This effect has alsobeen observed in a different polymeric matrix using a cross-linked side-chainpolysiloxane nematic polymer. For comparison, the pristine PDMS rubber inthe same experiment shows a minor stress response, two orders of magnitudesmaller than what is observed in Figure 6.2(a), which was attributed entirely tothe sample temperature change on IR irradiation. The temperature change byIR heating is unavoidable and reachesΔT ∼15 ∘Cmaximally in our setting.Thishighlights an important concept as if the response is due to the photon absorp-tion or the plain heat transfer. Although not presented in detail here, we havestudied the mechanical response purely as a function of temperature change.The temperature results are an order of magnitude smaller than the tempera-ture results in the case of IR stimulation. Hence, it is concluded that such aneffect does exist (i.e., the MWNT-loaded composite has a stronger mechanicalresponse compared to a pristine polymer at the sameΔT), but its magnitude isnegligible with respect to the direct IR-photon absorption mechanism.The change of actuation direction upon increasing sample extension is essen-

tially due to the nanotube alignment. A simple affine model of induced orien-tational order gives the biased probability distribution of tube axes as

P(𝜃) = 𝜆3∕2

(cos2𝜃 + 𝜆3∕2 sin2𝜃)3∕2

(6.1)

with the uniaxial stretching factor being 𝜆 = 1 + 𝜀.This corresponds to the datashown in Figure 6.1(c) and predicts the orientational order at relatively low pre-strains: Q ≈ 3∕5𝜀. Now, if we use the induced orientational bias, and average the(hypothetical) individual nanotube response, at the crossover strain of 𝜀∗ ≈ 0.1,the orientational order can be estimated as Q ∼ 0.06.We hypothesize that this individual response is essentially a contraction,

because this is how the better-aligned composite responds. It is relativelysimple to understand this for an initially rod-like tube, since, upon photon

�

� �

�


absorption, the nanotube may generate kink instabilities that decrease its netlength. Let us simply assume that each nanotube, under IR stimulus, undergoesa contraction by a factor Δ< 1 (proportional to radiation intensity, which ismaintained constant in our work), which is accompanied by a transverselyisotropic volume conserving expansion, 1∕

√Δ. This shows that a local strain

is created with the principal axes along the current nanotube orientation (atangle 𝜃 to the macroscopic z-axis; Figure 6.3a)

Λ(IR) =⎡⎢⎢⎣

1∕√Δ 0 0

0 1∕√Δ 0

0 0 Δ

⎤⎥⎥⎦

(6.2)

The projection of this local strain on the macroscopic axis of extension inthe sample (and force measurement) can be calculated as 𝜆z(IR) = Δcos2𝜃 +(1∕

√Δ)sin2

𝜃. When we average this local contribution using the probabilityof finding the nanotube at this orientation, P(𝜃), it produces an estimate ofthe effective stroke of actuation (𝜆z − 1) or, if multiplied by the correspondingYoung’s modulus, the effective applied stress, as shown in Figure 6.3(b).Such a model is indeed quite crude, since it ignores the effects of continuumelasticity and nanotube morphology. However, it is elastically self-consistentand has only one parameter Δ that carries all the underlying complexity of thenanotube problem in it.The orientational averaging is straightforward:

⟨𝜆2⟩ = ∫𝜋

0

[

Δcos2𝜃 +

(

1√Δ

)

sin2𝜃

]

P(𝜃) 14𝜋

sin 𝜃d𝜃d𝜑

≈ 13

(

Δ + 2√Δ

)

− 25𝜀

(

1√Δ

− Δ

)

(6.3)

Although a full analytic solution can be derived for the integral, it is more infor-mative to present its limit at a small-imposed prestrain, 𝜀. This shows the keyconcept that at low prestrain, 𝜀→ 0, the average actuation stroke of the disor-dered nanocomposite is positive, (𝜆z − 1), that is, the extension of its originallength. However, above the threshold prestrain, 𝜀*, this average deformationbecomes negative, that is, the contraction of its original length. It is rather sim-ple to find the threshold prestrain

𝜀∗ = 5(2 − Δ1∕2 − Δ)

6(1 + Δ1∕2 + Δ)(6.4)

in order to predict the crossover at 𝜀∗ ∼ 0.1, if the nanotube response factor,Δ, is about 0.8, that is, upon IR irradiation, the nanotube itself contracts byabout 20%. The value is higher than that expected, considering the previouslyobtained values of nanotube strains of only 1–2%. However, Figure 6.5(a) indi-cates that our proposition is not that of the lattice strain of nanotube walls but

�

� �

�


θ

Δ

Δ

(λz)

1

√

0.01

0.1 0.2

(a)

(b)

0.3 0.4

Applied prestrain ε0

–0.01

–0.02

–0.03

–0.04

Ave

rage s

troke (

λ z)–1

Figure 6.3 Scheme of local and macroscopic strains and the prediction of the actuation model.(a) Scheme illustrating how the distortion (kinking or undulation) of an individual tube,lying at an angle 𝜃 to the alignment axis, projects on the z-axis to contribute to themacroscopically uniaxial strain, Equation 6.2. (b) The result of theoretical modeling based onorientational averaging of local deformations from each nanotube, Equation 6.3; the dashedline shows the linear approximation at small prestrain 𝜀. Nanotube contraction factor ischosen, Δ = 0.8, as suggested by the crossover strain value 𝜀* ≈ 0.1. (Ahir and Terentjev [1].Reproduced with the permission of Nature Publishing Group.)

�

� �

�


a contortion of the total tube length. Although this has not yet been directlyobserved and reported in the literature, a similar effect of resonant undulationhas been shown (in simulation [12] and in experiment [13]) with applicationsof displacements beyond the linear regime. Although in our system the tubesrespond under totally different conditions where they are embedded in an elas-tic matrix under strain and absorb the IR photons, the overall distortion factorof 20%, suggested by the model, is perhaps quite reasonable.To clarify the points presented earlier, Figure 6.3(b) plots the full (nonex-

panded) result of orientational averaging of actuation stroke (⟨𝜆z⟩ − 1) fromEquation 6.3. The qualitative behavior is almost exactly reproduced, includ-ing the magnitude of the predicted actuation stroke (L0(IR)∕L0(0) − 1). Hence,it is likely that the orientational nature of the effect, with its change of actu-ation direction at a critical level of induced alignment, is captured correctly,while more work is clearly necessary to fully understand the individual nan-otube response to IR radiation that generates the phenomenological factor, Δ,used in this analysis.The strength of the photoactuator response, at a given radiation intensity, is

of the order of tens of kilopascals. In terms of stroke, this corresponds to actu-ation strains of 2–4%. The response expectedly increases at higher nanotubeloadings. The similar (thermal actuation) behavior is also observed when thesamples are heated by the same amount, but with an order of magnitude lowerthan amplitude.Understanding the nature of the actuatormechanisms in this systemcertainly

indicates further theoretical and experimental investigation. Many questionsremain unclear, such as the response of an individual nanotube, embedded ina polymer matrix, to IR photons. It is also not clear about the effect differenttypes of nanotubes would have, that is, smaller multiwall diameters, single-walltubes, and so on. Future investigations should also address the issue of hostmatrix and confirm its relatively negligible role in the actuation mechanism.With actuating materials already used in such widespread applications,

from micromanipulators to vibration control, the discovery of a structure thatcan respond to stimulations in both directions will open new possibilitiesand mean an important new step toward finding interesting applications fornanotube-based materials above and beyond improvements in existing carbonfiber technologies.

6.3 Fast Relaxation of Carbon Nanotubes in PolymerComposite Actuators

In this section, the relaxation of CNT polymer composites when stimulatedwith IR photons is presented [14]. The rate of the stimulated response isfaster than Debye relaxation, instead following a compressed-exponential law.

�

� �

�

6.3 Fast Relaxation of Carbon Nanotubes in Polymer Composite Actuators 187

However, the relaxation after turning off the light source follows the simpleexponential relaxation, as does the stimulated response at low nanotubeconcentration. In the following, we discuss possible models and hypotheses toexplain the fast photomechanical response. In this section, we report exper-imental investigations of kinetics response and relaxation of such polymernanocomposites under near-infrared (NIR) illumination. We also show thatthis response is significantly faster than the canonical Debye (exponential)relaxation. Meanwhile, the light-off relaxation is essentially a classic simpleexponential process.In order to study the kinetics of this response, and the relaxation with and

without the light, we shall examine the normalized stress ratio, Δ𝜎∕𝜎max, asillustrated in Figure 6.4. Figure 6.4(a) plots the normalized stress of 3wt%PDMS nanocomposite for a range of prestrains, 𝜀. The time axis is shiftedso that the photomechanical response starts at t = 0 s (light on) and reachessaturation at t ∼ 10–15 s. Actuation becomes marginally quicker as prestrain(and tube orientational ordering) increases. However, the different data areclose to each other in spite of the large difference in actual response, forexample, between 2% and 40% strains. The simultaneously measured changein sample temperature (also normalized, ΔT/Tmax) is shown on the same plotto emphasize the differences in the response rate.The behavior is indeed repeatable for all nanotube–polymer concentrations,

as the results shown in Figure 6.4(b) clearly demonstrate. Hence, using𝜀 = 20% for all samples, we ensure that the nanotubes are relatively wellaligned in the soft cross-linked elastomer matrix. For reference, Figure 6.4(b)also depicts the results for the pristine PDMS sample, (as expected, nophotomechanical response is observed) and the nanocomposite with lowtube concentration of about 0.02wt%. The notably slower response of thissample is in clear contrast to all other nanocomposites. This discrepancy willhinder a clear explanation for the observed effects. Apart from the lowestconcentration sample, the data shown in Figure 6.4(b) strongly suggest thatthe photomechanical actuation kinetics remains independent of nanotubealignment and concentration (above the percolation threshold).We fit the data with a compressed-exponential function 1 − exp[−(t∕𝜏)𝛽] to

examine the time dependence of the photoresponse effect.The quality of this fitand the important comparison with the classic exponential behavior are pre-sented in Figure 6.4(c). The two fitting parameters are the relaxation time, 𝜏≈ 5 s, and the exponential exponent, 𝛽 ≈ 2 s. These values were computed tobe nearly the same for all aligned composites with nanotube concentrationsabove the percolating threshold. Now, we would like to focus on themain effectand disregard a weak dependence of 𝜏 and 𝛽 on the applied prestrain. Such afast response of the system is a remarkable result and the main focus of thissection. One must appreciate that the individual photomechanical response ofa freestanding nanotube must proceed within a nanosecond timescale, if one

�

�

1

0.8

0.6

0.4

0.2

–5 0 5 10 15 20 25 30

0

0.2N

orm

aliz

ed te

mpera

ture

0.4

0.6

0.8

1

0

0.2

Norm

aliz

ed te

mpera

ture

0.4

0.6

0.8

1

2%

15%30%

40%

50%

Time (s)

(a) (b)

(d) (e)

(c)

Time (s)

–5 0 5 10 15

PDMS0.5 wt%1 wt%2 wt%3 wt%7 wt%

0.02 wt%

20

Time (s)

50 10 15 20

Time (s)

Norm

aliz

ed s

tress Δ

σ/σ m

ax

Norm

aliz

ed s

tress Δ

σ/σ m

ax

Δσ/σmax

Δσ/σmax

0

1

0.8

0.6

0.4

0.2

Norm

aliz

ed s

tress Δ

σ/σ m

ax

0

1

0.8

0.6

0.4

0.2

0

ε

2%

1

0.8

0.6

0.4

0.2

0

4%20%30%40%50%

0 10 20 03 04 50 60

Time (s)

1

0.8

0.6

0.4

0.2

00 10 20 30

0.02 wt% tubes

3 wt% C. Black

40 50 60

ε

Figure 6.4 (a) Normalized stress, versus time, which allows comparison of the response kinetics: The light-on response of 3 wt% composite atdifferent values of prestrain. The right y-axis shows the simultaneously measured, similarly normalized, change in temperature upon irradiation.(b) The light-on response of different composites, all measured at the same 20% prestrain. (c) Illustration of the data fit, for 3 wt% composite at20% prestrain. Experimental data is fitted by the compressed exponential (solid line) and the simple exponential (dashed line) functions todemonstrate the discrepancy. (d) The normalized stress relaxation of a 3 wt% nanocomposite illuminated at different prestrain, when the lightsource is switched off. (e) The light-on response of the composite with very low tube loading and also that of a sample with 3 wt% carbon black,both at 20% prestrains. (Ahir and Terentjev [1]. Reproduced with the permission of Nature Publishing Group.)

�

� �

�

6.3 Fast Relaxation of Carbon Nanotubes in Polymer Composite Actuators 189

assumes polaron excitation and relaxation [15]. The kinetics at the scale ofour observations is certainly due to the constraints of the rubbery matrix. Thepolymer would usually be expected to follow the classic Debye relaxation (𝛽= 1), if not slower, due to the mode coupling and viscoelasticity. This is notthe case in our experiments with a compressed exponent 𝛽 = 2 and a char-acteristic timescale of about 13 s. Moreover, the fast cooperative response isreproduced in both expansive (unaligned) and contractive (aligned) modes ofphotoactuation, suggesting a unique underlying mechanism for the bimodalphotomechanical actuation.When the light source is switched off, Figure 6.4(d), all the nanocompos-

ite materials in our range relax normally, following the classic e-t∕𝜏 law with𝜏 ≈ 5 s. The same normalized kinetics of the light-off relaxation is obtainedat all different values of prestrain, 𝜀. As a more detailed comparison to thefast light-on response illustrated, the plot in Figure 6.4(e) shows results froman identical experiment conducted on PDMS samples with trace amounts ofnanotubes (0.02wt%) and also with 3wt% carbon black. The response is evi-dently much slower in this case. Importantly, these curves superpose and alsofollow a simple exponential fit, 1 − e−t∕𝜏 with 𝜏 ∼ 10 s. Evidently, for the fasterresponse to take place, nanotube (and not carbon black) concentration needsto remain above the percolating threshold.In addition to the ideas based on the electronic structure of nanotubes, there

is a possibility to account for their large local deformation in a polymer matrix.A large (and rather fast) local tube heating is inevitable upon photon absorp-tion. In fact, there are reports of such an effect [16, 17], presumably based onthe incomplete reradiation of the absorbed energy. Assuming that the poly-mer chains are highly aligned in the vicinity of nanotubes due to the surfaceboundary anchoring, the local heating must generate local contracting strainalong the alignment axis. This is a classic thermodynamic effect of the uniaxialcontraction of a stretched rubber. Such a local strain can lead to Euler buck-ling instability of a rigid nanotube embedded in the elastic matrix, which canaccount for many aspects of the photoactuation.In order to comprehend the dynamics of such a response, assume that

the relaxation process is controlled by the overdamped balance of an elasticforce against viscous friction. To understand the fast response, one must takethe observed time dependence x ∼ exp[−𝛼t2], where x(t) is the relevant strainvariable and work backward to isolate the nature of the involved forces. Takingln(x)= −𝛼t2 and differentiating, one obtains the “kinetic equation” in theform of x = −(2𝛼t)x. The effective relaxation time is then defined as the ratioof the shear modulus, G, to the viscous coefficient 𝜂, from the force balancerelationship, Gx + 𝜂x = 0. In order to generate the compressed exponentialbehavior, this ratio [G/𝜂] must be a linear function of time, from the momentthe light was applied.

�

� �

�


Upon sudden local heating, the equilibrium balance between the chainalignment and the boundary conditions of the tube surface is distorted. Thisis because the entropy cost for chain stretching increases, which results in auniaxial contracting force exerted on the tube along its axis. The magnitudeof this force, in the leading order, is a linear function of the local temperatureincrease ΔT = T(t) − T0. If the temperature increases, then the contractingforce would also increase as a function of time (initially linearly with time).For small increments, at t → 0, we can write G = g0t, and the kinetic equationbecomes x = −[g0/𝜂]tx, exactly reproducing the results of our observations,with the effective relaxation time of 𝜏 = 𝜂/g0t. There are indeed many com-plications to this simple model. The real viscoelasticity of a polymeric systemwould make all of these arguments more complex. However, in the leadingorder, we would still expect to see the contraction dominated by the linear (ornear-linear) time dependence of the local rubber modulus.The fast compressed-exponential response was not observed in the light-off

relaxation, which agrees with our basic hypothesis. After the illuminationperiod, the temperature equilibrates throughout the sample, giving the averagemeasured temperature. The new balance of forces is reached and maintainedby the steady flux of heat from the irradiated tubes. When the light is turnedoff, both the viscosity and the modulus remain roughly constant (only weaklydependent on time), resulting in the simple Debye relaxation toward theoriginal local conformation of the elastomer, which was established at thecross-linking regime.The prescribed explanation based on the sharp local heating of nanotubes

captures many key features of our findings but also has some deficiencies. Thelight-on compressed-exponential response was not observed in two cases: (i) atconcentrations below overlap 𝜙c and (ii) at 𝜀 = 𝜀c, around the transition fromcompressive to contractive actuation. In (i), the kinetics could be dominatedby the bulk isotropic matrix between sparsely distributed nanotubes, while in(ii), the tubes of different orientations compensate each other’s local action ineffect, including the principal relaxation modes.Several phenomena, including the photoinduced polaron excitations con-

centrated near the tube defects or photogenerated charge redistribution, maywell coexist during irradiation and differentiating. Hence, detecting the domi-nant mechanism is not experimentally trivial. Electron microscopy techniquesmight be inherently unsuitable due to flooding of the π-conjugated tubes withelectrons. A definitive measurement of individual nanotube photoresponsewould involve irradiating a tube in a setup similar to single-chain atomic forcemicroscope studies [18].In conclusion, we have shown that elastomers filled with nanotubes respond

to light much faster than what classic relaxation predicts, following a universalcompressed-exponential law once above the percolation threshold. Thefavored explanation considers nanotubes as photon absorbers that locally

�

� �

�

6.4 Highly Oriented Nanotubes for Photomechanical Response 191

redistribute the energy, as the heat causing contraction of anisotropic polymerchains aligned near the nanotube walls. This demonstrates how nanotubes cancreate photomechanical properties in otherwise benign materials; the role ofthe nanotube–polymer interface is of utmost importance, and the rate of thephotoactuation response requires much further experimental and theoreticalinvestigation.

6.4 Highly Oriented Nanotubes for PhotomechanicalResponse and Flexible Energy Conversion

In this section, we present the photomechanical response of highly orientednanotubes inside elastomeric matrix. Elastomeric composites based onnanotube LCs that preserve the internal orientation of nanotubes have thepotential to create nanocomposites with anisotropic physical propertiesand flexible energy conversion. In this section, we demonstrate unique andreversible photomechanical response of this layered composite to excitationby NIR light at low nanotube mass fractions, using simple vacuum filtrationtechnique of fabricating nanotube LC films and utilizing a transfer process topoly(dimethyl)siloxane, wherein the LC arrangement is preserved. Upon exci-tation by NIR photons, with application of small or large prestrains, significantexpansion or contraction, respectively, of the sample occurs, that is continu-ously reversible and three orders of magnitude larger than in those observedin pristine polymer. Schlieren textures were noted in these LC composites,confirming long-range macroscopic nematic order of nanotubes within thecomposites. Order parameters of LC films were about Soptical = 0.51–0.58from dichroic measurements. Film concentrations, elastic modulus, andphotomechanical stress were all related to the nematic order parameter. Thephotomechanical stress was almost three times larger for the self-assembledLC nanotube actuator with similar nanotube concentration compared to thestress in actuator based on randomly oriented CNTs. Investigation into thekinetics of photomechanical actuation showed variation in the stretchingexponent 𝛽 with prestrains and concentration of nanotubes. Maximum pho-tomechanical stress of ∼0.5MPa/W and energy conversion of ∼0.085% wereachieved.The combination of properties, namely optical anisotropy, reversiblemechanical response to NIR excitation, and flexible energy conversion, allin one system makes nanotube LC elastomers an exciting material for softphotochromic actuation, energy conversion, and photo-origami applications.

6.4.1 Highly Oriented Nanotubes/Nanotube Liquid Crystals

Materials that flow as liquids and can order themselves macroscopically ascrystals are called LCs and hold great technological and commercialization

�

� �

�


potential [19]. LCs can also be found in nature, such as tobacco mosaic virus[20], proteins [21], and cells [12, 22]. Modern-day applications of LCs includepolymers such as Kevlar for bulletproof vest [13], and electro-optics in digitaland computer displays [23, 24].The uniqueness of LC is their tendency to alignin specific directions with macroscopic and long-range ordering. In recentyears, with the synthesis of nanotube LCs by Windle et al. [25], arrangementof nanotubes as LCs has become an interesting and attractive topic of studyfor the possibility of low-cost commercial applications based on self-alignednanotubes. LC nematic self-assembly of nanotubes, as well as graphene andother 2-D nanomaterials, presents interesting opportunities in developingmacroscopic nanocomposites with long-range order and unique anisotropicproperties. Energy-efficient photomechanical systems based on nanotubeLC-elastomers that combine anisotropic optical and thermal properties of thenanotube LCs and elasticity of the polymer network are yet to be explored.We have recently reported on LC of CNT films using simple vacuum filtrationand their subsequent applications as high-performance transistors [26]. Thinfilms of nanotube LCs with order parameters ranging from S = 0.1 to 0.5were successfully patterned into conducting channels of transistor devicesthat showed high on/off ratios of about 20,000, electron mobility values upto μe = 79 cm2/V s, and hole mobility values up to μh = 287 cm2/V s [26].Herein, we demonstrate elastomeric composites based on small amountsof single-wall nanotube (SWNT) LC films in PDMS with high orientationalorder, optical anisotropy, and reversible macroscopic mechanical response toNIR excitation. Further, we show strain-dependent flexible energy conversionbased on change in themicroscopic order parameter of nanotubes and stress tonanotube mass ratios, which are larger than all the nanotube/graphene-basednanocomposite-based light-driven actuators reported to date [1, 2, 16]. Theamount of nanotube used in this work is also ∼10,000 times smaller thanthat used in the previously reported electromechanical actuators based onnanotube–polymer composites [27], suggesting the importance of sponta-neous nanotube order for low-cost commercial applications. Further, themethods presented here may enable standardization of nanotube–compositefabrication processes, which are desperately needed for commercialization.Figure 6.5(A) presents the schematic of SWNTs LCs and the resulting

photomechanical composite actuator. Vacuum filtration technique was usedto obtain the nanotube LC films. The nanotube–surfactant solution, whenfiltered through the membrane, creates a concentration gradient as a resultof the change in fluid velocity across the membrane [26]. As the solutionis filtered off, and the concentration of the nanotube increased, nematicdomains form. As schematically presented in Figure 6.5(A), nanotube LCsare formed on an anodisc filter membrane. The LCs from the membrane aresubsequently transferred to a PDMS surface that is spin-coated on a glassslide. The membrane is then gently peeled off of the PDMS surface, leaving the

�

� �

�


PDMS

DispenseAlumina

membrane

Glass slide

(A)

(B)

LC-CNT

LC-CNT on

membrane

Fabrication of LC-CNT photomechanical actuator

Membrane

peeling1

st PDMS layer

1st PDMS layer 2

nd PDMS layer

Vacuum

Vacuum

3 4

21

SEM image of LC-CNT

0.1 0.2

0.40

0.50

0.60

Ord

er

para

mete

r, S

optical

0.3 0.4 0.5CNT concentration (μg/ml)

(g) (f)

Figure 6.5 Nanotube liquid crystal elastomer composite. (A) (1) Vacuum filtration is used todeposit carbon nanotubes (CNTs) onto an inorganic filter membrane. (2) PDMS isspin-coated on top of a glass slide. (3) The membrane consisting of LCs is pressed againstthe PDMS that resulted in complete transfer. (4) A second PDMS layer is spin-coated andpolymerized to enable LC being part of the polymeric network resulting in “Nanotube LCElastomer.” (B) SEM images of LC-CNTs: (a-1)–(a-3)): 0.01 μg/ml; (b-1)–(b-3)): 0.05 μg/ml;(c-1)–(c-3)): 0.1 μg/ml; (d-1)–(d-3)): 0.3 μg/ml; (e-1)–(e-3)): 0.5 μg/ml; Scale bars: Row 1:(a-1)–(e-1): 10 μm; Row 2: (a-2)–(e-2): 1 μm; Row 3: (a-3)–(e-3): 200 nm, (g) order parameterversus concentration and (f ) magnified image of (e-3). (C) Order parameter: Linearcorrelation between spatial frequency and optical order parameter. (D) Schlieren texturesand domain size analysis: Schlieren textures of nanotube LCs: Rotation of the polarizer by2.5∘ (92.5∘) resulted in enhanced contrast and better imaging of the Schlieren textures anddomain walls suggesting long-range order. Scale bar: 2 mm. (E) Domain size measurementsas a function of concentration inside the LC–polymer composites: (a) ∼0.01 μg/ml;(b) 0.05 μg/ml; (c) ∼0.1 μg/ml; (d) 0.3 μg/ml; (e) ∼0.5 μg/ml; (f ) Average domain size versusCNT concentration showing almost twice the decrease in domain size with increasingconcentration. Line is shown for eye guidance only. (Ahir and Terentjev [1]. Reproduced withthe permission of Nature Publishing Group.)(Ahir and Terentjev [1]. Reproduced with thepermission of Nature Publishing Group.)

�

� �

�


0.50

0.45

0.40

0.35

0.20

(E)(a) (b) (c)

(d) (e) (f)

(D)(C)

SF

FT

0.24

260 domains/mm2

8308 domains/mm2

14,367 domains/mm2

846 domains/mm2

1856 domains/mm2

6

5

4

3

210

0

0

100

200

300

20

30

4070

60

50

40

30

20

101

050 100 150

Domain size (μm2)

50 100 150

Domain size (μm2)

50 100 150

Domain size (μm2)

50

700

600

500

400

300

200

100

0100 150 0.1

A = –6.705*ln(Ci)+19.067

25

30

35

40

45

50

0.2 0.3 0.4 0.5

Domain size (μm2)

Ave

rage d

om

ain

siz

e (

μm2)

CNT concentration, Ci (μg/ml)

50 100 150

Domain size (μm2)

Dom

ain

count

Dom

ain

count

Dom

ain

count

Dom

ain

count

Dom

ain

count

0.28

SFFT

= 1.49 • Soptical

+ 0.01

r2 = 0.907

0.32 0.36

Soptical


oriented nanotubes on the surface. A second PDMS layer is then spin-coatedon the surface and polymerized to preserve the internal orientation of thefilm. This method also removes the difficulties arising in nanotube dispersionand fabrication process such as shear mixing [2], evaporative cross-linking[28, 29], and functionalization in acids [30], which are all challenging and canaffect the overall mechanical properties of the composites. Further, the lackof standards in the preparation of CNT-based nanocomposites makes themcurrently prohibitively expensive and hampers their commercialization [31].Composites based on nanotube LCs may become commercially viable becauseof the self-assembly of nanotubes as LCs with high anisotropic propertiesand order parameters. Anodisc alumina filter was used here due to the lowinteraction energy between SWNTs and the porous alumina surface, whichenables the film to be completely transferred from the alumina filter surface

�

� �

�


to PDMS (PDMS has a low surface energy of ∼19.8mJ/m2) [32]. While suchfilms have been used in the past to transfer random nanotube networks [32],we have used this technique for transferring the nanotube LC films and forpreserving the internal orientation of nanotubes in film.The full transfer of thefilm as shown is essential for preserving the LC state of the film to accomplishthe anisotropic properties. Since films of different concentrations also haveslightly different order parameters, above the isotropic to nematic transition,the method used here can result in composites with specific order parametersbased on the film concentrations and, therefore, specific properties. This maypotentially be helpful in standardizing a nanotube-based composite fabricationprocess.Figure 6.5(B) presents the SEM images of the nanotube LC films.Three rows

of images show the evolution of the microstructure at different concentrations.The columns represent the same concentration at different length scales. It isobserved that at lowmagnification (Figure 6.5(B)-a1), the lowest concentrationfilm (∼0.1 μg/ml) consisted of sets of nucleated nematic islands that wereloosely connected by few isotropic nanotubes in between. This is a charac-teristic two-phase behavior of a lyotropic nematic LC [19]. The minimumconcentration required for the formation of nematically ordered LC domainswas about 0.075 μg/ml [33]. Below this concentration, the films were purelyisotropic [26]. As the concentration of the nanotube in solution increased,nucleated nematic regions grow as it can be seen with the increase number ofislands in Figure 6.5(B)-b1 and c1. In Figure 6.5(B)-d1, the nucleated nematicislands become larger and close the gap between the adjacent islands. Finally,as shown in Figure 6.5(B)-e1, the films are continuous, bridge all the gaps, andform large nematic domains as in refining their own structure/self-healing toachieve final film morphology. Row 2 and Row 3 are the images at 1 μm and200 nm scales, respectively. In Row 2, the nematic-like LC texture of the filmis clearly observed with nanotubes oriented along a specific nematic director.It is clear that ±1/2 disinclinations are formed, confirming the topology ofthe nematic phase. In Row 3, it can be observed that the morphology of allfilms looks similar after the isotropic-nematic transition. For clarity, one ofthe images is enlarged to present the ordered arrangement of nanotubes.The inset in the image shows the concentration dependence of the localorder parameters above the isotropic–nematic transition. Some pores exist inbetween the bundles across all concentrations, and there is a twisting pattern ofindividual nanotubes due to rotation of the nanotube.The competing scenariosbetween translational and rotational entropies of nanotubes, thus, determinethe LC texture and order parameters. It is also seen that the orientation of thenanotubes during vacuum filtration occurs in bundles and not individually.This is due to the intertube attraction between the nanotubes. These bundlesare 10–20 nm in diameter. In Row 3, it is observed that irrespective of theconcentration at nanometer-length scales, most films have similar orientation

�

� �

�


in their nematic domains with slight variations in the order parameters. AnyLC anisotropy can be defined by its order parameter. In two dimensions, theorder parameter is given by

S = ⟨2cos2𝜃 − 1⟩ (6.5)

where the brackets denote the ensemble average over all angles. S, is, there-fore, at a maximum of 1 when all CNTs are aligned in the direction of 𝜃 and aminimum of −1 when all CNTs are perpendicular to 𝜃. Order parameter wascalculated for individual domains from the SEM images with the aid of 2-DFourier analysis similarly to the work by Bayan et al. for collagen fiber orienta-tion [33].These yielded values of S = 0.77–0.88 for the different LC films.Thesewere also compared to the values for randomly oriented films, which yieldedaverage order parameters of S = 0.06–0.24 (∼4–13 times smaller) for the sameconcentration.While the order parameter is rather simple to calculate from SEM images

using the 2-D FFT analysis, it is important to validate the results by measuringthe order parameters using polarization optical microscopy [34]. Hence,the order parameter of the nanotube LCs was verified using polarizationmicroscopy [34]. Films were transferred to a glass sample, and the orderparameter of the films was calculated using both FFT analysis and polarizationmicroscopy. The order parameter using polarized microscopy was evaluatedby the use of dichroic ratio Δ given as the ratio of absorbance that is paralleland perpendicular to the director. The absorbance is measured in parallel (A||)and perpendicular (A⟂) configurations using the Lambert–Beer law:

A = logIo − IdarkI − Idark

(6.6)

where Io is the intensity of light without any sample, I is the intensity of lightwith nanotube on the polymer, and Idark is the intensity with light blocked.Thedichroic ratio was calculated using the equation

Δ = A||A⟂

(6.7)

The order parameter is then given by the equation

Soptical =Δ − 12 + Δ

(6.8)

Figure 6.5(C) depicts the linear correlation between the order parametersfrom both the methods. The order parameters measured using polarizationmicroscopy was smaller than the order parameters measured using 2-D FFTby a factor of about 1.5. A linear relation between order parameters from bothtechniques was established:

SFFT = 1.49 × Soptical + 0.01 (6.9)

�

� �

�


The FFT technique although produced a higher order parameter is simpler toquantify based on SEM images of the nanotube LCs without additional exper-iments. This relationship is useful in calculating future order parameters ofnanotube films for their alignment, since one can obtain a realistic estimateof order parameter without resorting to transferring films onto glass slidesand performing additional polarization microscopy experiments. All the datareported from hereon indicate the corrected order parameter based on theoptical measurements. Thus, the aforementioned order parameter using FFTfor LC films can be corrected to Soptical = 0.51–0.58, and for randomly orientedfilms, Soptical = 0.04–0.16.Figure 6.5(D) presents the Schlieren textures of nanotube LC films suggesting

nematic orientational order in the macroscopic composite. At 2.5∘ rotation ofthe polarizer, due to enhanced contrast, the domain walls are visible whereeach domain aligned along a specific director. Measuring the domain sizeas a function of concentration yielded some interesting results, which arepresented in Figure 6.5(E)-a–f. The domain counts and size were calculatedusing the particle analysis function in NIH Image J software for the polymercomposites from the binary images [35]. The sizes of these domains wereanywhere from 1 to 150 μm2. It can be seen that with increase in concentration,the domain size decreases significantly. The number of domains per squaremillimeter (∼5–10 μm2 size) is seen to increase from 260 domains/mm2 at∼0.01 μg/ml to 14,367 domains/mm2 at ∼0.5 μg/ml. As the concentrationincreases, the film also spreads over a large area. However, this makes thedomain size smaller, suggesting a large number of directors for the formationof LCs in subsequent layers. As the nanotubes arrange in different layers,some of these nanotubes spontaneously become directors for the formationof individual domain, thereby making the process more localized, resulting insmaller domains as the film spreads over a large area.

6.4.2 Photomechanical Actuation of Oriented Nanotube Composites

Photomechanical responses of nanotube LC–polymer composites are pre-sented in Figure 6.6(A)-a–f. In Figure 6.6(A)-a, starting with a plain PDMSelastomer (Figure 6.6A-a) and progressing from ∼0.01 to ∼0.5 μg/ml con-centrations of nanotube–LC/PDMS composites (Figure 6.6A-f ), each plotshows the photomechanical response to ∼808 nm NIR illumination for fiveconsecutive cycles, each one being 60 s. Since the optical loss of PDMS in theNIR region is less than 0.5 dB/cm [36], negligible/zero response in the plainPDMS sample (Figure 6.6A-a) was expected. However, by a concentration ofabout 0.01 μg/ml, the photomechanical effect is clearly observable throughexpansion and contraction of the actuator. The inset in Figure 6.6(A)-b clearlyshows the expansion for low prestrains and contraction for moderate-to-highprestrains. All the composites with low prestrain values resulted in positive

�

�

(a) (b)

(e) (f) (g)

(c) (d)

10 Pre-strain

50%45%40%35%30%25%20%15%11%9%7%5%3%

300

Negligible stress

Plain PDMS 0.01 μg/ml

0.3 μg/ml 0.5 μg/ml

0.05 μg/ml 0.1 μg/ml

400

Time (s)

2001000 300 400

Time (s)

200

0

–0.20

–0.10–0.5

–1.0–1.5

–2.0

0.00

Ch

an

ge

in

str

ess (

kP

a)

20 40 60Time (s)

80100 0

0.0

0.5

Ch

an

ge

in

str

ess (

kP

a)

20 40 60Time (s)

80100

1000

300 400

Time (s)

2001000

300 400

Time (s)

200100

Order parameter vs stress

0 300 400

Time (s)

2001000

–25

–20

–15

Change in s

tress (

kP

a)

–10

–5

0

5

10

–25

–20

–15

Change in s

tress (

kP

a)

–10

–5

0

5

10

–25

–20

–15

Change in s

tress (

kP

a)

–10

–5

0

5

10

–25

–20

–15

Change in s

tress (

kP

a)

–10

–5

0

5

10

300 400 0.40

Photo

mechanic

al str

ess (

kP

a)

0.50 0.60

Order parameter, SopticalTime (s)

2001000

–25

–20

–15

Change in s

tress (

kP

a)

–10

–5

0

5

10

80 100

(b-1)

(a-1) (a-2)

(b-2)

Time (s)

6040200

–25

–20

–15

Ch

an

ge

in

str

ess (

kP

a)

–10

–5

0

5

10

80 100

Time (s)

6040200

–25

–20

–15C

ha

ng

e in

str

ess (

kP

a)

–10

–5

0

5

10

0

5

10

15

20

25

30

35

40

5

0

–5

–10

–15

–20

–25

Changed in s

tress (

kP

a)

(A) (B)

Figure 6.6 (A) Photomechanical responses of LC-CNT polymer composites. Prestrains from 3% to 50% were applied before NIR excitation: (a) plainPDMS; (b) 0.01 μg/ml; (c) 0.05 μg/ml; (d) 0.1 μg/ml; (e) 0.3 μg/ml; (f ) 0.5 μg/ml; (g) photomechanical response versus order parameterdemonstrating increased ordering leading to improved mechanical response of the composites. (B) Disordered versus ordered systems. (a-1) SEMimage of randomly oriented film (∼0.5 μg/ml concentration); (a-2) SEM image of LC nanotube film (∼0.5 μg/ml concentration with orderparameter S=0.6); (b-1) photomechanical stress change for randomly oriented film based actuator; (c-1) photomechanical stress change forLC-film-based actuator. (C) Kinetics of photomechanical actuation in nanotube liquid crystal elastomer. (a) Actuation kinetics; (b) relaxation kinetics;(c) variation of stretching exponent for both actuation and relaxation as a function of concentration of nanotube liquid crystals in elastomer;(d) variation of stretching exponent with prestrains. (D) Efficiencies of nanotube LC–polymer composites. (a) Optomechanical conversion factorversus concentration; (b) energy conversion efficiency versus concentration at different prestrains. (Fan et al. [3]. Reproduced with the permissionof IOP Science.)

�

�

(b)

(d)(c)

(b)(a)(a)

80

Actuation

Actuation

Energy conversion factor, ηM

ηM

= 0.45–0.49 exp(–9.23 Ci)

Conversion efficiency vs LC-CT concentration

Relaxation

Relaxation

kWW fittingkWW fitting

50403020100

Time (s)

60

0.1 0.2 0.3 10


20

0.8

Str

etc

hin

g e

xp

on

en

t (β

)

0.7

0.9

1.0

1.1

1.2

30 400.5

Co

nve

rsio

n e

ffic

ien

cy (

%)

–40

–30

–20

–10

0

10

Contraction

ExpansionPrestrain

100×10–4

0.1 0.2 0.3 0.4 0.5

20%

9%7%5%3%

30%

40%50%

0.40.30.20.1

Op

tica

l-m

ech

an

ica

l co

nve

rsio

n fa

cto

r

(ηM, M

Pa

•W

–1)

0.0

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

50

CNT concentration (μg/ml) CNT concentration (μg/ml) CNT concentration (μg/ml)0.4 0.5

0.6

0.7

0.8

0.9

1.0

1.1

1.2

τ = 5.40

τ = 5.21

τ = 7.11

τ = 6.88 β = 1.40

β = 0.96

β = 0.90

β = 0.91

4020

–1.0

–0.5

0.0

0.5

1.0

–1.0

–0.5

0.0

0.5

1.0

No

rma

lize

d c

ha

ng

e in

str

ess (

Δσ)

0

Time (s)

Expansion

region

Response in 3 %

prestrain

Contraction

region

Response in

40% prestrain

τ )Δσ = 1–exp [–(t β

]

τ )Δσ = 1–exp [–(t β

]

(C) (D)


�

� �

�


film expansion and thus positive-induced stress. Meanwhile, high prestrainvalues resulted in contraction (negative thermal expansion) and, thus, negativechange in stress. More concisely, weakly stretched composites show reversibleexpansion, while highly stretched composites show reversible contraction,which is an indication of rubbery elasticity. At ∼9% prestrain, the samplesexhibited zero stress or no photomechanical actuation [1]. This crossoverfrom small positive expansion to large negative expansion suggests rubberyelasticity at the thermoelastic inversion point [1]. The magnitude of the pho-tomechanical responsewas negligible (nomovement) for plain PDMS,+0.10 to−0.25 kPa for ∼0.01 μg/ml, +0.7 to –2.2 kPa for ∼0.05 μg/ml, +2.8 to −7.2 kPafor ∼0.1 μg/ml, +5.5 to −14.7 kPa for ∼0.3 μg/ml, and +8.0 to −22.8 kPa for∼0.5 μg/ml concentrations of nanotube LCs. Each plot in Figure 6.6(A) alsoshows the entire five-cycle response and shows the reproducibility from eachcycle to the next. Such photomechanical actuators have operated continuouslyin our laboratory for more than 3000 cycles without degradation [28].One interesting issue that naturally arises is the effect of nanotube ordering

on the photomechanical response. In order to investigate this, films with exactconcentrations (∼0.5 μg/ml) with similar nanotube purity were processed intoboth randomly oriented films (Soptical = 0.16) and surfactant-processed LCfilms (Soptical = 0.58). Wemaintained the thickness of the sample the same aftervacuum filtration and film transfer. Figure 6.6(B)–(a-1) and (a-2) present theSEM images of the randomly oriented and LC, respectively. The difference inthe morphologies can be easily seen with ordered arrangement of nanotubeshown in Figure 6.6(B)–(a-2). Subsequent testing of photomechanical responsesuggested that an almost three times smaller photomechanical response forrandomly oriented sample compared to the LC sample, Figure 6.6(B)–(b-1) and(b-2). This unambiguous result suggests that at a specific concentration, thephotomechanical response, the kinetics, and the energy transduction dependon the order of the nanotube in film. This may suggest that the overall pho-tomechanical response may be the sum of the individual nanotube–polymerresponses around the light spot. Using high-quality randomly oriented filmsonly resulted in lower response, suggesting that alignment is crucial forharvesting photomechanical properties.

6.4.3 Relaxation Behavior of Nanotube–Liquid Crystal Elastomers

Our studies have shown that nanotube LC elastomer composites demon-strated actuation kinetics that was fitted to the Kohlrausch–Williams–Watts(KWW) function for actuation Δ𝜎actuate(t) = 1−exp[−(t∕𝜏)𝛽] and relaxationΔ𝜎relax(t) = exp[−(t∕𝜏)𝛽] [3, 37]. Figure 6.6(C)-a and b presents the actuationand relaxation kinetics fitted to the KWW functions for expansion andcontraction, respectively. It is worthy to note that the time constants 𝜏 = 7 s foractuation and 𝜏 = 5 s for relaxation were calculated with stretching exponents

�

� �

�


𝛽actuation = 0.91 and 𝛽relaxation = 1.04. The stretched exponential functioncontains just two free parameters: the relaxation time 𝜏 and the fractional“stretching” exponent 𝛽, which satisfies 0<𝛽 ≤ 1. The upper limit of 𝛽 = 1corresponds to simple exponential decay or Debye relaxation, while lowervalues of 𝛽 are indicative of a more complicated nonexponential relaxationprocess or viscoelasticity [38]. The results here also suggest that upon NIRexcitation, heating the nanotubes and subsequent movement of polymerchains are highly complex processes. While 𝜏 is a material-sensitive param-eter, we studied how 𝛽 varies with nanotube concentration and prestrainsto investigate the topographical origin of 𝛽. We assumed that it would beconstant and independent of the nanotube concentrations and prestrains.Figure 6.6(C)-c and d presents the stretching exponent 𝛽 as a function ofnanotube concentration and prestrains, respectively. It can be seen that thestretching exponent 𝛽 is almost constant (𝛽 < 1) with increase in nanotubeconcentration for relaxation. In actuation, 𝛽 is seen to decrease overall, reachesa minimum of about 𝛽 = 0.8 at 0.2 μg/ml, and then goes back up to about 𝛽= 0.9 at 0.5 μg/ml, suggesting both short- and long-range interactions. Withincrease in prestrains, 𝛽 is also seen to vary in actuation. This variation of 𝛽may be due to change in microscopic order of the nanotubes, after stretchingresults in more complex chain movements and longer range of interactions ofthe disordered polymer when excited by NIR light. However, in both cases ofrelaxation, 𝛽 almost approaches unity at high concentrations and prestrains.The variation in 𝛽 can alsomean dynamic changes in the rheological propertiesof the sample with light excitation, prestrains, and nanotube concentrations.We tried to fit the actuation and relaxation with 𝛽 = 2, the results of whichwere poor in comparison with the previously used nanotube photomechanicalactuators [14]. Therefore, this study shows that the design of photomechanicalactuators as layered composite or nanocomposite encompassing similar mate-rials can have two different responses, and therefore, it is a highly complex butfascinating system.Not only the addition of nanotube LC to elastomers creates high-mechanical-

strength composites and enables photomechanical actuation, but it alsopotentially a viable system for energy harvesting. We calculated someoptical-to-mechanical conversion factors. Figure 6.6(D)-a presents theoptical-to-mechanical conversion factors as a function of concentration.This number is a measure of the stress generated to the power absorbedby the actuator light spot and has been reported in the past as a indicatorof photomechanical actuator performance [2, 39]. An extraordinary factorof ∼0.5MPa/W was measured for the nanotube LC elastomeric actuator.These numbers are close to those of the previously used CNTs and, morerecently, to those of the graphene-based photomechanical actuators of about0.5–10MPa/W [1, 2]. However, contrary to all previous works, the amountof CNT used in the present work was about 100–1000 times smaller.

�

� �

�


Our design is also a layered composite unlike previously used nan-otube/graphene photomechanical actuators, which were nanocomposites[1, 2]. Further, the mass fractions used were also about 10,000 times smallerthan those of the previously used electromechanical actuators based on nan-otube polymer nanocomposites [27]. This may suggest that instead of randommixing of nanotubes into polymer resulting in nanocomposites, high-qualitynanotubes, at low concentrations, that are self-assembled into LC with layereddesign may potentially lower the cost for commercial applications. This mayespecially be useful in thin-film transistors where accessing the extraordinaryproperties of nanotube LC may result in high electron mobility. We, therefore,believe the commercialization of nanotube products needs an understandingof the trade-offs between performance and material utilization.An important aspect of any actuator is its energy efficiency at convert-

ing external stimulus into useful work. Therefore, efficiency (𝜂) of thenanotube–LC composites to a known IR illumination source was evaluated.Figure 6.6(D)-b presents the efficiency as a function of nanotube–LC loading.The efficiency increased with an increase in concentration, which ranged from∼0.0015% (0.01 μg/ml at 50% prestrains) to ∼0.0045% (∼0.5 μg/ml at ∼50%prestrains). This is about three times the increase at such small nanotube–LCconcentration. Further, efficiencies were seen to be tunable with respectto strains. Stretching the rubber composite increases the efficiency due toessentially the increase in entropic force (rubber elasticity) [40]. However,the increase in efficiency is also related to increase in order parameters ofself-assembled nanotubes after stretching. Previously used nanocompositephotomechanical actuators based onCNTs have presented a change in inducedorder parameters using X-ray diffraction measurements upon stretching [1].For instance, for prestrain value of 𝜀 = 0.6 (60%), the induced orientationalorder, S, in nanotube–PDMS composite reaches as high as Sstretched = 0.29 froman unstretched value of about Sunstretched = 0.005. Although the change in orderparameter is large, these values are ∼4–5 times smaller than that observed inthe present work, with the existence of true LC actuators here [1]. Maximumenergy conversion efficiency of ∼0.085% was measured, which is about 1000times larger than the reported photothermal (8.5 × 10−5%) efficiency valuesfor PVDF-polymer-based light-driven actuators [39]. These values are alsosimilar to the recently reported energy conversion efficiencies of ∼0.03%for graphene-based photomechanical actuators [41]. All these observationsshow that as we stretch the composites, the self-assembled nanotubes shouldundergo further ordering in the direction of strain. Parallel and perpendicularautocorrelations from 0% strain to 92% strain suggest the possibility of domainsize increase along the y-direction (direction of strain). While the domain sizedecrease along the x-direction suggests a change in bundle size and orientationof the nanotube within the domain, the deformation of the individual domainssuggests localized change in distance between nanotube bundles and an

�

� �

�


Mass fraction comparsion

Our workPrevious

work50

40

30

20

10

10–4 10–3 10–2 10–1 100

CNT/graphene concentration (wt%)

S.V. Ahir et al., Physical Review Letters (2006)

S.v. Ahir et al., Nature Materials (2005)

S.Lu et al., Nanotechnology (2007)

J.Loomis et al., Applied Physics Letters (2012)

J.Loomis et al., Nanotechnology (21,2012)

J.Loomis et al., Nanotechnology (4,2012)

101

Change in s

tress (

kP

a)

Figure 6.7 Stress versus mass fraction comparisons: Logarithmic plot of mass fractions ofCNT/graphene versus stress suggesting superior performance of nanotube LC elastomerscompared to previous nanotube/graphene-based nanocomposite photomechanicalactuators. It should be noted that the present work has a layered composite structure unlikemost previous studies, which were nanocomposites. (Fan et al. [3]. Reproduced with thepermission of IOP Science.)

increase in order in the direction of strain in the viscoelastic PDMS matrix.This change in microscopic order of nanotube LC domains coupled withrubbery elasticity of the matrix should result in large contraction when excitedby NIR light and, thus, presents a unique actuation mechanism.Figure 6.7 presents the nanotube mass fractions, comparing our layered

LC photomechanical actuators to all other reported nanocomposite pho-tomechanical actuators composed of nanotubes and polymers [1, 2, 28,29, 42, 43]. Previously reported nanotube/graphene-based nanocompositephotomechanical actuators have used anywhere from ∼0.02 to ∼7wt% ofnanotubes and 0–5wt% of graphene in PDMS, respectively [1, 2, 28, 29, 42, 43].These are randomly oriented nanotube/graphene mixed inside the polymers asnanocomposites and do not show any optical anisotropy. Even after stretching,no LC ordering was reported in the previously used actuators. The massof CNT used in the previously used actuators for the weight percentage of

�

� �

�


nanotube/graphene mentioned earlier corresponds to ∼22–1100 μg to preparethe composites [1, 2, 29, 42, 43]. The present LC films reported here use only∼0.07–3.31 μg of CNT mass, which is about 100–10,000 times smaller. Ouractuator is a laminate with two layers of PDMS in between self-assemblednanotube LC layer. So while this is a different design from a nanocomposite,the amount of high-quality nanotube used is significantly smaller to achievea similar response. About 1mg/100ml of >99% purity nanotubes today costsabout $799. The 47-mm anodisc film/85mm film on MCE filter membrane,as reported here and in the recent past [26], needs 1ml of nanotube stocksolution. Meanwhile, we have been able to obtain 100 films (30–35mm inanodisc) using the 100ml stock solution. Each LC film resulted in 5–10 actu-ators of the previously mentioned size. Therefore, one can obtain anywherebetween 500 and 1000 actuators using 100ml nanotube stock solution, whichcost approximately anywhere between $0.799 and $1.50. Since there is noadditional processing for aligning the nanotubes, these films may be highlyuseful for applications such as thin-film transistors with high electron mobility[44] and nanopositioning systems at low cost [41].The work combines the anisotropic properties of LCs with flexibility of

elastomers, which was the approach for photochromic actuators in the past[45]. Our work also demonstrates the use of small amounts of nanotubesto achieve a large mechanical response compared to previously reportedactuators up to date [1, 2, 29, 42, 43]. The transfer process and the ability todefine composites with specific order parameters, which can be related toconcentration, elastic modulus, and photomechanical response, are importantto standardize nanotube composites based on LCs. Lack of standard process-ing techniques of CNT composites has hindered commercialization. Here,we propose a method consisting of simple and low-cost vacuum filtrationfollowed by a transfer process that preserves the orientation of the LC. Thepresent work could also be of significant interest to electromechanical actua-tion communities and can help create artificial muscles with other polymersemploying our processing technique, where highly anisotropic SWNTs withhigh conductivity and percolation pathways are preferred [27]. In the past,large amounts of SWNT were required (almost 0.1–18% w/w in Nafion ofrandomly oriented HiPco nanotubes [27]) to achieve high conductivity andsubsequent electromechanical actuation (macroscopic response of 4.5mm at18% w/w SWNT) [27]. This makes them prohibitively expensive. Comparedto this, our nanotube LC actuators use four orders of magnitude less massfractions of nanotube. Electromechanical actuators and energy conversiondevices based on low mass fractions using nanotube LCs could also pave theway for commercial development. It is also shown here that increasing the con-centration decreases the domain size, and the use of polarization microscopyto image dichroic nanotubes is convenient to develop maps of domain sizeand their counts. In this case, the domain size indicates the size of the aligned

�

� �

�

6.5 Photomechanical Actuation Based on 2-D Nanomaterial 205

CNTs against a specific director. These domains can be highly exciting inestimating the strains in composites and their nondestructive evaluation basedon polarization microscopy. Manufacturing nanotube LC composites withspecific order can also provide specific physical properties, thereby enablingstandardization of nanotube–polymer composite manufacturing processes.Additionally, eliminating process complexities such as acid treatment andfunctionalization of nanotubes in polymer composites makes our processenvironmentally friendly. Finally, it may be possible that structural laminatesbased on epoxies could use LC nanotubes as fillers that enable superstrongcomposites with order-dependent mechanical properties.Optomechanical conversion factor (𝜂m) of ∼97 kPa/W was calculated for

commercial polymers such as polyvinylidene fluoride (PVDF), which is aboutfive times lower than the value of ∼0.5MPa/W reported for nanotube–LCactuators [2]. Furthermore, the energy conversion efficiencies were about1000 times larger (8.5 × 10−5% for PVDF versus 0.085% for nanotube–LC)compared to PVDF [39]. Polymers containing cinnamic groups were reportedto be deformed and fixed into predetermined shapes such as elongated filmsand tubes, arches, or spirals by ultraviolet light illumination [46]. However,they can only be recovered to their original shape by irradiating UV light ofdifferent wavelength for 60min [46]. In comparison, the nanotube LC actua-tors relax to their original configuration, after light is switched off, and, thus,are completely reversible. The strain-dependent energy conversion would alsobe useful in energy scavenging fields using vibrational effects. Nanotube LCactuators as presented here show optical anisotropy, unique photomechanicalresponse, and tunable energy conversion in one system, which makes themhighly useful in the development of smart materials.

6.5 Photomechanical Actuation Based on 2-DNanomaterial (Graphene)–Polymer Composites

Since the discovery of graphene [47] in 2004, its incredible physical propertieshave been well documented, including thermal conductivity [48], mechanicalstrength [49], and quantum hall effect at room temperature [50]. Numerousarticles have appeared in the literature on graphene-based composites; mostof these actually used graphene-like sheets derived from graphite oxide (GO)or graphite intercalation compounds (GICs). GO- and GIC-derived fillersin polymer matrices can exhibit high electrical conductivities [51] and highYoung’s moduli [52] and easily be functionalized to tailor to the host polymerproperties [53]. While GO- and GIC-derived filler materials report electricalconductivities better than those reported for nanoclays [54], however, these areoften lower than the values reported for single-wall CNT fillers [55]. Secondly,graphene nanoplatelets (GNPs) and graphene nanoribbons (GNRs) have

�

� �

�


properties that are dependent on the number of layers, such as sat-urable absorption [56], linear monochromatic optical contrasts [57], andelectric-field-assisted band gaps [58], which are not exhibited in GO- andGIC-derived materials. These interesting properties call for investigation ofgraphene-based composite photomechanical actuators.Figure 6.8(A)-a presents the SEM image of GNPs deposited on a siliconwafer.

During close examination, the GNPs were verified to be <3 layers, with diam-eters in the range of 1–2 μm. Figure 6.8(A)-b shows Raman spectroscopy com-parison of plain PDMS polymer, GNP/PDMS composites, and CNT/PDMScomposites. The Raman peaks of the PDMS are seen in all the three curvesof Figure 6.8(A)-b; the 1582 and 2700 cm−1 peaks correspond to graphene.Theradial breathing mode (RBM), graphite (G), and CNT-2-D band (G′) are alsoclearly seen.The SEM and Raman spectroscopy confirmed high purity of thesesystems.Figure 6.8(B)-a shows the typical light-induced actuation and relaxation

kinetics during a 3% prestrain test. Five cycles of NIR illumination on for 60 s,followed by NIR illumination off for 30 s, were used for composites at differentlevels of prestrains.The light-induced responses became saturated by t = 5 s. Int = 5 s, after the light was switched off, the actuators underwent relaxation.Thisis significantly fast actuation compared to the previous reports on CNT/PDMSactuators where the actuation achieved saturation at t = 10–15 s [14]. Theactuation and relaxation response dependency on time was also examined,with experimental data fit to simple exponential functions (𝜏). The experimen-tal data was fitted to a simple exponential function of 1− exp[−(t/𝜏)], while therelaxation data was fitted to exp[−(t/𝜏)]. We observed Debye characteristicsin both actuation and relaxation responses [14]. Figure 6.8(B)-b shows theexperimental actuation response and associated fitting function when 𝜏 ≈ 1.7 s.Figure 6.8(B)-c shows the experimental relaxation responsewhen 𝜏 ≈ 2 s.Whennormalized to account for the magnitude of change in stress, actuation andrelaxation responses for all GNP/PDMS test composites were nearly identical.The fast actuation and relaxation of GNP/PDMS composites in both expansiveand contractive modes suggest unique actuation mechanism for actuatorsbased on GNP/PDMS. Unlike CNT/PDMS actuators that follow compressedexponential function for actuation [14], the GNP/PDMS actuators followsimple exponential function during actuation and relaxation as well as beingabout three times faster. Simple thermodynamic considerations of steady-stateheat flux from irradiated GNPs causing saturation of the photomechanicalresponse cannot explain the fast actuation and relaxation response.Figue 6.8(C)-a–f shows the photomechanical response to NIR illumination

for a single 60-s cycle at 340mW power. Since the optical loss of PDMS inthe NIR region is <0.5 dB/cm [36], the negligible response in the plain PDMSsample was expected. The photomechanical effect becomes clearly observablethrough the expansion and contraction of the actuator by 0.1wt%. At low

�

� �

�


(a)

Raman shift

1 μm

(b)

3000

2500

2000

Inte

nsity (

a.u

.)

1500

1000

500

100 600 1100 1600

Raman shift (cm–1)

2100 2600 3100

PDMSpeak

RBM

1582 cm–1 peak

2700 cm–1 peak,CNT-2D band or G′

band

Plain PDMS

GNP/PDMS

CNT/PDMS

PDMS

(A)

Figure 6.8 (A) GNP/PDMS photomechanical actuator characterization and testing: (a) SEMimage of GNPs, (b) Raman spectroscopy shift of PDMS, GNP/PDMS, and CNT/PDMS. (B)Photomechanical responses and actuation kinetics of GNP/PDMS composites: (a) Typicalphotomechanical stress response of GNP/PDMS actuator at 3% prestrains; (b–c) actuationand relaxation of GNP/PDMS actuator, respectively. (C) Comparison of increasingphotomechanically induced stress change in GNP/PDMS composites as a result of NIRillumination for increasing GNP concentrations: (a) Plain PDMS, (b) 0.1 wt% GNP, (c) 0.5 wt%GNP, (d) 1 wt% GNP, (e) 2 wt% GNP, and (f ) 5 wt% GNP. (D) Comparison of photomechanicallyinduced stress change in 2 wt% composites of various carbon forms: (a) Ggraphite oxide,(b) carbon nanotubes, (c) carbon black, and (d) graphene nanoplatelets. (Fan et al. [3].Reproduced with the permission of IOP Science.)

�

� �

�


Photomechanical response of GNP/PDMScomposite (2 wt%) at 3% prestrain

Illumination signal

Stress response

18

(a)

16

14

12

10

Change in s

tress,

Δσ (k

Pa)

Change in s

tress,

Δσ (k

Pa)

Change in s

tress,

Δσ (k

Pa)

8

6

4

2

0

0 100 200 300

Time (s)

400 500–2

16

12

8

4

0

16

12

8

4

0

0

1 – exp[–(t/τ)]

exp[–(t/τ)]

5 10 15

Time (s)

20 25

0 5 10 15Time (s)

20 25

τ ~ 1.7 s~

τ ~ 2 s~

(b) Actuation response

Relaxation response(c)

(B)


prestrains, positive-induced stress was observed, and high prestrain valuesresulted in compression and thus negative change in stress. More conciselystated, weakly stretched composites show reversible expansion, while highlystretched composites show reversible contraction [1].From GNP concentrations of 0.1 to 5wt%, (Figure 6.8C-b–f), the amplitude

of photoinduced stress was observed to increase, reaching almost four ordersof magnitude greater than that for the pristine PDMS polymer. The changefrom expansion to contraction in all samples is probably a result of orienta-tional effects, similar to those seen in CNT-based actuators [1, 59]. As appliedprestrains increase the overall test sample length, the GNPs within the PDMSpolymer become more aligned/rearranged with respect to one another. As aresult, the macroscopic magnitude of response is amplified.To compare the photomechanical responses of GNP/PDMS composites

with other forms of carbon/PDMS composites, test samples were fabri-cated for GO/PDMS, CNT/PDMS, and pyrolytic carbon black (CB)/PDMSusing identical fabrication methods. All samples underwent identical testingprocedures to the GNP/PDMS composites, Figure 6.8(D)a–d presents theseresults. All forms of carbon displayed similar photomechanically inducedexpansion/contraction stress responses as the GNPs, although with varyingmagnitudes. Of these samples, GO/PDMS (Figure 6.4a) showed the smallestamount of actuation, from +2 (3% prestrain) to −9 kPa (40% prestrain). Thisresult also shows that studies of graphene composites derived from GOmaterials are quite different from single-layer, bilayer, or few-layer graphene

�

� �

�


(C)

(a)

Photomechanically induced stress response of variousGNP/PDMS concentrations

Plain PDMS 0.1 wt% GNP

0.1 wt% GNP

30

20

10

0

–10

–20

–30

–40

–500 20 40

Time (s)

Negligible

response

60 80 100

Ch

an

ge

in

str

ess,

Δσ (k

Pa

) 30

20

10

0

–10

–20

–30

–40

–500 20 40

Time (s)

NIR illumination on

Expansion

Contraction

Direction of increasing

prestrain

NIR illumination off

60 80 100

Appliedprestrain:

Ch

an

ge

in

str

ess,

Δσ (k

Pa

)

30

20

10

0

–10

–20

–30

–40

–500 20 40

Time (s)

60 80 100

Ch

an

ge

in

str

ess,

Δσ (k

Pa

)

5 wt% GNP30

20

10

0

–10

–20

–30

–40

–500 20 40

Time (s)

60 80 100

Ch

an

ge

in

str

ess,

Δσ (k

Pa

)

2 wt% GNP30

20

10

0

–10

–20

–30

–40

–500 20 40

Time (s)

60 80 100

Ch

an

ge

in

str

ess,

Δσ (k

Pa

)

0.5 wt% GNP30

20

10

0

–10

–20

–30

–40

–500 20 40

Time (s)

60 80 100

Ch

an

ge

in

str

ess,

Δσ (k

Pa

)

(b)

(d)(c)

(e) (f)

3%

5%

7.5%

9%

15%

20%

25%

30%

35%

40%


samples.The magnitude of induced stress change in CNT/PDMS (Figure 6.4b)was slightly higher, from +4 (3% prestrain) to −10 kPa (40% prestrain). Finally,CB/PDMS (Figure 6.4c) exhibited a stress change from +7 (3% prestrain) to−14 kPa (40% prestrain). The GNP/PDMS response (Figure 6.4d) magnitudeof stress-induced change ranged from +14 to −36 kPa, an average of 4.5 timeslarger than that of GO/PDMS, 3.6 times that of CNT/PDMS, and 2.4 timesthat of CB/PDMS, for the same fabrication methods. The similar magnitudeof stress change seen in CNT/PDMS and CB/PDMS demonstrates that CNTsare not well dispersed in the PDMS matrix acting as particles similar to CBrather than tubular filaments due to short mixing times. Previous reportshave shown much larger stress increases for MWCNTs than seen here [14].The MWCNTs in such studies were shear mixed in a high shear laboratory

�

� �

�


Photomechanically induced stress response of variousPDMS composites (2 wt%)

Appliedprestrain:

3%

5%

7.5%

9%

15%

20%

25%

30%

35%

40%

Graphite oxide (GO)

Carbon black (CB) Graphene nanoplatelets (GNPs)

Carbon nanotubes (CNTs)(a) (b)

(d)(c)

(D)

30

20

10

0

–10

–20

–30

–40

–50

0 20 40

Time (s)

60 80 100

Change in s

tress,

Δσ (k

Pa)

NIR illumination on

Expansion

Contraction

Direction of increasing

prestrain

NIR illumination off

30

20

10

0

–10

–20

–30

–40

–50

0 20 40

Time (s)

60 80 100

Change in s

tress,

Δσ (k

Pa)

30

20

10

0

–10

–20

–30

–40

–50

0 20 40

Time (s)

60 80 100

Change in s

tress,

Δσ (k

Pa)

30

20

10

0

–10

–20

–30

–40

–50

0 20 40

Time (s)

60 80 100

Change in s

tress,

Δσ (k

Pa)


mixer for minimum of 24 h in the PDMS composite before fabricating theactuator. The resulting stress of 𝜎max ≈ 50 kPa at 8mW [1, 14] of absorbedpower (Peffective) gives rise to a calculated value of 𝜂M for CNTs ∼6.25MPa/W.Figure 6.9(A)-a shows a comparison of the photomechanical response for

three 2wt% GNP/PDMS samples at prestrains of 3% and 40%. As shown inthe figure, actuation and relaxation are relatively consistent throughout thesamples with marginal dependence on the prestrains applied, both in magni-tude of actuation and response time. Figure 6.9(A)-b shows the photomechan-ical stress response versus applied prestrain, with ±1 kPa standard deviationerror bars. At low prestrains (under 15%), the standard deviation is less than1.5 kPa, as prestrain increases, so does the standard deviation, reaching a max-imumof±5.4 kPa at 40% prestrain.This suggests that as prestrains increase, the

�

� �

�


30

(A)

20

10

0

–10

–20

–30

–40

–500 20 40

40% prestrain

3% prestrain

Actuation comparisons

(2 wt% GNP/PDMS)

Change in stress vs prestrain

(2 wt% GNP/PDMS)

Young’s modulus vs wt% GNP

Young’s

modulu

s,

E (

MP

a)

Sample 1

Sample 2

Sample 3

60 80

NIR on

NIR off

1000 5 10

2.25

2.00

1.75

1.50

1.250 1 2 3 4

E = 1.564(X 0.162)

5

wt% GNP

15 20 25 30

Ch

an

ge

in

na

tura

l le

ng

th,

ΔL

/L (

%)

35 40 45 50

Prestrain (%)Time (s)

Ch

an

ge

in

str

ess,

Δσ

(kP

a)

301.7

0.7

0.3

50 kPa

(b)

(c)

(a)

–1.3

–2.3

–3.3

20

10

0

–10

–20

–30

–40

–50

Ch

an

ge

in

str

ess,

Δσ

(kP

a)

Figure 6.9 (A) (a) Repeatability details of 2 wt% GNP/PDMS. Comparison of actuationresponse from three samples at 3% and 40% prestrains, and (b) average actuation responseversus prestrain with ±1 standard deviation error bars; (c) increase in Young’s modulus ofthe actuator with increase in GNP fraction. (B) Steady-state temperature measurements:Temperature decrease along the composites test samples as a result of distance from theillumination point. 0% prestrain dimension and temperature response (left side) arecompared with 40% prestrain dimensions and temperature response (right side). (Fan et al.[3]. Reproduced with the permission of IOP Science.)

orientational ordering of GNPs results in larger contraction of the actuator orbetter dispersed/aligned composites. Figure 6.9(A)-c presents the change inYoung’s modulus (E) as a function of GNP fractions dispersed in the polymer.A Young’s modulus increase of almost twice that of pristine PDMS elastomerwas obtained [60], and thus, higher weight percentage GNP actuators are alsostiffer. It is common knowledge that any type of energy conversion is associ-ated with heating effects with only a fraction of energy used to provide usefulwork. Due to this effect, the macroscopic steady-state temperature increase inthe actuators was studied.Laser irradiation on the GNP/PDMS composites was found to increase

the actuator temperature at the irradiation spot. The temperature decrease

�

� �

�


Illumination point

(spot size~3×3 mm)

Edge of test area

5 wt% GNP

40% prestrain

Increasing distance from

illumination point

a

a

b

a

b

bb

b

a

3 mm

35 mm

70 mm

3 mm

25 mm

50 mm

Sample dimensions at 0% prestrain

Sample dimensions at 40% prestrain

Increasing distance from

illumination point

0% prestrain

80

60

40

20

0

Tem

pera

ture

(°C

)

Temperature rise in PDMS composite samples

vs distance from laser source

1 wt% GNP

0.1 wt% GNP

1 wt% CNT

Plain PDMS

polymer

(B)


as a function of distance from the illumination source, therefore, was char-acterized for various GNP/PDMS composite concentrations. Figure 6.9(B)details this temperature rise for 0% prestrain (sample length 50mm) as wellas 40% prestrain (sample length 70mm). As expected, the highest temper-ature increase also occurs at the highest GNP concentration (5% GNP, 0%prestrain). A ∼75 ∘C temperature rise quickly diminishes as distance from theillumination point increases. As prestrain and thus sample length increase,the amount of graphene exposed to the fixed-size irradiation spot decreasesdue to orientational ordering; therefore, thermal effects also start to diminish.At 40% prestrain, the maximum temperature increase has fallen to ∼60 ∘C,a 15 ∘C change compared to the unstrained sample. It should be pointed outthat at large prestrains, the magnitude of the actuation increases while thetemperature rise in the actuator decreases. This outcome possibly indicatesthat thermal effects may not be the only effect contributing to the overallactuation mechanism.We believe that this response is a result of contributionsfrom electrostatic, elastic, polaronic, and thermal effects that cause thisimpressive overall photoresponse of GNP/PDMS composites similarly toCNT/PDMS composites [1, 14, 61].Exploring GNP/PDMS composites further, the maximum photomechanical

stress change corresponds to an optical-to-mechanical energy conversion

�

� �

�

6.6 Applications of Photomechanical Actuation in Nanopositioning 213

factor, 𝜂M of 7–9MPa/W. Polymers such as PVDF have significantly smallerphotomechanical response with calculated 𝜂M value of ∼97 kPa/W [62] thanazobenzene-containing polymers and gels [63]. While rapid photomechanicalactuation has been attained in azobenzene-containing polymers [64], prac-tical applications are difficult due to dependency on precise location of theazobenzene moieties in dark as well as photoplastic effects making it relaxslowly, in ∼10min [65]. Compared to these previously reported actuatorswith impressive characteristics, GNP/PDMS-based reversible contraction andexpansion enable fast actuation and relaxation. Thus, GNP-based materialscould be one of the best material choices for future mechanical actuators,thermoelectric-based devices, and solar energy converters.

6.6 Applications of Photomechanical Actuationin Nanopositioning

The addition of nanomaterials to polymers not only results in significant mate-rial property improvements but also assists in creating entirely new compositefunctionalities. While plenty of reports on photomechanical actuation exist,there has not been much in terms of applications. Here, we show that efficientlight absorption byGNPs and subsequent energy transduction to the polymericchains can be used to controllably produce significant amounts of motionthrough entropic elasticity of the prestrained composite useful for nanoposi-tioning applications. Using dual actuators, a two-axis submicron-resolutionstage was developed and allowed for two-axis photothermal positioning(∼100 μm per axis) with 120 nm resolution (feedback sensor limitation) and∼5 μm/s actuation speed. A PID control loop automatically stabilizes the stageagainst thermal drift, as well as random thermal-induced position fluctuations(up to the bandwidth of the feedback and position sensor). Maximum actuatorefficiency values of ∼0.03% were measured, approximately 1000 times greaterthan those recently reported for light-driven polymer systems.Nanopositioning systems have impacted areas as diverse as scanning probe

microscopy [66, 67], metrology [68, 69], advanced manufacturing [69, 70],biological processes [71], materials science [72], high-density storage/memory[73, 74], and precision machining [75, 76].Themost common nanopositioningsystems are capable of positioning below ∼1 nm and employ piezoactuatedflexure-based systems with feedback control [77, 78]. Nanopositioning systemsbased on ferroelectric actuators [79], shape memory alloys [80], magneticlevitation [81–83], laser interferometry [77, 84, 85], ultrasonic actuators [86,87], and electrothermal actuators [88] have also been demonstrated. Whilethese types of systems are impressive, new material development can resultin novel positioning systems that employ novel actuation technologies and

�

� �

�


can expand the scope of high-performance and low-cost nanopositioning andnanomanipulation applications.In general, light-driven actuators have several advantages including wire-

less/remote actuation, displacement control using IR intensity modulation(controlling number of photons), low noise, massively parallel actuation ofpositioner arrays from a single light source, and electrical/mechanical decou-pling. Previous research on photomechanical actuators has demonstratedapplications in microelectromechanical system (MEMS)-based microgrippersfor manipulation of microscale objects [89, 90], photonic switches [91],robotics [92], plastic motors [93], and adaptive micromirrors [94]. Recentresearch in CNT/elastomer [1, 14, 43, 61, 95] and graphene/elastomer pho-toactuators [2, 28, 29] has generated renewed excitement in photoactivematerials and their use as micro-/nano-optomechanical systems.

6.6.1 Principle of GnP/Elastomer Photothermal Actuation

Figure 6.10(A)-a presents a schematic showing the principle GNP/elastomerphotothermal actuation. Unrestricted elastomeric composites with grapheneadditives (sample [1]) have random polymeric chain arrangement/entanglement. Addition of weight (sample [2]) to the free end stretchesthe composites and induces a prestrain. As entangled polymeric chains arepulled into a more ordered arrangement, system entropy is reduced. When theprestrained composite sample is heated via IR illumination (sample [3]), light isabsorbed by theGNPswithin the polydimethylsiloxane (PDMS)matrix; opticalenergy is efficiently transduced into thermal energy through phonons in thesp2 graphene sheets [96]. Due to the high macroscopic thermal conductivity ofgraphene films (300W/mK) [97] and intimate dispersion of GNPs within thePDMS matrix, heat is percolated through the matrix, causing polymeric chaincontraction. Increasing polymeric chain temperature results in an associatedincrease in spring constant [98]. Following Hooke’s law, since there is constantforce applied (suspended weight), contraction in the composite occurs and liftsthe weight, resulting in usable work done through the system.The large ampli-tudes of actuation obtainable in graphene/elastomer and nanotube/elastomersystems are directly a result of the sp2 bonding in the nanocarbons. In general,strong covalent sp2 bonds result in efficient heat transfer by lattice vibrations inthe GNPs and therefore allow for large amplitudes of photothermal actuationaccomplished through heat transduction from the lattice into the polymericchains. This thereby enables a unique photothermal actuation mechanismin nanocarbon/elastomer composites. In pristine nanotubes and single-layergraphene, all bonds are strong sp2 covalent bonds that result in highly intrinsicthermal conductivity [48, 96, 99], while in reduced graphene oxide, bothsp2 and sp3 bonds can exist. Finally, amorphous carbon and diamond-likecarbon contain a large fraction of sp3 bonds that lower the system thermal

�

�

1

Photomechancial actuation Simplified single-axis nanopositioner layout

Reflector

Laser displacement sensor

Moving stage

GnP/PDMS

composite actuator

IR LED

(positive axis motion)

IR LED

(negative axis motion)

(a) (b)

2 3

IR LED

GnP/PDMS

composite

(A)

Weight

ΔL due to IR illumination (2–3)

represents usable work

ΔL

Figure 6.10 (A) Schematic: (a) photothermal actuation schematic. Three thin GNP/PDMS composite strips are mounted with their upper endsfixed to a rigid plate. The free end of sample [1] is unrestricted, while samples [2] and [3] have a weight attached, thus inducing prestrain into thecomposite actuator. [3] Illumination via an IR LED results in energy transduction to the polymeric chains, causing a contraction in the actuator(and thus usable work). (b) Simplified single-axis nanopositioner layout. A laser displacement sensor is used to measure stage position.Independently controlled diodes on either side of the stage allow for differential positive- or negative-axis stage motion. (B) (a) SEM of plain GNPpowder (not mixed with PDMS). (b) XPS data. (c) Changes in composite opacity as GNP loading is increased from 0 to 2 wt%. In all five slides, spintime is constant at 15 s. (d) Series of four slides with identical GNP loading (0.5 wt%), but decreasing spin times. (e) GNP/PDMS compositethicknesses are shown as a function of GNP wt% loading. Curves for a 15 s spin time as well as 90 s spin time are displayed. (f ) Compositethickness as a function of spin time is shown. Composite samples were fabricated with spin casting times between 15 and 90 s. (g) Sample GNPcomposite actuator (2 wt%) used in our two-axis nanopositioners. The first inset is a SEM image showing a corner of the composite strip, and thesecond inset is a detail of a GNP stack sticking out of the polymer. (Fan et al. [3]. Reproduced with the permission of IOP Science.)

�

�

XPS data

O1s

600 500 400

Binding energy (eV)

(a)(b)

GNP stacks

1 μm

300

290 288

C1s of GNP

C1s

––C C(sp2)

–C C(sp3)

286 284 282

Binding energy (eV)

Inte

nsity (

a.u

)

Inte

nsity (

a.u

)(B)

–C O

Constant GNP concentration (0.5 wt%)

75

mm

90 s0 wt% 0.1 wt% 0.5 wt%

(c) Constant spin time (15 s) (d)

1 wt% 2 wt% 60 s 30 s 15 s


�

�

300

250

200

150

100

50

000 1

300

250

200

150

100

50

00 50

T = 80 +300

et

20( )

100 150

Spin time (s)

15 s spin time

90 s spin time

Com

posite thic

kness (μm

)

Com

posite thic

kness (μm

)

GNP (wt%)

(g) GNP actuator

(e) (f)

GNP composite

detail

6 mm100 μm

1 μm

Thickness vs GNP (wt%) Thickness vs spin time


�

� �

�


conductivity [96]. All of these forms can be useful for application-specifictunable photothermal actuation amplitudes in carbon/elastomer composites.

6.6.2 Photomechanical-Actuation-Based Nanopositioning System

Figure 6.10(A)-b shows a simplified model of a single-axis photothermalnanopositioning system. A thin GNP/PDMS composite strip is held at afixed prestrain, a nanopositioning stage is mounted in the composite center,and IR light-emitting diodes (LEDs) are placed on either side for differentialcontrol via both positive and negative stage actuation (only one LED perside is shown for clarity, real system encompasses three LEDs per side orsix per axis). Illumination of the left LED results in localized heating ofthe GNP/PDMS composite around the light spot, causing polymeric chaincontraction in that region and subsequent stage translation to the left.Conversely, energizing the LED on the right side will cause the stage totranslate right. By dynamically modulating IR intensity, a force balance can beenacted on either side of the stage to maintain it in a desired displacement.However, even when the stage is at a set position, LED intensity is still beingcontinuously tuned via a proportional–integral–derivative (PID) control loop,which monitors stage position and compensates for thermal drift. This highlydynamic process achieves nanopositioning by controlling polymeric chainextension/contraction through rapidly modulating thermal energy input.Stage displacement is measured with the aid of high-speed laser displacementsensor (120 nm resolution) and a reflective element placed on the stage.

6.6.3 GNP/PDMS Actuator Fabrication and Characterization

Figure 6.10(B)-a presents the SEM image of GNPs deposited on a silicon wafer.GNPs were verified to be 3–5 layers, with diameters ranging from 1 to 2 μm.Figure 6.10(B)-b presents the photoelectron spectroscopy (XPS) of GNPs. TheXPS C1s spectra show the different binding energy values for different carbonbonds, namely sp3 C—C or C—H (285.5 eV), single C—O bonds (286.1 eV),and sp2 C—C bonds (284.5 eV). Figure 6.10(B)-b presents the high levels ofdeoxygenated sp2 carbon bonds that exist in GNPs, signaling highly intrinsicthermal conduction well suited for photomechanical actuators. The intensityof sp2 C=C is much higher than that of the sp3 C—C bonds and C—O bonds.GNPs are still slightly oxidized on the surface as one can see the oxidized C—Obonds.Investigating the effects of varying fabrication parameters, Figure 6.10(B)-c

presents the slides with increasing composite opacity with a correspondingincrease in GNP loading (from 0 to 2wt%). In all five slides, spin times and thuscomposite thickness were maintained constant. Conversely, Figure 6.10(B)-dpresents all composites with identical GNP loading (0.5wt%) but decreasingspin times. The longest spin time sample (90 s) is the thinnest and the most

�

� �

�


transparent, while the shortest spin time (15 s) is the thickest andmost opaque.Compared to carbon counterparts such as single-layer graphene (derived fromreduced graphene oxide) and CNTs, GNPs are easy to disperse uniformly andthus are ideal for use in advanced polymer composites.The high level of sp2 car-bon bonds, ease of dispersion in polymers via shearmixing, ability to spin-coat,and compatibility with MEMS lithographic processes make GNP/PDMS morefavorable for low-cost commercial manufacturing as compared to all othergraphene/polymer composites. Quantitatively, Figure 6.10(B)-e shows thegraph of composite thicknesses as a function of GNP wt% loading for 15 and90 s spin times. GNP wt% was not observed to have a significant impact onthickness, rather it was observed to be spin time and spin speed dependent,as shown in Figure 6.10(B)-f. Finally, Figure 6.10(B)-g presents a GNP/PDMScomposite actuator prior to mounting in the nanopositioner enclosure (a coinis shown for size comparison). The fabrication of the polymer compositesconsisted of shear mixing GNPs in PDMS, spin-coating large area thin films,polymerization, and cutting 50mm × 6mm actuator strips (strained 50% to75mm for use in nanopositioners).

6.6.4 Nanopositioner System Integration

Following fabrication and characterization of graphene/elastomer compositeactuators, they are integrated into a compact nanopositioning system incorpo-rating a laser-cut enclosure housing multiple IR LEDs, custom electronics, andcontrol software. Figure 6.11(A)-a shows an overview of a two-axis nanopo-sitioner (82mm × 82mm × 30mm). A removable cover plate encloses twoGNP/PDMS actuators and their respective positioning diodes. A cutout in thecover plate allows for mounting a floating stage on the actuators underneath.Indicating lights on the front (x-axis) and the left side (y-axis) provide a visualoperator tool by displaying positioning diode status. A single blue LED isilluminated on each axis, indicating that the nanopositioner is powered; sinceno other lights are lit, the floating stage is in the home position (0, 0). Detailsregarding the control system setup can be found in the SI.As shown in Figure 6.11(A)-b, the cover plate was removed, exposing two

GNP/PDMS composite actuators and their associated IR positioning LEDs.Six positioning diodes are mounted underneath each strained GNP/PDMScomposite actuator strip (12 total for the two-axis system). The diode arrange-ment allows for independent differential stage control along both the x andy axes. While positioning diodes and GNP actuators are mounted to the topof the printed circuit board (PCB), all other electronics are mounted on thebottom side. Figure 6.11(A)-c presents a clarified view of the nanopositionerwith the cover plate and GNP actuators removed to show the diode layout.This orientation shows differential arrangement between positive and negativepositioning diodes. Along their respective axes, IR LEDs closest to the home

�

� �

�


Cover plateFloating stage

(front side)

Positioning

diodes

Composite

actuator

(1 of 2)

(left side)

Indicating lights

(y axis)

Diode layout +y

y axis x axis

A +VDC

Indicating

LED

Positioning

LED

+PWM

Input

–PWM

Input

Load (NO)

Load (NO)

GND

Soild state

relay

B

C

C–x +x

–y

B A

Indicating

lights (x axis)

Nanopositioner (isometric) Cover plate removed (isometric)(a)(A)

(c) (d) Simplified control schematic

(b)

Figure 6.11 (A) Nanopositioner. (a) isometric of fully assembled nanopositioner. (b)Nanopositioner with cover plate removed, showing positioning diodes and dual GNP/PDMScomposite actuators. (c) Top view of the nanopositioner with the actuators removed, clearlyshowing diode layout. (d) Simplified control schematic for a single positioning diode. (B)Nanopositioner kinetics. (a) maximum stage displacement as a function of GNP wt% loadingfor coarse-adjust and fine-adjust positioning diodes. (b) Displacement kinetics as a functionof composite actuator thickness for a 1 wt% GNP sample. Useful displacement limit and timeare also indicated. (C) Resolution. Ordered and actual (a) x-axis and (b) y-axis positions. Stagewas ordered to sequentially move to positions +10, +20, +30 μm, and then −10, −10,−10 μm. Annotations [1] and [2] indicate position commands for 0 μm and −10 μm,respectively. (c) Sample positioning error (xactual − xordered) with micrometer scale. Spikes aredue to mismatch between actual and ordered positions when the control loop receives anew position request. (d) Sample positioning error (xactual − xordered) with nanometer scale.(e) Detail of position response and subsequent stage oscillations. Positive and negativeoscillation bounding curves are indicated by annotations [3] and [4], respectively. (Fan et al.[3]. Reproduced with the permission of IOP Science.)

position [origin of (0, 0)] were designated as “C,” those in the middle as “B,”and the diodes farthest away as “A.” Figure 6.11(A)-d presents the electricalschematic for a single positioning diode. A pulse-width modulation (PWM)signal drives a solid-state relay that turns the positioning diode on/off.Visible-wavelength-indicating diodes from the position sensor serve as anoperator feedback tool and are electrically connected in parallel and mounted

�

� �

�


200

Maximum displacement vs GNP wt%

(210 μm thick composite samples)

Composite thickness dependency

(1 wt% GNP samples)

90 μm thick

Useful displacement

limit

100

(a) (b)

(a)

(c) (d) (e)

(b)

75

50

A

A

B B

25

00 10 20 30

Time (s)

210 μm thick

Dis

pla

ce

me

nt

(μm

)

Time to reach useful

displacement limit

150

100

ma

xim

um

dis

pla

ce

me

nt

(μm

)

50

00 0.5 1 1.5 2 2.5

GNP wt%

Coarse adjust diode

xmax = δ[1 – e(–wt%/φ)]x = σ [1 – e(–t/τ)] + t/β

Fine adjust diode

Position response, y axis

Ordered position

40

30

20

10

0

Sta

ge

po

sitio

n (μm

)

–10

–20

–30

–40

15

10

5

0

–5

–10

–150 50 100 150 200

Time (s)

0 50 100 150 200

Time (s)

Actual position

Position response, x axis

Position error, y axis (μm scale) Position error, y axis (nm scale) Detail

2

11

1 2

500

250

0

Err

or

(nm

)

Err

or,

μm

(X

actu

al – X

ord

ere

d)

–250

–500

Time (s) Time (s)

0 50 100 150 200 0 15 30 45

2

3

4

21

3

4

21

0 μm position command

received

–10 μm position

command received

Negative oscillation

bounding curve in

microns = e(–t/τ) – 0.05

Positive oscillation

bounding curve in

microns = e(–t/τ) + 0.05

40

30

20

10

0

Sta

ge p

ositio

n (μm

)

–10

–20

–30

–400 50 100 150 200

Time (s)

(B)

(C)


externally on the nanopositioner to duplicate positioning diode intensity. Toprecisely control energy transduction to the polymeric chains and thus stageposition, continuous high-speed intensity tuning of each diode is required.

6.6.5 Kinetics of Photothermal Nanopositioners

In order to determine the stage kinetics, diode maximum displacement testswere conducted on various GNP wt% composites (of identical thicknesses).Figure 6.11(B)-a shows displacement testing results as a function of GNP wt%

�

� �

�


loading for two different positioning diode types (coarse-adjust, high-power IRdiode and fine-adjust, low-power diode). A discussion in the SI provides insightinto the advantages and trade-offs of selecting high-power versus low-powerIR diodes. While there was initially a sharp increase in the maximum displace-ment with increasing GNP concentration, loadings beyond ∼0.75wt% resultedin negligible additional stage travel demonstrating saturation of the photome-chanical response. The maximum displacement (xmax) responses as a functionof GNP wt% were fit to Equation 6.1 for both low- and high-power diodes,where 𝛿 is the displacement coefficient and𝜑 the load adjustment factor.Whilehigh-power IR diodes enable large actuator displacements, for a fixed numberof intensity steps, lower power diodes allow for finer temperature control andresolution, resulting in more accurate stage positioning.

xmax = 𝛿

[

1 − e(−wt%∕𝜑

)]

(6.10)

The composite thickness was also found to affect the overall stage kinetics.Figure 6.11(B)-b presents stage displacement responses for 90-μm- and210-μm-thick composites (the thinnest and thickest films obtained). Curvesshown were fit to the raw data using Equation 6.2, where t is the time, 𝜏 a timeconstant, and 𝛽 a drift adjustment factor.

x = 𝜎

[

1 − e(−t∕𝜏

)]

+ t∕𝛽 (6.11)

Thinner composites actuated faster yet had lower maximum displacementscompared to their thicker counterparts. With a fixed number of diode inten-sities available by our control setup, a composite with a larger displacementalso had lower resolution. Conversely, while thinner samples have smallermaximum displacements, they also have finer control steps. The curve fittingin Equation 6.2 to the experimental data shows that the actuation and relax-ation followed simple Debye characteristics and therefore a unique actuationmechanism [2]. Unlike CNT/PDMS actuators, which follow a compressedexponential function for actuation [2], the GNP/PDMS actuators were foundto not only follow a simple exponential function during both actuation andrelaxation but also exhibit about three times faster responses [2].

6.6.6 Useful Displacement versus MaximumDisplacement

While the curves in Figure 6.11(B)-a show displacement as a function of GNPwt%, these reported values are steady-state equilibrium positions reached onlyafter relatively long time periods (∼180 s). While possible to achieve these dis-placements, due to long wait times required, they are not necessarily usefulfor practical nanopositioning applications. Therefore, we set criteria to deter-mine useful displacement versus maximum displacement. This was achievedvia selection of a tripwire value for slope of the displacement versus time curve.

�

� �

�


It was determined that when the displacement curve slope equaled 2, it repre-sented the best balance between (i) quick actuation times, (ii) a large portionof maximum possible displacement, and (iii) still allowed ample additional dis-placement control for maintaining the stage in position. Therefore, taking thederivative of the displacement [Equation 6.2], setting it equal to 2, and solvingfor time (t) yields Equation 6.3.

t = −𝜏 × ln[𝜏

xmax(2 − 1∕𝛽)

]

(6.12)

Applying this constraint to the plots in Figure 6.11(B)-b, the curve for the thin(90 μm) composite sample reached a slope of 2 at 8.5 s, representing a displace-ment of ∼47.2 μm (∼50% of maximum displacement). For the thicker sample(210 μm), a slope of 2 was reached at 14 s, representing ∼58 μm displacementor ∼60% of maximum. Thereby, a two-axis nanopositioner outfitted with thethin composite would have a total travel of ∼94 μm per axis (2 × 47.2 μm) andthat outfitted with the thick composite would have a total travel of 116 μm peraxis (2 × 58 μm). Finally, actuation speed was calculated by dividing the usabledisplacement by the time required to reach those values yielding velocities of∼5.5 μm/s (90 μm“thin” composite) and∼4.1 μm/s (210 μm“thick” composite).These velocity values can be improved significantly through use of much thin-ner composite actuators (<1 μm).However, such ultrathin compositeswill neednew preparation steps as they are difficult to handle. One such solution is inte-gration of thinner films ontoMEMS substrates with on-chip positioning stagesto improve photothermal actuation velocity.

6.6.7 Accuracy and Resolution

The nanopositioners were given a sequence of ordered commands, andautomated stage response was monitored. Figure 6.11(C)-a and b showspositioner response to a command sequence for x and y axes. Stage responseis plotted against position command. Figure 6.11(C)-c–e shows the errorbetween ordered and actual position with scale bars in micrometer andnanometer, respectively. Large spikes are due to mismatch between actualand ordered positions upon new command receipt (annotations [1] and [2]).Figure 6.11(C)-e shows the error spike [1] as the new position commandis received and the subsequent decay of the bounding oscillations (curves[3] and [4]). The oscillations, although smaller than 120 nm, are a resultof filtering/averaging within the laser control head. True resolution of thedisplacement sensor could be less than 120 nm, and the manufacturer’sspecification on the laser displacement sensor used for stage position feedbackis 120 nm; therefore, we consider this to be the resolution limit for our nanopo-sitioner. In follow-on generations, incorporation of capacitive displacementto monitor stage position (vs the current laser displacement system) can

�

� �

�


potentially increase positioner resolution to sub-10 nm [100]. With increasedresolution, stage positioning accuracy can be improved through modificationssuch as (i) PID loop tuning, (ii) incorporation of coarse adjust as well asfine-adjust diodes into a single axis, and (iii) higher frequency PWM controlwith a larger number of intensity steps. These nanopositioners have beenmounted to an optical microscope (Nikon Eclipse L150) and have shownpositioning with nanometer-scale accuracy, suggesting that these could beuseful in nanomanipulation under optical microscopes.In summary, we have introduced a new class of nanopositioners that exploit

photothermal actuationmechanism in graphene–elastomer composites. Com-pared to electromechanical transduction, photomechanical/photothermalactuation offers an alternative way to couple energy into actuator structures,offering distinctive advantages such as wireless actuation, remote controlla-bility, electrical–mechanical decoupling, low noise, and scalability throughexisting MEMS processing technology. Unfortunately, few material systemshave been shown to exhibit photomechanical actuation properties andare often not compatible with complementary metal-oxide-semiconductor(CMOS)/MEMS processing techniques. For example, photomechanicalactuation has been achieved in PVDF [62], azobenzene [63], shape memorypolymers [46], LC elastomers [65], and chalcogenide glasses [101]. While atfirst it may seem that polymers are not good material system for positioningapplications due to thermal drifts, these drifts are corrected using the differ-ential dynamic photon-intensity modulation setup presented. Furthermore,ability to transduce photon energy into thermal energy that can changepolymer chain mobilities at rapid pace to produce reversible actuation makesthis system interesting to study. The uniqueness of our system – ability todynamically tune IR intensity/temperature to change mechanical stiffness ofthe polymer chains in localized composite areas – makes this nanopositionerhighly internally dynamic. One drawback of this work is that resolutionof the feedback system limits the nanopositioner resolution to ∼120 nm;however, this is still better than the ∼170 nm resolution recently reported fora polysilicon-based thermal nanopositioners [88].

6.7 Future Outlook

The photomechanical effects in polymer–nanotube and graphene compositesare an interesting and rich area of study. New 2-D nanomaterials such as MoS2are already making an impact in the area of light-driven actuators [102]. SuchMoS2-based polymer composites could enable wavelength-selective photoac-tuators [103]. Using photomechanical actuation, we demonstrated the world’sfirst nanopositioner with 120 nm resolution [41]. Previous applications of pho-tomechanical actuation in polymer nanotube composites includemicromirrors

�

� �

�

References 225

[94], micromotors [93, 104], microgrippers [61, 90], photothermal micropil-lar actuators [105], and microcantilevers [89]. The microgrippers have beenshown to grip small objects such as 10–16 μ polystyrene microspheres. Suchgrippers would be useful in biological applications to manipulate single cells.Similarly, the large rotational angle of micromirrors makes them useful in widevariety of optoelectronic applications.The photomechanical actuation of poly-mer/nanotube and graphene composites is reversible, with the stress levels intens of kilopascals, which is useful. Future applications could be in the area ofsoft robotics, origami, and programmable shapes based on light-induced actu-ation of such soft composites.

Acknowledgments

B.P would like to acknowledge NSF CMMI 1463869, DMR 1410678, and ECCS1463987. The authors thank Dr Nima Rahbar for helping with the manuscript.

References

1 Ahir, S.V. and Terentjev, E.M. (2005) Photomechanical actuation inpolymer-nanotube composites. Nature Materials, 4, 491–495.

2 Loomis, J., King, B., Burkhead, T. et al. (2012)Graphene-nanoplatelet-based photomechanical actuators. Nanotechnology,23.

3 Fan, X.M., King, B.C., Loomis, J. et al. (2014) Nanotube liquid crystalelastomers: photomechanical response and flexible energy conversion oflayered polymer composites. Nanotechnology, 25.

4 Huber, J.E., Fleck, N.A., and Ashby, M.F. (1997) The selection of mechan-ical actuators based on performance indices. Proceedings of the RoyalSociety of London A: Mathematical, Physical and Engineering Sciences,453, 2185–2205.

5 Bhattacharya, K. (2004) Microstructure of Martensite: Why It Forms andHow It Gives Rise to the Shape-Memory Effect, Oxford University Press.

6 Warner, M. and Terentjev, E. (2006) Liquid Crystal Elastomers, ClarendonPress.

7 Courty, S., Mine, J., Tajbakhsh, A.R., and Terentjev, E. (2003) Nematicelastomers with aligned carbon nanotubes: New electromechanical actua-tors. EPL (Europhysics Letters), 64, 654.

8 Gorga, R.E. and Cohen, R.E. (2004) Toughness enhancements inpoly(methyl methacrylate) by addition of oriented multiwall carbon nan-otubes. Journal of Polymer Science Part B: Polymer Physics, 42, 2690–2702.

9 Harris, P.J.F. (2004) Carbon nanotube composites. International MaterialsReviews, 49, 31–43.

�

� �

�


10 Charlier, J.C. and Michenaud, J.P. (1993) Energetics of multilayered carbontubules. Physical Review Letters, 70, 1858–1861.

11 Panyukov, S. and Rabin, Y. (1996) Statistical physics of polymer gels.Physics Reports, 269, 1–131.

12 Thompson, T.E. and Tillack, T.W. (1985) Organization of glycosphin-golipids in bilayers and plasma-membranes of mammalian-cells. AnnualReview of Biophysics and Biophysical Chemistry, 14, 361–386.

13 Ide, Y. and Ophir, Z. (1983) Orientation development in thermotropicliquid-crystal polymers. Polymer Engineering and Science, 23, 261–265.

14 Ahir, S.V. and Terentjev, E.M. (2006) Fast relaxation of carbon nanotubesin polymer composite actuators. Physical Review Letters, 96.

15 Verissimo-Alves, M., Capaz, R.B., Koiller, B. et al. (2001) Polarons incarbon nanotubes. Physical Review Letters, 86, 3372–3375.

16 Koerner, H., Price, G., Pearce, N.A. et al. (2004) Remotely actuatedpolymer nanocomposites – stress-recovery of carbon-nanotube-filledthermoplastic elastomers. Nature Materials, 3, 115–120.

17 Ajayan, P.M., Terrones, M., de la Guardia, A. et al. (2002) Nanotubes in aflash–ignition and reconstruction. Science, 296, 705.

18 Hugel, T., Holland, N.B., Cattani, A. et al. (2002) Single-molecule optome-chanical cycle. Science, 296, 1103–1106.

19 Gennes, P.G.D. and Prost, J. (1972) Physics of Liquid Crystals, OxfordScience Publications, Oxford.

20 Namba, K. and Stubbs, G. (1986) Structure of tobacco mosaic-virus at3.6-a resolution – implications for assembly. Science, 231, 1401–1406.

21 Schroder, H. (1977) Aggregation of proteins in membranes – example offluctuation-induced interactions in liquid-crystals. Journal of ChemicalPhysics, 67, 1617–1619.

22 Lazaris, A., Arcidiacono, S., Huang, Y. et al. (2002) Spider silk fibers spunfrom soluble recombinant silk produced in mammalian cells. Science, 295,472–476.

23 Kato, T. (2002) Self-assembly of phase-segregated liquid crystal structures.Science, 295, 2414–2418.

24 Vorflusev, V. and Kumar, S. (1999) Phase-separated composite films forliquid crystal displays. Science, 283, 1903–1905.

25 Song, W.H., Kinloch, I.A., and Windle, A.H. (2003) Nematic liquid crys-tallinity of multiwall carbon nanotubes. Science, 302, 1363.

26 King, B. and Panchapakesan, B. (2014) Vacuum filtration based forma-tion of liquid crystal films of semiconducting carbon nanotubes and highperformance transistor devices. Nanotechnology, 25.

27 Landi, B.J., Raffaelle, R.P., Heben, M.J. et al. (2002) Single wall carbonnanotube-Nafion composite actuators. Nano Letters, 2, 1329–1332.

�

� �

�

References 227

28 Loomis, J. and Panchapakesan, B. (2012) Dimensional dependence of pho-tomechanical response in carbon nanostructure composites: a case forcarbon-based mixed-dimensional systems. Nanotechnology, 23.

29 Loomis, J., King, B., and Panchapakesan, B. (2012) Layer dependentmechanical responses of graphene composites to near-infrared light.Applied Physics Letters, 100.

30 Peng, H.Q., Alemany, L.B., Margrave, J.L., and Khabashesku, V.N. (2003)Sidewall carboxylic acid functionalization of single-walled carbon nan-otubes. Journal of the American Chemical Society, 125, 15174–15182.

31 Ramanathan, T., Abdala, A.A., Stankovich, S. et al. (2008) Functionalizedgraphene sheets for polymer nanocomposites. Nature Nanotechnology, 3,327–331.

32 Zhou, Y.X., Hu, L.B., and Gruner, G. (2006) A method of printing carbonnanotube thin films. Applied Physics Letters, 88.

33 Bayan, C., Levitt, J.M., Miller, E. et al. (2009) Fully automated, quantita-tive, noninvasive assessment of collagen fiber content and organization inthick collagen gels. Journal of Applied Physics, 105.

34 Park, W.S. (2000) Determination of the order parameters of nematic liquidcrystals by infrared spectroscopy. Journal of the Korean Physical Society,37, 331–334.

35 Schneider, C., Rasband, W., and Eliceiri, K. (2012) NIH Image to ImageJ:25 years of image analysis. Nature Methods, 9, 671–675.

36 Cai, D.K., Neyer, A., Kuckuk, R., and Heise, H.M. (2008) Optical absorp-tion in transparent PDMS materials applied for multimode waveguidesfabrication. Optical Materials, 30, 1157–1161.

37 Williams, G. and Watts, D.C. (1970) Non-symmetrical dielectric relaxationbehaviour arising from a simple empirical decay function. Transactions ofthe Faraday Society, 66, 80–85.

38 Potuzak, M., Welch, R.C., and Mauro, J.C. (2011) Topological origin ofstretched exponential relaxation in glass. Journal of Chemical Physics, 135.

39 Sarkisov, S.S., Curley, M.J., Huey, L. et al. (2006) Light-driven actuatorsbased on polymer films. Optical Engineering, 45.

40 Sperling, L.H. (2006) Introduction to Physical Polymer Science, Fourth edn,Wiley.

41 Loomis, J., Fan, X.M., Khosravi, F. et al. (2013) Graphene/elastomercomposite-based photo-thermal nanopositioners. Scientific Reports, 3.

42 Ahir, S.V., Squires, A.M., Tajbakhsh, A.R., and Terentjev, E.M. (2006)Infrared actuation in aligned polymer-nanotube composites. PhysicalReview B, 73.

43 Lu, S.X. and Panchapakesan, B. (2007) Photomechanical responses ofcarbon nanotube/polymer actuators. Nanotechnology, 18.

�

� �

�


44 Lau, P.H., Takei, K., Wang, C. et al. (2013) Fully printed, high performancecarbon nanotube thin-film transistors on flexible substrates. Nano Letters,13, 3864–3869.

45 Mamiya, J. (2013) Photomechanical energy conversion based oncross-linked liquid-crystalline polymers. Polymer Journal, 45, 239–246.

46 Lendlein, A., Jiang, H.Y., Junger, O., and Langer, R. (2005) Light-inducedshape-memory polymers. Nature, 434, 879–882.

47 Novoselov, K.S., Geim, A.K., Morozov, S.V. et al. (2004) Electric field effectin atomically thin carbon films. Science, 306, 666–669.

48 Balandin, A.A., Ghosh, S., Bao, W.Z. et al. (2008) Superior thermal con-ductivity of single-layer graphene. Nano Letters, 8, 902–907.

49 Lu, Q., Arroyo, M., and Huang, R. (2009) Elastic bending modulus ofmonolayer graphene. Journal of Physics D-Applied Physics, 42.

50 Gusynin, V.P. and Sharapov, S.G. (2005) Unconventional integer quantumHall effect in graphene. Physical Review Letters, 95.

51 Dreyer, D.R., Park, S., Bielawski, C.W., and Ruoff, R.S. (2010) The chem-istry of graphene oxide. Chemical Society Reviews, 39, 228–240.

52 Gomez-Navarro, C., Burghard, M., and Kern, K. (2008) Elastic propertiesof chemically derived single graphene sheets. Nano Letters, 8, 2045–2049.

53 Bielawski, C.W., Potts, J.R., Dreyer, D.R., and Ruoff, R.S. (2011)Graphene-based polymer nanocomposites. Polymer, 52, 5–25.

54 Alexandre, M. and Dubois, P. (2000) Polymer-layered silicate nanocompos-ites: preparation, properties and uses of a new class of materials. MaterialsScience & Engineering R-Reports, 28, 1–63.

55 Thostenson, E.T., Li, C.Y., and Chou, T.W. (2005) Nanocomposites incontext. Composites Science and Technology, 65, 491–516.

56 Sun, Z.P., Hasan, T., Torrisi, F. et al. (2010) Graphene mode-locked ultra-fast laser. ACS Nano, 4, 803–810.

57 Casiraghi, C., Hartschuh, A., Lidorikis, E. et al. (2007) Rayleigh imaging ofgraphene and graphene layers. Nano Letters, 7, 2711–2717.

58 Zhang, Y.B., Tang, T.T., Girit, C. et al. (2009) Direct observation of awidely tunable bandgap in bilayer graphene. Nature, 459, 820–823.

59 Shaoxin, L. and Panchapakesan, B. (2007) Photomechanical responses ofcarbon nanotube/polymer actuators. Nanotechnology, 18, 8.

60 Armani, D., Liu, C. and Aluru, N. 1999. Re-configurable fluid circuits byPDMS elastomer micromachining, presented at the Micro Electro Mechan-ical Systems, 1999. MEMS ’99. Twelfth IEEE International Conference onOrlando, FL.

61 Lu, S.X. and Panchapakesan, B. (2005) Optically driven nanotube actua-tors. Nanotechnology, 16, 2548–2554.

62 Sarkisov, S.S., Curley, M.J., Fields, A. et al. (2004) Photomechanical effectin films of polyvinylidene fluoride. Applied Physics Letters, 85, 2747–2749.

�

� �

�

References 229

63 Seki, T. and Tamaki, T. (1993) Photomechanical effect in monolayers ofazobenzene side-chain polymers. Chemistry Letters, 22 (10), 1739–1742.

64 Tabiryan, N., Serak, S., Dai, X.M., and Bunning, T. (2005) Polymerfilm with optically controlled form and actuation. Optics Express, 13,7442–7448.

65 Ikeda, T., Mamiya, J., and Yu, Y.L. (2007) Photomechanics ofliquid-crystalline elastomers and other polymers. Angewandte ChemieInternational Edition, 46, 506–528.

66 Binnig, G. and Rohrer, H. (1982) Scanning tunneling microscopy. HelveticaPhysica Acta, 55, 726–735.

67 Binnig, G., Quate, C.F., and Gerber, C. (1986) Atomic force microscope.Physical Review Letters, 56, 930–933.

68 Yoshida, S. (1992) Nanometrology. Metrologia, 28, 433–441.69 Nakayama, K., Tanaka, M., Shiota, F., and Kuroda, K. (1992) Precision

physical measurements and nanometrology. Metrologia, 28, 483–502.70 Feinberg, B. (1973) Dimensional metrology in manufacturing – need for

relevance. Manufacturing Engineering & Management, 70, 27–30.71 Shao, Z.F., Yang, J., and Somlyo, A.P. (1995) Biological atomic force

microscopy: From microns to nanometers and beyond. Annual Reviewof Cell and Developmental Biology, 11, 241–265.

72 Brown, N.M.D. (1995) Scanning probe microscopies-applications in mate-rial science. Advances in Engineering Materials, 99–1, 161–168.

73 Esch, J. (1999) Prolog to high-density data storage based on the atomicforce microscope – an introduction to the paper by Mamin, Ried, Terris,and Rugar. Proceedings of the IEEE, 87, 1012–1013.

74 Mamin, H.J., Ried, R.P., Terris, B.D., and Rugar, D. (1999) High-densitydata storage based on the atomic force microscope. Proceedings of theIEEE, 87, 1014–1027.

75 Elmustafa, A.A. and Lagally, M.G. (2001) Flexural-hinge guided motionnanopositioner stage for precision machining: finite element simulations.Precision Engineering-Journal of the International Societies for PrecisionEngineering and Nanotechnology, 25, 77–81.

76 Schumacher, H.W., Keyser, U.F., Zeitler, U. et al. (2000) Controlledmechanical AFM machining of two-dimensional electron systems: fab-rication of a single-electron transistor. Physica E-Low-Dimensional Systems& Nanostructures, 6, 860–863.

77 Yeh, H.C., Ni, W.T., and Pan, S.S. (2005) Digital closed-loop nanoposi-tioning using rectilinear flexure stage and laser interferometry. ControlEngineering Practice, 13, 559–566.

78 Shim, J.Y. and Gweon, D.G. (2001) Piezo-driven metrological multiaxisnanopositioner. Review of Scientific Instruments, 72, 4183–4187.

�

� �

�


79 Weinacht, M., Martin, G., Tarakanov, E.A., and Yushin, N.K. (1995) Ferro-electric actuators with precision displacement control. Russian Ultrasonics,25, 30–33.

80 Ganor, Y., Shilo, D., Messier, J. et al. (2007) Testing system for ferromag-netic shape memory microactuators. Review of Scientific Instruments, 78.

81 Verma, S., Kim, W.J., and Gu, J. (2004) Six-axis nanopositioning devicewith precision magnetic levitation technology. IEEE-ASME Transactions onMechatronics, 9, 384–391.

82 Gu, J., Kim, W., and Verma, S. (2005) Nanoscale motion control with acompact minimum-actuator magnetic levitator. Journal of Dynamic Sys-tems Measurement and Control-Transactions of the Asme, 127, 433–442.

83 Verma, S., Shakir, H., and Kim, W.J. (2006) Novel electromagneticactuation scheme for multiaxis nanopositioning. IEEE Transactions onMagnetics, 42, 2052–2062.

84 Yeh, H.C., Ni, W.T., and Pan, S.S. (2002) Real-time motion control withsubnanometer heterodyne interferometry. International Journal of ModernPhysics D, 11, 1087–1099.

85 Lazar, J., Klapetek, P., Cip, O. et al. (2009) Local probe microscopy withinterferometric monitoring of the stage nanopositioning. MeasurementScience & Technology, 20.

86 Snitka, V. (2000) Ultrasonic actuators for nanometre positioning. Ultrason-ics, 38, 20–25.

87 Fan, K.C. and Lai, Z.F. (2008) An intelligent nano-positioning controlsystem driven by an ultrasonic motor. International Journal of PrecisionEngineering and Manufacturing, 9, 40–45.

88 Hubbard, N.B. and Howell, L.L. (2005) Design and characterization of adual-stage, thermally actuated nanopositioner. Journal of Micromechanicsand Microengineering, 15, 1482–1493.

89 Lu, S.X. and Panchapakesan, B. (2006) Nanotube micro-optomechanicalactuators. Applied Physics Letters, 88.

90 Lu, S.X., Liu, Y., Shao, N., and Panchapakesan, B. (2007) Nanotubemicro-opto-mechanical systems. Nanotechnology, 18.

91 Comstock, M.J., Levy, N., Cho, J. et al. (2008) Measuring reversible pho-tomechanical switching rates for a molecule at a surface. Applied PhysicsLetters, 92.

92 Uchino, K., Poosanaas, P., and Tonooka, K. (2001) Photostrictive actua-tors – new perspective. Ferroelectrics, 264, 1961–1966.

93 Yamada, M., Kondo, M., Mamiya, J.I. et al. (2008) Photomobile poly-mer materials: towards light-driven plastic motors. Angewandte ChemieInternational Edition, 47, 4986–4988.

94 Lu, S. and Panchapakesan, B. (2007) All-optical micromirrors fromnanotube MOMS with wavelength selectivity. Journal of Microelectrome-chanical Systems, 16, 1515–1523.

�

� �

�

References 231

95 Ahir, S.V., Terentjev, E.M., Lu, S.X., and Panchapakesan, B. (2007) Thermalfluctuations, stress relaxation, and actuation in carbon nanotube networks.Physical Review B, 76.

96 Balandin, A.A. (2011) Thermal properties of graphene and nanostructuredcarbon materials. Nature Materials, 10, 569–581.

97 Wu, H. and Drzal, L.T. (2012) Graphene nanoplatelet paper as alight-weight composite with excellent electrical and thermal conductivityand good gas barrier properties. Carbon, 50, 1135–1145.

98 James, B.E. and Mark, E. (2007) Rubberlike Elasticity, A Molecular Primer,Cambridge University Press.

99 Choi, S.U.S., Zhang, Z.G., Yu, W. et al. (2001) Anomalous thermal conduc-tivity enhancement in nanotube suspensions. Applied Physics Letters, 79,2252–2254.

100 Xia, S. and Nihtianov, S. (2012) Power-efficient high-speed andhigh-resolution capacitive-sensor interface for subnanometer displacementmeasurements. IEEE Transactions on Instrumentation and Measurement,61, 1315–1322.

101 Stuchlik, M., Krecmer, P., and Elliott, S.R. (2001) Opto-mechanical effectin chalcogenide glasses. Journal of Optoelectronics and Advanced Materi-als, 3, 361–366.

102 Fan, X.M., Khosravi, F., Rahneshin, V. et al. (2015) MoS2 actuators:reversible mechanical responses of MoS2-polymer nanocomposites tophotons. Nanotechnology, 26.

103 V. Rahneshin, F. Khosravi, D. A. Ziolkowska, J. B. Jasinski, and B.Panchapakesan, Chromatic mechanical response in 2-D layered transitionmetal dichalcogenide (TMDs) based nanocomposites, Scientific Reports vol.6, 2016.

104 Zhang, X., Yu, Z.B., Wang, C. et al. (2014) Photoactuators and motorsbased on carbon nanotubes with selective chirality distributions. NatureCommunications, 5.

105 Liu, X.Y., Wei, R.B., Hoang, P.T. et al. (2015) Reversible and rapid laseractuation of liquid crystalline elastomer micropillars with inclusion of goldnanoparticles. Advanced Functional Materials, 25, 3022–3032.

�

� �

�

233

7

Photomechanical Effects in Photochromic CrystalsLingyan Zhu1, Fei Tong1, Rabih O. Al-Kaysi2,3, and Christopher J. Bardeen1

1Department of Chemistry, University of California, Riverside, Riverside, CA, USA2Department of Basic Sciences, College of Science and Health Professions, King Saud bin AbdulazizUniversity for Health Sciences, Riyadh, Saudi Arabia3Ministry of National Guard Health Affairs, King Abdullah International Medical Research Center, Riyadh,Saudi Arabia

7.1 Introduction

Actuating structures that can move and manipulate small-scale objects havepotential applications in many areas, from defense to manufacturing tomedicine. Ideally, such actuators would function without being in physicalcontact with the controlling apparatus. Photons are in many ways the idealtools for controlling nanoscale noncontact actuators, since they can penetrateinto a wider variety of media and transport both energy and information.Molecules can transform light into motion by absorbing photons and under-going chemical reactions that can lead to breaking, bond formation, orgeometrical rearrangements. In many cases, the photochemical reactions canbe reversed by heating or by photoexcitation at a different wavelength, so themechanical action can be repeated.This chapter focuses on molecular crystal systems in which photochem-

ical changes generate mechanical motion on length scales greater than themolecular dimensions. In these materials, geometry changes associated withmolecular-level photochemical reactions couple together to drive meso- tomacroscopic deformations. Such “photomechanical crystals” represent a wayto directly convert photon energy into mechanical motion and can potentiallybecome active elements for photomechanical actuator devices. The use oforganized molecular materials to generate mechanical motion has receivedrenewed attention in recent years [1–3].


�

� �

�

234 7 Photomechanical Effects in Photochromic Crystals

7.2 General Principles for Organic PhotomechanicalMaterials

Photomechanical materials require a molecular photochemical element thatchanges light energy into mechanical motion. A molecule designed to undergoreversible chemical changes under light exposure, accompanied by a colorchange, is termed “photochromic,” and there is a large field of effort dedicatedto using such molecules in optical data storage. Molecular photochromismis a mature field that is covered in Chapter 2 as well as many review articlesand several books [4, 5]. Some representative molecular photochromic andphotoreactive systems are depicted in Figure 7.1. It should be emphasizedthat the capability of chemists to design new photoreactive compounds isessentially limitless.Although a photochemical reaction leads to a change in the geometry of an

individual molecule, in most cases, the effects of its motion cannot be observeddirectly, since the molecules are dissolved in an elastic medium (e.g., a liquid)and randomly oriented. Changes in molecular spectroscopic properties are theonly experimental indicators of the structural changes. Thus, the second chal-lenge is to harness the molecular-level motions arising from a photochemi-cal reaction in order to generate a mechanical response on measurable lengthscales. Typically, the reactive molecules must be organized or aligned in someway, so that they all “push” in the same direction. One common strategy fororganizing photoreactive molecules is to embed them in a host material suchas a liquid crystal polymer [7, 8]. However, perhaps, the easiest way to alignphotochromes is to let the molecules self-assemble themselves into a singlecrystal. The crystalline self-assembly approach to photomechanical materialsis the subject of this chapter.

7.3 History and Background

The field of organic solid-state photochemistry is more than 100 years old.A brief overview of early work in the field can be found in Schmidt’s reviewarticle on topochemistry [9]. After the initial observations, chemists quicklyrealized that the ordered environment inside a crystal provides a unique andwell-defined reaction geometry for chemical reactions. Photochemical reac-tions in crystal matrices could yield very different product distributions fromreactions that occurred in liquid solutions, for example. The systematic studyof crystal packing effects was pioneered by Schmidt et al. [10], who coined theterm “topochemistry” to describe how crystal packing predetermines the out-come of a photochemical reaction. The prototypical reaction, much studied bySchmidt and others, is the [2+ 2] photodimerization of cinnamic acid, shown in

�

� �

�

7.3 History and Background 235

N N

N N

hν′ or Δ

hν′ or Δ

hν

hν

hν′ or Δ

hν′ or Δ

hν

hν

hν

hν′ or Δhν

hνhν′

HO

COOH

COOH

α-Truxillic acid β-Truxillic acid

HOOC

R R

R R

R

R

HOOC COOH

or

or

O

OO

O

S SSS R3

R4

R1R2 R2

R3

R4

R1

O

O

O

O

OO

N

NH3

NH3

NH3

H3N

H3N

O

Co

O

N NH

N

H3N

H3N

NH3

NH3

NH3

2+2+

(a)

(b)

(c)

(d)

(e)

(f)

(g)

Figure 7.1 Examples of reversible photochromic reactions that have been used to drivephotomechanical motion in crystals. (a) trans–cis isomerization of azobenzene; (b) ringformation and cleavage reaction of diarylethene derivatives; (c) ring formation and cleavageisomerization of furylfulgide; (d) intramolecular hydrogen transfer reaction ofsalicyldienoanilines; (e) intramolecular linkage isomerization of a nitropentaaminecobalt (III)complex; (f ) [2+ 2] cycloaddition reaction of cinnamic acid; and (g) [4+ 4] cycloadditionreaction of anthracene derivatives. (Durr and Bouas-Laurent [4]. Reproduced with thepermission of Elsevier.)

�

� �

�


Figure 7.1(f ). In the solid state, UV light irradiation produces the 𝛼-truxillic acidphotodimer if the trans-cinnamic acid is crystallized in the 𝛼-polymorph, the𝛽-truxinic acid dimer if the cinnamic acid crystalizes in the 𝛽-polymorph, andthere is no photoreaction if the cinnamic acid crystallized in the 𝛾-polymorph[11, 12]. The use of crystal packing to preorder photochemical reactants rep-resents an interesting combination of crystal engineering and photochemistryand remains an active area of research [13–19]. Several comprehensive reviewshave been published on how crystal packing affects photochemical reactivity[20–23]. Recent work by MacGillivray et al. in expanding the scope and utilityof the [2+ 2] dimerization is particularly noteworthy [24]. Solid-state reactionshave also attracted interest as solvent-free “green chemistry” alternatives inorganic synthesis [25, 26].In addition to its synthetic utility, crystal photochemistry has also attracted

considerable interest from chemists trying to understand how the environmentaffects reactivity. For example, when a molecule reacts at one site within thecrystal, it can induce lattice strain that affects the reaction probability at othersites. McBride and Eckhardt have done extensive work on measuring sucheffects and developing theoretical models to understand them [21, 23, 27].The cooperative nature of photochemistry in crystals can lead to complicatedkinetic behavior, and the Johnson–Mehl–Avrami–Kolmogorov (JMAK) modelof nucleation is often invoked to analyze reaction rates [13, 28], althoughalternative approaches such as the Finke–Watzky model have also been usedsuccessfully [29, 30].During most of this period, however, the mechanical effects associated with

the photochemical reactions were regarded as more of a nuisance than as afeature. Photoreaction tends to create phase-separated regions composed ofproduct and reactant molecules within the reacting crystal [31]. The internalstrain resulting from the interface between different phases often leads tofracture and disintegration of the original crystal, as opposed to elastic defor-mation. When the photoproduct crystal disintegrates, it is rendered uselessfor single-crystal X-ray analysis. Kaupp used atomic force microscopy (AFM)to dramatically illustrate the disruption of a crystal surface that occurs dueto phase reconstruction during a solid-state photochemical reaction [32–34].Strategies such as using low intensities, irradiation in the long wavelengthtail of the absorption [35], and two-photon excitation [36] can alleviate theproblem of domain formation within a single crystal, but examples of macro-scopic single-crystal-to-single-crystal photochemical transformations remainrelatively uncommon. Thus, the same forces that we would seek to harness ina photomechanical system often lead to the destruction of the material itself,that is, crystal fracture and disintegration.A second factor is that many photochromes that have been traditionally used

as photomechanical elements are inactive in their crystal form. It is alwaysthe case that the dense, ordered packing in the crystal can inhibit geometry

�

� �

�


changes, and there is no guarantee that molecules that are photoreactive indilute solution will retain their reactivity in the solid state. Stilbene and azoben-zene are examples ofmolecules that easily photoisomerize in solution and poly-mers but are unreactive in crystal form due to steric constraints imposed bysurrounding molecules [37].Despite these difficulties, therewas early work suggesting that photomechan-

ical effects in molecular crystals might be something other than a destructiveforce. The first example of photomechanically responsive molecular crystalswas reported in 1982 by Abakumov and Nevodchikov [38]. The crystals werecomposed of a semiquinone complex of platinum group metals that gener-ated a free-radical complex when exposed to visible or near-infrared light.Thischange in molecular structure led to a reversible bending of the crystal by asmuch as 45∘. When the light was removed, the crystal reverted to its originalshape within 0.1 s. The bending action was ascribed to the radical-mediatedformation of dimerized Rh—Rh bonds between stacked complexes within thecrystal. From the results in that paper, it was not clear whether the bendingwas caused by a thermally induced chemical reaction or directly by the photo-chemistry. A decade later, in 1992, Abakumov and coworkers were involved ina study that clarified the chemistry of the rhodium semiquinone complexes butleft the question of the bending mechanism unanswered [39].After 1992, little attention was paid to the potential of molecular crystals as

photomechanical materials for more than a decade. Two main advances wererequired to initiate the current level of interest in photomechanical molecu-lar crystals. The first involved advances in crystal growth. As discussed earlier,larger crystals tend to disintegrate under illumination. But in 2002, Nakanishiand coworkers showed that while the photopolymerization of a diolefin deriva-tive resulted in the disintegration of bulk crystals, the same photoreaction didnot destroy nanocrystals [40]. In 2006, our group showed that although largercrystals composed of the anthracene derivative 9-tert-butylanthracene ester(9TBAE) shattered under illumination, crystalline nanorods composed of thesame molecule could expand up to 15% when irradiated by 365 nm light [6].This expansion was driven by a crystal-to-crystal [4+ 4] photodimerizationreaction and did not fragment the nanorods (Figure 7.2). This result suggestedthat such crystals could generate largemechanical displacements, possibly use-ful for photomechanical applications. The MacGillivray and Garcia-Garibaygroups have also reported that organic nano- and microcrystals can survivephotochemical transformations with their morphology intact [41, 42].The ability of small crystals to withstand chemical reactions is usually

ascribed to their high surface-to-volume ratio [6]. The idea is that any inter-facial strain buildup in the interior of the particle can be relieved at a nearbysurface, as opposed to fracture. A second possible factor is that the smalloptical path lengths in these particles lead to weak attenuation of the excitinglight. This leads to more uniform reaction throughout the crystal and prevents

�

� �

�


(a) (b)

Figure 7.2 (a) AFM image of a single-crystal 9TBAE nanorod before illumination and(b) after illumination with 365 nm. Scale bar is 6 μm. Note that the diameter of the rod inthe xy-plane appears greater than 200 nm due to its convolution with the broad AFM tip.(Al-Kaysi et al. [6]. Reproduced with the permission of American Chemical Society.)

the formation of large gradients of reacted versus unreacted molecules. Ifthe photoproduct is absorbing, the use of small crystals can also increase thephotochemical yield by minimizing the internal filter effect. A third factorthat could affect crystal survival is the fact that the mechanical properties of amolecular microcrystal can change in nontrivial ways during photoreaction,perhaps making the crystal more plastic and better able to withstand internalstresses [43]. It should be noted that a comprehensive theory of how fracturedepends on crystal size does not exist and that any conclusions we draw arequalitative at best. But it is clear that one key to exploiting the solid-statephotomechanical properties of many types of molecules is to reduce the crystaldimensions.During this same period, work on solid-state photochromic materials

proceeded, with notable progress in Irie’s group on the photoinducedring-opening/closing reaction of the diarylethene (DAE) family (Figure 7.1b)[44]. Several derivatives that could undergo the ring-opening/closing reactionin crystalline form were found [45]. Experiments demonstrated that crystalsurface features could be switched back and forth by exposure to UV followedby visible light [46]. In 2007, Irie’s group demonstrated that a molecularcrystal composed of the DAE derivative shown in Figure 7.3 could undergoreversible deformations upon exposure to different wavelengths of light[47]. Although the magnitude of the shape changes was small (<1%), whenvacuum evaporation was used to generate microcrystals on a surface, moredramatic deformations could be observed. Figure 7.4 shows an example of a10-μm-diameter molecular crystal rod grown in this manner interacting with

�

� �

�


10 μm 10 μm

Figure 7.3 Open- and closed-ring chemical structures and reaction scheme for adiarylethene derivative, 1,2-bis(2-ethyl-5-phenyl-3-thienyl)perfluorocyclopentene. Theimages illustrate the reversible deformation of a single crystal that can be switched backand forth using ultraviolet (365 nm) and visible (500 nm) light. A square single crystal of (1)with corner angles of 88∘ and 92∘ reversibly changes to a shape with corner angles of 82∘and 98∘. The crystal thickness was 570 nm. (Kobatake et al. [47]. Reproduced with thepermission of Nature Publishing Group.)

Figure 7.4 Movement of a gold microparticle by a rod-like diarylethene crystal afterirradiation with ultraviolet (365 nm) light. The gold microparticle is 90 times heavier thanthe rod-like crystal (250× 5× 5 μm) and appears in the images as a black spot. Theultraviolet-light-induced bending of the crystal could push the gold microparticle as far as30 μm. The exposure time of each frame was 500 μs (2000 frames per second), and thenumbers above the images are frame numbers. (Kobatake et al. [47]. Reproduced with thepermission of Nature Publishing Group.)

a gold microsphere. Upon photoexcitation, the rod flexes, pushing the sphereas far as 30 μm, indicating that a large force has been generated.It is now clear that the correct combination of photochrome and crystal

shape can elicit a photomechanical response from many different classesof molecular crystals. The field is currently in a discovery phase, with new

�

� �

�


materials and modes of action being characterized on a regular basis. Most ofthese materials have yet to be incorporated into functional actuator devices.Nevertheless, their small scale, hardness, response speed, and high densityof force-generating units make molecular crystals highly promising for pho-tomechanical applications. In the following, we review the different modes bywhich photoreactive crystals can generate mechanical action and then providea comprehensive summary of the molecular systems that have been exploredso far. We end the chapter with a section outlining current challenges andopportunities in the field.

7.4 Modes of Mechanical Action

7.4.1 Partial Reaction and Bimorph Formation

As with photomechanically responsive polymers, the most common mechan-ical deformation observed in molecular crystals is bending. Bending occurswhen the photochemical reaction is localized to the illuminated side of anoptically thick crystal, producing a bimorph structure. Bending is the result ofthe strain gradient that arises at the interface between two distinct chemicalphases comprised of the reactant (A) and product (B) molecules, as shown inFigure 7.5(a) [48, 49]. For a thick structure where the incident light experiencessignificant attenuation as it traverses the structure, one will end up with agradient of reacted and unreacted molecules. The stress between these tworegions is what provides the energy to drive a large-scale deformation of thestructure, for example, the bending shown in Figure 7.5(a).This type of motiondepends on the shape of the crystal, and a long flat crystal may seek to alleviateinternal strain by twisting. In fact, we have shown that both types of motioncan be observed using the same molecule, 9-methylanthracene (9MA), bychanging the crystal growth conditions to generate different shapes [50].In the bimorph approach, the location of the strain interface that drives the

motion is dictated by the illumination conditions. In this case, the motiondepends on the direction, intensity, and duration of the light exposure. Anadvantage of this strategy is that it provides a straightforward way to controlthe direction andmagnitude of the mechanical motion by controlling the light.One disadvantage is its sensitivity to the illumination conditions, which cannotbe easily controlled when the structure is located in a scattering medium.A second more important issue is that this strategy may fail when appliedto subwavelength structures where light attenuation within the structure isnegligible. The illumination requirements for bending may be relaxed if a pho-toreaction is somehow self-limiting. In this case, one can still end up with amixture of reactants and products even under uniform illumination conditionsand in very thin crystals. This mechanism may contribute to the twisting

�

� �

�

7.4 Modes of Mechanical Action 241

= Reactant = Photoproduct

(b)

(a)

hν

hν

hν

Figure 7.5 (a) Attenuation of exciting light leads to a gradient of reacted and unreactedmolecules. This forms a bimorph-type structure where the motion is driven by strainbetween the different phases. (b) Complete reaction of the crystal leads to a reconstructionto accommodate new packing arrangements of the product molecules and an overall shapechange of the crystal. (Al-Kaysi et al. [6]. Reproduced with the permission of AmericanChemical Society.)

observed in 9-anthracene carboxylic acid (9AC) and 4-chloro-cinnamic acid(4Cl-CA) microribbons described as follows. Although the reaction kineticscan be complex [28], in both crystals, the one-dimensional stacks impose astatistical limit on the ultimate dimerization yield, guaranteeing a mixture ofdimers and monomers within a reacted crystal [51, 52]. If these two speciesphase-separate, they may be able to generate a sort of built-in bimorphstructure even in very small crystals.

7.4.2 Complete Transformation and Crystal Reconfiguration

If the photochemical reaction A→B goes to 100% completion, it can stillgenerate a reconfiguration of the crystal that generates mechanical motion,as illustrated in Figure 7.5(b). Ideally, the photoproduct B is transparent tothe wavelength of light used to transform A so that the entire crystal canbe transformed. In this case, very high photochemical yields are achievable,but this can also lead to the disintegration of bulk crystals. But if this can beavoided, for example, by using small crystals, the different packing interactions

�

� �

�


between B molecules can lead to a reconstruction of the entire crystal lattice.This type of mechanism drives the photoinduced expansion of 9TBAEnanorods, for example [6, 53]. The concept is analogous to mechanically andthermally driven (thermosalient) phase transitions that can lead to crystalshape changes and motion [54]. The use of a photoinduced Martensitic phasetransformation to change the crystal shape might be useful for very smallstructures where light penetration is more uniform. The shape change willbe intrinsically determined by the crystal shape and molecular orientationsand not the illumination conditions. In our opinion, this mode of action isunique to molecular crystals and merits further exploration, although mostof the photomechanical results described in this chapter rely on the bimorphmechanism.

7.5 Photomechanical Molecular Crystal Systems

7.5.1 Intramolecular Photochemical Reactions

7.5.1.1 Ring-Opening/Closing ReactionsThe photostability and easy switching of the DAE concerted ring-opening/closing reaction has led many workers to exploit it for a wide variety of applica-tions [44, 55]. This reaction is categorized as P-type (photoreversible): the ringopening is initiated by UV light, while visible irradiation causes it to close upagain. Irie and coworkers have extensively explored the photomechanical prop-erties of this class of photochromic crystals. After their initial report on themechanical deformation of these materials in 2007 [47], they collaborated in astudy that showed that DAE crystals could also exhibit a pronounced photos-alient behavior when irradiated with moderately high-fluence laser pulses [56].In 2010, Morimoto and Irie reported on the performance of a cocrystal com-

posed of a DAE and perfluoronaphthalene [57]. When illuminated with UVlight, millimeter-scale crystals could bend away from the light and generate atip displacement of ∼200 μm. Irradiating with visible light from the same sidecaused the crystals to straighten again, and this could be repeated formore than200 cycles of UV–visible irradiation. Using high-power 355 nm laser pulses,they demonstrated that the response time of the crystal deformation was veryrapid – less than 5 μs. Furthermore, this rapid response was maintained evenat very low (5K) temperatures. Finally, they showed that these cocrystals couldlift gold microspheres whose mass was more than 200 times greater than thatof the crystal itself. From these measurements, they estimated that the pho-toreaction generated 44MPa of stress, comparable to piezoelectric crystals. Ina subsequent paper, Terao et al. showed that a cocrystal made from a mix-ture of two DAE derivatives had the photomechanical “horsepower” to act as alight-triggeredmicrorobotic arm that could turn a gear through several revolu-tions [58]. In all these examples, the photomechanical motion is driven by the

�

� �

�

7.5 Photomechanical Molecular Crystal Systems 243

interfacial strain between the reactant and photoproduct regions of the singlecrystal. The photoproduct absorbs strongly in the UV, leading to a photosta-tionary state that prevents formation of photoproduct uniformly throughoutthe crystal. This feature guarantees the formation of a bimorph structure forcrystals that are above a certainminimum thickness (in the range of 10 μm).Theroles of crystal packing and shape changes between reactant and product DAEmolecules in determining the bending have been investigated through X-raydiffraction measurements [59].In addition to Irie’s group, several other groups have worked with this class

of photochromes. Feringa and coworkers synthesized chiral DAE derivativesand prepared plate-like crystals with thicknesses less than 1 μm, via sublima-tion [60]. These chiral microcrystalline plates responded to 366 nm light byrolling instead of bending as reported by Irie. These same crystals could beunrolled to the flat form when exposed to >500 nm light. Kobatake’s grouphas also characterized the photomechanical action of DAE crystals. In 2011,they demonstrated a DAE derivative with a very low photocycloreversion(ring-opening) reaction quantum yield [61]. This allowed them to convert upto 90% of the ring-open form to the ring-closed form under 365 nm irradiation.These crystals showed strong bending and curling behavior during the conver-sion. Kobatake and coworkers have also shown that DAE plates can undergoa photoinduced twisting [62]. The direction of the twist could be controlledby changing the direction of light incidence with respect to the crystal axes,and they established that the twisting was due to bimorph formation withinthe plane of the crystal. To clarify the mechanism of bending for this classof crystals, Kobatake and coworkers have investigated the dependence of themechanical response on crystal thickness [63, 64] and irradiation wavelength[65], concluding that their results can be interpreted in terms of Timoshenko’sbimetal model [66].While bending and twisting motions provide the most visually dramatic

illustration of how photochemical reactions drive mechanical deformations,there exist other modes of action that are more subtle but equally interesting.In particular, the crystal surface properties can be modified in dramatic waysby photochromic reactions. In 2001, Irie et al. showed that the surface of a DAEsingle crystal could be reversibly deformed using light [46]. UV irradiationcaused the formation of ridges and valleys that almost completely disappearedwhen visible light was used to reverse the reaction. In 2006, the same groupshowed that a different DAE derivative could generate microfibrils under UVirradiation, which then disappeared under visible light [67]. These microfibrilswere small enough to dramatically affect the wettability of the crystal surface,changing the contact angle of a water droplet by 70∘ or more. The timescaleof the microfibril formation and recovery was quite slow, however, taking upto 24 h after light exposure, and the original contact angle could not be fullyrecovered. Athanassiou and coworkers formed molecular crystal microfibrils

�

� �

�


on polymer surface by dipping a polymer film into a solution of DAEmolecules[68]. UV and visible irradiation switched the fibrils between crystalline andamorphous forms, whose different mechanical stiffnesses imparted differentfriction coefficients to the surface. Thus, this material had a friction coefficientthat was tunable by light exposure. The use of light to switch the morphologyof the crystal surface, as opposed to its polarity, provides a new approach tomaking materials with controllable surface properties.Several groups have reported making nanocrystals [69] and nanowires [70]

composed of DAE derivatives. While these nanostructures still undergo thereversible photochromic reactions that characterize the larger crystals, therehave been no reports on these smaller crystals exhibiting observable pho-tomechanical responses. This may be due to the limited spatial perturbationthat results from the intramolecular DAE ring-opening/closing reaction. If thedimensional change is very small, then it may have to be summed over manylayers before it can generate a noticeable effect, that is, a large bend angle.A second explanation is that the low optical absorption of these nanostructuresprecludes the formation of a strong gradient (bimorph) between reactant andproduct molecules, as discussed in Section 7.4.Although the necessity of using UV light to initiate the ring-closing reac-

tion might hinder the functional application of this class of chromophores,Branda and coworkers have shown that the use of lanthanide-based upcon-verting nanoparticles allows one to use infrared light (980 nm) to drive thephotoswitching in both directions [71].This provides a way to make these pho-tochromes compatible with use in biological media, where near-infrared lightcauses much less photodamage compared to UV light.Besides the DAE class of photochromes, only one other type of intramolecu-

lar ring-opening/closing reaction has been used to generate a photomechanicalresponse in crystals. Plate-like microcrystals of furylfulgide could undergoa UV-induced ring-closing reaction that drives bending or curling motiontoward the light source [72]. As in the DAEs, this motion can be reversed byvisible light irradiation that induces the ring-opening reaction, and the motioncould be repeated over 200 cycles without loss of response.

7.5.1.2 PhotoisomerizationThe cis–trans photoisomerization reactionmay occur in stilbenes, cinnamates,imines, and azobenzenes. Mobility of the surrounding medium facilitates thischange in geometric configuration. In most cis–trans photoisomerizations,selection of the wavelength is crucial in determining the photochemical yield ofthe process, which in many cases is never 100% due to overlapping absorptionsof the geometric isomers (which leads to a photostationary state with amixtureof reactants and products). In confined environments, photoisomerizationcan still occur but often proceeds through a pathway that minimizes theconfigurational rearrangement of the molecule, for example, the hula-twist

�

� �

�


mechanism [73, 74], which typically does not lead to photomechanical changesin the molecular crystal.While not operative in crystals of unsubstituted azobenzene, it has been pos-

sible to observe photomechanical effects in crystals composed of azobenzenederivatives. Nakano first observed the formation of a surface relief grating ina cocrystal of an azobenzene derivative with ethyl acetate in 2008. The inter-pretation is that the cis–trans isomerization softened the crystal and allowedthe molecules to migrate to regions of high laser intensity, forming a periodicpattern on the crystal surface [75]. This type of photoinduced transition to aglassy state where molecules can be more mobile has also been observed insingle-component crystals of azobenzene derivatives [76].In order to observe macroscopic photomechanical motion such as bending,

Koshima et al. showed that crystals composed of amino-azobenzene deriva-tives could undergo up to 100 bending and unbending cycles under repeatedUV light irradiation [77]. After exposure to UV, a plate-like crystal wouldbend away from the light, then recover its original shape about 30 s after thelight was removed. This recovery time did not seem to correlate well withthe back-reaction of the trans→ cis isomerization (30min), suggesting thatthermal effects might play a role in the mechanical response. Barrett andcoworkers have prepared numerous azobenzene derivatives and extensivelycharacterized their crystal-to-crystal photoisomerizations and mechanicalresponse [37, 78, 79]. An example of how illumination controls the direction ofbending in these crystals is shown in Figure 7.6. Their work has demonstratedthat the use of halogen substitution allows this class of molecules to exhibitfast photomechanical motion with good cycling behavior and that derivativesthat allow visible light to drive the photomechanical motion can be prepared.Furthermore, they have used a cocrystallization strategy to generate multiple

cis-Crystal trans-Polycrystal

457 nm 457 nm

Figure 7.6 Bidirectional bending of a thin crystal composed of a cis-azobenzene derivativeusing 457 nm light. The arrow at the top of the figures indicates the direction of irradiation.(Bushuyev et al. [37]. Reproduced with the permission of American Chemical Society.)

�

� �

�


types of photomechanical crystals starting with a limited number of azoben-zene derivatives. This work has helped establish the azobenzene family as aviable alternative to the DAE family of photomechanical crystals.An interesting application of azobenzene photoisomerization was

demonstrated by Lee et al., who fabricated nanowires made from tris(4-((E)-phenyldiazenyl)phenyl)-benzene-1,3,5-tricarboxamide using the meniscus-guided solidification method [80]. These crystalline nanowires were eitheranchored vertically on a surface or grown on the tip of a glass microcapillarytube. A polystyrene strip attached to the same microcapillary tip provided ahard surface, which the photomechanical nanowire could press against. Byirradiating the nanowire from one side, it could bend toward the polystyrenestrip and grip the polymer microspheres (Figure 7.7). These photoactivatedmicrotweezers could be opened and closed by sequential UV and visible lightirradiation.The trans–cis photoisomerization of divinyl anthracene derivatives has also

been used to generate complex motions, as shown by the recent observationof photoinduced nanowire curling [81]. Using either the cis or trans isomerof an anthracene-9-(1,3-butadiene) derivative, dimethyl-2(3-(anthracen-9-yl)allylidene)malonate (DMAAM) that can be isomerized using visible light,we grew single-crystal nanowires. A single burst of visible light (475 nm),isomerizing 20–40% of the molecules within the nanowire initiated a rapidcollapse of the initially straight nanowires into a tightly coiled ball. Thisdramatic photoinitiated shape change does not rely on the details of thecrystal structure within the nanowire but rather on the generation of a mixedcrystal/amorphous phase that provides the internal energy to drive the shapechange. While the mechanistic details of this photomechanical responseremain to be worked out, it does show that the formation of mixed phases canlead to unexpectedly large and complex shape changes.Photoisomerization around a double bond is the most commonly used

unimolecular reaction to generate photomechanical effects but is by no meansthe only one. The salicylideneanilines are a class of compounds that undergo aphotoinduced intramolecular proton transfer followed by a double-bond shiftthat significantly changes the molecule geometry. Koshima et al. showed thatthis reversible solid-state photochemical reaction can drive crystal bendingand straightening, again with the response persisting over 200 UV–visible lightsequences [82]. The photomechanical response of this class of crystals wasshown to depend on the chirality of the molecules that composed the crystal,with achiral crystals showing a poorer response compared to enantiomericallypure crystals [83]. These crystals could also be harnessed to lift metal ringsthat were up to 300 times heavier than the crystals themselves.Recently, a cobalt-based coordination salt showed dramatic crystal jumping

behavior that is driven by a photoinduced linkage isomerization reaction [84].Naumov et al. have characterized these “photosalient” crystals, showing that

�

� �

�


Figure 7.7 A nanowire composed of a azobenzene derivative (left side) and a polystyrenenanowire (right side) are both attached to the tip of a glass microcapillary tube. UVirradiation of the azobenzene nanowire causes it to bend toward the polystyrene nanowire,clamping a microsphere between them. Scale bar, 20 μm. (Reprinted with permission fromRef. [80].)

the buildup of strain during a photochemical reaction could be released in asudden, violent shape change that could propel the crystal over considerabledistances (cm). Often, this jumping behavior was accompanied by crystal frag-mentation, and controlling this effect depends on controlling the crystal shapeand irradiation conditions. The different mechanical response modes of differ-ently shaped crystals are illustrated in Figure 7.8.

7.5.1.3 PhotodissociationSalzillo et al. showed that dinitroanthracene could absorb a photon andundergo a dissociation reaction where two NO molecules are lost toproduce anthraquinone [86]. This irreversible reaction generated large

�

� �

�


(a)

(b)

(c)

Figure 7.8 Various kinematic effects observed after UV irradiation of coordination metalcomplex crystals. (a) Rolling or flipping; (b) separation of a small fraction of the crystal,which propels the remaining portion of the crystal; (c) explosion or splitting of a crystal.(Medishetty et al. [85]. Reproduced with the permission of John Wiley and Sons.)

crystal deformations. The authors could follow the reaction in detail usinglow-frequency Raman spectroscopy to probe crystal lattice modes, while mon-itoring the reaction progress through the high-frequency molecular modes.They were able to deduce the information on the reaction pathway, providinga good example of how spectroscopic characterization of a photomechanicalprocess could provide detailed information on its mechanism.

7.5.2 Intermolecular Photochemical Reactions

7.5.2.1 [2+ 2] PhotodimerizationAs described earlier, the [2+ 2] photocycloaddition reaction is very wellstudied and was the basis for forming the topochemical rules of reactivity

�

� �

�


[10, 87, 88]. However, this photoreaction is not easily reversible and thusis of limited utility for actuator applications. But due to the existence ofmany types of molecules that undergo this reaction, it is still exploited forphotomechanical purposes and for demonstrating fundamental phenomena.Naumov et al. first showed that the benzylidinedimethylimidazoliones canundergo an intermolecular [2+ 2] photodimerization that causes crystal platesto bend by remarkably large angles (>90∘) without fracturing [89]. Later,Medishetty et al. showed that coordination of metal complexes containing astyrylpyridine ligand could “pop” under UV light due to [2+ 2] cycloadditionreactions that generated a strong photosalient effect [85]. Naumov’s grouphas also shown that benzylidenefuranone crystals can show variable modesof mechanical effects including jumping, bending, delamination, and crystalstriation, following UV irradiation that initiates the [2+ 2] photoreaction [90].The natural product 𝛼-santonin can also exhibit photosalient behavior drivenby a [2+ 2] photodimerization [91]. In this system, changing the molecule’sstructure through alkylation could perturb the molecular packing and com-pletely prevent the reaction from occurring, highlighting the importance ofcrystal engineering for intermolecular photochromic reactions.Zhang and coworkers have demonstrated the photoinduced bending of very

large (∼1 cm) crystals composed of trans-1,2-bis(4-pyridyl)ethylene cocrystal-lized with two water molecules per photochrome [92]. The remarkably largescale of the motion suggests that it is possible to fabricate macroscale pho-toactuators based on single molecular crystals. The reason this crystal remainsintact, while other crystals composed of similarmolecules shatter, was not clear.But the authors suggested that the use of methyl-benzyl-carboxylate tails leadsto greater molecular flexibility, while hydrogen bonding between the carboxy-lates and watermay enhance crystal cohesiveness.This work demonstrates thatthe problem of fragmentation can be solved, at least in some cases, but tun-ing themolecular structure and utilizing cocrystallization tomodify the crystalmechanical properties.In contrast to the large-scale crystals that undergo large bending motions,

the [2+ 2] photodimerization of 4Cl-CA showed a photomechanical responseonly when thin microribbons were fabricated [93]. This molecule crystallizesinto hydrogen-bonded stacks and undergoes a [2+ 2] photocycloadditionreaction across its double bond. Large crystals of 4Cl-CA clearly undergothis photoreaction when irradiated with 365 nm light but do not showany mechanical effects. But when crystalline microribbons of this moleculewere examined, they showed photoinduced twisting. In this case, the motionwas not reversible, and the dimer crystal product eventually decomposed intoan amorphous phase. The lack of response in the larger crystals was explainedqualitatively in terms of the larger crystals being less flexible and having greaterresistance to distortion. 4Cl-CA is an example of a molecule where the size

�

� �

�


and shape of the crystal determine whether a photomechanical response caneven be observed at all.

7.5.2.2 [4+ 4] PhotodimerizationMuch work in our own laboratory has centered on anthracene derivatives thatundergo an intermolecular [4+ 4] photodimerization to form a bridged pho-todimer with new bonds formed between the C9 and C10 carbons. While thisintermolecular reaction is not used as frequently in photochromic applicationsas the cis–trans and ring-opening unimolecular isomerizations described ear-lier, it serves to demonstrate many general aspects of molecular crystal pho-tomechanical materials. Thus, we describe our results on [4+ 4] systems insome detail in order to provide comprehensive examples of some of the generalconcepts described throughout this chapter. Specifically, we describe the casesthat illustrate how the photomechanical response is determined by crystal sizeand shape,molecular structure, and illumination conditions.These illustrationsare made in the context of both irreversible and reversible [4+ 4] photodimer-ization reactions.

Irreversible Solid-State [4+ 4] Photodimerization Reactions Our early work on9TBAE nanorods, described in Section 7.3, vividly demonstrated the impor-tance of crystal morphology by showing that solid-state photochemicalreactions could be harnessed to drive micron-scale displacements withoutthe fragmentation usually observed in larger crystals [6]. Later, a detailedinvestigation of a family of 9-substituted anthracene esters revealed an inter-esting aspect of their solid-state photochemistry [53]. After the monomercrystal is reacted, the photodimers rearrange into a metastable crystal formwhose structure is different from that formed when the dimers are crystallizedfrom liquid solution. Over the course of months, the intermediate solid-statereacted dimer crystal slowly converts into the lower energy solution-grownform. Since the photomechanical response is dictated by themetastable crystalstructure, this research illustrated that the photomechanical response canarise from the formation of nonequilibrium structures and can be difficultto predict based on equilibrium crystal structures. This result emphasizedthe importance of studying the fundamental dynamics of the solid-statechemical reaction in order to gain quantitative information on the mechanismof its photomechanical response. Finally, this system demonstrates that the∼100% conversion strategy shown in Figure 7.5(b) can lead to large shapechanges without bimorph formation.Thus, our studies of the anthracene estersillustrated several general concepts in the field of photomechanical crystals:the importance of crystal size and shape, the role of molecular packing, andthe ability to generate significant motion even in the absence of a bimorph.The importance of crystal shape was further explored by studying the

solid-state photodimerization of 9MA as a model system. The 9MA

�

� �

�


00

5

10

15

20

25

30

35

40

45

2 4 6 8 10 12 14 16 18 20 22

Fre

quency

Fre

quency

14

12

10

8

6

4

2

01.0 1.5 2.0

Aspect ratio

Aspect ratio

2.5 3.0

Fre

quency

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

Aspect ratio

14

12

10

8

6

4

2

0

5 μm

5 μm

(a)

(b)

Figure 7.9 SEM images and histogram analyses of (a) 9MA microneedles and (b) 9MAmicroribbons. The inset of (a) shows an enlarged histogram with a narrower bin size to showthe aspect ratio of the 9MA microneedles in more detail. Changing the crystal growthsolvent changes the crystal aspect ratio in a controlled way. (Kim et al. [50]. Reproducedwith the permission of American Chemical Society.)

photodimerization is an ideal model system to illustrate how heteroge-neous reaction kinetics and crystal shape can be used as design elementsfor the development of new photomechanical materials. By varying thecrystallization conditions, two different crystal shapes, microneedles andmicroribbons of 9MA, were obtained by the floating drop crystallizationmethod using different conditions (Figure 7.9). The microribbons twistedunder irradiation, while the microneedles bent. Examples of these shapechanges are shown in Figure 7.10. The net shape change after a long period ofirradiation was very small, but at intermediate times, the crystals were highlydeformed. The deformation was maximized at intermediate stages during thephotoreaction, while at the end point, the crystal had returned to its originalshape.9MA is an example of a molecule for which the dimensions of the reactant

and product crystals are not very different.Thus, unlike 9TBAE, these crystals

�

� �

�


0 s 4 s 8 s 20 s 48 s 68 s 91 s

0 s 12 s 24 s 36 s 56 s 80 s 108 s

(a)

(b)

Figure 7.10 Optical microscope images of bending and unbending of a 9MA microneedle(a) and twisting and untwisting of a 9MA microribbon (b) during 365 nm UV irradiation.Scale bars: 20 μm. (Kim et al. [50]. Reproduced with the permission of American ChemicalSociety.)

did not exhibit a significant expansion after∼100% conversion to product. Bothreactant and product phases, crystalline monomer and crystalline dimer, mustbe present to drive the mechanical deformation. One question is whether thereactant and product molecules are randomly distributed throughout the crys-tal or whether they form separate phases. To address this question, we usedNMR spin–lattice relaxation experiments to estimate the domain sizes [94].The very different T1 values of the protons in the 9MA and dimer (130 and17 s, respectively) allow the two species to be distinguished and the extent oftheir microscopic mixing to be assessed. Surprisingly, no significant changes inthe 9MA (monomer) or photodimer T1 recoveries were observed at any con-version factor, implying that even from the earliest stages of the reaction, thedomains were larger than the effective spin diffusion distance. The lack of spinexchange between the two domains indicated the formation of uniform dimerdomains even at very early stages of the photochemical reaction.In order to quantify how the simultaneous presence of reactant and product

phases drives crystal deformation, we analyzed the kinetics of the deformationand the photoreaction in single microneedles. The theoretical description oftwisting is complicated [95], so we only discuss the bending in detail. Fluores-cence was used to monitor the reaction progress. We used a focused 325 nm

�

� �

�


laser to initiate bending in individual 9MAmicroneedles while monitoring thefluorescence signal of the unreactedmonomer.The rod deformation was quan-tified by its curvature defined as 𝜅 = 1/𝜌, where 𝜌 is the radius of a circle, whichbest reproduces the shape of the deformed rod. The fluorescence signal beganto decay immediately after the light was switched on, with a time lag of sev-eral seconds before bending was observed. The curvature reached a maximumafter the fluorescence signal had decreased by ∼50%. The important point isthat the dynamics of the curvature change and the population conversion aredifferent, and there is no simple relation between the amount of dimer and theamount of curvature.The curvature 𝜅 can also be plotted versus the dimer frac-tion f dimer, where f dimer is defined as fdimer = 1 − fmon, and f mon is proportionalto the normalized fluorescence signal plot in Figure 7.11.The theory by Warner, originally developed to describe photoinduced bend-

ing in polymer beams, could be successfully applied to analyze the bending ofthe crystalline microneedles as well. According to this theory, the reduced cur-vature as a function of strain distribution within the beam is given by [48]

𝜅 ∝dstrain

w

[(dstrain

w+ 1

2

)

e−w

dstrain −dstrain

w+ 1

2

]

, (7.1)

where w is the beam thickness and dstrain is the characteristic exponential decaylength of the strain within the beam.The distribution of products was assumed

0.0 0.2 0.4 0.6 0.8 1.0

Dimer fraction

1.0

0.8

0.6

0.4

0.2

0.0

No

rma

lize

d c

urv

atu

re

Figure 7.11 Plot of curvature versus dimer fraction for the three 9-methylanthracenemicroneedles shown as squares (a), circles (b); and triangles (c). The curvature peaks at40–60% reaction, then decreases as the needles become predominantly dimer. (Kim et al.[50]. Reproduced with the permission of American Chemical Society.)

�

� �

�


to be exponential, characterized by a length, dreact. The fraction of dimers wasgiven by

fdimer =dreact

w

(

1 − e−w∕dreact

)

. (7.2)

We assumed that dreact and dstrain were linearly proportional. If we assume sim-ple first-order kinetics for themonomer→ dimer reaction, we could also derivean expression for fdimer as a function of time.

fdimer = (1 − e−kt). (7.3)

Equations 7.1–7.3 provided a way to connect 𝜅 (the bending curvature), fdimer,and time. Using these equations, we were able to calculate 𝜅(t) curves fordreact = dstrain and dreact = 2dstrain. For dreact = dstrain, the maximum occurs atfdimer = 0.345, earlier than what was observed experimentally. For dreact =2dstrain, the maximum occurs at fdimer = 0.548, a value that agreed more closelywith the experimental curves in Figure 7.11. The proportionality constant thatrelated dreact and dstrain must be close to unity. These results show that thedynamics of the bending motion are consistent with the evolution of a straingradient within the needle due to the preferential growth of dimer domains onone side and that the strain profile closely follows the spatial distribution ofdimers.

Reversible [4+ 4] Photodimerization Systems Although the 9TBAE and 9MA[4+ 4] photodimerizations can in principle be reversed by thermal activationor short-wavelength irradiation (254 nm), in practice, thermal “unzipping” isdetrimental to the crystal structure, and 254 nm irradiation cannot achieve100% back-reaction due to the formation of a photostationary state. To obtaina photomechanical molecular crystal that can be used multiple times, an easilyreversible solid-state photodimerization reaction is desired. As mentionedearlier, DAE and azobenzene derivative molecular crystals constitute P-typecrystals that require a second photon in order to switch the photoproductback to reactant. Molecular crystals that revert back by thermal processesare dubbed T-type. The reversible anthracene-based molecular crystals wediscuss in this section are all T-type, in which the photochromic reaction (andits accompanying mechanical motion) resets itself spontaneously at roomtemperature.Organic chemistry provides many opportunities to modulate the solid-state

photochemical reactivity of molecules. Simply removing the tert-butyl groupfrom 9TBAE yields 9-anthracene carboxylic acid (9AC), which crystallizesin a head-to-head “syn” arrangement [96], rather than the head-to-tail “anti”arrangement common in most 9-substituted anthracenes [97]. Although thesyn arrangement is often assumed to prevent the [4+ 4] photocycloaddition

�

� �

�


(a) (b) (c) (d)

(e) (f) (g) (h)

Figure 7.12 Single 200-nm-diameter nanorod of 9AC (∼60 μm long) briefly exposed to365 nm light in a 50% solution of phosphoric acid in water. The dotted circle shows theilluminated region (35 μm in diameter). The nanorod repeatedly flexes back and forth (afterUV illumination=panels b, d, f, h; after dark period=panels a, c, e, g). The time required torevert back is around 5 min at room temperature. Scale bar= 20.7 μm. (Al-Kaysi andBardeen [99]. Reproduced with the permission of John Wiley and Sons.)

reaction due to topochemical and steric factors, solid-state NMR mea-surements showed that 9AC does in fact undergo the [4+ 4] cycloadditionreaction characteristic of anthracenes in the solid state [98]. The photodimeris unstable at room temperature, however, and spontaneously reverts back tothe monomer state within a few minutes. We monitored the expansion andsubsequent contraction of individual 9AC rods using AFM [99]. After lightexposure and ∼100% conversion of an individual 9AC rod, it briefly expandsby 1–3% but then returns to its original shape.To induce more useful types of motion, we turned to the bimorph strategy

by using spatially localized photoexcitation. We found that a single rod, irra-diated in its central region, instantly bends under the influence of the lightbeam (Figure 7.12). After 2–5min in the dark, the bent rod returned to itsoriginal shape. This sequence could be repeated for multiple cycles, the rodbending under illumination and then straightening in the dark. Even thoughthe fluorescence recovery indicated that some monomer was lost during everyillumination cycle, this photodegradation did not appear to prevent the asso-ciated shape changes. It is likely that this could be improved by more rigorousexclusion of O2 from the sample, since electronically excited polyacenes areknown to undergo photoperoxidation. The photomechanical bending was alsoobserved for 9AC nanowires coated with a thin layer of silica, although theresponse time was significantly slower [100].This result showed it was possibleto have photomechanical elements encased in a protective shell, which may benecessary for some practical applications [101].To achieve a higher degree of spatiotemporal control over the nanorod

motion, we combined our reversible 9AC nanorods with two-photon

�

� �

�


photoexcitation (2PE) methods [102]. 2PE allows us to precisely localize theregion of reacted molecules in three-dimensional space and control the loca-tion and magnitude of the bend. It also benefits from the superior penetrationcharacteristics of the near-infrared as opposed to the ultraviolet light usedfor one-photon excitation. Reversible bends, all of the same magnitude andduration, could be induced at any point along a single nanorod. In addition,the bend angle could be described in terms of a simple kinetic model wherethe induced angle is proportional to the fraction of reacted monomers at thebend point. Thus, it was possible to control both the rate and magnitude ofthe bend by controlling the duration and intensity of light exposure. This workdemonstrated reversible bending in rods with diameters as small as 35 nm, aswell as the use of bending to induce translational motion of the rods across asurface.Finally, it was again possible to modify the photomechanical response of this

material by varying the crystal morphology. In the case of 9AC, we used amod-ified floating drop method to grow microribbons instead of nanorods [103].Instead of bending or expanding under uniform illumination, these crystalscould reversibly twist, as observed in the 9MAmicroribbons. The dependenceof the twist period on ribbon height and width could be described reasonablywell by classical elasticity theory, an encouraging sign that themechanical prop-erties of these crystals may be understood within the framework of traditionalengineering concepts.Although 9AC had become a workhorse in our laboratory, it became clear

that expanding the capabilities of our photomechanical molecular crystalnanostructures would require modifying the photochrome itself. We firstsystematically modified the 9AC molecule by replacing the H at the C-10position, directly across from the COOH group, with CH3, F, Cl, Br, and evenanother COOH, in an attempt to accelerate the photomechanical response byincreasing the rate of the backward dimer dissociation reaction [96].This workillustrated the challenges involved in the chemical modification approach.10-CH3-9AC crystallized in a completely different packing structure wherethe [4+ 4] dimerization became geometrically impossible. 10-Cl-9AC and10-Br-9AC crystallized into the same head-to-head stacking motif as 9ACbut showed no photochemical reactivity. 10-F-9AC also crystallized in thehead-to-head stacks, and only this molecule exhibited the same reversiblephotochemistry as 9AC. Unfortunately, the rate of dimer dissociation was atleast an order of magnitude slower for the 10-F-9AC crystals, leading to aslower overall cycling time.We had more success by placing substituents at other positions on the

anthracene ring. We synthesized four fluorinated derivatives of 9-anthracenecarboxylic acid (9AC) and characterized their properties [104]. Thespectroscopic behaviors and crystal structures of 4-fluoro-9-anthracenecarboxylic acid (4F-9AC), 2-fluoro-9-anthracene carboxylic acid (2F-9AC),

�

� �

�


10-fluoro-9-anthracene carboxylic acid (10F-9AC), and 2,6-difluoro-9-anthracene carboxylic acid (2,6d F-9AC) are all very similar to those of9AC. However, their photomechanical properties varied widely. 405 nm lightwas used to induce [4+ 4] photodimerization and a mechanical response incrystalline microneedles and ribbons. Both the photodimer dissociation rateand the mechanical recovery varied by more than an order of magnitude,with 4F-9AC exhibiting the most rapid recovery time, on the order of 30 s.Large 4F-9AC crystals remained intact after irradiation, without fragmenting,while microneedles could undergo more than 100 mechanical bending cycleswithout any signs of fatigue or deterioration (Figure 7.13). Nanoindentationmeasurements show that this crystal has a slightly reduced elastic modulusand a significantly reduced hardness, making it less brittle than the 9ACcrystal. Given the similarity of the crystal packing in all five molecules, theimproved photomechanical properties must arise from subtle changes inintermolecular interactions. These results demonstrate that it is possible tosignificantly improve the properties of photomechanical materials throughsmall modifications of the molecular structure.

7.5.3 Nonequilibrium Charge Distribution and Electronic Heating

Rather than use photons to generate chemical changes, one can also use thephoton energy to generate spatially separated charges. Structural changescan arise from the modified Coulomb interactions between the photoexcitedmolecules and their surroundings, rather than from rearrangement of thenuclei within a photoreacted molecule. Such deformations will persist duringthe lifetime of the charge-separated state. Zhang and Iijima used laser lightat 636 nm to induce reversible elastic deformations in long fibers consistingof bundles of multiwalled carbon nanotubes [105]. The detailed mechanism ofthe light-induced bending has yet to be elucidated but may be mediated byphotoinduced charge separation between nanotubes with different electronaffinities. Zewail and coworkers have utilized ultrafast electron microscopyto visualize nanoscale motion of photoexcited Cu-TCNQ charge-transfercrystals after pulsed excitation [106, 107]. This material is known to undergoa transition from a high- to low-impedance state under the influence ofboth DC electric fields and optical excitation. It appears that the low- andhigh-impedance forms are structurally different and that switching betweenthem can induce nanometer- to micrometer-scale expansion of the crystals(Figure 7.14).More recently, molecular crystals composed of molecules that can act as

electron acceptors, such as perylene diimide (PDI) and naphthalene diimide(NDI) derivatives, have demonstrated interesting photomechanical activity.Zhang et al. found that crystalline microribbons composed of PDI bilayerscould undergo a unique morphology change under 488 nm laser irradiation

�

� �

�


1.0

0.5

0.0

0 100 200 300 400 500 600 700

Time (s)

No

rma

lize

d fl

uo

resce

nce

(c)

(a) (b)0 s

5 s

60 s

120 s

420 s 25 s

15 s

5 s

1 s

0 s

Figure 7.13 Series of optical microscopy images of (a) 9AC and (b) 4F-9AC microcrystalsafter a 1 s exposure to 405 nm light causes them to deform. The 4F-9AC crystal completelyuntwists in 25 s, while the 9AC crystal requires 420 s to unbend. Both scale bars are 50 μm.(c) Fluorescence recovery curves of 4F-9AC (fastest, left), 2F-9AC (middle), and 9AC (slowest,right), showing the different photodimer dissociation rates for the different crystals. (Zhuet al. [104]. Reproduced with the permission of American Chemical Society.)

�

� �

�


(a)

(e)

(b) (c) (d)

OnOn OffOff

200 nm

(f)

150

100

50

0

W (nm)

0 10 20 30

Frame

40 50

Off

On

Figure 7.14 Ultrafast electron microscopy images of [Cu(TCNQ)] charge-transfer crystal inthe “off” structure (a and c; no pulsed-laser irradiation) and in the “on” structure (b and d;pulsed-laser irradiation) (scale bar= 500 nm). (e) Higher-resolution image of the crystal inthe “off” structure illustrating the gap that is closed by laser irradiation. (f ) Plot showing theresults of a sequence of “on” and “off” cycles. The channel width changed from 0(pulsed-laser irradiation) to 140± 5 nm (no pulsed-laser irradiation). (Flannigan et al. [106].Reproduced with the permission of John Wiley and Sons.)

[108]. Laser irradiation caused crystal layers to slide over each other, openingthe crystal up as a fan. Under electron beam exposure, the crystal reverted backto its original shape. Although the reaction mechanism of the light-inducedlayer sliding was not determined, the authors hypothesized that the layersliding to Coulombic forces generated by intermolecular electron transfer ledto repulsing interactions between the layers. Photoinduced electron transferhas also been used to drive chemical structure changes that lead to a photome-chanical response. For example, crystal bending has been observed for crystalscomposed of molecules that include an NDI acceptor and an alkylamine donor[109]. After undergoing a photoinduced electron-transfer reaction, the amineradical cation abstracts a proton from a neighboring amine, leading to theformation of a radical anion that distorts the stacking of the NDI derivatives.In this case, the formation of the stabile radical anion is not reversible. Whilethis approach to photomechanical materials is still at a very early stage, theuse of photoinduced charge transfer instead of photochemistry to drivemechanical motion may provide advantages in terms of response speed androbustness.

�

� �

�


7.6 Future Directions

7.6.1 Reaction Dynamics in Molecular Crystals

While the study of molecular photochemical reaction dynamics has reachedan advanced state, much less is known about how these dynamics change in asolid host material. Many groups have studied how the surrounding mediumaffects molecular reactivity, with the topochemical postulate being the mostwell-known outcome of these studies [10, 20–23]. But in photomechanicalmaterials, the photoreaction of a single molecule is only the first step. Boththe overall rate of the photoreaction and the spatial distribution of productswill affect the timescale and direction of the photomechanical response. Openquestions include whether reaction at a single site creates local strain thatfurther catalyzes the reaction, leading to spatially heterogeneous regions ofreactant and product molecules within the solid, or whether the reactionproceeds in a spatially homogeneous manner with reactant and product uni-formly mixed. After a significant number of molecules have reacted, it wouldbe interesting to know how quickly their individual motions synchronize todrive large-scale crystal shape changes. For example, does strain build up andthen relax by a sudden elastic shape change, or does the shape change proceedcontinuously in concert with the reaction progress? An interesting questionis whether photochemical reactions can initiate Martensitic phase transfor-mations in molecular crystals. This question was the subject of some debatein the 1970s [110, 111] and has received renewed attention as researchersattempt to understand how different crystal polymorphs transform into eachother [112, 113]. Finally, there is evidence that the crystal mechanical prop-erties change during the reaction [43, 90], which would complicate analysisof the motion. Such changes would presumably have a large effect on the sizethreshold where a crystal fractures instead of deforming [114].Some of these issues may be addressed by precisely controlling the excitation

conditions. For example, in the bimorph strategy (Figure 7.5a), it is necessary tocontrol the location and extent of photoexcitation within the photomechanicalstructure. Models can describe the bimorph mechanism of bending in molec-ular crystals, given a known spatial distribution of reactants and products [50,64, 115]. The ability to selectively convert any volume element within a struc-ture at any time would enable precise control of the deformation of a singlenanostructure. There now exist strategies for localizing the photoexcitation tosubwavelength regions in a three-dimensional sample [116]. Methods such asmultiphoton absorption [117] and stimulated emission depletion [118–120]have been used to drive photochemical reactions such as polymerization, andit is not a large step to envision using the same strategies to initiate photome-chanical reactions. The use of two-photon excitation to induce bends at differ-ent locations within a single nanorod provided a preliminary example of how

�

� �

�

7.6 Future Directions 261

this might be accomplished [102].The application of sophisticated microscopymethods to photoreactive systems should enable the generation ofmore preciseand complex motions in photomechanical structures.Beyond the spatial properties of the light, its temporal properties can also

affect the photomechanical response. Ideally, photoexcitation would convert100% of the molecules from one form to the other in as short a time as pos-sible. The development of techniques for the rapid and selective inversion ofmolecular populations is a problem in the field of quantum control [121].Therehave already been some intriguing results in the use of chirped femtosecondpulses to increase photoreaction yields in spiropyrans [122], as well as usingsequential absorption phenomena to drive spiro-oxazine reactions that do notoccur naturally in the crystalline state under low-fluence excitation [123]. Plas-monic effects can also be used to enhance photoreaction yields [124]. Novellaser excitation schemesmay help increase the rate and selectivity of themolec-ular photoreaction that powers the material’s photomechanical response.Given that molecular crystals can be characterized using X-ray and electron

diffraction, a realistic goal would be to achieve a predictive understandingof how crystal packing, orientation, and morphology, along with excitationconditions, work together to generate its photomechanical response. Thereis clearly room for both experimental and theoretical physical chemists tostudy these questions, although answering them may require considera-tion of many-body systems containing thousands of molecules. In the longrun, a quantitative understanding of how micron-scale motions arise frommolecule-scale geometry changes will be crucial for the rational design of newphotomechanical materials.

7.6.2 NewMaterials

There already exist a variety of photochemical reactions that can drivemechanical motion in the solid state. Nevertheless, there is still room forimprovement in the molecular reaction properties. In general, one wishes tohave a reaction that (i) proceeds rapidly and in high yield (to efficiently usethe photon energy); (ii) generates significant force; and (iii) can be repeatedmany times. A novel strategy to increase the yield of the reaction is to designa self-propagating chemical reaction with an effective quantum yield >1. Anexample of such a reaction in the solid state is the molecular decompositionof diphenyl-cyclopropenone derivatives to yield CO and diphenyl-acetylene,which was found to have a photochemical quantum yield of decomposition>4 [125]. In this case, the photochemical reaction releases gas, which leads tototal collapse of the crystal structure, but the concept may be generalizableto other reactions. A second area of concern is molecular photostability.The most robust photochromes (DAEs and fulgides) involve intramolecularring-opening/closing reactions. Reactions that generate larger distortions

�

� �

�


upon photoexcitation, such as the azobenzene cis–trans isomerization andthe anthracene [4+ 4] photodimerization, appear to be more susceptible toside reactions. A key challenge for chemists is to design molecules that exhibitlarge geometry changes upon photoexcitation but are sufficiently robust tosurvive over the course of many photocycles.Going beyond the molecular level, we must consider crystal shape and mor-

phology. In the previous sections, we have emphasized the role of crystal shapein determining the overall photomechanical response, for example, the bend-ing of 9MA microneedles and the twisting of 9MA microribbons. Our grouphas focused on the use of nano- and microstructured molecular crystals asphotomechanical elements for two reasons. First, as described earlier, thesestructures aremore robust and are able to survive photochemical reactions thatlead to disintegration of larger crystals. Second, we think that the advantages ofphotomechanical structures are most pronounced for small-scale applicationswhere electrical connections are impractical, for example, for structures func-tioning inside biological cells. The key to this approach is to develop ways togrow small molecular crystals with reasonably well-defined shapes and sizes.We have used solvent annealing in anodic aluminum oxide (AAO) templates tofabricate molecular crystal nanorods and controlled growth on flat substratesto generate microneedles and ribbons. In this latter approach, however, thecrystals still have a wide size distribution. One can tune the crystal growth con-ditions to obtain new shapes and better size distributions, but this is typicallyan empirical approach. For example, we have used acid-catalyzed hydrolysisof precursor molecules in aqueous surfactants such as sodium dodecyl sulfateto generate uniform microwires or microplates of photoactive molecules thatcould undergo photoinduced coiling and bending [126]. While there has beenencouraging progress in the creation of molecular crystal arrays in which boththe crystal size and orientation can be controlled to some degree [127–131],there is still a need for general methods that can generate molecular crystalswith uniform shapes and sizes.Finally, we briefly address the crystal mechanical properties. In general, we

desire a photomechanical material with a high elastic modulus so that it canapply pressure to an object without deforming itself [132]. This considerationmay favor molecular crystals that typically have higher elastic moduli forphotomechanical applications. On the other hand, the elasticity must be largeenough so that the material can respond to the internal strain generated by thephotoreactions without fracturing. The degree to which solid-state reactionsaffect crystal mechanical properties and how these properties affect fractureand photosalient behavior is just beginning to be explored.

7.6.3 InterfacingMolecular Crystals with Other Objects

A final challenge is to incorporate the photoactive materials into largercomposite structures. Many workers in the field draw analogies between

�

� �

�

7.6 Future Directions 263

photomechanical organic materials and biological muscle tissue, but musclesare usually attached to some framework (i.e., bones in the skeleton) in orderto provide a point of leverage. The integration of organic photomechanicalstructures into larger assemblies is a largely unexplored area.We have already mentioned the work by Lee et al. in which bendable

microwires were attached to a pipette tip and used a photoactivated tweezers[80]. This experiment and other experiments showing that photomechanicalcrystals can turn gears and move microspheres [47, 58] all provide prelim-inary demonstrations that it is possible to perform useful work with thesematerials. But there is no general strategy for incorporating photomechanicalelements into a mechanical device such as a walker or swimmer. The factthat molecular crystals can be dissolved in many solvents limits processingoptions.One possible strategy is to incorporate nanometer- to micrometer-sized

molecular crystals into a polymermatrix tomake a compositematerial that canbe processedmore easily. Koshima et al. used amagnetic field to align plate-likesalicylidene crystals in a silicone polymer [133]. This composite showed slightreversible bending under alternating UV and visible irradiation. Lan et al.used a slightly different approach to mix the photoisomerizable molecule2-hydroxynaphthylidene-1′-napthylamine with the polymer polyvinylidenefluoride-hexafluoropropylene [134]. Sahoo et al. obtained a photoactivecomposite by doping the azobenzene adduct phenylazophenylpalladiumhexafluoroacetylacetonate into the protein polymer sodium caseinate [135]. Inboth of these latter cases, dimethylformamide (DMF) was used as the commonsolvent to dissolve both themolecule and the polymer. As the DMF evaporated,the low solubility of the molecules in the polymer caused rod-shaped crystalsto precipitate out into the polymer host. Both crystal–polymer compositesexhibited fairly strong photomechanical responses in the form of bending ofthe polymer film. One can think of the embedded crystals as tiny mechanicalelements, analogous to photoisomerizable molecules.The direct interfacing of a photoswitchable molecular crystal with a second

species was demonstrated when Tsujioka et al. coated DAE crystals withvarious metals and showed that they retained their photochromism [136].To obtain a metal-coated photomechanical crystal into a function device,Kitagawa and Kobatake evaporated a thin (7 nm) gold film onto a DAE crystal[137]. By positioning the crystal between two conductive leads, they created atype of photoconductive switch. Light-induced bending and unbending of thecrystal caused it to reversibly move the gold film into contact with one of theleads. The current across the switch could be switched on and off using visibleand ultraviolet light, respectively. The ability to combine photomechanicaldeformation with metallic conductivity illustrates how composite materialscan lead to novel functionalities.

�

� �

�


7.7 Conclusion

The aim of this chapter is to provide the reader with a reasonably comprehen-sive introduction to photomechanical molecular crystals. Some major conclu-sions of this chapter include the following:1. There exist a large variety of photochemical reactions that can be harnessed

to drive photomechanicalmotion, especially if the dimensions of the crystalsare reduced to the micron range.

2. The character of the mechanical response can be strongly influenced by fac-tors such as crystal shape and illumination conditions.

3. The ability to vary bothmolecular structure and crystal shape provides waysto optimize the photomechanical performance.

4. Photomechanical crystals possess unique properties (high Young’s mod-ulus, rapid response times) that can potentially be exploited for actuatorapplications.

This area of research lies at the intersection of materials engineering, syntheticand physical chemistry, and optical physics. From a scientific point of view,the ability to simultaneously characterize the molecular-level structure, thephotochemical reaction kinetics, and the mechanical response of these mate-rials should eventually lead to a predictive understanding of their properties.Despite a slow beginning, the study of molecular assemblies as photomechan-ical elements is now firmly established and rapidly expanding. While theirfuture potential remains to be determined, it is clear that this field of researchwill lead to both new science and new technological applications for organicsolid-state materials.

Acknowledgments

This research was supported by the National Science Foundation grantDMR-1508099. R. O. Al-Kaysi acknowledges the support of KSAU-HS/KAIMRC through grants RC10/104 and King Abdulaziz City for Science andTechnology (KACST) through Grant AT-30-435.

References

1 Garcia-Garibay, M.A. (2007) Molecular crystals on the move: fromsingle-crystal-to-single-crystal photoreactions to molecular machinery.Angewandte Chemie, International Edition, 46, 2–5.

2 Nath, N.K., Panda, M.K., Sahoo, S.C., and Naumov, P. (2014) Ther-mally induced and photoinduced mechanical effects in molecular singlecrystals – a revival. CrystEngComm, 16, 1850–1858.

�

� �

�

References 265

3 Abendroth, J.M., Bushuyev, O.S., Weiss, P.S., and Barrett, C.J. (2015) Con-trolling motion at the nanoscale: rise of the molecular machines. ACSNano, 9, 7746–7768.

4 Durr, H. and Bouas-Laurent, H. (1990) Photochromism: Molecules and Sys-tems, Elsevier, New York.

5 Irie, M., Yokoyama, Y., and Seki, T. (2013) New Frontiers in Pho-tochromism, Springer, Japan.

6 Al-Kaysi, R.O., Muller, A.M., and Bardeen, C.J. (2006) Photochemicallydriven shape changes of crystalline organic nanorods. Journal of the Amer-ican Chemical Society, 128, 15938–15939.

7 Ikeda, T., Mamiya, J., and Yu, Y. (2007) Photomechanics ofliquid-crystalline elastomers and other polymers. Angewandte Chemie,International Edition, 46, 506–528.

8 White, T.J. and Broer, D.J. (2015) Programmable and adaptive mechanicswith liquid crystal polymer networks and elastomers. Nature Materials,14, 1087–1098.

9 Schmidt, G.M.J. (1971) Photodimerization in the solid state. Pure andApplied Chemistry, 27, 647–678.

10 Schmidt, G.M.J., Cohen, M.D., Donitz, J.D., and Hammond, C.S. (1976)Solid-State Photochemistry, Verlag Chemie, Weinheim.

11 Cohen, M.D., Schmidt, G.M.J., and Sonntag, F.I. (1964) Topochemistry.Part II. The photochemistry of trans-cinnamic acids. Journal of the Chemi-cal Society, 2000–2013.

12 Schmidt, G.M.J. (1964) Topochemistry. Part III. The crystal chemistry ofsome trans-cinnamic acids. Journal of the Chemical Society, 2014–2021.

13 Bertmer, M., Nieuwendaal, R.C., Barnes, A.B., and Hayes, S.E. (2006)Solid-state photodimerization kinetics of a-trans-cinnamic acid toa-truxillic acid studied via solid-state NMR. The Journal of Physical Chem-istry. B, 110, 6270–6273.

14 Fonseca, I., Hayes, S.E., and Bertmer, M. (2009) Size effects of aromaticsubstitution in the ortho position on the photodimerization kinetics ofa-trans cinnamic acid derivatives. A solid-state NMR study. PhysicalChemistry Chemical Physics, 11, 10211–10218.

15 Fonseca, I., Hayes, S.E., Blumich, B., and Bertmer, M. (2008) Temperaturestability and photodimerization kinetics of b-cinnamic acid and com-parison to its a-polymorph as studied by solid-state NMR spectroscopytechniques and DFT calculations. Physical Chemistry Chemical Physics, 10,5898–5907.

16 Nieuwendaal, R.C., Bertmer, M., and Hayes, S.E. (2008) An unexpectedphase transition during the [2+2] photocycloaddition reaction of cinnamicacid to truxillic acid: changes in polymorphism monitored by solid stateNMR. The Journal of Physical Chemistry. B, 112, 12920–12926.

�

� �

�


17 Khan, M., Brunklaus, G., Enkelmann, V., and Spiess, H.W. (2008) Tran-sient states in the [2+2] photodimerization of cinnamic acid: correlationof solid-state NMR and X-ray analysis. Journal of the American ChemicalSociety, 130, 1741–1748.

18 Jenkins, S.L., Almond, M.J., Atkinson, S.D.M. et al. (2006) The kinetics ofthe 2𝜋+2𝜋 photodimerisation reactions of single-crystalline derivativesof trans-cinnamic acid: a study by infrared microspectroscopy. Journal ofMolecular Structure, 786, 220–226.

19 Benedict, J.B. and Coppens, P. (2009) Kinetics of the single-crystal tosingle-crystal two-photon photodimerization of 𝛼-trans-cinnamic acid to𝛼-truxillic acid. The Journal of Physical Chemistry A, 113, 3116–3120.

20 Cohen, M.D. (1975) The photochemistry of organic solids. AngewandteChemie, International Edition, 14, 386–393.

21 McBride, J.M., Segmuller, B.E., Hollingsworth, M.D. et al. (1986) Mechani-cal stress and reactivity in organic solids. Science, 234, 830–835.

22 Ramamurthy, V. and Venkatesan, K. (1987) Photochemical reactions oforganic crystals. Chemical Reviews, 87, 433–481.

23 Luty, T. and Eckhardt, C.J. (1995) General theoretical concepts for solidstate reactions: quantitative formulation of the reaction cavity, stericcompression, and reaction-induced stress using an elastic multipole repre-sentation of chemical pressure. Journal of the American Chemical Society,117, 2441–2452.

24 MacGillivray, L.R., Papaefstathiou, G.S., Friscic, T. et al. (2008)Supramolecular control of reactivity in the solid state: from templates toladderanes to metal-organic frameworks. Accounts of Chemical Research,41, 280–291.

25 Kaupp, G., Herrmann, A., and Schmeyers, J. (2002) Waste-free chem-istry of diazonium salts and benign separation of coupling products insolid-state reactions. Chemistry – A European Journal, 8, 1395–1406.

26 Verman, M., Resendiz, M.J.E., and Garcia-Garibay, M.A. (2006) Large-scalephotochemical reactions of nanocrystalline suspensions: a promising greenchemistry method. Organic Letters, 8, 2615–2617.

27 Peachey, N.M. and Eckhardt, C.J. (1993) Energetics of organic solid-statereactions: lattice dynamics in the 2,5-distyrlpyrazine photoreaction. TheJournal of Physical Chemistry, 97, 10849–10856.

28 More, R., Busse, G., Hallmann, J. et al. (2010) Photodimerization ofcrystalline 9-anthracenecarboxylic acid: a nontopotactic autocatalytictransformation. Journal of Physical Chemistry C, 114, 4142–4148.

29 Finney, E.E. and Finke, R.G. (2009) Is there a minimal chemi-cal mechanism underlying classical Avrami-Erofe’ev treatmentsof phase-transformation kinetic data? Chemistry of Materials, 21,4692–4705.

�

� �

�

References 267

30 Salzillo, T., Zaccheroni, S., Valle, R.G.D. et al. (2014) Micro raman investi-gation of the photodimerization reaction of 9-cyanoanthracene in the solidstate. Journal of Physical Chemistry C, 118, 9628–9635.

31 Keating, A.E. and Garcia-Garibay, M.A. (1998) Photochemicalsolid-to-solid reactions, in Organic and Inorganic Photochemistry,1 edn, vol. 2 (eds V. Ramamurthy and K.S. Schanze), Dekker, New York,pp. 195–248.

32 Kaupp, G. (1992) Photodimerization of cinnamic acid in the solid state:new insights on application of atomic force microscopy. AngewandteChemie, International Edition, 31, 592–595.

33 Kaupp, G. (1992) Photodimerization of anthracenes in the solid state: newresults from atomic force microscopy. Angewandte Chemie, InternationalEdition, 31, 595–598.

34 Kaupp, G. and Plagmann, M. (1994) Atomic force microscopy and solidstate photolyses: phase rebuilding. Journal of Photochemistry and Photobi-ology A, 80, 399–407.

35 Enkelmann, V. (1998) Single-crystal-to-single-crystal transformations: thelong wavelength tail irradiation technique. Molecular Crystals and LiquidCrystals, 313, 15–23.

36 Harada, J., Nakajima, R., and Ogawa, K. (2008) X-ray diffraction analysisof photochromic reaction of fulgides: crystalline state reaction inducedby two-photon excitation. Journal of the American Chemical Society, 130,7085–7091.

37 Bushuyev, O.S., Tomberg, A., Friscic, T., and Barrett, C.J. (2013) Shapingcyrstals with light: crystal-to-crystal isomerization and photomechani-cal effect in fluorinated azobenzenes. Journal of the American ChemicalSociety, 135, 12556–12559.

38 Abakumov, G.A. and Nevodchikov, V.I. (1983) Thermo and photomechan-ical effects on crystals of a free-radical complex. Doklady Akademii NaukSSSR, 266, 1407–1410.

39 Lange, C.W., Foldeaki, M., Nevodchikov, V.I. et al. (1992)Photomechanical properties of rhodium(I)-semiquinonecomplexes. The structure, spectroscopy, and magnetism of(3,4-di-tert-butyl-1,2-semiquinonato)dicarbonylrhodium(I). Journal ofthe American Chemical Society, 114, 4220–4222.

40 Takahashi, S., Miura, H., Kasai, H. et al. (2002)Single-crystal-to-single-crystal transformation of diolefin derivatives innanocrystals. Journal of the American Chemical Society, 124, 10944–10945.

41 Bucar, D.K. and MacGillivray, L.R. (2007) Preparation and reactivity ofnanocrystalline cocrystals formed via sonocrystallization. Journal of theAmerican Chemical Society, 129, 32–33.

42 Kuzmanich, G., Gard, M.N., and Garcia-Garibay, M.A. (2009) Pho-tonic amplification by a singlet-state quantum chain reaction in the

�

� �

�


photodecarbonylation of crystalline diarycyclopropenones. Journal ofthe American Chemical Society, 131, 11606–11614.

43 Karunatilaka, C., Bucar, D.K., Ditzler, L.R. et al. (2011) Softening andhardening of macro- and nano-sized organic cocrystals in a single-crystaltransformation. Angewandte Chemie, International Edition, 50, 8642–8646.

44 Irie, M. (2000) Diarylethenes for memories and switches. ChemicalReviews, 100, 1685–1716.

45 Shibata, K., Muto, K., Kobatake, S., and Irie, M. (2002) Photocycliza-tion/cycloreversion quantum yields of diarylethenes in single crystals. TheJournal of Physical Chemistry A, 106, 209–214.

46 Irie, M., Kobatake, S., and Horichi, M. (2001) Reversible surface mor-phology changes of a photochromic diarylethene single crystal byphotoirradiation. Science, 291, 1769–1772.

47 Kobatake, S., Takami, S., Muto, H. et al. (2007) Rapid and reversible shapechanges of molecular crystals on photoirradiation. Nature, 446, 778–781.

48 Warner, M. (2004) Photoinduced deformations of beams, plates, and films.Physical Review Letters, 92, 134302/1–134302/4.

49 Corbett, D., Oosten, C.L.V., and Warner, M. (2008) Nonlinear dynam-ics of optical absorption of intense beams. Physical Review A, 78,013823/1–013823/4.

50 Kim, T., Zhu, L., Mueller, L.J., and Bardeen, C.J. (2014) Mechanism ofphoto-induced bending and twisting in crystalline microneedles andmicroribbons composed of 9-methylanthracene. Journal of the AmericanChemical Society, 136, 6617–6625.

51 Cohen, E.R. and Reiss, H. (1963) Kinetics of reactant isolation. I.One-dimensional problems. Journal of Chemical Physics, 38, 680–691.

52 Harris, K.D.M., Thomas, J.M., and Williams, D. (1991) Mathematical analy-sis of intra-stack dimerizations in reactive crystalline solids. Journal of theChemical Society, Faraday Transactions 2, 87, 325–331.

53 Zhu, L., Agarwal, A., Lai, J. et al. (2011) Solid-state photochemical andphotomechanical properties of molecular crystal nanorods composedof anthracene ester derivatives. Journal of Materials Chemistry, 21,6258–6268.

54 Skoko, Z., Zamir, S., Naumov, P., and Bernstein, J. (2010) The thermos-alient phenomenon. “Jumping crystals” and crystal chemistry of theanticholinergic agent oxitropium bromide. Journal of the American Chemi-cal Society, 132, 14191–14202.

55 Irie, M., Fukaminato, T., Matsuda, K., and Kobatake, S. (2014) Pho-tochromism of diarylethene molecules and crystals: memories, switches,and actuators. Chemical Reviews, 114, 12174–12277.

56 Colombier, I., Spagnoli, S., Corval, A. et al. (2007) Diarylethene micro-crystals make directional jumps upon ultraviolet irradiation. The Journal ofChemical Physics, 126, 011101/1–011101/3.

�

� �

�

References 269

57 Morimoto, M. and Irie, M. (2010) A diarylethene cocrystal that convertslight into mechanical work. Journal of the American Chemical Society, 132,14172–14178.

58 Terao, F., Morimoto, M., and Irie, M. (2012) Light-driven molecular-crystalactuators: rapid and reversible bending of rodlike mixed crystals ofdiarylethene derivatives. Angewandte Chemie, International Edition, 51,901–904.

59 Kuroki, L., Takami, S., Yoza, K. et al. (2010) Photoinduced shape changesof diarylethene single crystals: correlation between shape changes andmolecular packing. Photochemical & Photobiological Sciences, 9, 221–225.

60 Uchida, K., Sukata, S., Matsuzawa, Y. et al. (2008) Photoresponsive rollingand bending of thin crystals of chiral diarylethenes. Chemical Communica-tions, 326–328.

61 Kobatake, S., Hasegawa, H., and Miyamura, K. (2011) High-convertiblephotochromism of a diarylethene single crystal accompanying the crystalshape deformation. Crystal Growth & Design, 11, 1223–1229.

62 Kitagawa, D., Nishi, H., and Kobatake, S. (2013) Photoinduced twisting ofa photochromic diarylethene crystal. Angewandte Chemie, InternationalEdition, 52, 9320–9322.

63 Kitagawa, D. and Kobatake, S. (2013) Crystal thickness depen-dence of photoinduced crystal bending of 1,2-bis(2-methyl-5-(4-(1-naphtholyoxymethyl)phenyl)-3-thienyl)perfluorocyclopentene. Journalof Physical Chemistry C, 117, 20887–20892.

64 Kitagawa, D. and Kobatake, S. (2014) Crystal thickness dependenceof the photoinduced crystal bending of 1-(5-methyl-2-(4-p-vinylbenzoyloxymethyl)-phenyl)-4-thiazolyl)-2-(5-methyl-2-phenyl-4-thiazoyl)perfluorocyclopentene. Photochemical & Photobiological Sciences,13, 764–769.

65 Kitagawa, D., Tanaka, R., and Kobatake, S. (2015) Dependence of photoin-duced bending behavior of diarylethene crystals on irradiation wavelengthof ultraviolet light. Physical Chemistry Chemical Physics, 17, 27300–27305.

66 Timoshenko, S. (1925) Analysis of bimetal thermostats. Journal of theOptical Society of America, 11, 233–255.

67 Uchida, K., Izumi, N., Sukata, S. et al. (2006) Photoinduced reversible for-mation of microfibrils on a photochromic diarylethene microcrystallinesurface. Angewandte Chemie, International Edition, 45, 6470–6473.

68 Nanni, G., Ceseracciu, L., Oropesa-Nunez, R. et al. (2015) Tunable frictionbehavior of photochromic fibrillar surfaces. Langmuir, 31, 6072–6077.

69 Tagawa, N., Masuhara, A., Kasai, H. et al. (2010) Nanocrystallization ofdiarylethene and photochromic properties. Crystal Growth & Design, 10,2857–2859.

70 Patra, A., Metivier, R., Brisset, F., and Nakatani, K. (2012) Photochromicone-dimensional nanostructures based on dithienylethene: fabrication by

�

� �

�


light-induced precipitation and reversible transformation in the nanoparti-cle state. Chemical Communications, 48, 2489–2491.

71 Carling, C.J., Boyer, J.C., and Branda, N.R. (2009) Remote-control photo-switching using NIR light. Journal of the American Chemical Society, 131,10838–10839.

72 Koshima, H., Nakaya, H., Uchimoto, H., and Ojima, N. (2012) Photome-chanical motion of furylfulgide crystals. Chemistry Letters, 41, 107–109.

73 Liu, R.S.H. (2001) Photoisomerization by hula-twist: a fundamentalsupramolecular photochemical reaction. Accounts of Chemical Research,34, 555–562.

74 Liu, R.S.H., Yang, L.Y., and Liu, J. (2007) Mechanisms of photoisomer-ization of polyenes in confined media: from organic glasses to proteinbinding cavities. Photochemistry and Photobiology, 83, 2–10.

75 Nakano, H. (2008) Photoinduced surface relief grating formation on aco-crystal of 4-[bis(9,9-dimethylfluoren-2-yl)azobenzene and ethyl acetate.ChemPhysChem, 9, 2174–2176.

76 Nakano, H., Seki, S., and Kageyama, H. (2010) Photoinduced vitrificationnear the surfaces of single crystals of azobenzene-based molecular mate-rials with glass-forming ability. Physical Chemistry Chemical Physics, 12,7772–7774.

77 Koshima, H., Ojima, N., and Uchimoto, H. (2009) Mechanical motionof azobenzene crystals upon photoirradiation. Journal of the AmericanChemical Society, 131, 6890–6891.

78 Bushuyev, O.S., Singleton, T.A., and Barrett, C.J. (2013) Fast, reversible,and general photomechanical motion in single crystals of various azocompounds using visible light. Advanced Materials, 25, 1796–1800.

79 Bushuyev, O.S., Corkery, T.C., Barrett, C.J., and Friscic, T. (2014)Photo-mechanical azobenzene cocrystals and in situ X-ray diffractionmonitoring of their optically-induced crystal-to-crystal isomerization.Chemical Science, 5, 3158–3164.

80 Lee, J., Oh, S., Pyo, J. et al. (2015) A light-driven supramolecular nanowireactuator. Nanoscale, 7, 6457–6461.

81 Kim, T., Al-Muhanna, M.K., Al-Suwaidan, S.D. et al. (2013) Photo-inducedcurling of organic molecular crystal nanowires. Angewandte Chemie, Inter-national Edition, 52, 6889–6893.

82 Koshima, H., Takechi, K., Uchimoto, H. et al. (2011) Photmechanicalbending of salicylideneaniline crystals. Chemical Communications, 47,11423–11425.

83 Koshima, H., Matsuo, R., Matsudomi, M. et al. (2013) Light-driven bend-ing crystals of salicylidenephenylethylamines in enantiomeric and racemateforms. Crystal Growth & Design, 13, 4330–4337.

84 Naumov, P., Sahoo, S.C., Zakharov, B.Z., and Boldyreva, E.V. (2013)Dynamic single crystals: kinematic analysis of photoinduced crystal

�

� �

�

References 271

jumping (the photosalient effect). Angewandte Chemie, InternationalEdition, 52, 9990–9995.

85 Medishetty, R., Husain, A., Bai, Z. et al. (2014) Single crystals poppingunder UV light: a photosalient effect triggered by a [2+2] cycloadditionreaction. Angewandte Chemie, International Edition, 53, 5907–5911.

86 Salzillo, T., Bilotti, I., Valle, R.G.D. et al. (2012) Crystal-to-crystal pho-toinduced reaction of dinitroanthracene to anthraquinone. Journal of theAmerican Chemical Society, 134, 17671–17679.

87 Friscic, T. and MacGillivray, L.R. (2005) Single-crystal-to-single-crystal[2+2] photodimerizations: from discovery to design. Zeitschrift für Kristal-lographie, 220, 351–363.

88 Sonoda, Y. (2011) Solid-state [2+2] photodimerization and photopolymer-ization of a,w-diarylpolyene monomers: effective utilization of noncovalentintermolecular interactions in crystals. Molecules, 16, 119–148.

89 Naumov, P., Kowalik, J., Solntsev, K.M. et al. (2010) Topochemistry andphotomechanical effects in crystals of green fluorescent protein-like chro-mophores: effects of hydrogen bonding and crystal packing. Journal of theAmerican Chemical Society, 132, 5845–5857.

90 Nath, N.K., Runcevski, T., Lai, C.Y. et al. (2015) Surface and bulk effectsin photochemical reactions and photomechanical effects in dynamicmolecular crystals. Journal of the American Chemical Society, 137 (43),13866–13875.

91 Commins, P., Natarajan, A., Tsai, C.K. et al. (2015) Structure-reactivitycorrelations and mechanistic understanding of the photorearrangementand photosalient effect of 𝛼-santonin and its derivatives in solutions,crystals, and nanocrystalline suspensions. Crystal Growth & Design, 15,1983–1990.

92 Sun, J.-K., Li, W., Chen, C. et al. (2013) Photoinduced bending of a largesingle crystal of a 1,2-bis(4-pyridyl)ethylene-based pyridinium salt poweredby a [2+2] cycloaddition. Angewandte Chemie, International Edition, 52,6653–6657.

93 Kim, T., Zhu, L., Mueller, L.J., and Bardeen, C.J. (2012) Dependence of thesolid-state photomechanical response of 4-chlorocinnamic acid on crystalshape and size. CrystEngComm, 14, 7792–7799.

94 Takegoshi, K., Nakamura, S., and Terao, T. (1998) Solid-state photodimer-ization of 9-methylanthracene as studied by solid-state 13C NMR. SolidState Nuclear Magnetic Resonance, 11, 189–196.

95 Chen, Z., Majidi, C., Srolovitz, D.J., and Haataja, M. (2011) Tunable helicalribbons. Applied Physics Letters, 98, 011906/1–011906/3.

96 Zhu, L., Al-Kaysi, R.O., Dillon, R.J. et al. (2011) Crystal structures andphotophysical properties of 9-anthracene carboxylic acid derivatives forphotomechanical applications. Crystal Growth & Design, 11, 4975–4983.

�

� �

�


97 Cohen, M.D., Ludmer, Z., and Yakhot, V. (1975) The fluorescence prop-erties of crystalline anthracenes and their dependence on the crystalstructures. Physica Status Solidi (b), 67, 51–61.

98 Ito, Y. and Fujita, H. (1996) Formation of an unstable photodimer from9-anthracenecarboxylic acid in the solid state. The Journal of OrganicChemistry, 61, 5677–5680.

99 Al-Kaysi, R.O. and Bardeen, C.J. (2007) Reversible photoinduced shapechanges of crystalline organic nanorods. Advanced Materials, 19,1276–1280.

100 Al-Kaysi, R.O., Dillon, R.J., Zhu, L., and Bardeen, C.J. (2008) Templateassisted synthesis of silica-coated molecular crystal nanorods: fromhydrophobic to hydrophilic nanorods. Journal of Colloid and InterfaceScience, 327, 102–107.

101 Abumaree, M.H., Zhu, L., Bardeen, C.J. et al. (2011) Fabrication of aque-ous dispersed Taxol nanowires using modified anodic alumina oxidetemplates. RSC Advances, 1, 884–892.

102 Good, J.T., Burdett, J.J., and Bardeen, C.J. (2009) Using two-photon excita-tion to control bending motions in molecular-crystal nanorods. Small, 5,2902–2909.

103 Zhu, L., Al-Kaysi, R.O., and Bardeen, C.J. (2011) Reversible photoin-duced twisting of molecular crystal microribbons. Journal of the AmericanChemical Society, 133, 12569–12575.

104 Zhu, L., Tong, F., Salinas, C. et al. (2014) Improved solid-state photome-chanical materials by fluorine substitution of 9-anthracene carboxylic acid.Chemistry of Materials, 26, 6007–6015.

105 Zhang, Y. and Iijima, S. (1999) Elastic response of carbon nanotube bun-dles to visible light. Physical Review Letters, 82, 3472–3475.

106 Flannigan, D.J., Lobastov, V.A., and Zewail, A.H. (2007) Controllednanoscale mechanical phenomena discovered with ultrafast electronmicroscopy. Angewandte Chemie, International Edition, 46, 9206–9210.

107 Flannigan, D.J., Samartzis, P.C., Yurtsever, A., and Zewail, A.H. (2009)Nanomechanical motions of cantilevers: direct imaging in real space andtime with 4D electron microscopy. Nano Letters, 9, 875–881.

108 Zhang, Y., Peng, C., Zhou, Z. et al. (2015) Giant Morphological Changein Layered Microribbons Featuring Reversible Sliding of Stacking Layers.Advanced Materials, 27, 320–325.

109 Matsunaga, Y., Goto, K., Kubono, K. et al. (2014) Photoinduced colorchange and photomechanical effect of naphthalene diimides bearing alky-lamine moieties in the solid state. Chemistry - A European Journal, 20,7309–7316.

110 Mnyukh, Y.V., Panfilova, N.A., Petropavlov, N.N., and Uchvatova, N.S.(1975) Polymorphic transitions in molecular crystals – III. Journal ofPhysics and Chemistry of Solids, 36, 127–144.

�

� �

�

References 273

111 Williams, J.O. (1973) Phase transformations in molecular crystals. Journalof Materials Science Letters, 8, 1361–1362.

112 Anwar, J., Tuble, S.C., and Kendrick, J. (2007) Concerted molecular dis-placements in a thermally-induced solid-state transformation in crystals ofDL-norleucine. Journal of the American Chemical Society, 129, 2542–2547.

113 Zahn, D. and Anwar, J. (2013) Collective displacements in a molecularcrystal polymorphic transformation. RSC Advances, 3, 12810–12815.

114 Zaziski, D., Prilliman, S., Scher, E.C. et al. (2004) Critical size for fractureduring solid-solid phase transformations. Nano Letters, 4, 943–946.

115 Nath, N.K., Pejov, L., Nichols, S.M. et al. (2014) Model for photoinducedbending of slender molecular crystals. Journal of the American ChemicalSociety, 136, 2757–2766.

116 Hell, S.W. (2009) Microscopy and its focal switch. Nature Methods, 6,24–32.

117 Marder, S.R., Bredas, J.-L., and Perry, J.W. (2007) Materials for multipho-ton 3D microfabrication. MRS Bulletin, 32, 561–565.

118 Li, L., Gattass, R.R., Gershgoren, E. et al. (2009) Achieving 𝜆/20 resolutionby one-color initiation and deactivation of polymerization. Science, 324,910–913.

119 Scott, T.F., Kowalski, B.A., Sullivan, A.C. et al. (2009) Two-colorsingle-photon photoinitiation and photoinhibition for subdiffractionphotolithography. Science, 324, 913–917.

120 Andrew, T.L., Tsai, H.-Y., and Menon, R. (2009) Confining light to deepsubwavelength dimensions to enable optical nanopatterning. Science, 324,917–921.

121 Cao, J., Bardeen, C.J., and Wilson, K.R. (1998) Molecular “𝜋 pulse” fortotal inversion of electronic state population. Physical Review Letters, 80,1406–1409.

122 Konorov, S.O., Sidorov-Biryukov, D.A., Bugar, I. et al. (2003)Quantum-controlled color: chirp- and polarization-sensitive two-photonphotochromism of spiropyrans in the solid phase. Chemical Physics Letters,381, 572–578.

123 Suzuki, M., Asahi, T., and Masuhara, H. (2005) Cooperative photochemi-cal reaction mechanism of femtosecond laser-induced photocoloration inspirooxazine microcrystals. ChemPhysChem, 6, 2396–2403.

124 Nishi, H., Asahi, T., and Kobatake, S. (2012) Plasmonic enhancementof gold nanoparticles on photocycloreversion reaction of diarylethenederivatives depending on particle size, distand from the particle surface,and irradiation wavelength. Physical Chemistry Chemical Physics, 14,4898–4905.

125 Kuzmanich, G., Natarajan, A., Chin, K.K. et al. (2008) Solid-state photode-carbonylation of dipheylcyclopropenone: a quantum chain process made

�

� �

�


possible by ultrafast energy transfer. Journal of the American ChemicalSociety, 130, 1140–11441.

126 Al-Kaysi, R.O., Zhu, L., Al-Haidar, M. et al. (2015) Chemical reactionmethod for growing photomechanical organic microcrystals. CrystEng-Comm, 17, 8835–8842.

127 Briseno, A.L., Mannsfeld, S.C.B., Ling, M.M. et al. (2006) Patterningorganic single-crystal transistor arrays. Nature, 444, 913–917.

128 Liu, S., Mannsfeld, S.C.B., Wang, W.M. et al. (2009) Patterning ofalpha-sexithiophene single crystals with precisely controlled sizes andshapes. Chemistry of Materials, 21, 15–17.

129 Odom, T.W., Thalladi, V.R., Love, C.J., and Whitesides, G.M. (2002) Gen-eration of 30–50 nm structures using easily fabricated, composite PDMSmasks. Journal of the American Chemical Society, 124, 12112–12113.

130 Minemawari, H., Yamada, T., Matsui, H. et al. (2011) Inkjet printing ofsingle-crystal films. Nature, 475, 364–367.

131 Schiek, M., Balzer, F., Al-Shamery, K. et al. (2008) Organic molecular nan-otechnology. Small, 4, 176–181.

132 Harris, K.D., Cuypers, R., Scheibe, P. et al. (2005) Large amplitudelight-induced motion in high elastic modulus polymer actuators. Journal ofMaterials Chemistry, 15, 5043–5048.

133 Koshima, H., Matsudomi, M., Uemura, Y. et al. (2013) Light-driven bend-ing of polymer films in which salicylidenephenylethylamine crystals arealigned magentically. Chemistry Letters, 42, 1517–1519.

134 Lan, T. and Chen, W. (2013) Hybrid nanoscale organic molecular crystalsassembly as a photon-controlled actuator. Angewandte Chemie, Interna-tional Edition, 52, 6496–6500.

135 Sahoo, S.C., Nath, N.K., Zhang, L. et al. (2014) Actuation based onthermo/photosalient effect: a biogenic smart hybrid driven by light andheat. RSC Advances, 4, 7640–7647.

136 Tsujioka, T., Sesumi, Y., Takagi, R. et al. (2008) Selective metal depositionon photoswitchable molecular surfaces. Journal of the American ChemicalSociety, 130, 10740–10747.

137 Kitagawa, D. and Kobatake, S. (2015) Photoreversible current ON/OFFswitching by the photoinduced bending of gold-coated diarylethene crys-tals. Chemical Communications, 51, 4421–4424.

�

� �

�

275

8

Photomechanical Effects in Piezoelectric CeramicsKenji Uchino

International Center for Actuators and Transducers, Electrical Engineering and Materials Research Institute,The Pennsylvania State University, University Park, PA, USA

8.1 Introduction

The continuing thrust toward greater miniaturization and integration ofmicrorobotics and microelectronics has resulted in significant work towardthe development of piezoelectric actuators. One of the bottlenecks of thepiezoactuator is its necessity of the electric lead wire, which is too heavy fora miniaturized self-propelling robot of less than 1 cm3. The important reasonis a drastic reduction of the propelling friction force due to the increase inspecific area, that is, surface area/volume or weight ratio. “What if you, anexpert on actuators, could produce a remote-controlled actuator that wouldbypass the electrical lead?” To many people, “remote control” equals controlby electromagnetic waves (radio, light, or X-ray waves) or sound energy.Light-controlled actuators with piezoelectric ceramics require that lightenergy be transduced twice: first from light energy to electrical energy andsecond, from electrical energy to mechanical energy. These are “photovoltaic”and “piezoelectric” effects. A solar cell is a well-known photovoltaic device,but it does not generate sufficient voltage to drive a piezoelectric device; inother words, this combination fails due to the electric impedance mismatch.The key to success is to adopt a high-impedance photovoltaic effect (theso-called anomalous or bulk photovoltaic effect in piezoelectrics), which istotally different from the p–n junction-based solar cell.Thirty years after our discovery of the “photostrictive effect” [1] − which

directly converts the photonic energy to mechanical motion − it has recently

* Most of the figures in this chapter were reproduced from Chapter 5 of ‘Optical Nano and MicroActuator Technology’ edited by George K. Knopf, Yukitoshi Otani, 2012 by CRC Press with thepermission of CCC Republication.


�

� �

�

276 8 Photomechanical Effects in Piezoelectric Ceramics

started to draw significant attention for its potential usage in microactuationandmicrosensing applications.Optical actuators are also anticipated to be usedas the driving component in optically controlled electromagnetic-noise-freesystems.The photostrictive effect will also be used in fabricating a “photophon-ic” device, where light is transformed directly into sound from the mechanicalvibration induced by intermittent illumination at a human-audible frequency.As is well known, the original idea of “photophone” can be found in the patentsubmitted by Alexander Graham Bell in 1880, but our development seems tobe the first feasibility demonstration in practical solid-state devices.Our group studied the photomechanical effects mainly in ferroelec-

tric/piezoelectric polycrystalline materials based on lead zirconate titanate(PZT) for potential commercial applications. Lanthanum-modified PZT (i.e.,PLZT) ceramic is one of the most promising photomechanical materials dueto its relatively high piezoelectric coefficient and ease of fabrication. However,our previous studies have shown that for commercial applications, improve-ments in photovoltaic efficiency and response speed of the PLZT ceramicsare still essential. The improvement in photomechanical properties requiresconsideration of several parameters, such as material parameters, processingcondition and microstructure, and sample configuration and performancetesting conditions.This chapter reviews the theoretical background for the photomechanical

effect in piezoelectric ceramics first and, then, enhanced performance throughthe composition modification, sample preparation technique (thicknessand surface characteristics of the sample). Its potential future applicationsare briefly described finally. Since our group used “photostrictive” contin-uously rather than the terminology “photomechanical,” the author uses“photostriction” in this chapter.

8.2 Photovoltaic Effect

The photostriction phenomenon in piezoelectrics was discovered by Dr Brodyand the author, independently almost at the same time in 1981 [1, 2]. Owingto the lack of communication and delay of the publication review process, wedo not know exactly who discovered it first. In principle, photostrictive effectin our discovery arises from a superposition of the “bulk” photovoltaic effect,that is, generation of high voltage from the irradiation of light, and the conversepiezoelectric effect, that is, expansion or contraction under the voltage applied[1]. The photostrictive phenomenon has been observed in certain ferroelec-tric/piezoelectric materials. By doping suitable ionic species, the photovoltaiceffect is introduced in the material. The figure of merit (FOM) for photostric-tion magnitude is generally expressed as the product of photovoltage (electricfield), Eph, and the piezoelectric constant, d33, while the FOM for responsespeed is determined by the photocurrent (current density), Iph, as d33Iph/C (C:

�

� �

�

8.2 Photovoltaic Effect 277

capacitance of the photostrictive device). Therefore, for application purposes,enhancement and/or optimization of photostrictive properties requires consid-eration of both the terms in the FOM, that is, photovoltaic voltage and current,as well as its piezoelectric d constant. Recently, PLZT ceramics have gainedconsiderable attention due to their excellent photovoltaic properties, high d33,and ease of fabrication. We review the background of photovoltaic effect firstin this section.

8.2.1 Principle of the Bulk Photovoltaic Effect

8.2.1.1 “Bulk” Photovoltaic EffectWhen a noncentrosymmetric piezoelectric material (with some dopants)is illuminated with uniform light having a wavelength corresponding tothe absorption edge of the material, a steady photovoltage/photocurrent isgenerated [3]. Someone may be suspicious about the distinction betweenthe photovoltaic effect and pyroelectric effect (i.e., voltage/charge gen-eration due to the temperature change). Figure 8.1 demonstrates thedifference, where illumination responses of photovoltaic current are plot-ted under two different external resistances in 1.5mol% MnO2-doped0.895PbTiO3–0.105La(Zn2/3Nb1/3)O3 ceramic [4]. Mercury lamp illuminationon this ceramic sample slightly increased the sample temperature, leading tothe initial voltage peak (up to 8mV through 10MΩ resistor) for a couple of

R = 10 MΩ

1.82 MΩ

100 200

Vout

Light on Light off

Current source

I0 = 0.4 nA

8

6

4

2

0

Outp

ut vo

ltage (

mV

)

–2

–4

Time (s)

R

p

Hg lamp

Figure 8.1 Illumination responses of photovoltaic current for 1.5 mol% MnO2-doped 0.895PbTiO3-0.105 La(Zn2/3Nb1/3)O3 ceramic.

�

� �

�


Sample

Sensor

Displacementsensor

Millitrondisplacement meter

Oscilloscope

Highpressuremercury

lamp

IRblocking

filter(>700nm)

Bandpassfilter

(248–390nm)

Electrometer(Keithley 617)

voltampohm

Figure 8.2 Experimental setup for measuring photovoltaic and photostrictive effects.

tens of seconds. However, note that the output voltage is stabilized around4mV after the temperature stability was obtained. The magnitude of thesteady current is independent of the externally connected resistance. Whenthe illumination was shut off, the negative pyrocurrent was observed due to aslight temperature decrease again for tens of seconds. But, the output voltagebecame completely zero after the saturation, which verified that there was nojunction (piezoelectric ceramic–metal electrode) effect. The reader can nowclearly understand the difference between the photovoltaic and pyroelectriceffects from this demonstration. Note that we can eliminate the pyroelectriceffect when we use an IR-blocking filter for reducing the longer wavelengthlight intensity (refer to Figure 8.2).In some materials, the photovoltage generated is greater than the band-gap

energy and can be of the order of several kilovolts per centimeter. This phe-nomenon, thus referred to as the “bulk” or “anomalous” photovoltaic effect(APV), seems to be totally different from the corresponding phenomenon inthe p–n junction of semiconductors (e.g., solar battery) [5, 6].The APV effect isobserved primarily in the direction of the spontaneous polarization (PS) in theferroelectric material (refer to Section 8.2.2), and the generated photovoltageis proportional to the sample length along the PS direction.The origin of photovoltaic effect is not yet clear, even though several models

have been proposed on the mechanism of photovoltaic effect. The key featuresof the APV effect are summarized as follows:

�

� �

�


1. This effect is observed in a uniform crystal or ceramic having noncen-trosymmetry and is entirely different in nature from the p–n-junction effectobserved in semiconductors.

2. A steady photo voltage/current is generated under uniform illumination.3. The magnitude of the induced voltage is greater than the band gap of the

crystal.

Even our group previously proposed two models: current source model andvoltage source model, which are introduced in the following sections.

8.2.1.2 Experimental SetupPrior to the detailed discussion, the measuring setup is described here (refer toFigure 8.2). PLZT ceramic samples are cut into the standard sizes of 5× 5mm2

and polished to 1mm thickness. Note that there is thickness dependence of theexperimental results as discussed in Section 8.3.3.The samples are poled alongthe length (5mm) under a field of 2 kV/mm at 120 ∘C for 10min. The ceramicpreparation methods are described in Sections 8.2.3 and 8.3.3.The radiation from a high-pressure mercury lamp (Ushio Electric

USH-500D) is passed through infrared-cut optical filters in order to minimizethe thermal/pyroelectric effect. The light with the wavelength peak around366 nm, where the maximum photovoltaic effect of PLZT is obtained, isthen applied to the sample. A xenon lamp is alternatively used to measurethe wavelength dependence of the photovoltaic effect. The light sourceis monochromated by a monochromator to 6 nm HWHM (half-width athalf-maximum).The photovoltaic voltage under illumination generally reaches several

kilovolts per centimeter, and the current is on the order of nanoamperes. Theinduced current is recorded as a function of the externally applied voltage overa range of −100 to 100V, by means of a high-input impedance electrometer(Keithley 617). The photovoltaic voltage and current are determined from theintercepts of the horizontal and the vertical axes, respectively. An examplemeasurement is shown in Figure 8.3. The photovoltage (typically kilovolt)is estimated by the linear extrapolation method. Photostriction is directlymeasured by a differential transformer or an eddy current displacement sensor.

8.2.1.3 Current Source ModelTaking the necessity of both doping and crystal asymmetry into account, weproposed a current source model, as illustrated in Figure 8.4, which is basedon the electron energy-band model for (Pb,La)(Zr,Ti)O3 (PLZT) [7, 8]. Theenergy band is basically generated by the hybridized orbit of p-orbit of oxygenand d-orbit of Ti/Zr. The donor impurity levels induced in accordance withLa doping (or other dopants) are present slightly above the valence band (esti-mated from the photocurrent peak wavelength as shown in Figure 8.5). The

�

� �

�


Current (nA)

Photocurrent

Photovoltage

Voltage (v)

Conductivity

3

2

1

0–50 50 100

Figure 8.3 Photocurrentmeasured as a function ofapplied voltage underillumination.

Conduction band

Valence band

Lightillumination380 nm

3.26 eVEg = 3.3 eV

Figure 8.4 Energy band-gap model of excited electron transition from deep donor-impuritylevel in PLZT.

transition from these levels with an asymmetric potential due to the crystal-lographic anisotropy may provide the “preferred” momentum to the electron.Electromotive force is generated when electrons excited by light move in acertain direction of the ferroelectric/piezoelectric crystal, which may arisealong the spontaneous polarization direction. The asymmetric crystal exhibit-ing a photovoltaic response is also piezoelectric in principle, and therefore,a “photostrictive” effect is expected as a coupling of the bulk photovoltaicvoltage (Eph) with the piezoelectric constant (d).The photocurrent Jph varies in proportion to the illumination intensity I:

Jph = 𝜅𝛼I, (8.1)

where 𝛼 denotes the absorption coefficient and 𝜅 is a Glass constant (namedaccording to Glass’s contribution to the APV effect) [9]. On the other hand, thephotovoltage Eph shows saturation caused by a large photoconductive effect,

�

� �

�


Figure 8.5 Wavelengthdependence of photovoltaiccurrent in 0.895PT–0.105LZNand PLZT (3/52/48).

3

2

1

0300 400 500

Wavelength (nm)

PLZT3/52/48

0.895PT–0.105LZN

Photo

voltaic

curr

ent (μ

A/W

)

represented by

Eph = 𝜅𝛼I∕(𝜎d + 𝛽I), (8.2)

where 𝜎d is the dark conductivity and 𝛽 a constant relating to the photocon-ductivity.This model is validated:

1. The photovoltaic current is constant as shown in Figure 8.1, regardless ofthe externally connected resistance.

2. The photocurrent Jph is strongly dependent on the wavelength under con-stant intensity of illumination, suggesting a sort of band gap, as shown inFigure 8.5. A sharp peak is observed at 384 or 372 nm near the absorptionedge for 0.895PT–0.105LZN or PLZT (3/52/48), respectively. The donorlevel seems to be rather deep, close to the valence-band level.

3. The linear relationship of the photocurrentwith light intensity (Equation 8.1)is experimentally verified in Figure 8.6, where photoinduced short-circuitcurrent Jph (a) and the open-circuit electric field Eph (b) are plottedas a function of illumination intensity I for pure and MnO2-doped0.895PT–0.105LZN [4].

�

� �

�


0

1

0.8

Photo

voltage E

ph (

×10

5 V

/m)

Photo

curr

ent

J ph (

×10

–4 A

/m2)

J ph (

×10

–5 A

/m2)

0.6

0.4

0.2

02 4

Light intensity (×102 W/m2)

Mn-doped

Undoped

Mn-doped

Undoped

6 8 10

0

4

3

2

1

0

1.5(a)

(b)

1

0.5

2 4

Light intensity (×102 W/m2)

6 8 10

Figure 8.6 Short-circuit current Jph

(a) and open-circuit electric field Eph

(b) as a function of illuminationintensity I for pure and MnO2-doped0.895PT–0.105LZN.

8.2.1.4 Voltage Source ModelIn this model, the photovoltaic properties are attributed to the photocarriersand internal electric fields generated by near-UV illumination.The optical non-linearity of the second order, which is popularly introduced in ferroelectrics, isproposed as the origin of photoinduced dc field generation [10]. The expres-sion for the polarization of dielectrics, considering the nonlinear effect up tothe second order, is given by [11]

P = 𝜀0(𝜒1Eop + 𝜒2E2op), (8.3)

where 𝜀0 is the permittivity of vacuum, 𝜒1 the linear susceptibility, 𝜒2 thenonlinear susceptibility of the second order, and Eop the electric field of theillumination beam at an optical frequency (THz).In dielectrics, the value of the local electric field is different from the value of

the external electric field. For simplicity, the local field in dielectrics has beenapproximated using the Lorentz relation for a ferroelectricmaterial as Ref. [12]:

�

� �

�


Elocal = E + 𝛾P3𝜀0

, (8.4)

where E is the external electric field and 𝛾 the Lorentz factor. When an alter-nating electric field at an optical frequency is applied (i.e., light illumination),the average of the local electric field Elocal is not zero but can be calculated as

Elocal =16𝛾𝜒2E2

op. (8.5)

It must be noted that Equation 8.5 has been derived for a coherent propagationof the light wave at a single frequency. However, the condition of coherent illu-minationmay not be satisfied in our experimental conditions, where amercurylamp is used as a light source.The nonlinear effect will be affected by the degreeof coherence. Therefore, considering the depression of nonlinear effect due tothe incoherency, the expression for the effective dc field induced by incoherentlight source may be modified as:

Elocal = c1𝛾𝜒2(E2op)𝛽 , (8.6)

where c1 is a constant and 𝛽 a parameter expressing the depression effect. Thevalue of parameter 𝛽 is expected to lie between 0 and 1. Replacing the variableEop

2 with the intensity (Iop) (Ref. [11]), the following expression for the averageinduced (dc) field due to the incoherent light can be obtained:

Edc = Elocal = c2𝛾𝜒2(Iop)𝛽 , (8.7)

where c2 is a constant and Edc the effective dc field for photoinduced carriers.Note that the induced field, Edc, is proportional to the nonlinear susceptibilityas well as the Lorentz factor, 𝛾 .The photoconductivity can be obtained as a function of light intensity, Iop,

𝜎op = c3q𝜇

√

IopR, (8.8)

where q is the charge of the photocarrier, μ the carrier mobility, R the recom-bination rate of the carrier, and c3 a constant. Since the photocurrent is pro-vided by the product of the photoconductivity and the photoinduced dc field(Jph = 𝜎opEdc), we finally obtain

Jph = c4q𝜇𝛾𝜒2

√1R(Iop)

𝛽+ 12 , (8.9)

where c4 is another constant. Equations 8.8 and 8.9 provide a correlation forthe photovoltaic response of ferroelectrics on the basis of optical nonlinearity.Themodel validation and analysis aremade by the light intensity dependence

of photovoltaic properties.The experiments were conducted on PLZT 3/52/48samples with 1mm and 140 μm in thickness. Figure 8.7(a) shows the plot of

�

� �

�


(a) (b)

(c) (d)

0.1

1

1 10

Light intensity (mW/cm2)

: Experimental data

: Fitted by σop = 0.34 (Iop)0.54P

hoto

conductivity

(10

–12 Ω

–1)

0.1

1

10

1 10

Photo

voltage (

V/c

m)


: Experimental data: Fitted by Eph = 0.66 (Iop)

0.50

0.1

1

1 10

Photo

curr

ent (n

A/c

m)


: Fitted by Jph = 0.26 (Iop)0.96

: Experimental data

0.01

0.1

1

1 10 100

Photo

curr

ent (n

A/c

m)


: Experimental data: Fitted by Jph = 0.016 (Iop)1.3

Figure 8.7 Dependence of (a) photoconductivity, (b) photovoltage, and (c) photocurrent onillumination intensity in a PLZT 3/52/48 sample with 1 mm in thickness; (d) the result for asample with 140 μm in thickness.

photoconductivity (𝜎op) as a function of light intensity (Iop).The exponent relat-ing the photoconductivity and the light intensity was calculated to be 0.54.Thisis in good agreement with the value of 0.5 derived for the recombination pro-cess of the carriers (Equation 8.8). Note the difference fromEquation 8.2, wherewe assume the photoconductivity directly in proportion to the intensity (referto Figure 8.6). Figure 8.7(b) shows the experimental results of the open-circuitphotovoltage (Eph) as a function of light intensity. The photovoltage was foundto be proportional to the square root of the light intensity, leading to 𝛽 = 0.5(Equation 8.7). Figure 8.7(c) shows the results of short-circuit photocurrent(Jph) as a function of Iph.The parameter 𝛽 based on Equation 8.9 was calculatedto be 0.46, which is very close to the aforementioned 𝛽 value.The depression in𝛽 value can be attributed to the incoherent illumination of the mercury lamp.Note again that Jph is almost directly proportional to Iph, in accordance withEquation 8.1 (Figure 8.6).Investigation was further made in terms of the illumination coherency. Since

a partial coherence of light can be achieved in a very small area, an increase in 𝛽value is expected in thinner photovoltaic samples.The photocurrent measured

�

� �

�


as a function of light intensity in a very thin (140μm)PLZT sample (Figure 8.7d)resulted in the parameter 𝛽 (Equation 8.9) to be 0.80, which is higher than the 𝛽value of 0.46 in the thicker sample (1mm thickness). These results suggest thatthe parameter 𝛽 increases with a decrease in the thickness of photovoltaic sam-ple, due to higher coherency of illumination in thinner samples. This suggeststhat an enhancement in the photovoltaic properties may be achieved in a verythin sample or by using coherent illumination. As suggested already, we cannotconclude, at present, which model fits better for the experiments, the currentsource or the voltage source.

8.2.2 Effect of Light Polarization Direction

Effect of the light polarization direction on the photovoltaic phenomenon alsohelps with understanding the mechanism. Figure 8.8 shows the measuring sys-tem of the dependence of photovoltaic effect on light polarization direction

0

1.1

Polarizationdirection

Photovoltaicsample

(a)

(b)

Ps

θ

Polarizer Lens Mercurylamp

1.05

Photo

curr

ent change (

J/J 0

)

1

45 90

Polarizer rotation angle (°)

135 180

Photo

voltage c

hange (

V/V

0)

Figure 8.8 (a) Measuring system of the dependence of photovoltaic effect on lightpolarization direction. (b) Photovoltaic voltage and current as a function of the rotationangle.

�

� �

�


(a), and the photovoltaic voltage and current as a function of the rotation anglemeasured for the PLZT (3/52/48) polycrystalline sample (b).The rotation angle𝜃 was taken from the vertical spontaneous polarization direction. Even in apolycrystalline sample, both the photovoltaic voltage and current provide themaximum at 𝜃 = 0∘ and 180∘ and the minimum at 𝜃 = 90∘; this also indicatesthat the contributing electron orbit may be the p-d-hybridized orbit mentionedearlier (i.e., the perovskite Zr/Ti-O direction). This experiment is also impor-tant when the photostriction is employed to “photophones,” where the sampleis illuminated with the polarized light traveling through an optical fiber.

8.2.3 PLZT Composition Research

Since the FOM of the photostriction is evaluated by the product of thephotovoltatic voltage and the piezoelectric constant, that is, d⋅Eph, Pb(Zr,Ti)O3(PZT)-based ceramics are the primary focus because of their excellent piezo-electric properties, that is, high d values. Lanthanum-doped PZT (PLZT)is one of such materials with La3+ donor doping in the A-site, which is alsofamous as a transparent ceramic (good sinterability) applicable to electroopticdevices.PLZT (x/y/z) samples were prepared in accordance with the following

composition formula:

Pb1−xLax(ZryTiz)1−x∕4O3. (y + z = 1).

As discussed in Figure 8.12 in detail, the piezoelectric d coefficient exhibitsthe maximum around the morphotropic phase boundary (MPB) between thetetragonal and rhombohedral phases, our composition search was also focusedaround the MPB compositions. Figure 8.9 shows the photocurrent Jph for var-ious PLZT compositions with tetragonal and rhombohedral phases, plotted asa function of their remanent polarization Pr.

0.6

0.4

0.2

025

9/60/408/58/42

6/56/44

7/62/38

5/55/45

2/50/50

3/52/48

4/56/44

3/60/40

4/58/42

4/60/40

4/66/34

30 35

Remanent polarization (×10–2 C/m2)

PLZT (x/y/1–y)

Tetragonal

Rhombohedral

Ph

oto

volta

ic c

urr

en

t (n

A)

40 45

Figure 8.9 Interrelation of photovoltaic current with remanent polarization in PLZT family.

�

� �

�


1. Significantly large photocurrent is observed for the tetragonal compo-sition PLZT (3/52/48) [13]. This is the major reason why many data inthis paper were taken for this composition. The details are discussed inSection 8.3.2.

2. The relation Jph ∝ Pr first proposed by Brody [14, 15] appears valid for thePLZT system. Further, it is worth noting that the Pr value capable of pro-ducing a certain magnitude of Jph is generally larger in the rhombohedralsymmetry group than in the tetragonal group. The average remanent polar-ization exhibiting the same magnitude of photocurrent differs by 1.7 timesbetween the tetragonal and rhombohedral phases, which is nearly equal to√3, the inverse of the direction cosine of the [1 1 1] axis in the perovskite

structure. This suggests that the photoinduced electron excitation is relatedto the (0 0 1) axis-oriented orbit, that is, the hybridized orbit of p-orbit ofoxygen and d-orbit of Ti/Zr [16].

8.2.4 Dopant Research

Photovoltaic effect is caused by the dopant in a ferroelectric/piezoelectric crys-tal, as we discussed in Section 8.2.1. La3+ seems to be the primary dopant inPb(Zr,Ti)O3. Additional impurity doping on PLZT also affects the photovoltaicresponse significantly. Figure 8.10 shows the photovoltaic response for variousdopants with the same concentration of 1 at.% into the base PLZT (3/52/48)under an illumination intensity of 4mW/cm2 at 366 nm [8]. The dashed linein Figure 8.10 represents the constant power curve corresponding to the non-doped PLZT (3/52/48). Photovoltaic power is enhanced by donor doping ontothe B-site (Nb5+, Ta5+, W6+). On the contrary, impurity ions substituting atthe A-site and/or acceptor ions substituting at the B-site, whose ionic valences

5.0

K1+

Mg2+

Bi3+

Sn4+

Si4+

Ba2+

Na1+

Al3+

Y3+

Ta5+

W5+

Nb5+

Fe3+

Photo

curr

ent (n

A/c

m) 4.0

3.0

2.0

1.0

00.5 1.0

Photo-induced voltage (kV/cm)

1.5 2.0

Acceptor A-site

Undoped

Power-const.

Acceptor B-site

Donor A-site

Donor B-site

Figure 8.10 Photovoltaic response of PLZT (3/52/48) for various impurity dopants(illumination intensity: 4 mW/cm2).

�

� �

�


0.0

20

15

10

5

0

Photo

curr

ent (n

A/c

m)

Energ

y (μW

/cm

2)

0.2

Photocurrent

Voltage

Energy

0.4

Concentration of WO3 doping (at.%)

Tip

dis

pla

cem

ent (×

10 μ

m)

Photo

induced v

oltage (

×10

1 k

V/c

m)

0.6 0.8 1.0

Displacement

Figure 8.11 Photovoltatic current, voltage, power, and tip displacement of a bimorphspecimen as a function of dopant concentration in WO3-doped PLZT (3/52/48).

are small (1–4), degrade the effect on the performance. Figure 8.11 shows thephotovoltaic response plotted as a function of atomic percent of WO3-dopingconcentration [6]. Note that the maximum power is obtained at 0.4 at.% of thedopant, due to a significant enhancement in the current density.

8.3 Photostrictive Effect

8.3.1 Figures of Merit

The figures of merit for photostriction are derived here. The photostriction isinduced as a function of time, t, as

xph = d33Eph

(

1 − exp( −t

RC

))

, (8.10)

where xph is the photoinduced strain, d33 the piezoelectric constant of themate-rials, Eph the photovoltage, Iph the photocurrent, t the time, and R and C theresistance and capacitance, respectively, of the material.

1. For t≪ 1, we obtain

xph = d33Eph

( tRC

)

. (8.11)

Thus, the FOM for response speed should be provided by d33Eph

(1

RC

)

.

Taking the relation Iph=Eph

R, into account, this FOM is transformed to d33Iph

C.

Or, it can be given by d33Iph𝜀

(𝜀: permittivity).

�

� �

�

8.3 Photostrictive Effect 289

2. On the other hand, for t≫ 1, the saturated strain is provided by

xph = d33Eph. (8.12)

Thus, the FOM for the magnitude of strain is defined by d33Eph.

In order to obtain high photoinduced strain, materials with high d33 and Ephare needed. On the contrary, for high response speed such as photophonicapplications, materials with high d33, Iph, and low dielectric constant 𝜀 arerequired.

8.3.2 Materials Considerations

We reconsider the optimum compositions in the PLZT system from thephotostrictive actuator’s viewpoint. Figure 8.12(a), (b), and (c) shows contour

0

450

Rhombohedral RhombohedralTetragonal Tetragonal

Rhombohedral Tetragonal

338 443 267 0.44

0.72

1.11

0.36 0.21 0.68 0.44

1.19

1.19 1.18 2.05

0.85 1.31 1.16

1.45 0.71 0.90

0.83 1.01

0.16

6586971054749

961 901

252

468

392

422

457287

435

468

372

366

338 272

272

242

317 210 187

d33

(×10–12 m/V)

197

160

228

199 145

144 1025

126

951 916

864 1002 298 397

58/42

at.% PZ at.% PT at.% PTat.% PZy/1–y y/1–y

Eph = photovoltage (V/cm) Iph= photocurrent (nA/cm)

56/44 54/46 52/48 50/50 48/52 46/54 44/56 58/42 56/44 54/46 52/48 50/50 48/52 46/54 44/56

1

2

3

4at.%

La

5

6

(a) (b)

(c)

0

1

2

3

4at.%

La

5

6

0

at.% PTat.% PZ y/1–y

58/42 56/44 54/46 52/48 50/50 48/52 46/54 44/56

1

2

3

4at.%

La

5

6

2435

2.81

Figure 8.12 Contour maps of (a) photovoltatic voltage Eph, (b) photocurrent Iph, and (c)piezoelectric constant d33 in the PLZT (x/y/1− y) system.

�

� �

�


maps of photovoltatic voltage Eph, photocurrent Iph, and piezoelectric constantd33 on the PLZT (x/y/1− y) phase diagram, respectively [17]. There is theMPB between the rhombohedral and tetragonal phases around 52∼56% ofZr concentration y. As well known, the piezoelectric coefficient exhibits themaximum along the MPB. The photovoltaic effect is also excited aroundthe MPB. However, precisely speaking, the photovoltage was found to bemaximum at PLZT 5/54/46, while the maximum photocurrent was found atPLZT 4/48/52. In Figure 8.12(a) and (b), the solid circles indicate the locationof PLZT 3/52/48, which had been reported earlier to exhibit the maximumphotovoltage and current. In the finer measurement, the maximum photo-voltage and current have been found at different compositions of the PLZTsystem, both still being in the tetragonal phase. In conclusion, the FOM d33Ephis maximum for PLZT 5/54/46, while the maximum of the FOM d33Iph/C isfor PLZT 4/48/52. Refer to a similar composition study by Nonaka et al. [18].

8.3.3 Ceramic PreparationMethod Effect

8.3.3.1 ProcessingMethodFabrication and processing methods have been reported to profoundlyinfluence the photovoltaic properties and strain responses of PLZT ceramics[16, 19, 20]. This effect comes through the influence of processing methodson the microstructure and other physical properties such as density, porosity,and chemical composition. Ceramic materials with high density, low porosity,better homogeneity, and a good control of stoichiometry are desired forenhanced photovoltaic and photostrictive properties. Coprecipitation andsol–gel techniques are two of the chemical routes that have the inherentadvantage in producing high-density homogeneous ceramics with a greatercontrol of stoichiometry. Therefore, processes to fabricate photostrictiveceramics via chemical routes with suitable nonoxide precursors are attractive.PLZT ceramics prepared by sol–gel and coprecipitation techniques exhibitbetter photovoltaic and photostrictive properties as compared to the oxidemixing process [19, 20]. Ceramics prepared by solid-state reaction have com-positional variation and inhomogeneous distribution of impurities, whereasthe ceramics prepared by chemical synthesis exhibited high purity with goodchemical homogeneity at the nanometer scale.

8.3.3.2 Grain Size EffectEven when the composition is fixed, the photostriction still depends stronglyon the sintering condition or, in particular, grain size [16, 21]. Figure 8.13 showsthe dependence of the photostrictive characteristics on the grain size. As is wellknown, the piezoelectric coefficient d33 gradually decreaseswith decreasing thegrain size down to 1 μm range. On the contrary, photovoltage increases dras-tically with a decrease in grain size, and the photocurrent seems to exhibit the

�

� �

�


400

300

200

100

00 1 2

Grain size (μm)

Light intensity (40 W/m2) Light intensity (40 W/m2)

Ph

oto

volta

ge

(kV

/m)

Ph

oto

cu

rre

nt (n

A/m

)

Str

ain

(×

10

–4)

Grain size (μm)

3 4 0 1 2 3 4

400

350

300

d33 (

pm

/V)

120 1.6

1.2

0.8

0.4

0

90

60

30

0

Figure 8.13 Grain size dependence of photostrictive characteristics in PLZT (3/52/48).

maximum at around 1 μm.Thus, the photostriction exhibits a drastic increasesimilar to the photovoltage change. The smaller grain sample is preferable, if itis sintered to a high density.

8.3.3.3 Surface/Geometry DependenceSince the photostrictive effect is excited by the absorption of illumination in thesurface layer of ceramics, it is apparent that the surface geometry of the photo-strictive material will have a strong bearing on the generation of photocurrentand photovoltage. Using a sample thickness closer to the penetration depthwill ensure that the entire film will be active and efficiently utilized. We alsodiscussed on the light coherency for the “thin” sample shown in Figure 8.7.Therefore, investigation of photovoltaic response as a function of sample thick-ness is desired in determining the optimal thickness range with maximumphotovoltaic effect. In addition, studying the effect of surface roughness willprovide an insight into the absorption dependence of photostriction.In order to determine the optimum sample thickness, dependence of pho-

tovoltaic effect on sample thickness of PLZT (3/52/48) ceramics doped with0.5 at.%WO3 was examined [22]. Photovoltaic response was found to increasewith a decrease in sample thickness in PLZT ceramics (refer to Figure 8.14).A model was proposed in Figure 8.15 to explain and quantify the observed

influence of sample thickness on photovoltaic response [22], where theabsorption coefficient is assumed to be independent of light intensity andthe photocurrent density is taken to be proportional to light intensity. Thesample is assumed to comprise thin slices along the thickness direction of thesample. Figure 8.14 shows the plot between the normalized photocurrent (im)and sample thickness calculated for the external resistance (Rm = 200TΩ). Thecomputed result shows good agreement with the experimental data (◽ is forthe measured photocurrent, and ⋅ for the computed results from the proposedmodel). With increasing sample thickness, im increases, reaches a maxima,and subsequently decreases with the sample thickness. The decrease in im can

�

� �

�


0

2

4

6

8

10

12

0 200 400 600 800 1000

Proposed model

Experimental results

No

rma

lize

d p

ho

tocu

rre

nt

(nA

/cm

)

Sample thickness (μm)

Figure 8.14 Comparison of measured and computed normalized photocurrents withphotovoltaic coefficient (im/k) of 0.5 at.% WO3-doped PLZT (3/52/48).

Thickness

Length

Illumination

Width

IW

dt

Representedby

ii

im

im

i1 i2 R2 in Rn Rm

i0 R0 io Ro Rm

io– im

R1

Ri

...n...

Figure 8.15 Model to compute the dependence of photocurrent on sample thickness. Thesample was modeled as thin slices along the thickness direction and the correspondingcircuit diagrams are also shown.

�

� �

�


mainly be attributed to the dark conductivity (𝜎d). The optimum thickness (forthe present set of samples), which yields maximum photocurrent, is foundat 33 μm, which is close to the light (366 nm) penetration depth of the PLZT(absorption coefficient 𝛼 of PLZT (3/52/48) = 0.0252 μm−1 at 366 nm; theinverse of 𝛼 = 39 μm). The relatively low value of optimum thickness impliesthat the lower sample thickness will be expected to give better photovoltaicresponse.The effect of surface roughness on photovoltaic and photostrictive properties

was also examined in the PLZT sample, with different surface roughness valuesobtained by polishing to different surface finishes. The surface roughness wasmeasured by a profilometer (Tencor, Alpha-Step 200), and the average surfaceroughness was determined using the graphical center line method. The vari-ation of photovoltaic current with surface roughness is plotted in Figure 8.16[10]. The photocurrent increases exponentially with decreasing surface rough-ness. This is due to the fact that with an increase in surface roughness, thepenetration depth of the illumination decreases, while contributions frommul-tiple reflections increase. A model based on the effect of multireflection hasbeen proposed for two different shapes: a sine profile and a “V” profile rough-ness. In both these shapes, half of the up-down amplitude was taken as a rough-ness (r) and the cyclic distance period as a roughness pitch (g).The normalizedphotocurrents (im) computed for the aforementioned two surface profiles arealso plotted in Figure 8.16 as a function of surface roughness. A distance pitch(wavelength) of roughness at 1 μmgave the best fit for the experimental results,which is close to the size of the grain of this PLZT sample.

0

0.5

1

1.5

2

2.5

0 0.5 1 1.5 2 2.5 3 3.5 4

Norm

aliz

ed p

hoto

curr

ent

(nA

/cm

)

Surface roughness (μm)

Proposed sine and V profiles

Experimental results

Sine profile

V profile

Figure 8.16 Variation of photocurrent with surface roughness of 0.5 at.% WO3-doped PLZT.Comparison with the normalized computed photocurrent for the two surface profiles is alsomade.

�

� �

�


In conclusion, the optimum profile of the photostrictive PLZT actuator is afilm shape with a thickness around 30 μm and a surface roughness of less than0.2 μm.

8.4 Photostrictive Device Applications

In this section, we introduce the possible applications of photostrictionto photo-driven relay, a micro-walking machine, a photophone, and themicro-propelling robot, which are designed to function as a result of lightirradiation, having neither lead wires nor electric circuits. Refer to Ref. [23] forthe details of applications of photostrictive devices.

8.4.1 Displacement AmplificationMechanism

Since the maximum strain level of the photostriction is only 0.01% (one orderof magnitude smaller than the electrically induced piezostriction, and this cor-responds to 1 μm displacement from a 10mm sample), we need to consider asophisticated amplification mechanism of the displacement. We employed abimorph structure, which is analogous to a bimetal consisting of two metallicplates with different thermal expansion coefficients bonded together to gener-ate a bending deformation according to a temperature change.Two PLZT plates were glued back to back but were placed in opposite

polarizations, then connected on the edges electrically, as shown in Figure 8.17[8]. A purple light (366 nm) was shone to one side, which generated a photo-voltaic voltage of 7 kV across the length (along the polarization direction).Thiscaused the PLZT plate on that side to expand by nearly 0.01% of its length,while the plate on the other (unlit) side contracted due to the piezoelectriceffect through the photovoltage. Since the two plates were bonded together,the whole device bent away from the light. Figure 8.18 demonstrates the

Ps To e

xpand

To c

ontr

act

5 mm

0.4 mm

20 m

m

Irradiation of lightElectrode

Figure 8.17 Structure of the photo-drivenbimorph and its driving principle.

�

� �

�

8.4 Photostrictive Device Applications 295

150

100

50

0

–50Dis

pla

ce

me

nt

(μm

)

–100

–150Time (s)

Bimorph

5 10 15 25 30

Dummy Bimorph

Figure 8.18 Tip deflection of the bimorph device made from WO3 0.5 at.% doped PLZTunder a dual-beam control (illumination intensity: 10 mW/cm2).

tip deflection of the bimorph device made from WO3 0.5 at.% doped PLZTunder a dual-beam control (illumination intensity: 10mW/cm2). For this20-mm-long and 0.35-mm-thick biplate, the displacement at the edge was±150 μm, and the response speed was a couple of seconds.

8.4.2 Photo-Driven Relay

A photo-driven relay was constructed using a PLZT photostrictive bimorph asa driver, which consists of two ceramic plates bonded together with their polar-ization directions opposing each other (Figure 8.19) [8]. A dummy PLZT platewas positioned adjacent to the bimorph to cancel the photovoltaic voltage gen-erated on the bimorph.Utilizing a dual-beammethod, switchingwas controlledby alternately irradiating the bimorph and the dummy. The time delay of thebimorph that ordinarily occurs in the off process due to a low dark conductivitycould be avoided, making use of this dual-beam method; ±150 μm displace-ment was transferred to a snap action switch, with which on/off switching waspossible.The on/off response of the photo-driven relay was demonstrated witha typical delay time of 1–2 s.

8.4.3 Micro-walkingMachine

A photo-driven micro-walking machine was also developed using the pho-tostrictive bimorphs [24]. It was simple in structure, having only two PLZTbimorph legs (5mm × 20mm × 0.35mm) fixed to a plastic board, as shown inFigure 8.20. When the two legs were irradiated with purple light alternately,

�

� �

�


Contact

Leaf spring

Moving piece Operatingdirection

Beam 2

Beam 1

Base

Dual beam method

PLZT dummy

PLZT bimorph

Ps

Ps

Figure 8.19 Structure of the photo-driven relay.

Purple colorirradiation

Purple colorirradiation

Proceeding direction

Figure 8.20 Photo-driven micro-walking machine made of two photostrictive bimorphs.Alternating irradiation provides a walking motion.

the device moved similarly to an inchworm. The photostrictive bimorphas a whole was caused to bend by ±150 μm as if it averted the radiation oflight. The inchworm built on a trial basis exhibited rather slow walking speed(several tens of micrometers per minute), since slip occurred between thecontacting surface of its leg and the floor. The walking speed can be increasedto approximately 1mm/min by providing some contrivances such as the use ofa foothold having microgrooves fitted to the steps of the legs.

�

� �

�

8.4 Photostrictive Device Applications 297

8.4.4 “Photophone”

The technology to transmit voice data (i.e., a phone call) at the speed of lightthrough lasers and fiber optics has been advancing rapidly. However, the end ofthe line – interface speaker – limits the technology, since optical phone signalsmust be converted from light energy to mechanical sound via electrical energyat present. The photostriction may provide new photoacoustic devices.Photomechanical resonance of a PLZT ceramic bimorph has been success-

fully induced using chopped near-ultraviolet irradiation, having neither elec-tric lead wires nor electric circuits [25]. A thin cover glass was attached onthe photostrictive bimorph structure to decrease the resonance frequency soas to easily observe the photoinduced resonance. A dual-beam method wasused to irradiate the two sides of the bimorph alternately with an optical chop-per; intermittently with a 180∘ phase difference.Themechanical resonance wasthen monitored by changing the chopper frequency. Figure 8.21 shows the tipdisplacement of the thin-plate-attached sample as a function of chopper fre-quency. Photoinduced mechanical resonance was successfully observed. Theresonance frequency was about 75Hz with the mechanical quality factor Qmof about 30. The maximum tip displacement of this photostrictive sample wasabout 5 μm at the resonance point. Although the sound level is low, the exper-iment promises photostrictive PLZT bimorphs as photoacoustic components,or “photophones,” for the next optical communication age.

8.4.5 Micro-propelling Robot

A new application of highly efficient, photostrictive PLZT films on flexiblesubstrates has been conceived for usage in the new class of small vehicles

Figure 8.21 Tip deflectionof the bimorph devicemade from WO3 0.5 at.%doped PLZT under adual-beam control(illumination intensity:10 mW/cm2).

50

6

5

Dis

pla

cem

ent

(μm

)

4

3

2

1

060 70 80

Frequency (Hz)

90 100

�

� �

�


1. Initial stage

2. Illumination on

Illumination

Nail

3. Illumination off

30 μm50 μm

PLZT film

Move

Nail

Polarization

direction Transparent

electrode

30 μm50 μm

PLZT film

Nail

Moving direction

Polarization

direction Transparent

electrode

30 μm

50 μm

PLZT film

NailMove

(a)

A

B

(b)

Flexible substrate

Flexible substrate

Flexible substrate

Figure 8.22 (a) Schematic diagram of an arch-shaped photoactuating film device and (b) itstriangular top shape.

for future space missions [26]. Micro-propelling robot can be designed intoarch-shaped photoactuating composite films (unimorph type) with a triangu-lar top (Figure 8.22). In order to maximize the photostrictive properties of thesample, the sample thickness was determined to be 30 μm.This device is drivenat their resonance mode under an intermittent illumination. Photoactuatingfilms may be fabricated from PLZT solutions and coated on one side of asuitable flexible substrate, which will then be designed to have a curvatureof 1 cm−1. A slight difference in length/width between the right and left legsis designed in order to provide a slight difference between their resonance

�

� �

�

8.5 Concluding Remarks 299

frequencies. This facilitates the control of the device in both clockwise andcounterclockwise rotations (i.e., right and left steering). A light chopperoperating at a frequency close to resonance can be used to illuminate thedevice, in order to maximize the vibration of the bimorph, which will thenprovide the capability to turn by applying different resonance frequencies atthe two legs.

8.5 Concluding Remarks

Photomechanical/photostrictive actuators can be driven only by the irradiationof light, so that they will be suitable for use in actuators, to which lead wires canhardly be connected because of their ultrasmall size or of their employed con-ditions such as ultrahigh vacuum or outer space. The photostrictive bimorphswill also be applicable to “photophones.” Also note their remote control capa-bility without being interfered by electromagnetic noise. Figure 8.23 summa-rizes the “response speed” improvement of the photostrictive bulk ceramic andof the device in the sequence of year and the key technology developed inour research center. Compared to the speed at 1 h at the discovery age withPZT, two-orders-of-magnitude improvement (up to 10 s) has been achieved inmaterials, and even photoinduced resonance in an audible frequency range wasrealized in the devices. The new principle actuators have considerable effectsupon the future micromechatronics.

Bulk ceramic

Responsivity (s)

Year

Device

Discovery in PZT

PLZT 3/52/48

Doping concentrationdevice designing

(resonance usage)

PLZT 4/48/52

Sol–gelGrain size controlDonor doping

80

84

88

92

96

00

104 102 100 10–2

Figure 8.23 Response speed improvement of the photostrictive bulk ceramic and of thedevice in the sequence of year and the key technology development.

�

� �

�


References

1 Uchino, K. and Aizawa, M. (1985) Photostrictive actuators using PLZTceramics. Japanese Journal of Applied Physics, 24, 139–141.

2 Brody, P.S. (1983) Optomechanical bimorph actuator. Ferroelectrics, 50, 27.3 Fridkin, V.M. (1979) Photoferroelectrics, in Solid-State Sciences, vol. 9(eds M. Cardona, P. Fulde, and H.-J. Queisser), Springer-Verlag, New York,pp. 85–113.

4 Uchino, K., Miyazawa, Y., and Nomura, S. (1982) High-voltage photovoltaiceffect in PbTiO3-based ceramics. Japanese Journal of Applied Physics, 21(12), 1671–1674.

5 Uchino, K. (1996) New applications of photostriction. Innovations in Mate-rials and Research, 1 (1), 11–22.

6 Chu, S.Y. and Uchino, K. (1994) Impurity doping effect on photostriction inPLZT ceramics. Journal of Advanced Performance Materials, 1, 129–143.

7 Uchino, K., Aizawa, M., and Nomura, S. (1985) Photostrictive effect in(Pb,La)(Zr,Ti)O3. Ferroelectrics, 64, 199.

8 Tanimura, M. and Uchino, K. (1988) Effect of impurity doping onphoto-strictive in ferroelectrics. Sensors and Materials, 1, 47–56.

9 Glass, A.M., von der Linde, D., and Negran, T.J. (1974) Highvoltage bulkphotovoltaic effect and the photorefractive process in LiNbO3. AppliedPhysics Letters, 25, 233.

10 Poosanaas, P., Tonooka, K., and Uchino, K. (2000) Photostrictive actuators.Mechatronics, 10, 467–487.

11 Hecht, E. (1987) Optics, with contributions by, in , 2nd edn (ed. A. Zajac),Addison-Wesley Publishing, Massachusetts, pp. 44, 81–104, 610–616.

12 Kittel, C. (1996) Introduction to Solid States Physics, 7th edn, John Wiley &Sons, Inc., New York, p. 388.

13 Uchino, K., Miyazawa, Y., and Nomura, S. (1983) Photovoltaic effect in fer-roelectric ceramics and its applications. Japanese Journal of Applied Physics,22, 102.

14 Brody, P.S. (1973) Large polarization-dependent photovoltages in ceramicBaTiO3 + 5wt.% CaTiO3. Solid State Communications, 12, 673.

15 Brody, P.S. (1975) High voltage photovoltaic effect in barium titanate andlead titanate-lead zirconate ceramics. Journal of Solid State Chemistry, 12,193.

16 Sada, T., Inoue, M., and Uchino, K. (1987) Photostriction in PLZT ceramics.Journal of the Ceramic Society of Japan, 95, 499–504.

17 Poosanaas, P. and Uchino, K. (1999) Photostrictive effect inlanthanum-modified lead zirconate titanate ceramics near the morphotropicphase boundary. Materials Chemistry and Physics, 61, 31–41.

�

� �

�

References 301

18 Nonaka, K., Akiyama, M., Takase, A. et al. (1995) Nonstoichiometry effectsand their additivity on anomalous photovoltaic efficiency in lead lan-thanum zirconate titanate ceramics. Japanese Journal of Applied Physics,34, 5380–5383.

19 Poosanaas, P., Dogan, A., Prasadarao, A.V. et al. (1999) Effect of ceramicprocessing methods on photostrictive ceramics. Advanced PerformanceMaterials, 6, 57–69.

20 Poosanaas, P., Dogan, A., Prasadarao, A.V. et al. (1997) Photostriction ofsol–gel processed PLZT ceramics. Journal of Electroceramics, 1, 105–111.

21 Sada, T., Inoue, M., and Uchino, K. (1987) Photostrictive effect in PLZTceramics. Journal of the Ceramic Society of Japan, 5, 545–550.

22 Poosanaas, P., Dogan, A., Thakoor, S., and Uchino, K. (1998) Influence ofsample thickness on the performance of photostrictive ceramics. Journal ofApplied Physics, 84 (3), 1508–1512.

23 Uchino, K. (1997) New applications of photostrictive ferroics. MaterialsResearch Innovations, 1, 163–168.

24 Uchino, K. (1989) Micro walking machine using piezoelectric actuators.Journal of Robotics and Mechanism, 124, 44–47.

25 Chu, S.Y. and Uchino, K. (1995) Proceedings of the 9th International Sym-posium on the Applications of Ferroelectrics, State College, PA, p. 743.

26 Thakoor, S., Morookian, J.M., and Cutts, J.A. (1996) The Role of Piezoceram-ics Microactuation for Advanced Mobility. Conference of the Proceedings of10th IEEE International Symposium on the Applications of Ferroelectrics, 1,pp. 205–211.

�

� �

�

303

9

Switching Surface Topographies Based on LiquidCrystal Network CoatingsDanqing Liu and Dirk J. Broer

Department of Chemical Engineering and Chemistry, Institute for Complex Molecular Systems,Technical University of Eindhoven, Eindhoven, Netherlands

9.1 Introduction

The surface topography of materials is usually defined as the local deviations ofa surface from a (close to) perfectly flat plane. When applied in a coating or ona thin film, the topography of a surface substantially affects the properties of amaterial such as friction [1, 2], human perception during touch [3], biologicalinteractions such as biofouling [4], or the interaction with flowing fluid matter[5]. In biology, we find many examples of living species that benefit from theirspecific and often unique topographic patterns at their surfaces [6].This can bestatic patterns as, for instance, found on the leaves of the lotus flower, repellingwater and dirt particles [7], or the feet of a gecko providing stickiness to the sur-faces to beat gravity [8]. However, dynamic topographic structures also appearin nature and can, for instance, be found in mammals. A well-known exampleis the pilomotor reflex on the skin of mammals, which creates insulation undercold conditions and provides protection by scaring away predators when thebody appears larger. Furthermore, many nanoscopic effects are also known as,for instance, found in the studies of cell proliferation at an active surface topog-raphy [9] or themotion of cilia constructs in the respiratory systemofmammalsto transport and expel liquid and dirt [10].Many studies have been devoted to the usage of static surface topographies

fabricated by wrinkling [11, 12], (photo)embossing [13, 14], or lithography[15]. In most cases, these structures are static. It is both a scientific and atechnical challenge to make these surface structures switchable, that is, theycan be turned “on” and “off” on the command of an external trigger. Whentopographical switching is controlled, new appealing applications come withinreach, such as switchable lenses and gratings, particle transport at solid or


�

� �

�

304 9 Switching Surface Topographies Based on Liquid Crystal Network Coatings

wet surfaces, and fluid transport if wave-like surface deformation can beestablished. In addition, the mechanics at or between sliding surfaces canbe controlled, and one gets access to control properties such as friction,stickiness, and adhesion (or release) simply by altering between the absenceand presence of micrometer-sized protrusions.In a series of papers, we propose the use of liquid crystal (LC) networks to

induce the formation of surface topographies [16–21]. The driving principleis the light-triggered trans-to-cis isomerization of a copolymerized azoben-zene molecule added in a relatively low concentration to the network. Fromthe literature, it is known that a focused polarized laser spot induces surfacewrinkling in azobenzene-modified linear polymer films as driven by photosoft-ening and pressure as generated by the formation of free volume [22, 23].Thesesystems form more or less permanent protrusions. Only by “overwriting” by asecond illumination, these protrusions can be removed. For some applications,this might be favorable. For other applications, it is desirable that the surfacestructures disappear as soon as the trigger, the light source, is switched off.Theelastic properties of the polymer network then provide the desired reversibilityto the systems where the structures evolve and disappear by switching the lightsource on and off. An additional advantage of the use of photopolymerized LCnetworks is that one can chose between a variety of alignment strategies, whichleads to the formation of preprogrammed surface structures.

9.2 Liquid Crystal Networks

The history of the formation of cross-linked LC networks goes back to the1960s, where different authors suggested polymerizing LCs in their mesophaseto realize highly ordered polymers [24, 25]. Initiating the polymerization bythermal decomposition of a peroxide or azo-based initiators (thermosetting)of LC diacrylates was reported to yield 3D cross-linked polymers with a strongoptical anisotropy. The molecular order of the monomer phase was retainedat high temperature and often above the decomposition temperature [26–29].However, the use of heat to initiate polymerization often conflicts with the tem-perature range of the LC phases of the reactive LCs. The first reports to uselight to initiate the bulk (solventless) polymerization of reactive LCs relate tomonoacrylates that form linear LC side-chain polymers [30–34]. Real fixationofmolecular order often does not occur as phase transitionsmight occur duringthe formation of these linear polymers. For this reason, the bulk photopoly-merization of polyfunctional LCmonomers (photosetting LCs) became impor-tant. The photoinitiated free-radical polymerization of monolithically alignednematic diacrylates, in the present literature often referred to as reactive meso-gens (RMs), produces a stable polymer networkwith a texture and order similarto those of the monomer [35–38].

�

� �

�

9.2 Liquid Crystal Networks 305

The principle of this process as well as some examples of often used LCmonomers is schematically shown in Figure 9.1. In practice, monomer mix-tures are formulated to fine-tune the properties in monomeric state (meltingtemperature, width and type of the LC phase, and viscosity) as well of thefinal polymer (modulus, glass transition, and refractive indices). An importantfeature, and an enormous advantage of this technique, is that the alignment ofthe LC monomer can be locally controlled by alignment films at the substrates[39, 40]. Rubbed polyimide provides planar alignment, which is uniaxial whenthe rubbing direction is parallel at both sides of the LC monomer or twistedwhen they are orthogonal.The use of a combination of rubbing and a surfactantprovides splayed alignment. The use of photoalignment layers gives access toeven more complex orientation patterns. The addition of chiral, often reactive,molecules rotates the molecules in the direction perpendicular to their longaxes of which the pitch can be accurately adjusted by the concentration of thechiral molecule [38, 41].The mechanical properties of the LC networks are, apart from their

anisotropic nature, of the same order as those of the isotropic acrylate net-works. This means that the modulus and strength are similar and dependstrongly on the molecular parameters such as cross-link density and the ratiobetween stiff and flexible units [42]. The LC monomers shown in Figure 9.1(c)form glassy polymers with the moduli of a few gigapascals and a glass transi-tion temperature (Tg) between 60 and 120 ∘C. Copolymerizing them with themonoacrylates as shown in Figure 9.1(d) reduces the Tg and the modulus inthe rubber plateau found at temperatures above Tg.An interesting feature is that the thermal expansion of the uniaxially oriented

LC networks is highly anisotropic [43]. The thermal expansion measured par-allel to the molecular alignment is close to zero well below Tg and becomesnegative around and above Tg. When measured perpendicular to the molec-ular alignment, the thermal expansion is unusually high and does not reach aplateau above Tg as is seen for isotropic polymer networks. The volume ther-mal expansion that can be calculated from the linear thermal expansion coeffi-cients exhibits a normal value. This effect is attributed to a change of the orderparameter upon heating of the sample, although the transition to isotropic isnot reached. In fact, a change of order parameter of around 0.7 at room tem-perature to 0.6 at 150 ∘C is responsible for this behavior. Increasing the lengthof the alkylene spacer between the aromatic core and the acrylate moiety, forexample, to 11 methylene units results in a somewhat larger decrease of theorder parameter, a lower Tg, and a larger anisotropy of the thermal expansion.The effects of this anisotropic expansion behavior on films with gradients inmolecular orientation are well described in a number of applications and canbe a simple bending for a splayed configuration to complex, preset, deformationfigures [44–46].

�

� �

�


hν

(a)

O

O

O

O

O

O(CH2)6O

O

Cr - (SC - 88 -) 101- N - 155 - I

O

O

O

O

O(CH2)11O

Cr - 77 - SC - 177 - N - 135 - I

11O

O O

O

O(CH2)

O(CH2)

3O

O

O

O

O

O(CH2)3O

O

Cr - 73 - N - 129 - I

O

O(CH2)6O

O

O

O

O

O(CH2)6O

O

Cr - 86 - N - 116 - I

(c)

(b)

Twisted nematic

Splayed nematicChiral-nematic

O(CH2)6O

Figure 9.1 (a) Schematic representation of the formation of a liquid crystal network.(b) Besides uniaxial alignment, the LCs can be ordered in twisted, splayed, or chiral-nematicconfigurations, for example, by using surface techniques or chiral additives. Some examplesof (c) LC diacrylates and (d) monoacrylates often used in polymerizable LC formulations.

�

� �

�



(d)

O

O(CH2)6O OH

Cr - (N - 44) - 76 - I

O

O(CH2)6O O

O

OC6H13

Cr - 54 - N - 67 - I

9.2.1 Photoresponsive Liquid Crystal Networks

As detailed in Chapter 5, the principle of photoactivated deformation of LCnetwork films has many parallels with the deformation induced by heatingof structured LC network films. In this case, a copolymerized azobenzenemolecule disturbs the order when it transforms from its trans state to the cisstate [47, 48]. Figure 9.2(a) shows an example of an azobenzene molecule thatcan be copolymerized with the LC monomers shown in Figure 9.1. The transstate of thismolecule complieswith the rod-like character of the LCmonomers,and the order parameter of the network maintains its original value between0.6 and 0.7. When converted to the cis state, the bended configuration of theazobenzene disturbs the molecular order in the network, and the film tendsto contract along the orientation direction and expands orthogonal to that.A small percentage of about 2–6wt% is sufficient to obtain considerable pho-tomechanical response. Employing the ability to hierarchically align and orientLC systems, photomechanical responses in the splayed configuration (schemat-ically shown in Figure 9.2b) yield large deflections as the material contracts atthe planar surface and expands at the surface with homeotropic alignment.This photoinduced bending is nicely demonstrated in inkjet-print cilia of

this material with planar orientation at the top and homeotropic orientationat the bottom (Figure 9.2c and d) [49]. When these cilia are exposed to UVlight, they bend. The bending demonstrated in Figure 9.2(e) is recorded forfilms immersed in water to minimize temperature effects. Meanwhile, otherexamples have been published where shape changes from planar to curvedshapes of enormous complexity have been demonstrated via origami type offolding and unfolding [50, 51].

9.2.2 Photoinduced Surface Deformation

After having demonstrated large light-induced geometrical changes in pho-tosensitized LC networks, the question of whether geometrical changes canalso be made visible in films that strongly adhered to a rigid substrate comesup. This principle is schematically illustrated in Figure 9.3. An initially highly

�

� �

�


UV

Vis, Δ

(a)

(b)

(c) (d) (e)

R =

Figure 9.2 Photoinduced shape changes. (a) A copolymerized monomer with anazobenzene moiety undergoes a trans-to-cis conversion when exposed to UV light.(b) When embedded in a splayed LC network, it induces contraction at one side of the filmand expansion at the opposite side. (c and d) LC monomers containing azobenzenemonomers can be inkjet-printed on a pattern of sacrificial polyvinyl alcohol, which afterpolymerization of the LC monomers can be removed, yielding partly freestanding cilia.(e) When these cilia are exposed to UV light, they bend to a curved state. When the light isswitched off, they bend back to close to flat.

Figure 9.3 The principle of surface actuation where the initially highly ordered state of anLC network is disturbed and the film surface protrudes by changes of the packing of therod-like moieties at less ordered locations.

�

� �

�


ordered state is disturbed by the action of the trans-to-cis transformation ofthe azobenzene. The film is restricted in its in-plane dimensional changes bythe substrate. Eventual deformation by volumetric changes should escape intothe third dimension perpendicular to the film surface.For a first demonstration of this effect, we have chosen a coating based on

a chiral-nematic network with planar orientation, that is, the orientations ofthe helix axes are perpendicular to the film surface as schematically shown inFigure 9.4(a). This configuration has the advantage that in-plane stresses areminimized when the local order is decreased. The periodic orthogonal aver-age orientation of themolecules balances the order-parameter-related in-planecontraction and expansion to a large extent. The monomers used are shown inFigure 9.4(b). The ratio between mono- and diacrylates gives the right balancebetween rigidity of the surface for sufficient wear resistance and the plastic-ity needed for deformation. Chirality is induced by 3.4wt% chiral diacrylate,which gives a pitch of the chiral-nematic helix of 660 nm.Thephotosensitivity isachieved by the presence of 2wt% azobenzenemonomer. Photopolymerizationis initiated with light >400 nm, avoiding early conversion of the azobenzene toits cis state. Actuation is performed with a mercury lamp strongly emitting,among other emission lines, 365 nm light. The formation of the protrusion ismeasured by interference microscopy and is found to be around 10% of theinitial thickness for a coating thickness smaller than 10 μm. For thicker coat-ings, the penetration depth of the UV light starts to play a role, and the relativemodulation depth becomes smaller.A number of control experiments were carried out to complement this

examination, as summarized in Figure 9.5. To demonstrate the importanceof LC order of the network, a sample was prepared from exactly the samecomposition but polymerized at an elevated temperature in the isotropic stateof the monomer mixture. This isotropic network was actuated in a similarway as mentioned earlier. Figure 9.5(a) and (b) shows the deformation duringactuation. The height of the protrusion was <0.5% of the initial thickness. Inanother control experiment, we exchange the photomechanical-responsiveazobenzene with a stable dye molecule. For this experiment, Tinuvin waschosen, which is a commercial photostabilizer often added to improve onthe photodegradation of polymers such as polycarbonate. It has about thesame absorbance as azobenzene, but it does not show any photomechanicaleffects. As shown in Figure 9.5(c), indeed the position of the absorption bandas well its absorbance was very similar as that of the azobenzene in its initialtrans state. When exposed to UV light, light absorption leads to some heatingof the sample, which results in the formation of protrusion with a heightof around 2% of the initial thickness (Figure 9.5d and e). In fact, this is aworst-case experiment as the absorption of the azobenzene drops as soonas it is converted to the cis state. Nevertheless, one must conclude that localthermal heating might also contribute to local expansion of the film. But the

�

� �

�


3

0

2

1

0

500 1000 1500 2000 2500

Length (μm)

He

igh

t (μ

m)

0

3

2.5

0.5

2

1.5

1

OO

O O

O

O

O

OO

O

OO

O

O

O

O

OO

O O C H

O

OO O

O O

O

O

O

O

O

O

H

H

O

O

O

O

O OO

O

O

H

H O O

OOO

O

(d)

(c)

(b)

(a)

UV

Figure 9.4 (a) Mask exposure of a chiral-nematic coating results in local deformation of thefilm. (b) The addition of 5 wt% chiral LC monomer to the monomer mixture induces theformation of the molecular helices perpendicular to the film surface, and the addition of2 wt% azobenzene monomer makes the film photosensitive. (c) The height of the resultingprotrusions is analyzed by interference microscope and is around 10% of the initial filmthickness. (d) The three-dimensional image of the surface shows the correspondence withthe mask that is used for patterned exposure.

�

� �

�


0.8

0.6

0.4

Ab

so

rba

nce

0.2

0300 350 400

Wavelength (nm) Length (μm)

He

igh

t (μ

m)

450 5000

3

(e)

(d)

(b)(a)

(c)

2

1

01000 2000 3000

He

igh

t (μ

m)

332

10

2

1

0 Length (μm)0 500

TinuvinAzobenzene

1000

μm

3

2

1

0

μm

Figure 9.5 (a) Interference microscopy measurements show 3D image of (b) surfacetopographies formed in isotropic films containing azobenzene and its surface profile.(c) Comparison of the absorbance of azobenzene and stable Tinuvin dye added to exchangethe azobenzene to determine heating effects. (d) Interference microscopy measurementsshow the 3D image and (e) the corresponding surface profile when the Tinuvin containingsample is UV-exposed similarly to the azobenzene-containing coatings.

contribution is smaller than the effect of the photomechanical response of theazobenzene-modified films. The presence of LC order is a necessity to obtainappreciable mechanical response.

9.2.3 Photoinduced Surface Deformation Preset by PatternedDirector Orientation

As discussed in Section 9.1, the formation of LC polymeric networks byin situ polymerization of LC monomers provides the possibility to createfilms or coatings with director patterns. Anticipating that the molecular orderand the director also determine the topography of a light-activated surface,a number of experiments with different director patterns were carried out. Ina first set of experiments, we compared a line pattern with alternating stripesof chiral-nematic alignment and isotropic orientation, as schematically shownin Figure 9.6(a). These samples can be simply prepared by photopolymerizinga chiral-nematic sample, with the same components as discussed in theprevious section, by exposing through a line mask. After polymerization of the

�

�

(a)

(b)

(c)

Figure 9.6 Line-patterned coatings with locally different director profiles. (a) A coating with alternating stripes chiral-nematic order next toisotropic order deforms from a flat state (b) to a deformed state 9c) under exposure to UV light. (b, c) Interference microscopic images takenbefore and during exposure, which show a modulation depth of around 10% relative to the initial coating thickness. (d) A coating with alternatingstripes of planar chiral-nematic order next to homeotropic order. (e) The cross section of the coating measured by interference microscopemeasure prior to UV exposure. The small corrugations, enlarged in the inset, originate from an imprinted ITO pattern. (f ) The same film underexposure to UV light. The planar chiral-nematic area expands relative to the homeotropic area with a modulation depth of around 20%. (See colorplate section for the color representation of this figure.)

�

�

20.2

0.150.1

0.050

1.5

1

0.5

0

1.5

1

0.5

0

0 100 200

200 400 600 800 1000 12000

300 400 500 600 700 800 900 1000 1100 1200

0 100 200 300 400 500 600 700 800 900 1000 1100 1200

Height (µm)2

Height (µm)

(d)Position (µm)

(f)

Position (µm)

(e)


�

� �

�


exposed area, the sample is heated to a temperature well above the transitiontemperature to the isotropic state of the monomer mixture, and a second floodexposure is provided, which cures the isotropic region. During polymerization,care is taken that the azobenzene moieties remain in their trans state by usinga mercury lamp provided with an optical filter that rejects light <400 nm.Upon actuation of this film by UV light, the azobenzene is converted both in

the chiral-nematic area and in the isotropic area to its cis state. There is somedifference in the absorbance. In the planar chiral-nematic state, the azoben-zenes are oriented with the axes of their transitionmoment predominantly par-allel to the electrical vector of the incoming UV light, whereas in the isotropicarea, the distribution is random. This could lead to minor differences in theeffectiveness in which UV light is absorbed. But the dominant difference comesfrom the fact that LC order is a requirement to generate an excess of free vol-ume, as discussed in the previous section. Consequently, when the sample isexposed, the areas with the LC order expand much more than the isotropicarea.The difference in height, further denoted asmodulation depth,maximizesat around 10% of the initial film thickness.This means that starting with an ini-tial film thickness of 5 μm, the ridges formed in the LC-ordered area are around0.5 μm elevated from the disordered area.Figure 9.6(b) and (c) shows interference microscopic images of the surface

before and during exposure, respectively. After switching off the light, thefilm relaxes back to the initial state. Careful observation of the transition atthe edge of the ridges shown in Figure 9.6(c) reveals some crack formation.This is inherent to the stresses built up between the two areas and also to thefabrication method, which enables some material transport during the firstexposure as caused by reaction-induced diffusion. This also explains why thefilm as prepared is not completely flat but shows some initial ridges as well(Figure 9.6b). But these are two orders of magnitude smaller than that of thestructures formed during exposure.When we prepare a similar sample, but now with homeotropically aligned

area rather than isotropic area alternating with planar chiral-nematic area,we see an amplification of the modulation depth. The director pattern of thecoating with chiral-planar area next to vertically oriented area is schematicallyshown in Figure 9.6(d). A director pattern such as this can be made bybringing the monomer mixture in between two substrates. One substrate hasa continuous indium tin oxide (ITO) conductive layer and a planar polyimideorientation layer.The opposite substrate has a line-patterned ITO coating witha planar polyimide orientation layer. By applying 70V across the 10-μm-thickfilm of the LC monomer mixture, the planar molecular helix locally unwindsand gives a homeotropic orientation of the LC monomer molecules. In thisconfiguration, the LC film is photopolymerized and the director pattern isfrozen-in.

�

� �

�


After removal of the top substrate, the film was adhered only to the lowersubstrate. Figure 9.6(e) shows the surface profile with small indentations ofaround 60 nm caused by the imprint of the ITO layer.The imprints correspondto the locations of the homeotropic orientation. When the film is actuated byUV light, the planar chiral-nematic area expands to form the ridges shown inFigure 9.6(f ). In this case, we have the effect of the expansion of the planarchiral-nematic area as discussed earlier. However, there is also the tendencyof the uniaxial homeotropic domains to contract parallel to the director andexpand perpendicular to that in the lateral direction. This effect comes on topof the volumetric changes. As a result, the modulation depth of these filmsupon actuation hasmaximum values of around 20% of the initial film thickness,depending on the light intensity and concentration of azobenzene.Despite this large surface deformation, no cracks could be observed at the

edges of the ridges. This is explained by the fact that photopolymerization wascarried out homogeneously without material transport during the polymer-ization process, meaning that there are no areas of depleted monomer withreduced strength.A set of experiments with a more random director pattern are presented

in Figure 9.7. In Figure 9.7(a), we schematically show the director pattern ofa so-called fingerprint texture. The helix axes of a chiral-nematic mixture arechosen to be parallel to the substrate surface. This can be obtained by care-fully balancing the surface anchoring forces using homeotropic alignment filmsand/or surfactants in combination with a well-controlled film thickness. Theaxes of the molecular helices deviate from being straight and tend to take arandom planar order to form the fingerprint textures. Upon actuation with UVlight, the same principle of surface deformation is achieved as was the case forthe film depicted in Figure 9.6(d): at the position of planar order, the film willexpand perpendicular to the surface, and at the position of perpendicular order,the film tends to contract in vertical direction and expand in the lateral horizon-tal directions. Directly after polymerization of a 4-μm-thick film, the surfaceshows aminor surfacewrinkling as shown in Figure 9.7(b).Thesewrinkles, withfeature heights typically <20 nm, are attributed to theMarangoni effect causedby surface tension gradients between planar and homeotropic oriented areas,although some minor effects of anisotropic polymerization shrinkage cannotbe excluded. Upon actuation of this film, a random, fingerprint-like, surfacestructure is being formed with maximum feature heights of around 8 μm, cor-responding to ∼20% of the initial film thickness. The hills of the topographicstructures correspond to the bright areas observed between crossed polariz-ers, the valley to the black areas, indeed proving that expansion takes place atthe planar area and shrinkage at the homeotropic area.Another approach to stimulate random orientation of the director is to rely

on the formation of polydomain films as schematically shown in Figure 9.7(c).

�

�

Homeotropic

Planar

(a) (b)hν

Pf

(b)

UV

Δ

0.6

0.4

0.2

0

−0.2

−0.4

−0.6

Heig

ht (μ

m)

0.6

0.4

0.2

0

−0.2

−0.4

−0.6

Heig

ht (μ

m)

0 01 02 03 04 50 0 01 02 03 04 50

Length (μm) Length (μm)

Figure 9.7 Randomly patterned coatings with locally different director patterns. (a) A fingerprint pattern, which forms in a chiral-nematic coatingwith the helix axes aligned parallel to the substrate surface. (b) The fingerprint expands at the locations where the rod-like molecular units areoriented parallel to the surface and shrink at the positions where they are aligned perpendicular, resulting in a modulation depth of around 20% ofthe initial coating thickness. (c) An illustration of a coating with a polydomain pattern. (d) Here also, the domains with (close to) planar orientationexpand, whereas the domains with (close to) homeotropic orientation shrink. (See color plate section for the color representation of this figure.)

�

�

(c)(d)

hνΔ

0.6

0.4

0.2

000 200 4000 200 300100 400

−0.2

−0.4

−0.6 Length (μm)Length (μm)

389 μm292 μm 389 μm

292 μm

μm2

1.5

1

0.5

0

He

igh

t (μ

m)

0.6

0.4

0.2

−0.2

−0.4

−0.6

He

igh

t (μ

m)


�

� �

�


Thedomain size can be controlled to some extent by blending immisciblemate-rials in the LCmonomer mixture, which is distributed as small droplets. In thiscase, a fluorinated acrylate monomer is being used in a concentration of 2wt%.In the processing stage, for example, during spin coating, the droplets are liquidas themixture itself does not affect the rheological properties of the monomersin solution. During polymerization, they polymerize simultaneously with theLCmonomers.The resulting domains have different director patterns, which allrespond differently when being actuated. The largest response with the forma-tion of the highest structures is being obtained from the domains with parallelorientation to the substrate. The domain with homeotropic orientation mightshrink, whereas the domains with tilted orientation respond in between thetwo extremes depending on the tilt angle. The result is that the surface, uponactuation, deforms from close to flat to a surface with random corrugations asshown by the interference microscopic image of Figure 9.7(d).Although both approaches shown in Figure 9.7 result in random randomly

distributed corrugated structures, there is an important difference. The heightof the structures of the fingerprint textures is roughly the same over the wholesurface, whereas the height of the structures obtained from the polydomainsamples is random. This manifests itself when the friction force is recorded oftwo surfaces sliding against each other [19].When the polydomain surfaces areactuated from flat to corrugated, the friction force increases because of inter-locking of the structures in comparison with sliding two pieces of sandpaperagainst each other. But when the fingerprint structures are actuated, the fric-tion force decreases. The tops of the structures are randomly distributed andare of equal height. This prevents interlocking, and the friction force becomesless because of a decreased contact area.

9.2.4 On theMechanism of Surface Deformation

The bending deformation of azobenzene-modified LC network films can belargely explained by changes of the order parameter of the network when theazobenzene undergoes its transformation from the trans to the cis isomer incombination with a gradient over the cross section of the film. The bendedazobenzene moiety does not comply with the molecular organization of themolecular rods and disturbs the orientation of its neighboring molecular units.The gradient can be either in light absorption as caused by the absorption ofthe azobenzene itself or by an added absorber, in modulus as, for instance,caused by a gradient in cross-link density or by a gradient of the director(cf. Figure 9.1b). However, the formation of protrusions as discussed in theprevious sections cannot be fully explained by this principle. For instance, inthe case of the locally exposed chiral-nematic film, the protrusion must becaused by a volume increase under the assumption that no lateral displacementof matter takes place. Lateral displacement is prohibited in the case of firmadhesion of the coating to the substrate and a finite elasticity caused by the

�

� �

�


cross-links. In the case of chiral-nematic alignment, lateral stresses related tochanges of local order are even largely suppressed by the alternating lateraldirector pattern, which alternates positive and negative mechanical responses.Volumetric changes in combination with azobenzene-containing polymershave been reported before [52, 53], but mostly for polymers with higherazobenzene content. The effects demonstrated here already occur at azoben-zene concentrations as low as 2wt% of the azobenzene monomer where thefree volume generated by the conformational change of the azobenzene cannotprovide sufficient volume increase to fully explain the observed dimensionalchanges of the chiral-nematic films shown in Figure 9.4.To explain these volumetric increases further, we performed density mea-

surements by immersing the chiral-nematic polymer film shown in Figure 9.4in a salt water solution. When the density of the film is just higher than thatof the solution, the sample sinks to the bottom (Figure 9.8). When in this case,the UV lamp is switched on, the sample starts floating as soon as the densityof the film samples decreases below the density of the solution. By repeatingthis experiment for many different salt concentrations, an estimate could bemade for the density change. From these measurements, the density of the LCnetwork film was estimated to be 1.217 g/cm3. Upon UV exposure, the densitydecreases to 1.106 g/cm3, indeed corresponding to the 10% volume increasethat can be derived from the height of the protrusions obtained by the maskexposure.The experiment shown in Figure 9.8 contrasts the response of this material

to a control sample, the density of which remains unchainged. In this film, the

(a) (b) (c) (d) (e)

Figure 9.8 Density change in a chiral-nematic film containing azobenzene and Tinuvin;(a) before UV exposure, both the Tinuvin and azobenzene film are at the bottom of flask;(b) snapshot of films during exposure showing the azobenzene film to float and the Tinuvinfilm to remain at the bottom; and (c−e) after removal of UV light, the azobenzene film startsshrinking to reach its initial position at the bottom. The Tinuvin-modified film remained atthe bottom.

�

� �

�


azobenzene has been exchanged with Tinuvin (see also Figure 9.5c). Also here,it was added as control experiment to estimate the influence of eventual heat-ing of the sample as absorption of the light occurs. In fact, it is a worst-caseexperiment as the azobenzene bleaches on the fast transition from trans to cis,meaning that the absorption decreases immediately as soon as the UV sourceis switched on – something that does not occur with the Tinuvin. However,despite the fact that the sample with the Tinuvin will heat up more than thesample with azobenzene, it remains at the bottom of the container, demon-strating that heating, and the corresponding density decrease, plays a minorrole here. It is the photomechanical effect of the azobenzene that causes themajor part of the density decrease.From the foregoing, it appears that the generation of volume is caused by

the trans-to-cis conversion of the azobenzene but that further explanation isrequired to explain the size of the effect as it is unlikely that 2wt% azoben-zene monomer, roughly corresponding to 2 vol%, will generate 10 vol% volumeincrease by its transition. Further information can be found in the kineticsof the on- and off-reactions. As soon as the light source is switched on, thesystem responds within 10–20 s to reach its maximum height. This is demon-strated in Figure 9.9 in an experiment at a relatively low UV intensity. Whenthe UV source is removed, the protrusion disappears almost completely within10 s. However, when the trans-to-cis conversion is measured by UV–vis spec-troscopy, the chemical relaxation back from cis to trans appears to take hours,which means that there is a controversy between the kinetics of the chemi-cal reaction of the azobenzene and the volumetric response of the protrusion.

70

60

50

40

30

20

10

00 10 20 30 40

Time (s)

50 60 70 80

Heig

ht (n

m)

UV on UV off

Figure 9.9 Mechanical response measured as a height change of the thin film versus timeupon actuation by 365 nm light of intensity 78 mW/cm2 and a subsequent relaxation indark. The film thickness is 4 μm, and the corresponding density decrease is 1.3%.

�

� �

�


More careful analysis of the volumetric relaxation seems to reveal that there aretwo relaxation regions: a fast one directly after the light is switched off, whichbrings back the height to about 20% of its maximumwithin seconds and a slowone that seems to takemuch longer. It is anticipated that the fast process relatesto free-volume relaxation and is mainly determined by the viscoelastic proper-ties of the polymer network.The slow relaxation corresponds to the kinetics ofthe cis-to-trans reaction on a timescale of hours. The free volume apparentlyis not maintained by the cis state of the azobenzene, and the correspondingreduced order parameter of the LC network, but needs a process of continuousactivation by light.A further remarkable observation is that the effect of the formation of

protrusions can be enhanced by coexposing the sample with 455 nm lightnext to 365 nm light. The 365 nm light excites the trans state of azobenzene.The 455 nm source is normally used for the back reaction from trans to cisas its coincides with the maximum of the cis absorption band. By blending,the oscillation of the oscillatory trans–cis reaction is stimulated. Figure 9.10illustrates that, at discrete intensity ratios, the height of the protrusions ismaximized. Apparently, it is the oscillation of the azobenzene compoundthat creates the free volume. As soon as the oscillation is deactivated whenthe light is switched off, the protrusion largely disappears. However, one

0 0.2

Ratio light intensities 455 nm/365 nm

0.4 0.6 0.8 1.0

14

12

10

8

6

4

2

0

Re

lative

mo

du

latio

n d

ep

th (

%)

100 mW/cm2

200 mW/cm2

300 mW/cm2

0

1.4

1.2

1

0.8

0.6

0.4

0.2

0200 400

Length (μm)600

0 200 400Length (μm)

600

Heig

ht (μ

m)

Heig

ht (μ

m)

1.4

1.2

1

0.8

0.6

0.4

0.2

0

Figure 9.10 Modulation depth measured at a mask exposed chiral-nematic polymer filmunder various illumination conditions. The 365 nm LED light intensity was chosen to be 100,200, and 300 mW/cm2. The 455 nm LED light was added in different intensities in ratiosvarying between 0 (455 nm LED switched off) and 1 (455 nm intensity equal to 365 nmintensity). The insets show the actual interference microscopy structures measured with andwithout 455 nm light.

�

� �

�


should realize that the azobenzene is then still predominantly in its cisstate, which slowly converts back to trans. This leads to the conclusion thatthe dynamics in the network creates the free volume rather than the strokeof the azobenzene molecule itself. This is further confirmed by a controlexperiment with azobenzene molecules, which are single sided modified withan acrylate group. The volume that this molecule creates is only a fraction ofthat of the diacrylate.

9.3 Conclusions

LC networks that change shape under the influence of an external trigger areconsidered to be of potential importance for soft robotic devices. Light astrigger is attractive as it can be applied remotely without the need to fabricatecomplex electrode structures or contacts with fluid chemicals or gases. Inthis chapter, we have demonstrated that LC networks can also generatesurface deformation when adhered as a coating to a rigid substrate. The samemolecules that generate geometrical changes in films can create volumetricchanges in the coatings. The mechanism is, however, at least partly, different.The changes in order parameter induced by photoinduced trans-to-cis con-formations in azobenzene derivatives create intermolecular free space, which,for instance, appears from a reduction in density. However, this free volumeis only maintained as long as the azobenzene units remain into a continuousoscillation dissipating its energy to the polymer network. That also explainsthat only 2wt% of added azobenzene monomer, corresponding to roughly2 vol%, creates free volume as much as 10 vol%.When coatings aremadewith patterned director profiles, the effect of volume

increase and order-parameter-related geometrical changes are combined.Thisenables the topographic structures to reach heights, which are 20% of the coat-ing thickness, depending on the coating thickness, azobenzene concentration,and light intensity. The director patterns can be regular as line gratings, morerandom in the form of fingerprints or random spikes with irregular heights.The heights of the structures that form under light exposure are larger whenthe azobenzene molecules are actuated simultaneously in their trans and cisstates.

References

1 Archard, J.F. (1974) Surface topography and tribology. Tribology, 7,213–220.

2 Sayles, R.S. and Thomas, T.R. (1978) Surface topography as a nonstationaryrandom process. Nature, 271, 431–434.

�

� �

�

References 323

3 Skedung, L., Arvidsson, M., Chung, J.Y. et al. (2013) Feeling small: exploringthe tactile perception limits. Scientific Reports, 3, 2617.

4 Shivapooja, P., Wang, Q., Beatriz, O. et al. (2013) Bioinspired surfaces withdynamic topography for active control of biofouling. Advanced Materials,13, 1430–1434.

5 Zhou, G. and Shi-Chune, Y. (2011) Effect of surface roughness on laminarliquid flow in micro-channels. Applied Thermal Engineering, 31, 228–234.

6 Kasemo, B. (2002) Biological surface science. Surface Science, 500, 656–677.7 Lafuma, A. and Quere, D. (2003) Superhydrophobic states. Nature Materi-

als, 2, 457–460.8 Autumn, K., Liang, Y.A., Hsieh, S.T. et al. (2000) Adhesive force of a singlegecko foot-hair. Nature, 405, 681–685.

9 Mendes, P.M., Cellular nanotechnology: making biological interfacessmarter, Chemical Society Reviews 42, 9207—9218 (2013).

10 Stannard, W. and O’Callaghan, C. (2006) Ciliary function and the role ofcilia in clearance. Journal of Aerosol Medicine, 19, 110–115.

11 Rodríguez-Hernández, J. (2015) Wrinkled interfaces: taking advantage ofsurface instabilities to pattern polymer surfaces. Progress in Polymer Science,42, 1–41.

12 Vandeparre, H. and Damman, P. (2008) Wrinkling of stimuloresponsivesurfaces: mechanical instability coupled to diffusion. Physical Review Letters,101, 124301–124304.

13 Sánchez, C., De Gans, B.J., Kozodaev, D. et al. (2005) Photoembossingof periodic relief structures using polymerization-induced diffusion: acombinatorial study. Advanced Materials, 17, 2567–2571.

14 Paul, K.E., Breen, T.L., Aizenberg, J., and Whitesides, G.M. (1998) Masklessphotolithography: embossed photoresist as its own optical element. AppliedPhysics Letters, 73, 2893–2895.

15 Berendsen, C.W.J., Škeren, M., Najdek, D., and Cerný, F. (2009) Super-hydrophobic surface structures in thermoplastic polymers by interfer-ence lithography and thermal imprinting. Applied Surface Science, 255,9305–9319.

16 Liu, D., Bastiaansen, C.W.M., Den Toonder, J.M.J., and Broer, D.J. (2012)Photo-switchable surface topologies in chiral nematic coatings. AngewandteChemie, International Edition, 51, 892–896.

17 Liu, D., Bastiaansen, C.W.M., Den Toonder, J.M.J., and Broer, D.J. (2012)Light-induced formation of dynamic and permanent surface topologies inchiral-nematic polymer networks. Macromolecules, 45, 8005–8012.

18 Liu, D. and Broer, D.J. (2014) Self-assembled dynamic 3D fingerprints inliquid-crystal coatings towards controllable friction and adhesion. Ange-wandte Chemie, International Edition, 53, 4542–4546.

19 Liu, D. and Broer, D.J. (2014) Light controlled friction at a liquid crystalpolymer coating with switchable patterning. Soft Matter, 10, 7952–7958.

�

� �

�


20 Liu, D., Liu, L., Onck, P.R., and Broer, D.J. (2015) Reverse switching of sur-face roughness in a self-organized polydomain liquid crystal coating. PNAS,112, 3880–3885.

21 Liu, D. and Broer, D.J. (2015) New insights into photoactivated volume gen-eration boost surface morphing in liquid crystal coatings. Nature Communi-cations, 6, 8334.

22 Yager, K.G. and Barret, C.J. (2009, Chapter 4) Amorphous azobenzenepolymers for light-induced surface patterning, in Smart Light-ResponsiveMaterials: Azobenzene-Containing Polymers and Liquid Crystals (eds Y.Zhao and T. Ikeda), John Wiley & Sons, Inc, Hoboken, NJ.

23 Kulikovska, O., Gharagozloo-Hubmann, K., Stumpe, J. et al. (2012) Forma-tion of surface relief grating in polymers with pendant azobenzene chro-mophores as studied by AFM/UFM. Nanotechnology, 23, 485309–485323.

24 Wendorff, J.H., Scattering in liquid crystalline polymer systems in LiquidCrystalline Order in Polymers, A. Blumstein), New York, Academic Press, ,Chapter 1, 1–41 (1967).

25 DeGennes, P.G. (1969) Possibilites offertes par la reticulation de polymeresen presence d’un cristal liquide. Physics Letters, 28A, 725–726.

26 Strzelecki, L. and Liebert, L. (1973) Sur la synthèse de quelques nouveauxmonomères mésomorphes. Polymérisation de la p-acryloyloxybenzylidènep-carboxyaniline. Bulletin de la Société Chimique de France, 2, 597–605.

27 Bouligand, Y., Cladis, P., Liebert, L., and Strzelecki, L. (1974) Study ofsections of polymerized liquid crystals. Molecular Crystals and LiquidCrystals, 25, 233–252.

28 Clough, S.B., Blumstein, A., and Hsu, E.C. (1976) Structure and thermalexpansion of some polymers with mesomorphic ordering. Macromolecules,9, 123–127.

29 Arslanov, V.V. and Nikolajeva, V.I. (1984) Fixation of the structure of aliquid-crystalline monomer based on aromethine by polymerisation in themesophase. Vysokomolekulyarnye Soedineniya, Seriya A, 26, 208.

30 Shannon, P.J. (1984) Photopolymerization in cholesteric mesophases.Macromolecules, 17, 1873–1876.

31 Broer, D.J., Finkelmann, H., and Kondo, K. (1988) In-situ photopolymer-ization of oriented liquid-crystalline acrylates, 1. Preservation of molecularorder during photopolymerization. Makromolekulare Chemie, 189, 185–194.

32 Broer, D.J., Mol, G.N., and Challa, G. (1989) In-situ photopolymerization oforiented liquid-crystalline acrylates, 2 Kinetic aspects of photopolymeriza-tion in the mesophase. Makromolekulare Chemie, 190, 19–30.

33 Hoyle, C.E., Chawla, C.P., and Griffin, A.C. (1988) Photopolymerization ofa liquid crystalline monomer. Molecular Crystals and Liquid Crystals, 157,639–650.

�

� �

�

References 325

34 Hoyle, C.E., Chawla, C.P., and Griffin, A.C. (1989) Photoinitiated polymer-ization of a liquid crystalline monomer: order and mobility effects. Polymer,30, 1909–1912.

35 Broer, D.J., Boven, J., Mol, G.N., and Challa, G. (1989) In-situ photopoly-merization of oriented liquid-crystalline acrylates, 3. Oriented polymernetworks from a mesogenic diacrylate. Makromolekulare Chemie, 190,2255–2268.

36 Broer, D.J., Hikmet, R.A.M., and Challa, G. (1989) In-situ photopolymeriza-tion of oriented liquid-crystalline acrylates, 4. Influence of a lateral methylsubstituent on monomer and oriented network properties of a mesogenicdiacrylate. Makromolekulare Chemie, 190, 3201–3215.

37 Broer, D.J., Mol, G.N., and Challa, G. (1991) In-situ photopolymerizationof oriented liquid-crystalline acrylates, 5. Influence of the alkylene spaceron the properties of the mesogenic monomers and the formation andproperties of oriented polymer networks. Makromolekulare Chemie, 192,59–74.

38 Broer, D.J. and Heynderickx, I. (1990) Three-dimensionally-ordered polymernetworks with a helicoidal structure. Macromolecules, 23, 2474–2477.

39 Lub, J. and Broer, D.J. (2011, Chapter 1) Densely crosslinked liquid crys-tal networks by controlled photopolymerization of ordered liquid crystalmonomers – properties and applications, in Cross-Linked Liquid CrystallineSystem (eds D.J. Broer, G.P.C. Crawford, and S. Zumer), Taylor & FrancisGroup, LLC, Boca Raton, FL.

40 Broer, D.J. (1995) Creation of supramolecular thin film architectures withliquid-crystalline networks. Molecular Crystals and Liquid Crystals Scienceand Technology, Section A, 261, 513–517.

41 Lub, J., Broer, D.J., Hikmet, R.A.M., and Nierop, K.G.J. (1995) Synthesisand photopolymerization of cholesteric liquid crystalline diacrylates. LiquidCrystals, 18, 319–326.

42 Hikmet, R.A.M. and Broer, D.J. (1991) Dynamic mechanical properties ofanisotropic networks formed by liquid crystalline acrylates. Polymer, 32,1627–1632.

43 Broer, D.J. and Mol, G.N. (1991) Anisotropic thermal expansion of denselycrosslinked oriented polymer networks. Polymer Engineering and Science,31, 625–631.

44 Mol, G.N., Harris, K.D., Bastiaansen, C.W.M., and Broer, D.J. (2005)Thermo-mechanical responses of liquid-crystal networks with a splayedmolecular organization. Advanced Functional Materials, 15, 1155–1159.

45 Warner, M., Modes, C.D., and Corbett, D. (2010) Curvature in nematicelastica responding to light and heat. Proceedings of the Royal Society ofLondon A, 466, 2975–2989.

46 De Haan, L.T., Schenning, A.P.H.J., and Broer, D.J. (2014) Programmedmorphing of liquid crystal networks. Polymer, 55, 5885–5896.

�

� �

�


47 Yu, Y., Nakano, M., and Ikeda, T. (2003) Photomechanics: directed bendingof a polymer film by light. Nature, 425, 145.

48 White, T.J., Serak, S.V., Tabiryan, N.V. et al. (2009) Polarization-controlled,photodriven bending in monodomain liquid crystal elastomer cantilevers.Journal of Materials Chemistry, 19, 1080–1085.

49 Van Oosten, C.L., Bastiaansen, C.W.M. and Broer, D.J., Printed artificialcilia from liquid-crystal network actuators modularly driven by light, NatureMaterials 8, 677–682 (2009).

50 McConney, M.E., Martinez, A., Tondiglia, V.P. et al. (2013) Topographyfrom topology: photoinduced surface features generated in liquid crystalpolymer networks. Advanced Materials, 25, 5880–5885.

51 Ware, T.H., McConney, M.E., Wie, J.J. et al. (2015) Voxelated liquid crystalelastomers. Science, 347, 982–984.

52 Barrett, C.J., Natansohn, A.L., and Rochon, P.L. (1996) Mechanism of opti-cally inscribed high-efficiency diffraction gratings in azo polymer films. TheJournal of Physical Chemistry, 100, 8836–8842.

53 Natansohn, A., Rochon, P., Meng, X. et al. (1998) Molecular addressing?Selective photoinduced cooperative motion of polar ester groups in copoly-mers containing azobenzene groups. Macromolecules, 31, 1155–1161.

�

� �

�

327

10

Photoinduced Shape ProgrammingTaylor H. Ware

Department of Bioengineering, The University of Texas at Dallas, Richardson, TX, USA

In architecture, a guiding design principle is that form must follow function;this concept is not, however, limited to architectural structures [1]. In thenatural and synthetic worlds, examples abound of organisms and devices thatchange shape as the desired function changes [2, 3]. It is with this example thatmaterials scientists have long been intrigued with the possibility of materialscapable of exhibiting shape change. This goal has been adopted by manyscientists under the umbrella of stimuli-responsive materials [4]. While such adescription can be broadly applied to many property changes as a function ofa change in environmental conditions, shape change has received significantattention. Mechanically active materials are perhaps most intriguing whenthe material does not simply replace a machine but instead offers significantadvantages in terms of functional performance, property, or mechanismof operation. One particular opportunity afforded by recent advances inmaterials science is the ability to use light to directly or indirectly trigger shapechange [5]. In this chapter, we explore the demonstrated and potential usesof light as it relates to the ability to program and trigger shape-responsivebehavior. This chapter briefly reviews the key classes of shape-responsivematerials, framing the unique characteristics afforded by light.Shape change in a material can be broadly classified into two categories:

one-way shape memory and two-way shape memory. It should be noted,however, that in the scientific literature, there are numerous uses of theseterms, and as such, each must be defined for clarity. Throughout this chapter,the term shape-changing material is used to describe any material that exhibitsa programmable and anisotropic shape change in response to a stimulus.One-way shape memory is defined here as the ability to temporarily store ametastable shape and subsequently recover a globally stable shape [6]. Thisphenomenon is widely observed in polymeric materials and is essentially


�

� �

�

328 10 Photoinduced Shape Programming

temporal control of the elastic recovery seen in elastomers. Two-way shapememory is defined as a reversible shape change that occurs spontaneouslyin response to a change in environmental conditions. Upon returning to theoriginal environmental conditions, the original shape is also regained. Thisbroad definition could include effects such as isotropic thermal expansionor swelling of a polymer network in a good solvent, but in this chapter, suchwidely observed effects are not discussed. It is the goal of this chapter toelucidate the unique roles that light may play at each stage of the process inshape-changing materials. Before directly reviewing the ways that researchershave used light, there are several high-level design considerations that areapplicable to many shape-changing materials systems to discuss.In the design of shape-changing materials for a particular application, it is

first important to define the type of shape change that is needed. A particularshape change, without regard to its reversibility or required stimulus, can beclassified by the desired changes in each direction. For example, swelling ordeswelling of a uniformly cross-linked polymer network generates a volumechange but does not exhibit anisotropy.The recovery of an elastomer deformedalong a single axis, in contrast, is isochoric but is highly anisotropic. While themechanism through which shape change is achieved determines the volumeconserving or preserving nature of the deformation, the anisotropy of theshape change is widely dependent on material design, including processingconditions, or the delivery of the stimulus. In general, this anisotropy canarise either from anisotropy in the stimulus or through intrinsic anisotropyin the material. Examples of each possibility are discussed in this chapter.Light provides, perhaps, the most opportunity to design an anisotropicstimulus. When designing an anisotropic stimulus, it should be noted that thestimulus can change in a number of different domains. Spatial and temporalselectivity is a readily achievable advantage of light, as compared to heat orsolvent. Furthermore, the stimulus can be delivered from a distance in a trulynoncontact manner. It should be noted, however, that optically responsivematerials have additional aspects that must be considered, especially regardingabsorption at the stimulus wavelength. An inherent trade-off exists where,as absorption increases the effective stimulus, however, the power of thestimulating radiation decreases exponentially through the material thickness.This relationship can be expressed quantitatively with the Beer–Lambert law[7]. A modified version of this law is presented in Equation 10.1.

PP0

= e−𝜀bc. (10.1)

In this equation, the fraction of transmitted radiant power ( PP0) is related

exponentially to the molar absorptivity of the absorbing medium (𝜀), the path

�

� �

�

10.1 One-Way Shape Memory 329

length (b), and concentration (c). For optically thick materials, this relationshipensures that anisotropic stimulus is observed through the material thickness.The details of this consideration are discussed in relation to particular materialcharacterization. Furthermore, other unique opportunities provided withlight, such as polarization-controlled shape response, are discussed. Finally,special attention is required to differentiate photochemical mechanisms,where the primary mechanism of shape change does not rely on changing thesample temperature, from photothermal mechanisms. Both mechanisms, andcombinations thereof, have been used extensively in shape-changing materialsand are discussed in this chapter. Having introduced the relevant nomenclatureand design criteria, the three types of shape change are briefly reviewed.

10.1 One-Way Shape Memory

One-way shape memory is a phenomenon that has been widely reported inpolymers and metals and, to some extent, in ceramics, although the mecha-nisms responsible for this behavior differ [8–10]. As photoresponsive one-wayshape memory metals and ceramics have not been widely reported in theliterature, the focus of this work is on shape memory polymers (SMPs). Itshould be noted that SMP does not refer to a particular class of polymer but,instead, denotes a material that has a variety of necessary characteristics andhas undergone several processing steps [11]. As a result, an incredibly widevariety of chemistries can be used as SMPs, including thermosets, thermo-plastics, and supramolecular polymers. These materials can be single-phaseor phase-separated and might be amorphous, semicrystalline, or liquidcrystalline in nature [12]. A number of books and reviews on the broader topicof shape memory extensively review the topic of thermally induced shapememory in polymeric materials; here, we only provide brief context of thesematerials and processes to illuminate the behavior of photoresponsive SMPs.These processing steps for a traditional thermal SMP are shown in Figure 10.1.Briefly, the material is deformed at an elevated temperature. This deformationis then fixed while the material is cooled. Upon unloading, this deformation ismaintained.The efficacy of this strain retention is quantified as the shape fixityratio. This stored strain remains in the sample until the material is heated andthe original shape is recovered. Shape recovery ratio is a quantitative measureof the recovered strain. The term SMP has now been used to describe suchan extensive list of distinct polymeric materials; it is beyond the scope of thischapter to provide an exhaustive listing. Broadly, these materials must containtwo basic characteristics: cross-links and switching segments. Switchingsegments serve to fix the imposed deformation. Most commonly, switching

�

�

(A) (B)

(C)

Extension

and

coolingShape B

Shape A

200

150

100

50

0

2040

6080

*(a)*(b)

1000

200

400

Normaliz

ed forc

e (F/(E

*A))600

800

Deformation

Recovery

Unloading

Fixing

Temperature (°C)

Str

ain

(%

)

Shape B

50 mm 50 mm

Heating

Ttrans

Ttrans

Ttrans

Netpoint Switching segment, relaxed

Switching segment, elongated and fixed

°C

°C

°C

Figure 10.1 Shape memory polymers are thermally responsive materials that have been programmed into a metastable state; this state can befixed indefinitely by cooling. Upon heating, entropic elasticity drives recovery from the metastable state to the globally state (A). (Lendlein andKelch [10]. Reproduced with the permission of John Wiley and Sons. ) The shape memory cycle, both programming and recovery, is compared tothe behavior of an elastomer. An elastomer subjected to similar programming conditions is shown for comparison (b); note the lack of shapefixing upon cooling in the elastomer (B). (Liu et al. [6]. Reproduced with the permission of Royal Society of Chemistry. ) Shape memory polymerscan be incorporated into composite structures and used to generate complex shape change (C). (Felton et al. [13]. Reproduced with thepermission of Royal Society of Chemistry. )

�

� �

�


is achieved through crystallization or vitrification of the polymer networkbut can also refer to other mechanisms such as reversible cross-links, eitherphysical or chemical. Cross-linking segments, as defined in this context, serveto enable recovery from the metastable shape to the globally stable shape,and both physical and chemical cross-links are widely utilized. These activematerials have been proposed for a wide variety of smart devices, includingself-folding origami (Figure 10.1) [13]. In the context of photomechanicalone-way shape memory, examples from the literature of how light can beutilized to modulate shape memory behavior are discussed. Broadly, thesestrategies can be grouped into photothermal or photochemical; we begin withphotothermal examples.

10.1.1 Photothermal

Photothermal behavior, while mechanistically similar to ambient heating,presents several unique advantages and challenges as compared to ambientheating. Of primary concern are the coupling and dissipation of the desiredlight energy as heat. This can be achieved through utilization of the intrinsicabsorption, dispersion of absorbing fillers or dyes, or by coating absorbingmaterials on the surface of the material through techniques such as inkjetprinting. Dye-doped SMPs were among the first reported photothermallytriggered SMPs. Maitland et al. doped a commercially available polyurethaneSMP with an IR-absorbing dye [14]. This material was then fabricated intoa variety of smart devices designed to treat stroke or thrombosis [15, 16].This work provided the first system level design of photothermally triggeredshape memory, especially with regard to efficient light delivery and coupling tothe material. Specifically, these designs demonstrated an important criterionfor many photothermally activated smart polymers; light energy could bedelivered in such a way that ambient heating is minimal. This is particularlycritical in sensitive biological environments, such as the vasculature. Maitlandet al. further developed this concept to enable photoresponsive SMP foamsfor the treatment of aneurisms [17], as seen in Figure 10.2. Besides usingsmall-molecule dyes, a number of other fillers can also be used to provideoptical absorption in SMPs.SMP composites provide another versatile route to enabling photothermal

SMPs. These materials face many of the same benefits as dye-doped SMPs,but it should be noted that these composites may also be designed to improverecovery stress or electrical conductivity. Koerner et al. first demonstratedphotothermally triggered recovery of an SMP using the absorption of infraredlight by carbon nanotubes [18]. In this case, the nanotubes were used asmultifunctional fillers in a thermoplastic elastomer matrix providing absorp-tion, electrical conductivity, and increased recovery stress. Furthermore, thispaper demonstrated that localized recovery can be observed, as expected, for

�

� �

�


A

(a)

(b)

(c)

B C D

E F G H

Figure 10.2 The shape memory effect in polymers can be triggered using a photothermalstimulus. This remote triggering of this shape change has been proposed for two-way shapememory in a variety of biomedical applications, including a shape memory polymer foamdesigned to fill an embolism. Time-lapse photographs of a foam undergoing laser-triggeredrecovery in a model embolism (a). (Maitland et al. [17]. Reproduced with the permission ofSPIE. ) Photothermal heating of an SMP has also been proposed as a thrombectomy devicecapable of being inserted in a small form factor and undergoing subsequent shape change(b). (Small et al. [16]. Reproduced with the permission of Optical Society of America. )Photothermal triggering of shape memory behavior in carbon-nanotube-filled elastomersleads to complex shape recovery due to anisotropic heating (c). (Koerner et al. [18].Reproduced with the permission of Nature Publishing Group. )

highly absorbing materials, despite the presence of absorbing filler throughoutthe matrix. The result is a bending recovery shown in Figure 10.2. Sincethis initial investigation, carbon nanotubes have been used in a variety ofdistinct SMPs [19–21]. The one-way shape memory effect of Nafion has beenwidely investigated due to its broad thermal transition; this transition enablesprogramming of multiple metastable shapes [22]. Using photothermal heating

�

� �

�


of tuned intensity in the programming steps enables control of the recoverytemperature of each metastable shape [23]. This demonstration underlines thepoint that photothermal stimuli are not limited to triggering shape recoverybut can also be incorporated to enable specific processing conditions. Inaddition to carbon nanotubes, photoresponsive SMPs have been realized usingother absorbing fillers, such as carbon black, gold nanorods, and graphene.Gold nanorods are one particularly interesting example of an optical

absorber that provides additional functionality. The surface plasmon reso-nance is both highly wavelength and polarization selective [24, 25]. Theseparticles have been utilized in both amorphous and semicrystalline SMPs[26–28]. Zhang et al. aligned the gold nanorods in the sample through drawingthe composite [29]. The resulting composite only recovered strain whenilluminated with polarization aligned along the drawing direction.This uniquecharacteristic demonstrates another unique advantage to photoresponsivematerials as compared to ambient heating and is a good example of the abilityof light to act as a smart stimulus. Presumably, this strategy could be generatedwith a wide variety of dichroic composites that absorb at wavelengths acrossthe electromagnetic spectrum. Toward this direction, Clarke and coworkersrecently demonstrated that gold nanorods could be used to trigger hydropho-bic SMPs despite poor compatibility, further broadening the applicabilityof these results [30]. Besides the work directly related to the synthesis ofphotoresponsive materials and composites, significant progress has beenmade using spatially controlled photothermal stimulus to create complexone-way shape memory responses [31].In shape-responsive materials, spatial localization of the mechanical

response can lead to complex 3D shape transformations. There are twoestablished routes to generating this response: programmed absorption (smartmaterial) or programmed illumination (smart stimulus), or combinationsthereof. Furthermore, this can be extended through spatially controlledabsorption of a given wavelength. Work by Dickey and coworkers hasdemonstrated the utility of programmed absorption in one-way SMPs [32]. Acommercially available biaxially stretched polystyrene sheet, sold as a children’stoy named “Shrinky-Dink,” combined with printed absorbers (inks), can bedirected to fold from flat into complex 3D shapes. As shown in Figure 10.3, theprinted ink, localized to select regions, heats when exposed to IR light whilethe substrate is transparent to this stimulus. This local photothermal stimulusleads to a spatially and through-thickness heterogeneous shape recovery. Inshort, a printed line when heated with IR light generates a fold in the polymersheet, when the glass transition temperature of the substrate is exceeded [34].Three-dimensional shapes such as a box can be readily generated (Figure 10.3),but origami-inspired structures require directional and sequential control offolding. By printing on both sides of the sheet, the direction of the fold, theso-called mountain and valley folds, can be introduced. Cho and coworkers

�

� �

�


(a) (d)

(b) (e)

(c)

(g)

(i)

1

1

1

1

1

1

1

1

1

1

2

2

2

2

3

3

33

3

2

2

21

0

Transparency (%)

50

100

111

4

4

4

4

44

1

(h)

(f)

Figure 10.3 The photothermally triggered shape memory effect can be used to createobjects that morph from flat to complex 3D shapes by localizing light absorption. Byprinting IR-absorbing black ink on prestretched polystyrene, mountain and valley folds canbe introduced (a–c). By making arrays of these folds, origami-inspired shape memory can beobtained (d–f ). (Liu et al. [32]. Reproduced with the permission of Royal Society ofChemistry. ) Complex arrays can be generated, leading to complex 3D structures such as anicosahedron (g–i). Time-lapse images of photothermally triggered recover of an icosahedronis shown in (j). (Creative Commons License http://creativecommons.org/licenses/by/4.0/legalcode). (Lee et al. [33]. Reproduced with the permission of Royal Society of Chemistry. )

�

� �

�


43 (°C)(j) 70 (°C) 106 (°C)

25 (°C)

109 (°C)

27 (°C)

112 (°C) 117 (°C)

30 (°C) 31 (°C) 32 (°C)

30 (°C)


demonstrated sequential folding of complex geometric shapes, such as icosa-hedrons [33]. The result is a versatile platform for irreversible morphing ofpolymer substrates from flat to complex origami-inspired shapes based onlocalized absorption. In addition to modifying nonabsorbing polymer matriceswith absorbing fillers or coatings, intrinsically absorbing materials have beenused for photothermally triggered SMPs.Polymerization of polymer networks with intrinsic absorption can be

used to effectively trigger a shape change upon irradiation. Kumpfer andRowan have polymerized metal–ligand complexes, such as 4-oxr-2,6-bis(N-methylbenzimidazolyl), with olefinic monomers to generate materials thatheat in response to light [35]. In these materials, the metal complexesphase-separate to form hard domains capable of serving as reversible physicalcross-links. Upon irradiation with UV light, these hard domains preferentiallyabsorb, heat, and soften. As a result, the elastomeric matrix retains strainwhen deformed under UV light, and subsequent exposure to UV light, in theabsence of an applied load, leads to recovery of the permanent shape. Similarcontrol may also be possible with other supramolecular physical cross-linkssuch as ureidopyrimidinone, which forms an array of hydrogen bonds andintrinsically absorbs UV light, and has previously been used in self-healing andshape memory materials [36–38].In summary, light can be used to trigger the one-way shape memory effect

in a wide variety of materials. There are two key components to a materialthat may be used as a photothermally triggered SMP: an absorber to a desiredwavelength and a material with both cross-links and a switchable fixing phase.

�

� �

�


In some cases, the absorption is intrinsic to the polymer, but through theuse of absorbing fillers or dyes, nearly any traditional thermal SMP can beconverted to a photothermally triggered SMP. However, this modification is farfrom trivial and requires careful consideration of the methods of processingand triggering the shape change. Nonetheless, photothermal SMPs may holdpromise for remotely triggering biomedical devices in the vasculature or otherdifficult-to-reach locales and creating self-folding structures. Future effortswill likely continue on two fronts: devices and materials chemistry. Due tothe plethora of potential SMP chemistries and demonstrated additives thatcan efficiently transduce light of a variety of wavelengths to heat, it is theauthor’s opinion that engineers will continue to probe applications using thisrelatively accessible form of noncontact activation. This may be particularlylikely in applications where ambient heating is not tolerable. The second areaof expected advances will be in new materials chemistry in polymers andpolymer composites that combine photothermal activation of the one-wayshape memory effect with other functional properties. This likely will includethe synthesis of new monomers and materials. The strategy of copolymerizingactive monomers in the polymeric material can also be used to drive shapememory in processes not based on temperature change.

10.1.2 Photochemical

A wide variety of photochemical mechanisms have been explored to introducephotoinduced stimulus response, including self-healing, self-cleaning, andshape memory [39]. Although the focus of this portion of this chapter isone-way shape memory, it should be noted that these stimuli-responsivebehaviors have often interrelated requirements and often use similar chemicalmechanisms.There are broadly two classes of photochemical reactions used inSMPs. The first, and more common, is to use reversible exchange or additionreactions to dynamically break and form covalent cross-links. The second is touse photoisomerization of a chromophore in the network. It should be notedthat both of these mechanisms do not proceed through thermal mechanisms,but that through absorption of the incident radiation temperature rise mayoccur concurrently and alter the photochemical reaction and, as such, mustbe carefully considered. It is first worth considering exactly where in the shapememory cycle light can be used as a stimulus. In the traditional thermal shapememory cycle about a phase transition in a polymer, heat is first combinedwith a mechanical deformation.The heat is then removed and the deformationis fixed, until heat is reapplied. In reversible photochemical reactions, lightcan be used to replace or fix a mechanical deformation. This is typicallyachieved through the formation of new cross-links or rearrangement ofexisting cross-links within the material. The altered cross-links can serve tostabilize the deformed chain conformation. Light can then also be used to

�

� �

�


reverse these reactions and break these altered cross-links. Light can alsobe used to replace the mechanical deformation; this direct photomechanicaleffect is discussed in the two-way shape memory portion of this chapter.From a photochemistry perspective, we first consider dynamic exchange andaddition reactions utilized in SMPs.Reversible binding groups can be used to serve as the fixing phase in shape

memory materials. This fixing phase prevents the entropically driven elasticrecovery of the polymer network. The first photochemically triggered SMPswere reported by Lendlein et al., where reversible dimerization of cinnamicgroups were either polymerized or doped within a network containing photo-stable cross-links (Figure 10.4) [40]. By exposing a deformed sample to UV light(>260 nm), dimers form and subsequently fix the deformation. Conversely,this deformation will relax after exposure to UV light of shorter wavelengths(<260 nm). This more energetic optical stimulus cleaves the cross-linkingdimers and allows for the entropically driven recovery toward to the shapedictated by the photostable cross-links. It should be noted that this mechanismis fundamentally different from utilizing a thermally induced phase transitionsuch as the glass transition or melting to fix mechanically programmed strainwhere shape is fixed by limiting the mobility of the polymer chain through aphase transition. This initial example, however, led to reasonably low shapefixity, <50%, and programming and recovery irradiations that took over anhour. However, this initial demonstration catalyzed a new direction in pho-tochemically activated one-way SMPs. Other efforts have also demonstratedthe use of other cinnamides as well as coumarin and anthracenes that are alsocapable of reversible dimerization [42, 43]. Importantly, these photochemicallyactive materials can operate over a wide temperature range, which may enableuse in applications without a constant working temperature. Alternatively,thermal insensitivity can enable materials that are by design both thermallyand photoresponsive. For instance, Wang et al. used coumarin moieties tostore deformation in a hyperbranched polymer where the glass transitioncould be used to fix a third shape. The resulting triple shape material containsboth photonically and thermally “memorized” shapes. This decoupling isnot generally possible in photothermally triggered shape memory materials.Although these chemistries often require exposure of the sample to shortwavelength UV light, which is quickly attenuated by the polymer matrix, andmany other materials may limit its practicality, this work served to catalyzea new direction in photochemical SMPs. More recently, a wide variety ofchemistries that utilize more accessible optical stimuli have been described.Covalently cross-linked networks that contain specifically designed labile

moieties have been designed such that rearrangement of the network can beachieved in response to a stimulus. This class of material adds a degree ofthermoplastic character to covalently cross-linked materials and could leadto multifunctional or recyclable thermosets. Many of these mechanisms are

�

�

Stretching and

photo fixing

UV λ > 260 nm

Photocleaving

UV λ > 260 nm

a

(A)

(C)

(B)

a

a

b

b

b

c

c c

Removing

external

stress

Figure 10.4 Photochemical modifications of polymer structure can be used to design light-responsive shape memory polymers. By exposure tolong-wave UV light, a cinnamyl-containing polymer, cross-links are formed fixing a distinct shape (b). Upon exposure to short-wave UV light,these cross-links are broken and the original shape (a) is recovered. These shapes include those resulting from tensile deformation (A) andcomplex bending deformation (B). The reversible formation of these cross-links within the polymer network is shown in (C). (Lendlein et al. [40].Reproduced with the permission of Nature Publishing Group. ) Photo-origami can be accomplished through the process shown in (D). Thisprocess relies on a photoinduced fixing of strain through reconfigurable cross-links. A six-sided box was demonstrated (E) and a model developedthat matched experimental observations (F–H). (Ryu et al. [41]. Reproduced with the permission of AIP Publishing LLC. )

�

�

Uniaxially strain to εx = ε1

(D) 1a

1b

(E) (F) (G) (H)

Uniaxially strain to εy = ε1

Irradiate through mask 2

Irradiate through mask 1a,

Uniaxially strain to εx = ε1 < ε1,

Irradiate through mash 1b

Remove load

Stamp box perimeterRemove load


�

� �

�


featured in the self-healing polymer literature and are often described as cova-lent adaptable networks [44]. A variety of thermally induced rearrangementmechanisms have been explored including the Diels–Alder cycloadditionreaction, disulfide exchange, and transesterification [45–48]. In these dynamicnetworks, light can be used to trigger a change in the globally stable shapeof the polymer after initial polymerization. Several mechanisms exist forgenerating exchangeable cross-links with light, including networks containingallyl sulfides or trithiocarbonates [49, 50]. Allyl sulfide-containing networkshave received particular attention within SMPs. Bowman et al. incorpo-rated allyl sulfides into amorphous networks formed through the radicalpolymerization of polyfunctional thiols and an allyl disulfide-containingdivinyl ether. Radical-mediated reorganizations break the existing bonds andreorganize new cross-links, permanently fixing the applied deformation. Byirradiating a stretched sample, recovery is limited in the irradiated regions,and complex recovery can be observed [51, 52]. For example, by irradiatingan optically thick sample from one side under an imposed load, the samplebends away from the irradiation after the load has been released. By patterningthe illumination, photo-origami can be achieved as demonstrated by Qi andcoworkers (Figure 10.4) [41, 53]. The adaptable nature of these materials isreliant on the presence of a latent radical source within the material, such asa photoinitiator. While this can limit the number of times a material can bereprogrammed, it provides a convenient mechanism to control the desiredwavelength of irradiation that activates the exchange of cross-links withinthe material. Furthermore, in photopolymerized networks, care should betaken that during polymerization, the radical source is not exhausted. Thiscan be addressed through the introduction of two initiators with sufficientlydifferent absorption profiles (e.g., visible and UV). Trithiocarbonates are atype of reversible addition–fragmentation chain-transfer agents that do notnecessarily require a photoinitiator to induce rearrangement in dynamiccovalent networks [49]. These moieties undergo direct cleavage. In addition tostrategies that generate dynamic covalent bonds in response to light, a varietyof photoinduced SMPs have taken advantage of the dynamic and reversiblenature of supramolecular interactions.Molecular photoisomerization is a powerful tool that has been widely used

throughout the field of stimuli-responsive materials [54–56]. The key to thismechanism is a chromophore capable of a rearrangement at the molecularlevel that leads to a change in shape. Perhaps, the most widely referencedexample of such an isomerization is the interconversion of trans and cisisomers around a double bond. This occurs in a wide variety of moleculesthat contain carbon–carbon double bonds, including maleic acid and stilbenederivatives [57, 58]. Perhaps, the most widely used in materials chemistry,however, are derivatives in azobenzene. In 1966, Merian observed the photo-contraction of nylon fibers dyed with an azobenzene derivative and attributed

�

� �

�


it to the structural change in the azobenzene chromophore [59]. In thisinitial demonstration, the observed deformation was, however, exceedinglysmall (∼0.1% strain). Since that time, significant attention has been paid toazobenzene as a photomechanical transducer, and it has been incorporatedinto a wide variety of polymeric systems. It is important to note that in thecase of azobenzene, the trans state is generally thermodynamically stable,but the cis isomer can be induced through illumination with UV light [60].The trans isomer can be recovered through exposure to visible light or byheating. As such, the chromophore acts as a molecular motor and can generatemacroscopic shape change. The behavior of azobenzene in response to avariety of light stimuli is summarized in Figure 10.5. Based on the pioneeringwork by Ikeda and coworkers, polarization-directed bending using UV lightand recovery using visible light have been widely used to generate reversibleshape change [62]. These two-way shape memory materials will be discussedin detail later, but first we will discuss using photoisomerization to controlone-way shape memory behavior.Illumination of azobenzene-containing polymer networks with blue-green

light has been used to replace a variety of steps within the traditional one-wayshape memory cycle. Blue-green light excites both the trans–cis isomeriza-tion and the cis–trans isomerization. By using polarized light, the net resultof this molecular cycling is that the chromophores orient the incoming lightperpendicular to the electric field vector [63, 64]. In both amorphous poly-imides and liquid-crystalline acrylic networks, this leads to bending or twisting,thus potentially eliminating the need to provide external loading to program aSMP [65, 66]. By controlling the network structure or processing conditions,the degree to which this bend is fixed can be controlled [67]. As such, theisomerization of azobenzene leads to a fixed state, analogous to a thermallyprogrammedmaterial.The role of network structure and composition in shapefixing has been subsequently studied [68]. In addition to material composi-tion and structure, it was found that thermal history, specifically the introduc-tion of free volume, of the glassy network is critical. Namely networks withlarge free volume, generated through quenching above the glass transition, iscritical to shape fixing. The globally stable shape can be recovered upon heat-ing or through exposure to circularly polarized light. Furthermore, it has beenshown that this photo fixing can be combined with thermally programmedshapes [69]. This is another example of the utilization of materials in nonequi-librium states in shape memory materials. Additionally, White and coworkershave demonstrated that light can be used to fixmechanically induced deforma-tion in azobenzene-containing glassy liquid crystal networks, as demonstratedin Figure 10.5 [61, 70]. Importantly, this shape memory cycle is completelyisothermal and uses a relatively benign wavelength of light, which may be par-ticularly attractive for a variety of applications. It should be noted, however, thatthe failure strain of the material is quite low, which limits the type of deforma-tion that can be fixed.

�

�

NN

NN

hv

hv ′, kT

O

O

O

O

OO

O

O

O

O

O

O

NN

OO

O

O

(a) (b)

(c)

(i) (ii)

(vi)

(iii) (iv)

(v)

Figure 10.5 Isomerization of azobenzene can be used to generate all-optical shape memory effect in liquid-crystalline polymers. Theisomerization of azobenzene from the trans isomer to the cis isomer can be accomplished with UV light and can be reversed with visible light orheat (a). Azobenzene-containing monomers can be polymerized with nonresponsive liquid-crystalline monomers to generate glassy liquidcrystal polymer networks (b). The resulting materials exhibit light-activated shape memory where (i) permanent shape, (ii) mechanicaldeformation, (iii) photo fixing, and (iv) shape retention (in the absence of light) are observed. Exposure to right-handed circularly polarized lightunlocks the photo-fixed state (v), allowing for the recovery of the permanent shape (vi) (c). (Lee et al. [61]. Reproduced with the permission ofRoyal Society of Chemistry. )

�

� �

�

10.2 Two-Way Shape Memory 343

In summary, photochemical activation of shape memory has experienceda decade of advances following a seminal paper by Lendlein et al. [40]. Theseresearch efforts have largely mirrored efforts in photo-healable polymers, aheavily researched area of stimuli-responsive polymers [39]. The materialschemistries that can be employed are still somewhat limited as comparedto photothermal two-way SMPs; however, this class of materials holds greatpromise as the shapememory cycle can be performed completely isothermally.This may be particularly attractive in biomedical applications where tissueheating may not be tolerable. However, the use of UV light, especially shortwavelengths (<300 nm), is particularly limiting for many applications. It isexpected, however, that strategies will continue to emerge from distinct areasof polymer chemistry, as was observed with reversible addition–fragmentationchain-transfer agents, such as allyl disulfides and trithiocarbonates, and willbe repurposed toward photochemical SMPs. The isomerization of azoben-zene provides a photochemical trigger that can be used to generate shapememory in response to a benign stimulus, blue-green linearly polarized light.Photoisomerization is widely used in two-way SMPs.

10.2 Two-Way Shape Memory

Two-way shape memory broadly refers to reversible shape changes in amaterials system. Often, these materials are categorized as artificial muscles,but these materials are not limited to simple mimicking of their naturalcounterparts. There are a wide variety of interactions that can be used toinduce two-way shape memory, such as thermal expansion, swelling, piezo-electricity, magnetostriction, electrostriction, photostriction, molecular shapechange, and a variety of phase transitions [71–79]. Each of these interactionsis reversible and can lead to dramatic shape changes in response to a stimulusif the composition and processing of the material are carefully controlled. Ascompared to shape memory materials, the key distinction is the driving forceto return from the activated state to the original state upon the removal ofthe stimulus. Such effects can be observed in polymers, metals, and ceramics[9]. Perhaps, the most widely used examples of two-way shape memory arepiezoelectric ceramics and shape memory alloys, which have found appli-cations from microelectromechanical systems to implantable devices suchas stents [80, 81]. In this work, we focus on how reversible shape changescan be induced using light as the stimulus. The majority of this section dealswith polymeric and carbonaceous materials as examples of these materialsare relatively abundant in the literature; however, molecular crystals andphotostrictive ceramics are also discussed. In addition to the discussion ofthe fundamental mechanisms of the photoresponsive material, opportunitiesto leverage mechanical design of two-way shape memory materials to direct

�

� �

�


and enhance the photomechanical effect are discussed. Furthermore, bothphotothermal and photochemical mechanisms of generating shape change arediscussed. Before beginning to review photoinduced two-way shape memorymaterials, we discuss the common issues and aspects that should be consideredin order to leverage a dimensional change at the molecular level for directedshape change.The capacity to convert light energy into mechanical work is a fascinating

possibility. Several factors must be considered when considering the use ofphotoinduced two-way shape memory. Among these factors are the max-imum reversible shape change, maximum stress generated, and the typesof shape change (torsion, bending, stretching) that can be generated. Ingeneral, however, shape change must be directed to be of use to the engineer.The process of directing the two-way shape memory effect often requires theintroduction of anisotropy in the material. In piezoelectric materials, this isoften referred to as poling the material; while in shape memory alloys, this iscalled training [82, 83]. Similarly, in ordered polymers, the ordered phase mustbe aligned [84]. In each case, the goal is to align the microscopic shape changeof the material over a large area. The method of alignment, often electric field,mechanical deformation, magnetic field, or self-assembly, limits the range ofpotential shapes that can be achieved, and the limitations for each class ofphotoresponsive materials are discussed in this chapter. It should be noted thattraining can be accomplished through structural anisotropy in a composite,such as might be provided by a fiber in a polymer composite. Finally, strategiesthat utilize spatial heterogeneity in a material that responds isotopically arediscussed.

10.2.1 Photothermal

Photothermal stimulus can be used to induce a number of reversible shapechanges including order–disorder transitions, swelling, and simple coefficientof thermal expansion. There are several important aspects when consideringphotothermal two-way shape memory. These two-way shape memory materi-als require control of not only the efficient transduction of incident radiationto heat but also the removal of the light stimulus after triggering the two-wayshape memory material, which ultimately leads to cooling, causing the mate-rial to revert to its original shape. This process highlights the importance ofnot only delivering energy to the system but also the ultimate removal of thisenergy. It is often the removal of this energy that limits the operational fre-quency of photothermal two-way shape memory materials. It is also worthnoting that a certain degree of stimulus is required. Minimal “on–off ” controlis required; however, spatial control of stimulus can be used to direct heteroge-neous two-way shapememory.The photothermal activatedmaterials reviewedhere will be contextualized by considering each of these factors. We begin by

�

� �

�


discussing polymeric photothermal two-way shapememorymaterials based onreversible swelling/deswelling.Thermoresponsive gels have been widely studied as dynamic materials, in

particular with regard to biological applications including drug delivery andtissue engineering [85]. Much of this work has utilized materials based onpoly(n-isopropylacrylamide) (PNIPAM) [86]. This material, which was firstsynthesized by Sprecht et al. in the 1950s, exhibits a lower critical solutiontemperature (LCST) in water in a physiologically relevant temperature range(∼32 ∘C for the homopolymer) [87, 88]. Since this initial discovery, significantattention has been paid not only to the behavior of the linear polymer insolution but also to the behavior of hydrogels, typically formed by the copoly-merization of n-isopropylacrylamide with a difunctional cross-linker, such asN,N ′-methylene-bis-acrylamide [86]. These gels share the thermoresponsivenature of the linear polymers, but due to the cross-linked nature of thepolymer chains, the change from hydrophobic to hydrophilic is manifested asa reversible swelling. It should be noted that swelling-based two-way shapememory materials require consideration of the mass transport required toinduce shape change, which is a function of chemistry, microstructure, andmacroscopic geometry of the two-way shape memory material. Furthermore,as the stimulus response is inherently isotropic in nature, spatial hetero-geneity must be utilized to direct the two-way shape memory material intocomplex forms. Due to the intense investigation of these thermoresponsivePNIPAM-based gels, it is perhaps not surprising that several examplesof photothermally triggered shape changes in these materials have beendescribed.As PNIPAM does not exhibit efficient optical absorption in the visible or

near-infrared spectrum, a key component of a photothermally triggered gelis the addition of an efficient absorber. This was first demonstrated usinga trisodium salt of copper chlorophyllin, a polymerizable chromophorethat absorbs in the visible spectrum (488 nm), by Suzuki and Tanaka [89].Cylindrical samples exhibited a 50% decrease in diameter upon irradiation,which was related to the local temperature of the sample and not to thephotochemical changes in the sample. By controlling the irradiation area, theisotropic two-way shape memory was observed to be spatially heterogeneous.Since this initial example, a variety of absorbers have been utilized, includingmany types of dyes, carbonaceous materials, and nanoparticles. Nayak andLyon covalently attached a highly efficient photon-to-heat converter, namelymalachite green, to colloidal cross-linked PNIPAM microparticles [90]. Thenecessary photothermal stimulus was found to strongly depend on the bathtemperature as well as the concentration of dye within the particles. It isimportant to note that due to the aqueous operating environment, dyes thatcovalently attach to the polymer network must be used. This restriction can

�

� �

�


be avoided through the use of nanomaterials trapped in the polymer networkthrough in situ polymerization.Significant advances in the synthesis of nanomaterials with tunable optical

properties have directly enabled a wide number of photothermally trig-gered gels. Carbonaceous materials, including graphene, graphene oxide,and nanotubes, have been shown to be efficient near-IR absorbers in ther-mally responsive hydrogels. Javey and coworkers demonstrated PNIPAMsingle-walled carbon nanotube two-way shape memory materials withtunable response time [91]. These two-way shape memory materials wereused to create self-folding origami-like structures, including a cube and aflower, by localizing these responsive materials to the hinges of a scoredpolymer substrate. Nakashima and coworkers showed that thermally inducedswelling–deswelling in PNIPAM carbon nanotube hybrid gels can be usedto generate shape changes for at least 1200 cycles (Figure 10.6) [92]. Takeiand coworkers fabricated NIPAM with CNTs capable of bending in responseto sunlight [94]. This two-way shape memory material was designed in sucha way that it is packaged into a sealed container allowing for operation inair, potentially enabling applications such as solar-powered mechanicallyresponsive curtains. However, this interesting two-way shape memory mate-rial is relatively slow (14min) to reach the maximum bending angle of 210∘.Efficient absorption of light is not, however, limited to carbon nanotubes butis observed in a wide variety of carbon-based nanomaterials.Graphene oxide combines optical absorption with relatively simple dis-

persion in aqueous solutions. This provides ample opportunity for grapheneoxide to be incorporated into hydrogels. Near-IR absorption of grapheneoxide has been utilized to trigger two-way shape memory. Jiang and coworkerspolymerized methacrylate functionalized graphene oxide sheets into a PNI-PAAM hydrogel [95]. The photothermally responsive gel deswelled by a factorof 15 over 75 s. Qu and coworkers used a porous graphene oxide NIPAMhydrogel to reversibly capture cells that could be released after exposure toan IR laser [96]. Wagner dispersed unfunctionalized graphene oxide sheetswithin a PNIPAM matrix [97]. As the graphene oxide was not chemicallyattached to the hydrogel, the LCST was unaffected by the incorporation ofthis material. The resulting material photothermally responds to blue light(460 nm, 1.5mW/cm2) over 90 s. Reduced graphene oxide has also beencombined with a distinct polymer matrix, elastin-like proteins that also exhibitan LCST phase transition, to generate bending and crawling two-way shapememory materials (Figure 10.6) [93]. Nanomaterials, besides those based oncarbon, have also been extensively used.The geometry ofmetal nanoparticles can be used to tune the surface plasmon

of the particle, leading to controllable, efficient absorption. West and cowork-ers incorporated silica nanoparticles with thin gold shells into PNIPAM-basedhydrogels. By tuning the gold shell thickness, the absorption of the particle

�

� �

�


0

1

0.9

0.8

0.7d/d

0

0.6

0.5

0.4200 400 600 800 1000 1200

Cycle number

(a)

(b) (c)

(1) Apply laser10

150 140 150 140

11 10 11

10

150 140 150 140

11 10 11

(2) Curl under

(3) Uncurl, push forward

(4) Return to initial state

Figure 10.6 Photothermal two-way shape memory of thermoresponsive gels producesdurable and potentially complex two-way shape memory materials. Carbonnanotube/NIPAM composites exhibit two-way shape memory in response to an NIR laser.This two-way shape memory can be repeated at least 1200 times (a). (Fujigaya et al. [92].Reproduced with the permission of John Wiley and Sons .) By localizing the irradiationsource, an elastin composite hydrogel two-way shape memory material can be used togenerate complex and controllable motion, such as a hand with individually addressablefingers (b) or a crawler (c). (Wang et al. [93]. Reproduced with permission from AmericanChemical Society.)

was tuned from 700 to 1050 nm, across the “water window” where the livingtissue is transparent [98, 99]. Gold nanoshells were more recently used bySukhishvili and coworkers to create photothermally responsive layered hydro-gels [100]. The absorption between the layers was controlled using alternatingdeposition of PNIPAM-grafted gold nanoshell, PINAPAM-grafted solid

�

� �

�


gold particles, or unresponsive polymers. Complex two-way shape memorymaterials were fabricated using different nanoparticles, allowing for selectivephotothermal two-way shape memory of certain layers, based on the irra-diation wavelength. The resulting composites could be designed to exhibitspatially heterogeneous shape change or true uniaxial two-way shape memory(Poisson ratio of 0) (Figure 10.7). Among the most powerful systems of plas-monic nanoparticles are gold nanorods, where the absorption of the particlecan be widely tuned by controlling the size of the nanorods. Kumacheva andcoworkers incorporated gold nanorods into PNIPAM microgels and showedrepeatable deswelling of the gel in response to irradiation with a laser at 810 nm[102]. Katayama and coworkers used poly(ethylene glycol)-functionalized goldnanorods to both induce localized shape change and release a model drug,a rhodamine-labeled dextran [102]. Serpe and coworkers have also usedphotothermal triggering of NIPAM–gold nanoparticle composites to createmultifunctional devices with both optical and drug release functionalities [103,104]. Microgels were used as the active layer of etalons, a photoresponsiveoptical element that reflects selective wavelengths of light tuned by thedistance between two partially reflective surfaces. Photothermal two-wayshape memory can also be used as a platform for reconfigurable two-wayshape memory materials. Hayward and coworkers demonstrated the utilityof patterned irradiation to direct the buckling of thin sheets into a library ofshapes with response times of ∼2 s for 25-μm-thick sheets [105]. This bucklingcan be used to generate directed motion by sweeping the stimulus acrossthe sample. This work highlights the complex buckling mechanics witnessedby compliant freestanding films that are subject to spatially complex stimuliresponse. A variety of nanoparticles with optical absorption can be used tophotothermally heat thermoresponsive gels.Besides metallic nanoparticles, oxide nanoparticles can also be utilized

to generate photothermal stimulus response. Iron oxide nanoparticles havebeen widely used for inductive heating in an alternating magnetic field, butthese particles also exhibit strong optical absorption. Hayward and coworkersutilized PNIPAM gels loaded with iron oxide nanoparticles to introducelocalized control of surface topography [101]. Thermoresponsive hydrogelsconstrained to a patterned surface have been shown to undergo reversiblecreasing at the free surface of the gel [106]. In the high-temperature (irradi-ated) state, the creases open on the gel surface. By localizing the irradiation,these creases can be selectively opened. Swelling and deswelling processeseach lasted 3–4min. Hayward et al. further demonstrated the utility of thisapproach by patterning the gel surface with distinct chemical patterns thatcould be locally obstructed through creasing (Figure 10.7). This may enabledynamic substrates for cell cultures that reversibly display different chemicalpatterns. Shi and coworkers created an all-polymer two-way shape memorymaterial by incorporating polypyrrole nanoparticles to create a fast-actuating

�

� �

�


Light

Dark

(h)

(g)

(g) (e)

546 nm 785 nm

546 nm 785 nm

(f)

(a) (b) (c)

(i) (j)

200 μm

200 μm 200 μm 200 μm

200 μm 200 μm 200 μm

200 μm 200 μm

Figure 10.7 Spatially patterned hydrogel photothermal two-way shape memory materialscan undergo complex mechanical responses to a photothermal stimulus. Layerednanocomposite hydrogels can be activated into complex shapes by limiting two-way shapememory to a subset of the layers in the gel. This is achieved through the use ofnanoparticles with different absorption spectra (a–f ). (Zhu et al. [100]. Reproduced with thepermission of American Chemical Society.) Hydrogels bound to a pattern surface undergo areversible creasing instability that can be triggered photothermally (g). The resultingsurfaces can reversibly display or hide chemical functionalities in predetermined patterns(h, i). Within a sample, subsets of these patterns can be activated using localized irradiation(j). (Yoon et al. [101]. Reproduced with the permission of John Wiley and Sons.)

�

� �

�


microvalve [107]. Upon irradiation within 2 s, deswelling led to the openingof a microchannel that would autonomously close 6–8 s after removal ofthe stimulus. Photothermally responsive hydrogels show great promise forapplications such as microfluidics, cell culture substrates, and implantabledevices, but in many applications, the environment is too dry to permitstability of these two-way shape memory materials. Furthermore, the use ofmass transport limits the dimensions of the two-way shape memory materialif the operating frequency is a critical performance factor. With this in mind,we now discuss polymeric two-way shape memory materials that are notbased on swelling and deswelling but on order–disorder transitions triggeredphotothermally.A variety of thermal phase transitions beyond the LCST have been used to

generate reversiblemechanical deformations in polymer and polymer compos-ite two-way shapememorymaterials.These include order–disorder transitionssuch as melting from a liquid-crystalline (nematic or smectic) phase or froma semicrystalline phase to an isotropic (or lower order) phase. Chemical orphysical cross-links can then be used to direct the reordering (upon cooling)such that the polymer regains the original shape. If the molecular order isprogrammed on a macroscopic level, large, reversible, and anisotropic strainscan be observed. This was first demonstrated in uniaxially aligned nematicnetworks. This deformation can approach 400% strain upon heating in lightlycross-linked nematic elastomers, where the mesogen is contained in the mainchain [108]. Since these thermally responsive materials were first described byFinkelmann et al., several examples of photothermally triggered liquid crystalelastomers have been described. Terentjev and coworkers demonstrated thatmultiwalled carbon nanotubes can be used to trigger reversible photothermaltwo-way shape memory in main-chain liquid crystal elastomers and surpris-ingly in aligned carbon nanotube composites with poly(dimethyl siloxane)[109, 110]. Chen and coworkers used single-walled carbon nanotubes inliquid crystal elastomers with side-on mesogens aligned through uniaxial hotdrawing [111]. Strains of ∼30% were observed in composites with 0.2wt%nanotubes. Jiang and coworkers used photothermal two-way shape memoryin similar nanocomposites to generate a heliotropic, sun-tracking mechanism[112]. The shape change in these materials, however, is not limited to uniaxialdeformations. Kohlmeyer and Chen demonstrated self-folding hinges thatcould be used to obtain origami structures [113]. These hinges consist of abilayer of an inactive silicone-based elastomer and a single-walled carbonnanotube doped and a side-chain liquid crystal elastomer. Several structureswere demonstrated, including an inchworm walker and a gripper, by control-ling the geometry of the photoresponsive film. These structures are shownin Figure 10.8. In addition to the mechanical alignment of the liquid crystalelastomers already discussed, certain liquid crystal polymers can be aligned

�

� �

�


40

35

30

25

20

1510 20 30 40 50 60

Sample–lamp distance (mm)

SWNT-LCE

(c)

(a) (b)

(d)

(e)

(f)

PC

PC

Silicone

Silicone

θ

θ

Θ(°)

Figure 10.8 Liquid crystal polymer networks can be triggered reversibly usingphotothermal stimulus to form complex two-way shape memory materials. Usingphotoalignment, the molecular order of liquid-crystalline polymer networks can bepatterned. When an IR-absorbing dye is included in the network, complex photothermaltwo-way shape memory is observed (a). For instance, a defect pattern can be introduced,which reversible morphs from flat to conical in nature. The opening angle of the conedepends on the intensity of the photothermal stimulus (b) [114]. Liquid crystalelastomer – silicone rubber bilayers that undergo reversible bending (c) can be fabricated.By incorporating carbon nanotubes, photothermal stimulus can be applied, and theresulting composite can be formed into a crawler (d) capable of using a notched surface (e)to generate locomotion (f ) [113]. (Broer and Mol [115]. Reproduced with the permission ofJohn Wiley and Sons.)

using patterned surfaces. Broer and Mol pioneered the use of surface align-ment to generate polymer networks with spatially patterned liquid-crystallineordering [115]. These materials largely consist of glassy, densely cross-linkednetworks formed from the polymerization of nematic diacrylates. Thesematerials do not undergo large-scale order–disorder transitions upon heatingbut, instead, gradually reduce the order over a wide temperature range (until

�

� �

�


thermal degradation) [116], resulting in anisotropic shape change, contractionalong the alignment axis (nematic director), and expansion perpendicular tothe alignment axis. When combined with optical patterning methodologies,films with complex order profiles can be synthesized, leading to spatiallycomplex shape changes without the need for patterned stimulus or the useof bilayers. The resulting two-way shape memory materials were shown tobe photothermally responsive after incorporating an IR-absorbing dye [114].The resulting complex shape changes are shown in Figure 10.8. Liquid crystalelastomers will be revisited in the discussion of photochemically triggeredtwo-way shape memory materials. Having discussed polymeric two-wayshape memory materials based on phase transitions, two-way shape memorymaterials based on photothermal expansion in the absence of a transition willbe explored.As thermal expansion is present in all materials and is normally isotropic

in nature, this chapter does not attempt to review how photothermal effectsmay be present in nearly any materials system. Instead, we focus on the designschemes that utilize the often small strains associated with thermal expansionto yield anisotropic and directed reversible shape changes. As was discussedin swelling-based two-way shape memory materials, isotropic volume changescan be directed using bilayer structures that bend in response to a photother-mal stimulus. Tsukruk and coworkers demonstrated that a polymer–siliconmicrocantilever is a sensitive photothermal shape-changing structure [117].As the polymer film (polystyrene) has a much larger coefficient of thermalexpansion as compared to the silicon cantilever (Δ𝛼 ≥ 200× 10−6 K−1), bendingof approximately 2 nm/mK is observed. These cantilevers could present a newtype of IR sensor, based on photothermal two-way shape memory materials,with unprecedented thermal resolution. However, two-way shape memorymaterials are not confined to small stroke shape changes. Baughman andcoworkers used coiling to amplify the thermal expansion found in drawn poly-mer fibers, such as fishing line [71]. The drawing process leads to a fiber thatcontracts in response to light. Photothermal tensile two-way shape memory of7% in less than 1 s against a stress of 25MPa was reported. The illuminatingsource was a 250W incandescent lamp. These examples indicate the potentialof using photothermally induced shape changes even in the absence of a phasetransformation. Despite this potential, photothermally responsive materialshave several potential disadvantages. Generically, two-way shape memorymay be induced by changes in the ambient temperature even in the absenceof light. Furthermore, photothermal heating of the two-way shape memorymaterial can lead to deleterious heating of sensitive environments such as thetissue around an implanted device. Photochemically activated materials, bycomparison, offer a distinct route to two-way shape memory.

�

� �

�


10.2.2 Photochemical

A variety of photochemical reactions can be used to generate reversible shapechanges in polymers and ceramics. In this chapter, the opportunities providedby these materials are reviewed. The synthetic versatility that is inherent topolymer chemistry provides a rich platform to incorporate photochemicallyresponsive moieties and, as such, has received the most attention in the liter-ature and will continue to be the focus of this chapter. After the pioneeringwork by Merian on the photomechanical effects of an azobenzene-dye-dopedpolymer fiber, molecular isomerization was recognized as a powerful tool toinduce macroscopic shape change [59]. While the trans–cis isomerization ofazobenzene remains a commonly usedmolecular transducer in photochemicaltwo-way shape memory materials, a wide variety of chromophores have beenexplored, including ring-opening reactions such as conversion of spiropyranto merocyanine and the dissociation reactions such as those of leuco deriva-tives [118–120]. Each of these photochemical reactions will be discussed. Inaddition, strategies to amplify and direct these molecular-level changes intomacroscopic shape changes will be emphasized. First, photochemical two-wayshape memory based on the isomerization of azobenzene will be discussed.Azobenzene, and most derivatives of azobenzene, predominately exists in

the rod-like trans isomer in the absence of stimulus [121]. Upon exposureto UV light, isomerization to the bent cis isomer efficiently occurs, leadingto a decrease in molecular length of approximately 0.35 nm. Furthermore,azobenzene is dichroic and therefore can be aligned using trans–cis–transisomerization, resulting in a statistical buildup of the isomer in responseto polarized light. This mechanically active chromophore has been deriva-tive to form monomers that can be incorporated into semicrystalline,liquid-crystalline, and amorphous polymers. It should be noted that thechemical nature of the substituents changes the photochemical behavior,absorption, and relative stability of both isomers, with respect to azobenzene,but a large number of derivatives exhibit controllable photoisomerization.The details of this rich photochemistry are reviewed elsewhere in this book;here, we focus on two-way shape memory materials that utilize this molecularswitch, beginning with amorphous materials.As the azobenzene moiety is stable under a wide variety of conditions,

it can be incorporated through a wide variety of polymerization tech-niques. Eisenbach was among the first to design a monomer containingazobenzene to systematically control the photomechanical response ofazobenzene-containing polymers [122]. When a diacrylate derivative ofazobenzene was copolymerized with ethyl acrylate, the resulting elastomerwas shown to reversibly contract by 0.2% upon irradiation with light withwavelengths below 365 nm. This deformation could be recovered by exposureto longer (>365 nm) wavelength light. The achievable photomechanical strain

�

� �

�


was later increased to greater than 1% by Matejka et al., by increasing theconcentration of the photochromic monomer in butyl-acrylate-containingnetworks [123]. These effects are not limited to elastomers, however. Agoliniand Gay reported that semicrystalline polyimides based on diaminoazoben-zene exhibited a 0.5% contraction when exposed to UV light [124]. This workwas later expanded to include a wide range of polyimides that contain azoben-zene, including both amorphous and semicrystalline materials. Tan, White,and coworkers focused on the bending of polyimide beams in response tolinearly polarized blue-green (442 nm) light [66]. In response to this stimulus,azobenzene undergoes both trans–cis and cis–trans isomerization. The resultis a statistical buildup of oriented trans isomers normal to the electric fieldvector of the incident light. As a result, the unstructured, amorphous materialsexhibit directionally controlled bending based on the incident polarization oflight. Furthermore, by controlling the polyimide structure, relaxation behaviorcould be controlled to be immediately reversible, recovered immediately uponstimulus removal, or metastable, with deformation fixed indefinitely untilanother stimulus is applied [125]. Further studies elucidated the importanceof not only phase and composition but also processing conditions, such asaging and drawing on the photomechanical response [68, 126]. Drawinggenerates anisotropy in the material and can lead to larger unidirectionalbending. Wie et al. used drawn azobenzene-containing polyimide laminatedto polyvinylidene fluoride to convert periodic irradiation of blue light at up to5Hz to electrical energy [127].The rod-like nature of azobenzene chromophores is similar to that of many

calamitic liquid-crystalline monomers. As a result, liquid-crystalline materialshave been developed where the liquid-crystalline phase is used to amplify ordirect the photomechanical response. Finkelmann et al. and Terentjev et al.demonstrated that by copolymerizing azobenzene-containing monomers innematic liquid crystal elastomers, the nematic order of the sample could bereduced through irradiation with UV light [128, 129]. The reduction in orderamplifies the shape change of the chromophore. Terentjev et al. showed thatby maintaining a sample temperature just below the nematic-to-isotropictransition temperature and irradiating with UV light, strains greater than 100%could be triggered. By exposure to green light heat, azobenzene isomerizes tothe globally stable trans configuration, and the film recovers the initial shapeupon cooling to the original reference temperature. Ikeda and coworkerslater demonstrated a nematic elastomer consisting entirely of monofunc-tional and difunctional azobenzene-containing monomers [62]. The resultingpolydomain (unaligned) films exhibited directionally controlled bending inresponse to polarized UV light, with bending observed along the electric fieldvector of the linearly polarized light. Ikeda and coworkers later showed thatmonodomain (uniaxially aligned) liquid crystal elastomers exhibit preferential

�

� �

�


bending along the nematic director. It is the combination of these mechanisms,controlled anisotropic stimulus response, and amplification of photomechan-ical strain through cooperative motion of nonresponsive mesogens thathas generated significant attention to liquid-crystalline photomechanicaltwo-way shape memory materials. As liquid-crystalline materials are reviewedin detail elsewhere in this book, here we review the design strategy to usethese photomechanical materials as functional two-way shape memorymaterials.Azobenzene-containing photomechanical two-way shape memory materials

have been used to demonstrate both tensile and bending reversible shapechanges [118, 130]. While tensile two-way shape memory materials have beenproposed as a type of artificial muscle, demonstrations of bending-basedtwo-way shape memory materials have been more prevalent in the scientificliterature.This likely can be attributed to the efficient absorption of UV light inazobenzene-containing samples. As a result, many of the studied compositionsare optically thick, which leads to preferential stimulus response at the surfaceof the material and ultimately bending. Examples of these bending two-wayshape memory materials are shown in Figure 10.9. Broer and coworkersprinted microcantilevers of azobenzene-containing nematic liquid crystalpolymers that bend to a radius of 0.22mm in response to UV light and recovertheir initial shape upon irradiation with visible light [131]. These beams couldact as artificial cilia capable of pumping fluids in microfluidics. However, themeasured response times were slower than 1Hz and would likely need tobe increased to several hertz to induce meaningful fluid control. Ikeda andcoworkers demonstrated a plastic motor, with an azobenzene-containingliquid crystal elastomer laminated to a polyethylene film as a mechanicallyresponsive belt used to drive two pulleys [132]. The “fuel” for this motor istwo light sources, one UV and one visible, irradiating distinct parts of the beltgenerating a continuous rotation. It should be emphasized that this motor isboth battery and contact free. This work also demonstrated a liquid crystalelastomer film, glued into a cylinder shape, capable of translating itself, byrotation, upon sequential radiation with UV and visible light. Palffy-Muhorayand coworkers demonstrated that a single light source can be used to introducetranslation through irradiation with a visible laser [133]. A side-chain liquidcrystal elastomer was doped with an azobenzene-based dye and cut into adisk shape. Upon irradiation, the disk bends. If the irradiation is nonuniform,the anisotropic bending propels the floating sample along the surface of thewater away from the light source. It is, however, unclear with regard to therelative importance of photochemical and photothermal processes in thismaterial. By combining both photothermal and photochemical responsesin densely cross-linked liquid crystal networks along with mechanical res-onance of a cantilever, high-frequency oscillators were demonstrated by

�

� �

�


Ultraviolet onVis

Belt

UV

Axle

Axle

White spot

as a marker

0 s 16 s

8 s 24 s

Pulleys

Alignment

direction

Light off

(a)

(b)

(e)

(c)

(d)Visible on

t = 0 4 6 10 s

Ultraviolet + Visible on

Ultraviolet + Visible light Ultraviolet light Light off

Figure 10.9 Liquid crystal polymer networks are capable of photochemical two-way shapememory when polymerized with azobenzene-containing monomers. Printed cilia-liketwo-way shape memory materials (a) exhibit photoinduced two-way shape memory inresponse to sequential exposure to UV and visible light (b). (van Oosten et al. [131].Reproduced with permission from Nature Publishing Group.) A continuous circular film canbe used to power a light-activated motor in response to continuous UV and visible light(c and d). (Yamada et al. [132]. Reproduced with permission from John Wiley and Sons.)Azobenzene-doped liquid crystal elastomers were also demonstrated to swim across asurface in response to continuous irradiation, until the sample left the illuminated area.(Camacho-Lopez et al. [133]. Reproduced with the permission of Nature Publishing Group .)

White et al. These oscillators can be designed to exhibit both large deflectionand high oscillation frequencies (110∘ and 23Hz) [134]. As was previouslymentioned, photochemical two-way shape memory materials are not limitedto azobenzene-containing materials.Spiropyran undergoes a ring-opening reaction in response to UV light that

can be utilized to generate reversible shape change. Smets first observedphotomechanical behavior of poly(methacrylates)-containing pendantspriropyran chromophores [135]. Depending on the method of incorporatingthe chromophore in the material, either photoexpansion or photocontractioncan be observed. In both the cases, the strain is reversible either by leavingthe film in the dark or by irradiating with green light [119]. In addition

�

� �

�


to the length change associated with this ring-opening reaction, there is asignificant change in polarity, with the merocyanine species being significantlymore polar. Some results suggest that this change in polarity could lead toaggregation of chromophores doped within the polymer network, which maycomplicate the photomechanical response observed in these systems [136].This change in polarity can also be used to modulate the swelling of gelswith UV light. One common issue among these materials is the use of UVlight and blue light, particularly for biomedical applications. Living tissueeffectively scatters or absorbs these wavelengths of light, which complicatesdelivery of the stimulus and may lead to tissue damage. Lu and coworkerssynthesized multiblock copolymers containing rigid polyarylamide blocksthat contain benzocyclobutene and poly(ethylene oxide) blocks [137, 138].A thermal treatment cross-linked the polyarylamide blocks while removingthe poly(ethylene oxide) blocks, resulting in a nanofibrillar structure. Theresulting material responded photomechanically over the visible and near-IRspectrum. In response to near-IR irradiation, the film reversibly bends. Themechanism of this bending is unclear, but it was used to create a photogenera-tor by laminating the photoresponsive material to polyvinylidene fluoride thatresponds at up to 5Hz. The fact that this material responds to wavelengthswithin the water window, where the tissue is reasonably transparent, couldenable future biomedical applications.Although the focus of this chapter has been shape memory and two-way

shape memory in photoresponsive polymeric systems, photomechanicaleffects have been observed in nonpolymeric materials as well, for example,ceramics with noncentrosymmetric crystal structure ceramics, such aslanthanum-modified lead zirconate titanate [139]. This effect is an effectivesuperposition of the photovoltaic and piezoelectric effects in a single material.By mechanically designing a cantilever and irradiating with a chopped stimu-lus, beams can be driven at audible frequencies (75Hz) and the deformationamplified using mechanical resonance [140]. It should be noted that thesetwo-way shape memory materials have moduli more than an order of magni-tude higher, near 60GPa, than the stiffest photomechanically active polymertwo-way shape memory materials, near 2GPa, for azobenzene-containingpolyimides. However, realizable strains are exceedingly small by comparison,for example, a 20-mm-long cantilever exhibits maximal tip displacement of0.15mm. Organic crystalline materials are also capable of photomechanicaltwo-way shape memory and are reviewed in depth elsewhere in this book.A number of anthracene derivatives have been crystallized into microwires,plates, or ribbons [141]. These materials exhibit programmed reversible shapechanges based on photodimerization in the crystalline phase, leading toanisotropic shape change coupled with a phase transition from one crystallinephase to another or from crystalline to amorphous. In certain examples, fastresponse times (>1 s) are observed. Bending, twisting, and even jumping

�

� �

�


in response to UV light have been demonstrated. This interesting class ofmaterials combines the fast response times seen in ceramics with some ofthe compositional flexibility of polymeric materials. However, it should benoted that the inherently brittle nature of many molecular crystals limits thedimensions of the two-way shape memory material to the microscale.

10.3 Summary and Outlook

Photothermal and photochemical mechanisms can be used to drive shapechange in a wide variety of materials, from polymers to ceramics. Reversibleand irreversible shape changes have been described. These changes can betriggered using light from UV, visible, and near-IR spectra. These activematerials require control not only at the molecular level, photochemistryand polymer chain conformation, but also through to the continuum-levelmechanics. However, through control of these aspects, a variety of complex,programmed, and noncontact stimuli-responsive devices have been demon-strated, from microjumpers to light-powered motors. However, future designchallenges remain, from materials chemistry to mechanical design. Continuedefforts to design smart materials that respond to the near-IR transparencywindow in the human tissue may enable remote deployment of implantabledevices. Materials capable of responding to solar radiation may provide uniqueopportunities for solar tracking and responsive architectural structures.Ultimately, photomechanical shape-changing materials stand poised to, incertain situations, replace complex and heavy mechanisms without the needto require onboard power or power-receiving electronics.

References

1 Greenough, H. and Small, H.A. (1958) Form and Function: Remarks onArt, Design and Architecture, University of California Press, Berkeley.

2 Russell, B., Motlagh, D., and Ashley, W.W. (2000) Form follows function:how muscle shape is regulated by work. Journal of Applied Physiology, 88(3), 1127–1132.

3 Sedmera, D. (2005) Form follows function: developmental and physio-logical view on ventricular myocardial architecture. European Journal ofCardio-Thoracic Surgery, 28 (4), 526–528.

4 Roy, D., Cambre, J.N., and Sumerlin, B.S. (2010) Future perspectivesand recent advances in stimuli-responsive materials. Progress in PolymerScience, 35 (1), 278–301.

5 Koerner, H., White, T.J., Tabiryan, N.V. et al. (2008) Photogenerating workfrom polymers. Materials Today, 11 (7), 34–42.

�

� �

�

References 359

6 Liu, C., Qin, H., and Mather, P. (2007) Review of progress inshape-memory polymers. Journal of Materials Chemistry, 17 (16),1543–1558.

7 Walter, B. (1889) Ueber den Nachweis des Zerfalles von Moleculargruppenin Lösungen durch Fluorescenz-und Absorptionserscheinungen. Annalender Physik, 272 (2), 518–532.

8 Matsumura, T., Nakamura, T., Tetsuka, M. et al. (eds) (1999) Shape Mem-ory Ceramics, MRS Proceedings, Cambridge University Press, Cambridge,MA.

9 Otsuka, K. and Wayman, C.M. (1999) Shape Memory Materials, Cam-bridge University Press, Cambridge, MA.

10 Lendlein, A. and Kelch, S. (2002) Shape-memory polymers. AngewandteChemie, International Edition, 41 (12), 2034–2057.

11 Behl, M. and Lendlein, A. (2007) Shape-memory polymers. MaterialsToday, 10 (4), 20–28.

12 Mather, P.T., Luo, X., and Rousseau, I.A. (2009) Shape memory polymerresearch. Annual Review of Materials Research, 39, 445–471.

13 Felton, S.M., Tolley, M.T., Shin, B. et al. (2013) Self-folding with shapememory composites. Soft Matter, 9 (32), 7688–7694.

14 Maitland, D.J., Metzger, M.F., Schumann, D. et al. (2002) Photothermalproperties of shape memory polymer micro-two-way shape memorymaterials for treating stroke. Lasers in Surgery and Medicine, 30 (1), 1–11.

15 Maitland, D.J., Wilson, T., Metzger, M., and Schumann, D.L.(eds) (2002)Laser-Activated Shape Memory Polymer Microtwo-Way Shape MemoryMaterials for Treating Stroke. International Symposium on BiomedicalOptics, International Society for Optics and Photonics.

16 Small, W. IV,, Wilson, T., Benett, W. et al. (2005) Laser-activated shapememory polymer intravascular thrombectomy device. Optics Express, 13(20), 8204–8213.

17 Maitland, D.J., Small, W., Ortega, J.M. et al. (2007) Prototypelaser-activated shape memory polymer foam device for embolic treatmentof aneurysms. Journal of Biomedical Optics, 12 (3), 030504.

18 Koerner, H., Price, G., Pearce, N.A. et al. (2004) Remotely actuatedpolymer nanocomposites—stress-recovery of carbon-nanotube-filledthermoplastic elastomers. Nature Materials, 3 (2), 115–120.

19 Ni, Q.-Q., Zhang, C.-S., Fu, Y. et al. (2007) Shape memory effect andmechanical properties of carbon nanotube/shape memory polymernanocomposites. Composite Structures, 81 (2), 176–184.

20 Cho, J.W., Kim, J.W., Jung, Y.C., and Goo, N.S. (2005) Electroactiveshape-memory polyurethane composites incorporating carbon nanotubes.Macromolecular Rapid Communications, 26 (5), 412–416.

�

� �

�


21 Jin Yoo, H., Chae Jung, Y., Gopal Sahoo, N., and Whan, C.J. (2006)Polyurethane-carbon nanotube nanocomposites prepared by in-situ poly-merization with electroactive shape memory. Journal of MacromolecularScience, Part B, 45 (4), 441–451.

22 Xie, T. (2010) Tunable polymer multi-shape memory effect. Nature, 464(7286), 267–270.

23 Kohlmeyer, R.R., Lor, M., and Chen, J. (2012) Remote, local, and chemicalprogramming of healable multishape memory polymer nanocomposites.Nano Letters, 12 (6), 2757–2762.

24 Eustis, S. and El-Sayed, M.A. (2006) Why gold nanoparticles are moreprecious than pretty gold: noble metal surface plasmon resonance and itsenhancement of the radiative and nonradiative properties of nanocrystalsof different shapes. Chemical Society Reviews, 35 (3), 209–217.

25 Link, S. and El-Sayed, M.A. (1999) Spectral properties and relaxationdynamics of surface plasmon electronic oscillations in gold and silvernanodots and nanorods. The Journal of Physical Chemistry B, 103 (40),8410–8426.

26 Hribar, K.C., Metter, R.B., Ifkovits, J.L. et al. (2009) Light-induced tem-perature transitions in biodegradable polymer and nanorod composites.Small, 5 (16), 1830–1834.

27 Zhang, H., Xia, H., and Zhao, Y. (2012) Optically triggered and spatiallycontrollable shape-memory polymer-gold nanoparticle composite materi-als. Journal of Materials Chemistry, 22 (3), 845–849.

28 Zhang, H., Xia, H., and Zhao, Y. (2014) Light-controlled complex deforma-tion and motion of shape-memory polymers using a temperature gradient.ACS Macro Letters, 3 (9), 940–943.

29 Zhang, H., Zhang, J., Tong, X. et al. (2013) Light polarization-controlledshape-memory polymer/gold nanorod composite. Macromolecular RapidCommunications, 34 (19), 1575–1579.

30 Abbott, D.B., Maity, S., Burkey, M.T. et al. (2014) Blending withnon-responsive polymers to incorporate nanoparticles into shape-memorymaterials and enable photothermal heating: the effects of heterogeneoustemperature distribution. Macromolecular Chemistry and Physics, 215 (23),2345–2356.

31 Liu, Y., Genzer, J., and Dickey, M.D. (2015) “2D or not 2D”:shape-programming polymer sheets. Progress in Polymer Science, 52,79–106.

32 Liu, Y., Boyles, J.K., Genzer, J., and Dickey, M.D. (2012) Self-folding ofpolymer sheets using local light absorption. Soft Matter, 8 (6), 1764–1769.

33 Lee, Y., Lee, H., Hwang, T. et al. (2015) Sequential folding usinglight-activated polystyrene sheet. Scientific Reports, 5, 16544.

�

� �

�

References 361

34 Mailen, R.W., Liu, Y., Dickey, M.D. et al. (2015) Modelling of shape mem-ory polymer sheets that self-fold in response to localized heating. SoftMatter, 11 (39), 7827–7834.

35 Kumpfer, J.R. and Rowan, S.J. (2011) Thermo-, photo-, andchemo-responsive shape-memory properties from photo-cross-linkedmetallo-supramolecular polymers. Journal of the American ChemicalSociety, 133 (32), 12866–12874.

36 Biyani, M.V., Foster, E.J., and Weder, C. (2013) Light-healable supramolec-ular nanocomposites based on modified cellulose nanocrystals. ACS MacroLetters, 2 (3), 236–240.

37 Ware, T., Hearon, K., Lonnecker, A. et al. (2012) Triple-shape memorypolymers based on self-complementary hydrogen bonding. Macro-molecules, 45 (2), 1062–1069.

38 Li, J., Viveros, J.A., Wrue, M.H., and Anthamatten, M. (2007)Shape-memory effects in polymer networks containing reversibly asso-ciating side-groups. Advanced Materials, 19 (19), 2851–2855.

39 Habault, D., Zhang, H., and Zhao, Y. (2013) Light-triggered self-healingand shape-memory polymers. Chemical Society Reviews, 42 (17),7244–7256.

40 Lendlein, A., Jiang, H., Jünger, O., and Langer, R. (2005) Light-inducedshape-memory polymers. Nature, 434 (7035), 879–882.

41 Ryu, J., D’Amato, M., Cui, X. et al. (2012) Photo-origami—bending andfolding polymers with light. Applied Physics Letters, 100 (16), 161908.

42 Wu, L., Jin, C., and Sun, X. (2010) Synthesis, properties, and light-inducedshape memory effect of multiblock polyesterurethanes containingbiodegradable segments and pendant cinnamamide groups. Biomacro-molecules, 12 (1), 235–241.

43 Nagata, M. and Yamamoto, Y. (2009) Synthesis and characterization ofphotocrosslinked poly (𝜀-caprolactone) s showing shape-memory prop-erties. Journal of Polymer Science Part A: Polymer Chemistry, 47 (9),2422–2433.

44 Guimard, N.K., Oehlenschlaeger, K.K., Zhou, J. et al. (2012) Currenttrends in the field of self-healing materials. Macromolecular Chemistry andPhysics, 213 (2), 131–143.

45 Inoue, K., Yamashiro, M., and Iji, M. (2009) Recyclable shape-memorypolymer: poly (lactic acid) crosslinked by a thermoreversible Diels–Alderreaction. Journal of Applied Polymer Science, 112 (2), 876–885.

46 Raquez, J.M., Vanderstappen, S., Meyer, F. et al. (2011) Design ofcross-linked semicrystalline poly (𝜀-caprolactone)-based networks withone-way and two-way shape-memory properties through Diels–Alderreactions. Chemistry: A European Journal, 17 (36), 10135–10143.

�

� �

�


47 Michal, B.T., Jaye, C.A., Spencer, E.J., and Rowan, S.J. (2013) Inherentlyphotohealable and thermal shape-memory polydisulfide networks. ACSMacro Letters, 2 (8), 694–699.

48 Pei, Z., Yang, Y., Chen, Q. et al. (2014) Mouldable liquid-crystalline elas-tomer two-way shape memory materials with exchangeable covalentbonds. Nature Materials, 13 (1), 36–41.

49 Amamoto, Y., Otsuka, H., Takahara, A., and Matyjaszewski, K. (2012)Changes in network structure of chemical gels controlled by solvent qual-ity through photoinduced radical reshuffling reactions of trithiocarbonateunits. ACS Macro Letters, 1 (4), 478–481.

50 Scott, T.F., Schneider, A.D., Cook, W.D., and Bowman, C.N. (2005)Photoinduced plasticity in cross-linked polymers. Science, 308 (5728),1615–1617.

51 Fairbanks, B.D., Singh, S.P., Bowman, C.N., and Anseth, K.S. (2011) Pho-todegradable, photoadaptable hydrogels via radical-mediated disulfidefragmentation reaction. Macromolecules, 44 (8), 2444–2450.

52 Scott, T.F., Draughon, R.B., and Bowman, C.N. (2006) Actuation incrosslinked polymers via photoinduced stress relaxation. Advanced Materi-als, 18 (16), 2128–2132.

53 Mu, X., Sowan, N., Tumbic, J.A. et al. (2015) Photo-induced bendingin a light-activated polymer laminated composite. Soft Matter, 11 (13),2673–2682.

54 Schloter, S., Hofmann, U., Strohriegl, P. et al. (1998) High-performancepolysiloxane-based photorefractive polymers with nonlinear optical azo,stilbene, and tolane chromophores. Journal of the Optical Society of Amer-ica B, 15 (9), 2473–2475.

55 Sekkat, Z. and Dumont, M. (1991) Polarization effects in photoisomeriza-tion of azo dyes in polymeric films. Applied Physics B, 53 (2), 121–123.

56 Xie, S., Natansohn, A., and Rochon, P. (1993) Recent developments in aro-matic azo polymers research. Chemistry of Materials, 5 (4), 403–411.

57 Waldeck, D.H. (1991) Photoisomerization dynamics of stilbenes. ChemicalReviews, 91 (3), 415–436.

58 Nozaki, K. and Ogg, R. Jr., (1941) cis-trans Isomerizations. I. The mecha-nism of a catalyzed isomerization of maleic acid to fumaric acid*. Journalof the American Chemical Society, 63 (10), 2583–2586.

59 Merian, E. (1966) Steric factors influencing the dyeing of hydrophobicfibers. Textile Research Journal, 36 (7), 612–618.

60 Freundlich, H. and Heller, W. (1939) The adsorption of cis-andtrans-azobenzene. Journal of the American Chemical Society, 61 (8),2228–2230.

61 Lee, K.M., Koerner, H., Vaia, R.A. et al. (2011) Light-activated shapememory of glassy, azobenzene liquid crystalline polymer networks. SoftMatter, 7 (9), 4318–4324.

�

� �

�

References 363

62 Yu, Y., Nakano, M., and Ikeda, T. (2003) Photomechanics: directed bendingof a polymer film by light. Nature, 425 (6954), 145.

63 Fuhrmann, T.O., Kunze, M., Lieker, I., Stracke, A.,and Wendorff, J.H. (eds)(1996) Light-Induced Isomerization and Photo-Orientation in Azo and Stil-bene Polymers. SPIE’s 1996 International Symposium on Optical Science,Engineering, and Instrumentation, International Society for Optics andPhotonics.

64 Makushenko, A., Neporent, B., and Stolbova, O. (1971) Reversible ori-entational photodichroism and photoisomerization of aromatic azocompounds. I. Optics and Spectroscopy (USSR), 31, 557–564.

65 White, T.J., Serak, S.V., Tabiryan, N.V. et al. (2009) Polarization-controlled,photodriven bending in monodomain liquid crystal elastomer cantilevers.Journal of Materials Chemistry, 19 (8), 1080–1085.

66 Wang, D.H., Lee, K.M., Yu, Z. et al. (2011) Photomechanical responseof glassy azobenzene polyimide networks. Macromolecules, 44 (10),3840–3846.

67 Lee, K.M., Tabiryan, N.V., Bunning, T.J., and White, T.J. (2012) Pho-tomechanical mechanism and structure–property considerations in thegeneration of photomechanical work in glassy, azobenzene liquid crystalpolymer networks. Journal of Materials Chemistry, 22 (2), 691–698.

68 Lee, B.K.M., Koerner, H., Wang, D.H. et al. (2012) Tailoring the Photome-chanical Response of Glassy, azobenzene-functionalized polyimides byphysical aging. Macromolecules., 45 (18), 7527–7534.

69 Lee, K.M. and White, T.J. (2011) Photomechanical response of compos-ite structures built from azobenzene liquid crystal polymer networks.Polymers, 3 (3), 1447–1457.

70 Wie, J.J., Lee, K.M., and White, T.J. (2014) Thermally and optically fixableshape memory in azobenzene-functionalized glassy liquid crystalline poly-mer networks. Molecular Crystals and Liquid Crystals, 596 (1), 113–121.

71 Haines, C.S., Lima, M.D., Li, N. et al. (2014) Artificial muscles from fishingline and sewing thread. Science., 343 (6173), 868–872.

72 Sidorenko, A., Krupenkin, T., Taylor, A. et al. (2007) Reversible switchingof hydrogel-actuated nanostructures into complex micropatterns. Science,315 (5811), 487–490.

73 Masmanidis, S.C., Karabalin, R.B., De Vlaminck, I. et al. (2007) Multifunc-tional nanomechanical systems via tunably coupled piezoelectric actuation.Science, 317 (5839), 780–783.

74 Tan, X. and Baras, J.S. (2004) Modeling and control of hysteresis inmagnetostrictive two-way shape memory materials. Automatica, 40 (9),1469–1480.

75 Pelrine, R.E., Kornbluh, R.D., and Joseph, J.P. (1998) Electrostriction ofpolymer dielectrics with compliant electrodes as a means of actuation.Sensors and Actuators A: Physical, 64 (1), 77–85.

�

� �

�


76 Harris, K.D., Cuypers, R., Scheibe, P. et al. (2005) Large amplitudelight-induced motion in high elastic modulus polymer two-way shapememory materials. Journal of Materials Chemistry, 15 (47), 5043–5048.

77 Chung, T., Romo-Uribe, A., and Mather, P.T. (2008) Two-way reversibleshape memory in a semicrystalline network. Macromolecules, 41 (1),184–192.

78 Behl, M., Kratz, K., Noechel, U. et al. (2013) Temperature-memorypolymer two-way shape memory materials. Proceedings of the NationalAcademy of Sciences, 110 (31), 12555–12559.

79 Thomsen, D.L., Keller, P., Naciri, J. et al. (2001) Liquid crystal elas-tomers with mechanical properties of a muscle. Macromolecules, 34 (17),5868–5875.

80 Duerig, T.W., Melton, K., and Stöckel, D. (2013) Engineering Aspects ofShape Memory Alloys, Butterworth-Heinemann, Oxford.

81 Uchino, K. (1997) Piezoelectric Two-Way Shape Memory Materials andUltrasonic Motors, Springer Science & Business Media, New York.

82 Huang, W. (2002) On the selection of shape memory alloys for two-wayshape memory materials. Materials & Design, 23 (1), 11–19.

83 Jaffe, B. (2012) Piezoelectric Ceramics, Elsevier, Amsterdam.84 Küpfer, J. and Finkelmann, H. (1991) Nematic liquid single crystal elas-

tomers. Die Makromolekulare Chemie. Rapid Communications, 12 (12),717–726.

85 Klouda, L. and Mikos, A.G. (2008) Thermoresponsive hydrogels inbiomedical applications. European Journal of Pharmaceutics and Bio-pharmaceutics, 68 (1), 34–45.

86 Schild, H.G. (1992) Poly (N-isopropylacrylamide): experiment, theory andapplication. Progress in Polymer Science, 17 (2), 163–249.

87 Sprecht, E., Neuman, A., and Neher, H. (1956) US Patent 2,773,063, Rohm& Haas.

88 Scarpa, J.S., Mueller, D.D., and Klotz, I.M. (1967) Slowhydrogen-deuterium exchange in a non- alpha-helical polyamide. Journalof the American Chemical Society, 89 (24), 6024–6030.

89 Suzuki, A. and Tanaka, T. (1990) Phase transition in polymer gels inducedby visible light. Nature, 346 (6282), 345–347.

90 Nayak, S. and Lyon, L.A. (2004) Photoinduced phase transitions in poly(N-isopropylacrylamide) microgels. Chemistry of Materials, 16 (13),2623–2627.

91 Zhang, X., Pint, C.L., Lee, M.H. et al. (2011) Optically-andthermally-responsive programmable materials based on carbonnanotube-hydrogel polymer composites. Nano Letters, 11 (8), 3239–3244.

92 Fujigaya, T., Morimoto, T., Niidome, Y., and Nakashima, N. (2008) NIRlaser-driven reversible volume phase transition of single-walled carbon

�

� �

�

References 365

nanotube/poly (N-isopropylacrylamide) composite gels. Advanced Materi-als, 20 (19), 3610–3614.

93 Wang, E., Desai, M.S., and Lee, S.-W. (2013) Light-controlledgraphene-elastin composite hydrogel two-way shape memory materials.Nano Letters, 13 (6), 2826–2830.

94 Yamamoto, Y., Kanao, K., Arie, T. et al. (2015) Air ambient-operatedpNIPAM-based flexible two-way shape memory materials stimulatedby human body temperature and sunlight. ACS Applied Materials &Interfaces, 7, 11002–11006.

95 Lo, C.-W., Zhu, D., and Jiang, H. (2011) An infrared-light responsivegraphene-oxide incorporated poly (N-isopropylacrylamide) hydrogelnanocomposite. Soft Matter, 7 (12), 5604–5609.

96 Li, W., Wang, J., Ren, J., and Qu, X. (2013) 3D graphene oxide–polymerhydrogel: near-infrared light-triggered active scaffold for reversible cellcapture and on-demand release. Advanced Materials, 25 (46), 6737–6743.

97 Breuer, L., Raue, M., Kirschbaum, M. et al. (2015) Light-controllable poly-meric material based on temperature-sensitive hydrogels with incorporatedgraphene oxide. Physica Status Solidi (a), 212, 1368–1374.

98 Sershen, S.R., Mensing, G.A., Ng, M. et al. (2005) Independent opticalcontrol of microfluidic valves formed from optomechanically responsivenanocomposite hydrogels. Advanced Materials, 17 (11), 1366–1368.

99 Sershen, S., Westcott, S., Halas, N., and West, J. (2002) Independentoptically addressable nanoparticle-polymer optomechanical composites.Applied Physics Letters, 80 (24), 4609–4611.

100 Zhu, Z., Senses, E., Akcora, P., and Sukhishvili, S.A. (2012) Programmablelight-controlled shape changes in layered polymer nanocomposites. ACSNano, 6 (4), 3152–3162.

101 Yoon, J., Bian, P., Kim, J. et al. (2012) Local switching of chemical patternsthrough light-triggered unfolding of creased hydrogel surfaces. AngewandteChemie, 124 (29), 7258–7261.

102 Das, M., Sanson, N., Fava, D., and Kumacheva, E. (2007) Microgels loadedwith gold nanorods: photothermally triggered volume transitions underphysiological conditions. Langmuir, 23 (1), 196–201.

103 Jones, C.D., Serpe, M.J., Schroeder, L., and Lyon, L.A. (2003) Microlensformation in microgel/gold colloid composite materials via photother-mal patterning. Journal of the American Chemical Society, 125 (18),5292–5293.

104 Islam, M.R., Irvine, J., and Serpe, M.J. (2015) Photothermally inducedoptical property changes of poly (N-isopropylacrylamide) microgel-basedetalons. ACS Applied Materials & Interfaces, 7 (43), 24370–24376.

105 Hauser, A.W., Evans, A.A., Na, J.H., and Hayward, R.C. (2015) Photother-mally reprogrammable buckling of nanocomposite gel sheets. AngewandteChemie, International Edition, 54 (18), 5434–5437.

�

� �

�


106 Trujillo, V., Kim, J., and Hayward, R.C. (2008) Creasing instability ofsurface-attached hydrogels. Soft Matter, 4 (3), 564–569.

107 Jadhav, A.D., Yan, B., Luo, R.-C. et al. (2015) Photoresponsive microvalvefor remote actuation and flow control in microfluidic devices. Biomicroflu-idics, 9 (3), 034114.

108 Wermter, H. and Finkelmann, H. (2001) Liquid crystalline elastomers asartificial muscles. e-Polymers, 1, 111–123.

109 Ahir, S.V. and Terentjev, E.M. (2005) Photomechanical actuation inpolymer-nanotube composites. Nature Materials, 4 (6), 491–495.

110 Torras, N., Marshall, J.E., Zinoviev, K.E. et al. (2014) Gas-pressuremolding-based fabrication of smart two-way shape memory materialsfrom nematic liquid-crystalline elastomer. Macromolecular Materials andEngineering, 299 (10), 1163–1169.

111 Yang, L., Setyowati, K., Li, A. et al. (2008) Reversible infrared actuation ofcarbon nanotube–liquid crystalline elastomer nanocomposites. AdvancedMaterials, 20 (12), 2271–2275.

112 Li, C., Liu, Y., Huang, X., and Jiang, H. (2012) Direct sun-drivenartificial heliotropism for solar energy harvesting based on aphoto-thermomechanical liquid-crystal elastomer nanocomposite.Advanced Functional Materials, 22 (24), 5166–5174.

113 Kohlmeyer, R.R. and Chen, J. (2013) Wavelength-selective, ir light-drivenhinges based on liquid crystalline elastomer composites. AngewandteChemie, International Edition, 52 (35), 9234–9237.

114 de Haan, L.T., Sánchez-Somolinos, C., Bastiaansen, C.M.W. et al. (2012)Engineering of complex order and the macroscopic deformation of liquidcrystal polymer networks. Angewandte Chemie, International Edition, 51,12469–12472.

115 Broer, D.J. and Mol, G.N. (1991) Anisotropic thermal expansion of denselycross-linked oriented polymer networks. Polymer Engineering & Science,31, 625–631.

116 Liu, D. and Broer, D.J. (2014) Liquid crystal polymer networks: prepa-ration, properties, and applications of films with patterned molecularalignment. Langmuir, 30, 13499–13509.

117 LeMieux, M.C., McConney, M.E., Lin, Y.-H. et al. (2006) Polymericnanolayers as two-way shape memory materials for ultrasensitive thermalbimorphs. Nano Letters, 6 (4), 730–734.

118 Barrett, C.J., Mamiya, J.-I., Yager, K.G., and Ikeda, T. (2007)Photo-mechanical effects in azobenzene-containing soft materials. SoftMatter, 3 (10), 1249–1261.

119 Florea, L., Diamond, D., and Benito-Lopez, F. (2012) Photo-responsivepolymeric structures based on spiropyran. Macromolecular Materials andEngineering, 297 (12), 1148–1159.

�

� �

�

References 367

120 Wei, J. and Yu, Y. (2012) Photodeformable polymer gels and crosslinkedliquid-crystalline polymers. Soft Matter, 8 (31), 8050–8059.

121 Kumar, G.S. and Neckers, D. (1989) Photochemistry ofazobenzene-containing polymers. Chemical Reviews, 89 (8), 1915–1925.

122 Eisenbach, C.D. (1978) Effect of polymer matrix on the cis-trans isomer-ization of azobenzene residues in bulk polymers. Die MakromolekulareChemie, 179 (10), 2489–2506.

123 Matejka, L., Dušek, K., and Ilavský, M. (1979) The thermal effect in thephotomechanical conversion of a photochromic polymer. Polymer Bulletin,1 (9), 659–664.

124 Agolini, F. and Gay, F.P. (1970) Synthesis and properties of azoaromaticpolymers. Macromolecules, 3 (3), 349–351.

125 Lee, K.M., Wang, D.H., Koerner, H. et al. (2012) Enhancement of pho-togenerated mechanical force in azobenzene-functionalized polyimides.Angewandte Chemie, 124 (17), 4193–4197.

126 Wang, D.H., Wie, J.J., Lee, K.M. et al. (2014) Impact of backbone rigidityon the photomechanical response of glassy, azobenzene-functionalizedpolyimides. Macromolecules., 47 (2), 659–667.

127 Wie, J.J., Wang, D.H., Tondiglia, V.P. et al. (2014) Photopiezoelectriccomposites of azobenzene-functionalized polyimides and polyvinylidenefluoride. Macromolecular Rapid Communications, 35 (24), 2050–2056.

128 Finkelmann, H., Nishikawa, E., Pereira, G., and Warner, M. (2001) A newopto-mechanical effect in solids. Physical Review Letters, 87 (1), 015501.

129 Hogan, P., Tajbakhsh, A., and Terentjev, E. (2002) UV manipulation oforder and macroscopic shape in nematic elastomers. Physical Review E, 65(4), 041720.

130 Li, M.H., Keller, P., Li, B. et al. (2003) Light-driven side-on nematicelastomer two-way shape memory materials. Advanced Materials, 15,569–572.

131 van Oosten, C.L., Bastiaansen, C.W.M., and Broer, D.J. (2009) Printed arti-ficial cilia from liquid-crystal network two-way shape memory materialsmodularly driven by light. Nature Materials, 8 (8), 677–682.

132 Yamada, M., Kondo, M., Mamiya, J.I. et al. (2008) Photomobile poly-mer materials: towards light-driven plastic motors. Angewandte Chemie,International Edition, 47 (27), 4986–4988.

133 Camacho-Lopez, M., Finkelmann, H., Palffy-Muhoray, P., and Shelley, M.(2004) Fast liquid-crystal elastomer swims into the dark. Nature Materials,3 (5), 307–310.

134 White, T.J., Tabiryan, N.V., Serak, S.V. et al. (2008) A high frequencyphotodriven polymer oscillator. Soft Matter, 4 (9), 1796–1798.

135 Smets, G. (1983) Photochromic phenomena in the solid phase, in UnusualProperties of New Polymers, Springer, New York, pp. 17–44.

�

� �

�


136 Athanassiou, A., Kalyva, M., Lakiotaki, K. et al. (2005) All-opticalreversible actuation of photochromic-polymer microsystems. AdvancedMaterials, 17 (8), 988–992.

137 Ugur, G., Chang, J., Xiang, S. et al. (2012) A near-infrared mechanoresponsive polymer system. Advanced Materials, 24 (20), 2685–2690.

138 Shen, X., Viney, C., Johnson, E.R. et al. (2013) Large negative thermalexpansion of a polymer driven by a submolecular conformational change.Nature Chemistry, 5 (12), 1035–1041.

139 Poosanaas, P., Tonooka, K., and Uchino, K. (2000) Photostrictive two-wayshape memory materials. Mechatronics, 10 (4), 467–487.

140 Chu, S.-Y. and Uchino, K. (1995) Photostrictive effect in Plzt-based ceram-ics and its applications. Ferroelectrics, 174 (1), 185–196.

141 Kim, T., Zhu, L., Al-Kaysi, R.O., and Bardeen, C.J. (2014) Organic pho-tomechanical materials. ChemPhysChem, 15 (3), 400–414.

�

� �

�

369

11

Photomechanical Effects to Enable DevicesM. Ravi Shankar

Department of Industrial Engineering, University of Pittsburgh, Pittsburgh, PA, USA

11.1 Introduction

Conventionally, active mechanical systems manipulate within a predefinedworkspace using articulated joints, hinges, and rotor elements. This configu-ration underpins the vast majority of macroscale machinery and mechanisms.Typically, these systems are characterized by a multimaterial construction andoperate by energizing the actinic elements within a control loop. A parallelthrust on “Compliant Mechanisms” has emerged [1], which offers the abilityto realize manipulation in otherwise monolithic constructs. Manipulating thegradation of mechanical properties, cross sections, and stiffness, mechanismshave been realized in integral structures. The inherently scale-free natureof the mechanics underpinning the design space allows for these ideas tobe translated across length scales, ranging from microdevices to aerospacestructures [2].Incorporation of activematerials, whosemicrostructure design enables them

to function as a machine element [3], provides an opportunity for creatingfundamentally new classes of monolithic machines. In “hard materials” suchas shape memory alloys, design of microstructured domains has been pursuedas a route for achieving complex actuation modes in integral thin films [3].The ability to program the actuation by microstructure design offers theability to achieve ultrafine spatial modulation of actuation, even though theactinic/powering energy source is modulated at larger length scales. Hence,spatially articulated actuation can emerge without requiring the delivery ofthe actinic energy to the individual elements such as hinges or joints. Whileoffering mechanical robustness, utilization of materials such as shape memoryalloys is limited by complexities in programming the microstructures, as wellas their integration in micromachined systems. Furthermore, the materials are


�

� �

�

370 11 Photomechanical Effects to Enable Devices

not typically capable of large morphing strains, which places a bound on theirutility in compliant mechanisms that aim to manifest significant mechanicalmorphing.Translating these ideas in soft polymeric materials can offer a frame-

work for creating new classes of systems. Soft actuators fabricated fromstimuli-responsive polymers are useful for the fabrication of soft machines thatare compliant and generate robust, multimodal deformation.The ability to spa-tiotemporally tune the response when actuating and manipulating is central toimplementingmechanisms and actuators that conform to complex geometries,including those found in the natural world. An array of underlyingmechanismshas been considered for realizing this promise in polymers, including thermalcontraction in drawn semicrystalline polymers [4], thermal shape memoryin distinct classes of covalently and physically cross-linked polymers [5], andthermally/photoresponsive liquid-crystalline polymers [6–8].Liquid-crystalline polymers networks (LCN) offer exciting opportunities for

tuning mechanical responses via facile microstructure patterning and exploit-ing the coupling between the changes in the nematic order and themacroscopicstrains generated by the cross-linked polymer network. Prescribing a priornematic director and then triggering a transition in the magnitude of theorder parameter can, in principle, offer a route for prescribing the strain-fieldpoint-by-point [6]. Patterned LCN that are responsive to heat can be coaxedto undergo multiple actuation modes, including the generation of bendingstrains/curvatures when heated [9], creation of chiral structures (helicoids andspirals) [10], localization of strains in folds, and emergence of large out-of-planedeformation by exploiting topological defects [6]. The ideas used in thermo-mechanically responsive LCN can be extended without loss of generality inphotoresponsive liquid crystal polymers – such as azobenzene-functionalizedliquid crystalline polymers networks (ALCN). For example, topological defectscan be utilized in glassy, ALCN to generate out-of-plane deformation usinglight [11]. Manipulation of topographies at finer length scales has been accom-plished using patterned chiral nematic networks that offer opportunities fortuning friction on-demand using light [12, 13].Among the wide range of azobenzene-functionalized polymers that have

been studied, glassy ALCN are promising. In GPa-modulus ALCN, tensof kilojoules per cubic meter of photomechanical work and a reasonablephotomechanical stress (∼1MPa) can be generated. This has been realized bycross-linking acrylate-based reactive mesogens with photoactive mesogensthat contain the azobenzene in the rigid portion of the molecule [14]. Thestiffness of these materials is particularly useful for the design of actuators byvirtue of their ability to overcome larger blocking stresses, albeit during thegeneration of small magnitudes of photomechanical strains (few percent).The conversion of light into mechanical work in ALCN is ultimately deter-

mined by the kinetics of trans–cis isomerization and/or trans–cis–trans

�

� �

�

11.2 Analog Photomechanical Actuators 371

reorientation of the azobenzene. Irradiation with UV ( around 375 nm) isknown to induce isomerization of trans-to-cis [8, 15].The associatedmolecularcontraction is transduced into macroscopically observed contractile strainsat the irradiated surface. The gradient in the strains through the thickness,in cantilever-like samples, manifests the characteristic bending toward theactinic light. Irradiation with blue light (440–450 nm), however, involvesabsorption by both the cis and trans states and results in a buildup of theazochromophores perpendicular to the direction of polarization of the actiniclight [8]. Such reorientation also manifests photoactuation.Utilization of photoresponsive polymers for actuators requires the ability to

achieve to-and-fro actuation. Much of the current research has focused onirradiating cantilevers fabricated from these materials and characterizing thedeflection. However, a few efforts have examined multiplexed irradiation forinducing photostrains and their subsequent erasure. Ikeda’s work has shownthat trans–cis isomerization with UV in polydomain ALCN films can be subse-quently erased by exposure to visible light [16] by inducing cis–trans reversion.The reversion of the cis state can also occur in the dark at room temperature,after several hours. White’s group has shown that the trans–cis–trans reori-entation under irradiation with polarized 440–450 nm light, which otherwiseresults in persistent photostrains and a “photofixed state,” can be erased usingthe same light source but with circular polarization [17]. But for the circularpolarization treatment, “photo-fixed” strains from 440 to 450 nm irradiationremain stable, persist even after the actinic light has been turned off and areunrecoverable even after prolonged storage at room temperature [17]. How-ever, the strain reversion that underpins realization of to-and-fro actuation islimited by the same reaction kinetics that are ultimately traceable to the effi-ciency of isomerization of the azobenzene in the macromolecular network.

11.2 Analog Photomechanical Actuators

The effect of intensity of light on the photomechanical deformation in ALCNgenerally reveals a monotonic increase in the magnitude and rate of accu-mulation of the strains with increasing intensity. Figure 11.1 shows how thecurvature/bending angle increases with increasing intensity of light for a fixedduration (10min) of irradiation with 445 nm light. The angle between thedirection of the polarization and the nematic director offers another controlvariable for modulating the photomechanical response. The continuous vari-ation of the response as a function of the irradiation parameters offers a routefor “analog” control of the photoactuation. Modulating the intensity and polar-ization (simultaneously or individually) is a vehicle for tuning the magnitudeof the resulting bending curvatures and twist. The analog mode can offer finegranularity in the spatiotemporal control of the photomechanical response.

�

� �

�


200

100

0

–100

–200

TN

Be

nd

ing

an

gle

(°)

0 100 200

Intensity (mW/cm2)

300

0° to 90°

0° to 90°

90° to 0°

F B

90° to 0°

F B

Figure 11.1 Effect of intensity of 445 nm laser light on the bending of twisted nematicALCN when the polarization is set parallel to the long axis of the cantilever. The orientationof the direction of polarization with respect to the nematic director determines if thecantilever bends toward the light (0∘ to 90∘) or away from it (90∘ to 0∘) sample. F is thesurface that is irradiated, and B is the nonirradiated back. The hatched line illustrates theorientation of the nematic director. Vertical direction is the long axis of the cantilever. (Wieet al. [18]. Reproduced with the permission of Royal Society of Chemistry.)

Integration ofmaterials in photomechanical devices that utilize the attributesof the light to control mechanical manipulation has been pursued across lengthscales. At the submillimeter length scales, light-induced bending has beenmodulated in cilia-like structures [19]. Translating the ability to actuatewith light into more complex actuation modes, for example, in a roboticmanipulator, will require fine-tuning of the response to achieve the desiredtrajectories of motion. For example, in Ref. [20], a robotic arm was designedto pick, move, and place a pay load in an ALCN composite at the centimeterlength scale (Figure 11.2). Although implementable in a specific application,control of such systems in generic applications is challenging.The elegance of utilization of photonic energy (or any other radiant energy

source) for driving a mechanical system in analog mode is undercut by chal-lenges in the implementation of feedback control. A control loop requires theutilization of embedded strain sensors to correct for actuation errors, whichfeeds back to the actinic light source.This implementation will be complex andlimits the utilization of these materials in practical applications.

�

� �

�

11.3 Discrete-State (Digital) Photomechanical Actuators 373

Light off

Light on

Joints

(a)

(b)

Hand

Arm

Wrist

Fingers

20 mm

5 m

mLight on

Light on

22 s 6 s

Light off 5 s15 s0

Figure 11.2 A light-driven robotic manipulator. (Cheng et al. [20]. Reproduced with thepermission of Royal Society of Chemistry.)

11.3 Discrete-State (Digital) PhotomechanicalActuators

A strategy to circumvent the need for feedback control can be accomplished byutilizing photoresponsive materials in configurations that involve discretizedmechanical states that are also multistable; that is, the actuator remains in oneof the mechanical states until subjected to irradiation. When subjected to a

�

� �

�


sufficient quantum of irradiation, the actuator switches to another discretestate. This simplifies control – although the discretization of the states elimi-nates the ability to achieve a spatially continuous set of states.While this idea isstill in its infancy, the reduction of the challenge of control to the achievementof a certain threshold of irradiation, especially when seeking to exploit theutility in remote, contactless actuation applications can be attractive.

11.3.1 Binary Actuators

A prototypical implementation of a digital photomechanical actuator is thebuckled arch. In Ref. [21], a photoreponsive sample was created by excisingstrips 1mm in width from azobenzene-functionalized polyimide films (15 μmin thickness).The strips were clamped at their ends.This sets up a rigid bound-ary condition at the two ends that are initially Lo apart. The clamped samplewas mounted on a linear positioning system.When the positioning system wasused to reduce the distance between the glass slide-clamped ends to a distanceof L, the sample buckles to assume the shape of an arch as illustrated in theinset in Figure 11.3. Note that this arch is symmetrically bistable with respectto being concave or convex when observed from the top. Typically, Lo valuesin the range of 10–15mm were considered, and L∕Lo values in the range of0.97–0.99 were used.Figure 11.4 illustrates the progressive deformation of an arch that is illumi-

nated from the bottom at the midpoint of the arch with a 445 nm laser witha beam diameter of 5mm and intensity of 18mW/cm2 with polarization setalong the long axis of the arch (e1 in Figure 11.4a). For the first 8.4 s of irra-diation, the arch undergoes gradual deformation to increase its strain energyin excess of that in Figure 11.4(a). However, upon reaching a critical config-uration as shown in Figure 11.4(b), it spontaneously undergoes an ultrafastsnap-through involving velocities of > 100 mm∕s to reach the second bistableconfiguration (Figure 11.4c). A substantial kinetic energy release occurs duringthis transition, which inverts the arch geometry shown in Figure 11.4(c) rel-ative to that shown in (a). Figure 11.4(d) shows a high-speed image sequenceillustrating the snap-through recorded from the edge of the instability (i.e., forthe events occurring between Figure 11.4b and c). See supplementary movie,detailed modeling, and analysis in Ref. [21]. It is found that the approach to thelimit point (Figure 11.4b) following the irradiation is dependent on the intensityof the laser. The time required to approach the critical configuration as shownin Figure 11.4(b)was found to range from1 to 10 swhen the intensitywas variedfrom 10 to 100 mW∕cm2.Following this demonstration in the polyimide, the generic nature of the

underlying idea allowed it to be employed in ALCN by creating analogousarch-shaped geometries. A similar progression of deformation leading upto the ultrafast mechanics of the instability is observed. In addition, bothpolyimide and monodomain ALCN samples allow for bidirectional actuation,by switching the illumination to trigger snap-through from either direction.

�

� �

�


Mirror

Flip mirror

LinearPolarizer

1/2 Waveplate

Beam elevator

Mirror

Mirror

e2

e11 mm

Lo

2h

L

θR

Mirror

Mirror

Illumination from top

Illumination from bottom

Sample

445 m

m

Figure 11.3 Apparatus for irradiating arches created from photoactive materials from thetop and bottom using a blue-green laser with polarization set parallel to the long axis of thearch. (Shankar et al. [21]. Reproduced with permission from Figure 2 of PNAS, 2013, 47,18792–18797.)

Figure 11.5 shows a monodomain LCN sample that was first illuminatedfrom the bottom to trigger a snap-through upward. Then the sample wasilluminated from the top, to trigger a downward snap-through. Recall thatthe arch remains in either of the bistable configurations, unless irradiated. Inaddition, illuminating the arch as shown in Figure 11.5(e) from the top, after ithas snapped through, can restore the system to that shown in Figure 11.5(a),thus achieving bidirectional switching. In enabling ultrafast actuation betweenthese binary states, a range of technological opportunities become possible.

11.3.2 Latency of Binary Actuators and Repetitive Actuation

While the snap-through event itself was ultrafast as illustrated in Figure 11.4(d),the approach to instability was still quasi-static and occurred over severalseconds at intensities <100mW/cm2. As seen in Figure 11.6, the timefor approaching the instability (tcr) scaled inversely with the irradiation

�

� �

�


Original geometry

Geometry at the limitpoint after irradiationG

rad

ua

l d

efo

rma

tio

n to

ap

pro

ach

in

sta

bili

ty~

1-10

s d

ep

en

din

g o

n in

ten

sity

Ultra

fast

sn

ap

-th

rou

gh

~10

ms

1.7

mm

Geometry after snap-through

(d)tcr

tcr

+3.4 ms

tcr

+6.8 ms

tcr

+10.2 ms

(a)

(b)

(c)

hv

hv

hv

e2

e2

Figure 11.4 Photoinduced snap-through in azobenzene-functionalized polymers.(a) Geometry before irradiation. (b) The edge of instability. (c) Inversion of the geometryfollowing ultrafast snap-through. The ruler in the figure is in millimeter. (d) High-speedimage sequence showing the progress of snap-through. The scale bar is 3 mm. (Shankaret al. [21]. Reproduced with permission from Figure 3 of PNAS, 2013, 47, 18792–18797.)

�

� �

�


Illuminationfrom bottom

Illuminationfrom top

Small gradual deformation

(a) (b)

(d) (c)

(e) (f)

(h) (g)

Small gradual deformation

Snap-through–upward

Snap-through–downward

ton

445 nm 445 nm

445 nm

445 nm

445 nm

445 nm

445 nm445 nm

tcr = ton + 0.7 s

tontcr = ton + 4.1 s

Figure 11.5 Bidirectional snap-through in monodomain ALCN sample. (a) Originalgeometry that is illuminated from the bottom; (b) geometry at the limit point; (c) onset ofsnap-through upward; and (d) the final geometry. (e) Illumination of the sample from thetop; (f ) the geometry at the limit point; (g) the onset of downward snap-through; and (h) thefinal geometry after the to-and-fro snap-through. (Shankar et al. [21]. Reproduced withpermission from Figure 5 of PNAS, 2013, 47, 18792–18797.)

intensity (I). The higher the intensity of light, the faster the actuatorapproaches the edge of instability and undergoes snap-through. Instead of thestraight strips studied earlier, specimens akin to that illustrated in Figure 11.6can be used to reduce the deformation and associated latency for triggering the

�

� �

�


Straight strips

e2

e1

W αW

W

Lo/2

Lo/2Lo

LCPN (L/Lo = 0.99)

Slope = –1

10

1

10 100

I (mW/cm2)

t cr –

t on (

s)

1000

Polyimide (L/Lo = 0.99)

Fa

bric

atio

n

?

Arc

h g

eo

me

tryT-shaped strips

Figure 11.6 Modulating the onset of snap-through via mechanical design.

snap-through. The idea is that when the T-shaped sample is held rigidly at itsends and buckled, it will lead to an asymmetric arch geometry, unlike that seenin the previouswork.This is because the sample has a larger bending stiffness inthe wider section (𝛼W) compared to the thinner section. When this geometryis irradiated near the center of the arch, the asymmetric geometry can undergosnap-through with a reduced latency. For proof of concept, a T-shaped samplewith 𝛼 = 2 and L/Lo ∼ 0.99 was studied. It is found that tcr for the onset ofinstability in the T-shaped sample is about a third of that for the symmetricarch. In fact, if we cut out a T-sample with construction paper, buckle it into anarch, and push down on the center, it can be seen that it ismuch easier to inducesnap-through in the T than in the symmetric arch from the straight strip.It is noteworthy that the approach to the instability occurs via gradual defor-

mation during the accumulation of the photostrains.Then, when the geometryreaches a critical state (the limit point), ultrafast snap-through ensues.

�

� �

�


Note that the approach to instability with the 445 nm irradiation occurs viaaccumulation of the photofixed strain that is incremental and essentiallyquasi-static. This strain is not recoverable, unless irradiated with light ofcircular polarization [17]. Figure 11.6 shows that the time to approach theinstability for a given geometry is also a function of the intensity. But, it doesnot appear that the photostrains required for approaching the instabilityneed to be accumulated in one continuous irradiation event. Since, the strainaccumulation is quasi-static via photo fixing, the approach to instability canoccur in discrete steps or in an articulated manner, that is, it is possibleto utilize the original geometry of the arch and subject it to some 445 nmirradiation a priori such that it is primed to a geometry that is closer to thelimit point. This idea is illustrated in Figure 11.7, where the original arch (a) ispositioned closer to the instability as shown in Figure 11.7(b). The geometryshown in Figure 11.7(b) requires less strain to reach the limit point and triggerthe instability when irradiated again. The shape shown in Figure 11.7(b) isexpected to retain its geometry even when the irradiation is turned off, due

Original geometry

(a) (b)

(d) (c)

445 nm 445 nm

Intermediate primed state photofixed even after 445 nm is

445 nm445 nm

ee2

Treatment with 445 nmlight to prime the

sample closer to thelimit point

Limit point reached faster from theprimed state

Irra

dia

te w

ith 4

45 n

m w

hen

snap-t

hro

ugh n

eeded in

short

ord

er

Geometry after snap-through

Spontaneous ultrafastsnap-through from the

limit point

Figure 11.7 Priming the sample for photomechanical snap-through using photo fixing with445 nm irradiation where (a) the original geometry is irradiated for a short time to create(b) an intermediate primed state, which is closer to limit point compared to the prior arch.Then, when needed, the primed state will be irradiated, and it will reach (c) the limit pointquickly, and (d) snap-through can ensue.

�

� �

�


to the permanence of the photo-fixing effect (which is only recoverable whentreated with circular polarization, cf. [17]). When ultrafast actuation is desired,the geometry shown in Figure 11.7(b) can be irradiated, and it will reachthe limit point (Figure 11.7c) much more quickly because it requires smalleramount of strain to reach the instability compared to the original arch inFigure 11.7(a). Therefore, snap-through can be triggered in short order as andwhen needed (Figure 11.7c–d).An important challenge to be overcome is the inevitable loss of photore-

sponsiveness of the samples after multiple actuation (snap-through) cycles.Snap-through in bistable actuators is triggered due to the accumulation ofthe photoinduced strains, which in the case of the 445 nm irradiation occursvia the trans–cis–trans reorientation of the mesogens perpendicular to thepolarization of the light. The contractile strains generated as a result deformthe sample to the edge of instability, and then, the ultrafast actuation occurs.After the reorientation during irradiation, the mesogens predominantlyremain in the new state, leading to persistent photostrains. The progressivedepletion of mesogens that can be reoriented after multiple irradiations wouldeventually lead to the sample becoming photomechanically unresponsive,eventually. However, it has been shown that irradiation with light of circularpolarization eliminates photo-fixed strains and restores the photoactuationbehavior [17]. This observation allows for integration with aforementionedideas aimed at modulating the latency of the actuators to engineer bistableactuators that can be toggled multiple times. This is achievable by eliminatingexhaustion during repetitive to-and-fro snap-through via erasure of priorphoto fixing using circular polarization. To use this first trigger snap-throughakin to that shown in Figure 11.5 from one side using linear polarized light,but after the snap-through is complete, the irradiated zone will be treated withlight of circular polarization of suitable intensity. Through such multiplexedirradiation, the “memory” of prior actuation can be essentially eliminated,which can allow the actuator to respond without significant hysteresis in thesubsequent actuation cycles.

11.3.3 Multistable Implementations

The degrees of freedom of a photomechanical system composed of bistableactuators can be multiplied using assemblies of the individual actuators. Thesimplest implementation of bistable system is in surfaces with topographiesthat can be switched remotely to modulate optical, thermal, or aerodynamicproperties. For example, the design of micromirror arrays that can themselvesbe triggered with light is illustrated in Figure 11.8 [22]. Here, it is possible toimagine bistable structures that can also act as mirrors, whose state can beswitched using illumination with a laser beam.When it is in one of the bistablestates (state 1), themicromirror illuminates a pixel on a projection screen, while

�

� �

�


Incidentlight

(a) (b)

(c)

Incidentbeams

1

1

1

1

1

1

0

1

1

0

0

01

1

0

1

Silv

er

coating

Triggeringlight on

Triggeringlight on

Pro

ject

edbe

am

Pro

ject

edbe

ams

Photo

actu

ata

ble

mirro

r arr

ay

Ultrafast

Ultrafast

Ultrafastpositioning

hv

State “1” State “0”

Figure 11.8 (a) A silvered arch that can be triggered with 445 nm light to undergo ultrafastsnap-through and switch between reflective (1) and nonreflective states (0). Triggering lightmust also have a polarization that is set parallel to the long axis of beam for this to occur.(b) An array of such phototriggered micromirrors can be used to project beams from thoseincident on them. (c) A potential unit cell of a morphing structure that is composed of a“tripod” made using photomechanical materials that can undergo snap-through. Theannular table atop the tripod can adopt various configurations depending onwhether/which of the arch-shaped legs are snapped through. The reorientation of the tablewhen one of the arches undergoes snap-through is also shown.

in the other, it cannot (state 0). Each of these pixel states can be modified onthemicromirror array in an ultrafastmanner using a triggering light beam.Thissystem offers a contactless morphable topography composed of discrete com-binations of the individual actuators.The ultrafast switching elements can form unit cells with discrete states that

constitute light-driven binary robotic systems [23]. Figure 11.8(c) illustrates a

�

� �

�


“tripod” that is composed of legs fabricated from the arch-shaped elements thatcan undergo photomechanical snap-through. The normal vector to the annu-lar table atop this tripod can adopt a discrete set of directions (or points inorientation space) depending on which of the bistable state each of the archesadopts. If all the legs are snapped-through, the table can move vertically by anamount equal to the stroke length of the actuators. Otherwise, the normal vec-tor of the table can adopt 7 different directions (1 prior state (or when all thelegs are actuated) + 3 directions when one of the legs is actuated + 3 when twoof the three are actuated). By arranging a series of these tripod-table structuresone atop another, the number of achievable statesmultiplies.This would form arather unique robotic arm – a photoactuatable morphing structure composedof individual binary actuators that can undergo ultrafast actuation as envisagedin Figure 11.8(c). This can present opportunities for systems such as reconfig-urable antennas, which can be toggled between the discrete states using light.In addition to such configurations, it is possible to synthesize multistable

elements from bistable structures. Literature on compliant mechanisms isreplete with examples [1] where, say, tristable mechanisms can be createdfrom bistable actuators. Figure 11.9(a) illustrates the “Double Young TristableMechanism” [24], which can be adapted into a photomechanical mechanismwhen the flexible sections marked by “*” in the figure are replaced witha photoresponsive polymer arch. The actuation of this mechanism in a

End-effector

(b)

(a)

Second stableposition Third stable

position

As fabricated position

Figure 11.9 Engineering tristable mechanisms using bistable elements. (Chen and Du [24].Reproduced with the permission of ASME.) Reproduced with permission from Figure 11 ofJournal of Mechanisms and Robotics, 2013, 5, 011007 and (Chen et al. [25]. Figure 2 ofJournal of Mechanisms and Robotics, 2010, 2, 014501.)

�

� �

�


(a) at e1 (b) at e2

(c) at e3(d) at e4

Figure 11.10 Stable positions of a rotary device created with individual arch-shapedelements. Dark gray lines illustrate rigid elements, while light gray lines illustrate flexibleelements. Replacing the flexible elements with a photoresponsive polymer will allow for thedevelopment of a light-driven stepper motor. (Oh and Kota [26]. Reproduced with thepermission of ASME.)

coordinated manner can lead to the achievement of tristability. Assembliesfabricated from the tristable actuators can offer more degrees of freedomthan those from the binary actuators. An alternative approach is illustrated inFigure 11.9(b), which illustrates two arch-shaped actuators that are connectedorthogonally [25]. Again, if the arches marked by “*” in the figure are replacedwith the photoresponsive materials and snap-through was utilized, it will offeran alternative route for fabricating a tristable element. It is also possible tointegrate bistable actuators to drive a multistable rotary system (Figure 11.10)[26]. Here, replacing the green elements with photoresponsive polymers,which can toggle between bistable states when exposed to light, can allow forcreating an indexable rotary mechanism – a light-driven stepper motor.The ideas presented here can be applied without loss of generality to engi-

neer light actuated mechanisms, including morphing robotic structures thatsnap in a digital manner from one configuration to another, surfaces that havequantized shapes, and stepper motors that click between discrete rotationalpositions – all of which are triggered and controlled with light.These ideas can

�

� �

�


enable photomechanical robotic systems and mechanical logic that offer pre-cise manipulation while offering a range of configurations and manipulationopportunities. Realization of these ideas will face challenges when the opticalpath to the individual actuators from the light source is occluded. Integration offiber optic elements with the bistable actuators will be required for engineeringsuch photomechanical machines.

11.3.4 Beyond Bistable, Buckled Rods

An intriguing form of multistability occurs in helical lattice structures akin tothat illustrated in Figure 11.11 [27].This mechanismmimics the bacteriophageT4 virus’s tail sheath that can deform to produce a large displacement dur-ing the attachment to a host cell. By utilizing composite layup methods, stripsare fabricated and are bolted to one another to allow them to remain attachedto one another. The material essentially uses the bolts as hinges about whichrotation can occur. Figure 11.11(e) illustrates the center line of the interactinghelices. It has been shown that this device is multistable, illustrating a stableextended mode and two possible collapsed geometries, which are illustratedin Figure 11.11(a)–(d). It is possible to exploit the ability of twisted nematicstrips fabricated from ALCNmaterials to create spirals whose twist and radiusare controllable via the intensity of irradiation [18]. Consider twisted nematicstrips, which are characterized by the nematic directormaking an angle of+45∘with the long axis that has been irradiated with UV to create a helix. Anotherstrip that is characterized by the nematic director making an angle of −45∘creates a helix of the opposite chirality. The helices can be integrated to cre-ate the helical lattice akin to that in Ref. [27]. When the strains are allowed to

(a) (b) (e)

x

z

y

ξ1

ξ2

R

(c) (d)

Figure 11.11 Multistability of lattices constructed using intersecting helices that mimics thetail sheath of the bacteriophage T4. (Pirrera et al. [27]. Reproduced with the permission ofElsevier.)

�

� �

�


relax, the strips will seek to become straight and, thus, generate a shape akin tothat shown in Figure 11.11(a). However, when this structure is irradiated, it willtransform into an intermediate shape akin to that shown in Figure 11.11(b) andthen snap into the stable state as shown in Figure 11.11(c)/(d). Thus, a bistablehelical mechanism that can transform into two geometries and generate a largestroke length during the actuation can result.The ability to hardcode customizable actuation characteristics and

geometric transformations by patterning complex nematic directors ispossible in the ALCN samples. Combining this with the opportunities for“Overcurvature-Driven Origami” that was set forth in Ref. [28] can enablethe creation of three-dimensional volume enclosing frames from prior flatgeometries. It is possible to enable the ultrafast self-assembly of structuressuch as those illustrated in Figure 11.12(a) and (b), starting from a prior flatring. This transformation underpins a range of consumer products such asself-deploying tents, goal posts, and laundry baskets. If these ideas can be

(a)

10

8

6

4

2

00 0.01

50 μm

Δ

Top Bottom

R

Photostrain

R (

mm

)

0.02

Op = 1.33

3

A0

A0

A1

A1

A2

A2

A3

A3

A4

A4

2

1

0

0 1

2

A1O

p = 2.33

Op = 2.63

Op = 3

(b)

(c)

(e)

Bend and

move ends

together

Move ends

together

(d)

ζ

α

Figure 11.12 (a) Transformation of a flat ring into a range of geometries as a function of theovercurvature (Op). (b) Utilization of the underlying mechanics of overcurvature inconsumer products. (c) Nematic director field that can emulate these mechanics inliquid-crystalline polymers. Reproduced with permission from Figures 1 and 2 of [28].(d) Progressive supercoiling of a ring that is subjected to twist. (e) Creating hinges bybending a metallic measuring tape that is characterized by a preset curvature along theshort axis. Bending such a tape will create localized folds, which can be manipulated bymodifying the fixturing at the ends. (Mouthuy et al. [28]. Reproduced with the permission ofMacmillan Publishers Limited.)

�

� �

�


implemented in suitable mechanical designs fabricated from ALCN, it maybe possible to engineer novel self-deploying systems that start in a smallform factor until irradiated. Then, they deploy into a range of elementsautonomously. Consider a material with a preset curvature 𝜅. If fabricated ina cantilever-like beam and released, it would spontaneously curve up. Instead,if a ring is created with an intrinsic strain and then released, it would adopt arange of geometries shown in Figure 11.12(a). The mathematical details of thetransition from a flat shape before release to a range of these geometries can bewritten in terms of the overcurvature Op =

√1 + (𝜅R)2, where the transition

to various values of Op results in the various geometries. The mathematicalunderpinnings of these transitions can be found in Ref. [28].It is possible to use photopatterning techniques akin to that in Ref. [11] to

create a film with a nematic structure that is illustrated in Figure 11.12(c). Ifa straight strip of a twisted-nematic ALCN sample is made with the nematicdirector on the top parallel to the long axis and the bottom parallel to theshort axis and irradiated with UV, it would curl upward [7]. Now, if thestrains are confined to a ring with a radius R, the principles of Overcurvaturein Ref. [28] come into play, and depending on the geometry, the flat ringwould transform into a range of geometries – thus enabling “PhotoinducedOvercurvature-Driven Origami” [28].Ideas of overcurvature assume the lack of twist in the samples, instead of rely-

ing on the mismatch between the curvature prebiased onto the sample and theintrinsic curvature the material wishes to adopt. However, an array of super-coiled geometries becomes possible if a finite amount of twist was introducedin these closed geometries. Research on the mechanics of supercoiling in DNAplasmids [29, 30] shows that a transformation of the geometries from a ringto an “8”-shaped structure and subsequently a hierarchically twisted geometrycan emerge as illustrated in Figure 11.12(d). This is a system where the trans-formation of a flat ring into the “8”-shape (A0 − A1) transformation involvesan instability, where an abrupt change is observed in the Δ − plot, whenthe twist generated by the material exceeds

√3 A

rC. A and C are the flexural

and torsional rigidities of the cross section, and r is the radius of the ring [31].Past this instability-induced transformation, subsequent transformation occurquasi-statically as thematerial undergoes progressive supercoiling. New designopportunities can also emerge by coopting the phenomenon of “subdomain-ing” observed in DNA, which can emerge from immobilizing sections of theplasmid, for example, via interaction with the cell wall [32]. Substructures canemerge, embedded within the supercoiled shape, to offer new classes of geo-metric transformations beyond that in the ring with an intrinsic twist akin tothat shown in Figure 11.12(d).

�

� �

�

11.4 Photomechanical Mechanisms and Machines 387

11.4 Photomechanical Mechanisms and Machines

The desire for assembling 3D shapes from 2D forms using principles ofOrigami has motivated studies aimed at inducing programmable folding inmaterials. Illustrative examples include utilizing multimaterial constructionand Joule heating to induce bending [33, 34], exploiting shape memory effect[35], inducing localized stress relaxation via light-induced cleavage of polymerchains [36], inducing localized shrinkage in “Shrinky Dinks” via localizedthermal absorption [37], and using solvent/heat in patterned liquid-crystallineelastomers [6], which are patterned to localize the bending strains. The abilityto create folds that are controllable using light in azobenzene-functionalizedliquid crystal elastomers can offer a powerful tool for creating mechanismsthat emerge in otherwise monolithic materials. The ability to transform 2Dobjects into functional 3D forms that can manipulate in an articulated mannerat the folds can become a critical element of a photomechanical mechanism.An alternative route involving mechanical design to active hinges using tape

springs has been considered extensively in the aerospace structure literature tocreate folded trusses, deployable/morphing elements, and so on [38]. By modi-fying the boundary conditions, it is possible to modulate the localization of thebending in tape springs to create a range of geometries. This is a simple routefor creatingmechanisms that canmanipulate under the influence of an impulse,without requiring individual control and directed actuation of the hinges. Theidea of creating hinges with a tape spring can be visualized via the schematicillustrated in Figure 11.12(e). The measuring tape is characterized by a curva-ture along the short axis. Bending this tape will attempt to induce a curvaturealong the long axis. However, this will entail the creation of a finite Gaussiancurvature with a significant energy penalty. The system can reduce this energyby accommodating the bending via localized folds with the Gaussian curvaturerestricted to the corners at the edges of the fold. It can be seen from commonexperience that it is possible tomanipulate the hinges by simplymoving the endconditions – thus offering a simple route for creating a jointed mechanism ina monolithic object. This approach can offer interesting design opportunitiesin ALCN that have been preformed to endow a transverse curvature along theshort axis of a sample in the shape of a strip. Imposing photostrains that gener-ate a curvature along the long axis of this material will lead to analogous effectsand lead to the creation of a folded structure. Controlling the level of the photo-strain and the resulting curvature can result in a photomechanical mechanismwhere the hinges and their location as well as the resulting manipulation canbe controlled directly using light. Extending the idea of a tape spring to a plateleads to the utilization of “corrugated shells” and plates that offer opportunities

�

� �

�


for morphing structures [39]. These include the ability to achieve multistabil-ity interspersed by ultrafast transition and also the ability to achieve a range ofgeometries depending on the imposed boundary conditions [40].Integrating a range of the actuator designs outlined in this chapter can

offer an array of functional designs that can self-assemble and manipulatein a programmable manner. The programming is integrated in the material’smicrostructure and by mechanical design. The collective actuation in anintegrated photomechanical mechanism characterized by a range of theseactuators operating in conjunction with one another may require a range ofdesign optimizations to achieve the geometric transformations and desiredmanipulation. Utilizing opportunities enabled by high-frequency oscillationin ALCN [41] and the generation of walking gaits [42] and operation ofpulley-like mechanisms [43] in conjunction with morphing and manipulationcan enable fundamentally new photomechanical machinery.

References

1 Howell, L.L. (2001) Compliant Mechanisms, John Wiley & Sons, Inc., NewYork.

2 Kota, S., Hetrick, J.A., Osborn, R., Paul, D., Pendleton, E., Flick, P., andTilmann, C. (2003) Design and Application of Compliant Mechanisms forMorphing Aircraft Structures. Smart Structures and Materials, Interna-tional Society for Optics and Photonics, pp. 24–33.

3 Bhattacharya, K. and James, R.D. (2005) The material is the machine. Sci-ence, 307 (5706), 53–54.

4 Haines, C.S., Lima, M.D., Li, N., Spinks, G.M., Foroughi, J., Madden, J.D.,Kim, S.H., Fang, S., de Andrade, M.J., Göktepe, F. et al. (2014) Artificialmuscles from fishing line and sewing thread. Science, 343 (6173), 868–872.

5 Liu, C., Qin, H., and Mather, P. (2007) Review of progress in shape-memorypolymers. Journal of Materials Chemistry, 17 (16), 1543–1558.

6 Ware, T.H., McConney, M.E., Wie, J.J., Tondiglia, V.P., and White, T.J.(2015) Voxelated liquid crystal elastomers. Science, 347 (6225), 982–984.

7 Lee, K.M., Bunning, T.J., and White, T.J. (2012) Autonomous, hands-freeshape memory in glassy, liquid crystalline polymer networks. AdvancedMaterials, 24 (21), 2839–2843.

8 Lee, K.M., Tabiryan, N.V., Bunning, T.J., and White, T.J. (2012) Pho-tomechanical mechanism and structure-property considerations in thegeneration of photomechanical work in glassy, azobenzene liquid crystalpolymer networks. Journal of Materials Chemistry, 22 (2), 691–698.

9 Mol, G.N., Harris, K.D., Bastiaansen, C.W., and Broer, D.J. (2005)Thermo-mechanical responses of liquid-crystal networks with a splayedmolecular organization. Advanced Functional Materials, 15 (7), 1155–1159.

�

� �

�

References 389

10 Sawa, Y., Ye, F., Urayama, K., Takigawa, T., Gimenez-Pinto, V., Selinger, R.L.,and Selinger, J.V. (2011) Shape selection of twist-nematic-elastomer ribbons.Proceedings of the National academy of Sciences of the United States ofAmerica, 108 (16), 6364–6368.

11 McConney, M.E., Martinez, A., Tondiglia, V.P., Lee, K.M., Langley, D.,Smalyukh, I.I., and White, T.J. (2013) Topography from topology: pho-toinduced surface features generated in liquid crystal polymer networks.Advanced Materials, 25 (41), 5880–5885.

12 Liu, D. and Broer, D.J. (2014) Self-assembled dynamic 3d fingerprints inliquid-crystal coatings towards controllable friction and adhesion. Ange-wandte Chemie International Edition, 53 (18), 4542–4546.

13 Liu, D., Bastiaansen, C.W., den Toonder, J.M., and Broer, D.J. (2012)Photo-switchable surface topologies in chiral nematic coatings. AngewandteChemie International Edition, 51 (4), 892–896.

14 van Oosten, C.L., Harris, K.D., Bastiaansen, C.W.M., and Broer, D.J. (2007)Glassy photomechanical liquid-crystal network actuators for microscaledevices. European Physical Journal E, 23, 329–336.

15 Elias, A.L., Harris, K.D., Bastiaansen, C.W.M., Broer, D.J., and Brett, M.J.(2006) Photopatterned liquid crystalline polymers for microactuators.Journal of Materials Chemistry, 16, 2903–2912.

16 Yu, Y., Nakano, M., and Ikeda, T. (2003) Directed bending of a polymer filmby light. Nature, 425, 145.

17 Lee, K.M., Koerner, H., Vaia, R.A., Bunning, T.J., and White, T.J. (2011)Light-activated shape memory of glassy, azobenzene liquid crystallinepolymer networks. Soft Matter, 7 (9), 4318–4324.

18 Wie, J.J., Lee, K.M., Smith, M.L., Vaia, R.A., and White, T.J. (2013) Tor-sional mechanical responses in azobenzene functionalized liquid crystallinepolymer networks. Soft Matter, 9 (39), 9303–9310.

19 van Oosten, C.L., Bastiaansen, C.W., and Broer, D.J. (2009) Printed artificialcilia from liquid-crystal network actuators modularly driven by light. NatureMaterials, 8 (8), 677–682.

20 Cheng, F., Yin, R., Zhang, Y., Yen, C.C., and Yu, Y. (2010) Fully plasticmicrorobots which manipulate objects using only visible light. Soft Matter,6 (15), 3447–3449.

21 Shankar, M.R., Smith, M.L., Tondiglia, V.P., Lee, K.M., McConney, M.E.,Wang, D.H., Tan, L.S., and White, T.J. (2013) Contactless, photoinitiatedsnap-through in azobenzene-functionalized polymers. Proceedings of theNational academy of Sciences of the United States of America, 110 (47), 18792–18 797.

22 Holmes, D.P. and Crosby, A.J. (2007) Snapping surfaces. Advanced Materi-als, 19 (21), 3589–3593.

�

� �

�


23 Hafez, M., Lichter, M.D., and Dubowsky, S. (2003) Optimized binarymodular reconfigurable robotic devices. IEEE/ASME Transactions onMechatronics, 8 (1), 18–25.

24 Chen, G. and Du, Y. (2013) Double-young tristable mechanisms. Journal ofMechanisms and Robotics, 5 (1), 011 007.

25 Chen, G., Aten, Q.T., Zirbel, S., Jensen, B.D., and Howell, L.L. (2010) Atristable mechanism configuration employing orthogonal compliant mecha-nisms. Journal of Mechanisms and Robotics, 2 (1), 014 501.

26 Oh, Y.S. and Kota, S. (2009) Synthesis of multistable equilibrium compli-ant mechanisms using combinations of bistable mechanisms. Journal ofMechanical Design, 131 (2), 021 002.

27 Pirrera, A., Lachenal, X., Daynes, S., Weaver, P.M., and Chenchiah, I.V.(2013) Multi-stable cylindrical lattices. Journal of the Mechanics and Physicsof Solids, 61 (11), 2087–2107.

28 Mouthuy, P.O., Coulombier, M., Pardoen, T., Raskin, J.P., and Jonas, A.M.(2012) Overcurvature describes the buckling and folding of rings fromcurved origami to foldable tents. Nature Communications, 3, 1290.

29 Thompson, J.T., van der Heijden, G.M., and Neukirch, S. (2002) Supercoil-ing of DNA plasmids: mechanics of the generalized ply. Proceedings of theRoyal Society of London Series A, 458, 959–985.

30 Coleman, B.D. and Swigon, D. (2000) Theory of supercoiled elastic ringswith self-contact and its application to DNA plasmids. Journal of Elasticityand the Physical Science of Solids, 60 (3), 173–221.

31 Zajac, E. (1962) Stability of two planar loop elasticas. Journal of AppliedMechanics, 29 (1), 136–142.

32 Leng, F., Chen, B., and Dunlap, D.D. (2011) Dividing a supercoiled DNAmolecule into two independent topological domains. Proceedings of theNational academy of Sciences of the United States of America, 108 (50), 19973–19 978.

33 Hawkes, E., An, B., Benbernou, N., Tanaka, H., Kim, S., Demaine, E., Rus,D., and Wood, R. (2010) Programmable matter by folding. Proceedings ofthe National academy of Sciences of the United States of America, 107 (28),12 441–12 445.

34 Raviv, D., Zhao, W., McKnelly, C., Papadopoulou, A., Kadambi, A., Shi, B.,Hirsch, S., Dikovsky, D., Zyracki, M., Olguin, C. et al. (2014) Active printedmaterials for complex self-evolving deformations. Scientific Reports, 4, 7422.

35 Mao, Y., Yu, K., Isakov, M.S., Wu, J., Dunn, M.L., and Qi, H.J. (2015)Sequential self-folding structures by 3d printed digital shape memorypolymers. Scientific Reports, 5, 13616.

36 Ryu, J., D’Amato, M., Cui, X., Long, K.N., Qi, H.J., and Dunn, M.L. (2012)Photo-origami-Bending and folding polymers with light. Applied PhysicsLetters, 100 (16), 161 908.

�

� �

�

References 391

37 Liu, Y., Boyles, J.K., Genzer, J., and Dickey, M.D. (2012) Self-folding of poly-mer sheets using local light absorption. Soft Matter, 8 (6), 1764–1769.

38 Seffen, K. and Pellegrino, S. (1999) Deployment dynamics of tape springs.Proceedings of the Royal Society of London Series A, 455, 1003–1048.

39 Norman, A., Golabchi, M., Seffen, K., and Guest, S.D. (2009) Multistabletextured shell structures. Advances in Science and Technology, 54, 168–173.

40 Norman, A., Seffen, K., and Guest, S. (2009) Morphing of curved cor-rugated shells. International Journal of Solids and Structures, 46 (7),1624–1633.

41 White, T.J., Tabiryan, N.V., Serak, S.V., Hrozhyk, U.A., Tondiglia, V.P.,Koerner, H., Vaia, R.A., and Bunning, T.J. (2008) A high frequency pho-todriven polymer oscillator. Soft Matter, 4 (9), 1796–1798.

42 Yamada, M., Kondo, M., Miyasato, R., Naka, Y., Mamiya, J.I., Kinoshita, M.,Shishido, A., Yu, Y., Barrett, C.J., and Ikeda, T. (2008) Photomobile poly-mer materials-various three-dimensional movements. Journal of MaterialsChemistry, 19 (1), 60–62.

43 Yamada, M., Kondo, M., Mamiya, J.I., Yu, Y., Kinoshita, M., Barrett, C.J.,and Ikeda, T. (2008) Photomobile polymer materials: towards light-drivenplastic motors. Angewandte Chemie International Edition, 47 (27),4986–4988.

�

� �

�

393

12

Photomechanical Effects in Materials, Composites, andSystems: Outlook and Future ChallengesTimothy J. White

Dayton, OH, USA

12.1 Introduction

By now, it goes without saying that photomechanical effects in materials andcomposites are a complex interplay of photochemistry (and heat transferin some instances), polymer chemistry and physics, mechanics, optics, andengineering.The preceding chapters detail in extraordinary fashion the diverseapproaches and research focused on converting light into mechanical output.This chapter ends this book by looking toward the opportunities over thehorizon awaiting this community and general topic of research.The overarching theme of this chapter (and book as a whole) is improving

the effectiveness of transduction, which can be dependent on the propertiesof the input energy, the efficiency of conversion at the microscale (molecularscale), and system design and engineering of the elements that not only con-vert but also possibly amplify this energy into the desired output. This chapterseparates these discussions although it should be acknowledged that, in anyeventual application, they are inherently intertwined.

12.2 Outlook and Challenges

12.2.1 Breadth and Depth

As research on photomechanical materials and composites takes the initialsteps toward examinations and implementations of these materials intosystems, the already diverse range of subject to study in this topical area willrequire not only breadth but also an increasingly greater depth. In other words,


�

� �

�

394 12 Photomechanical Effects in Materials, Composites, and Systems

the next stage in the “life” of research in this topic will inherently require closelycoupled and truly collaborative research across disciplines to encompass thebreadth and depth of knowledge. The recent literature extends into thesedirections, building off the fundamental structure–property–performancestudies that have been detailed extensively throughout this text. As perfor-mance is increasingly characterized, the quantification and classification ofwhether a material or composite is a “good” photomechanical material willbe adequately assessed. The leading researchers in this community know thata bent cantilever is simply an illustration of a mechanical response and not ahallmark of utility. New entrants into the field should not think that any signof deflection equates to utility.Research of these materials and composites will increasingly require “con-

vergence,” when disparate disciplines collide [1]. Those trained in the physicalsciences such as chemistry and physics or those trained in optics, mechanics,and engineering are only partially grounded to the subject. Some, by happen-stance ormotivation, will uniquely assimilate two ormore of these to assimilatea more complete skillset – but it is hard to imagine a researcher or researchgroup with the complete set of diverse skills at the depth necessary to realizecomplete systems.

12.2.2 Beyond Bending: Mechanics Implementations

The photoresponsive materials described throughout this book substantiallyattenuate light. Light attenuation across the sample thickness results in eitherphotochemically generated or photothermal strain profiles to match. Whethera material bends is simply a question of how stiff it is, which is a product ofthe modulus of the material and the geometry of the sample, most notably thethickness [2]. By simply employing the nascent thermal expansion coefficient(or photochemistry) and a method to induce cross-sectional absorption gradi-ents, almost any material can be made to bend. If it does not bend, it will bendif the thickness or physical dimension of the sample is adjusted appropriately.Bending does not indicate utility nor is it a good means of differentiating

performance. Bending is a visualization of the mechanical output – which hashelped editors, reviewers, and peer readers in the initial stages of this researcharea confirm what sometimes has seemed surprising to the unacquainted. Asthe community develops and the research matures, we must develop robustexperimental strategies, methods, and metrics to assess candidate materialsystems or composites. Thankfully, the actuator community, most notablyProf. Michael Ashby, has painstakingly detailed comparisons of broad classesof actuator designs, materials, and approaches [3]. A canon to assess thesematerials does exist, and our community must embrace and contextualizeresults in the broader language spoken and employed by the mechanics and

�

� �

�

12.2 Outlook and Challenges 395

1e+6

Magnetostrictive Magnetostrictive

Hydraulic

Electric cylinder

Muscle

Moving coil

Moving coil

Shape memory

Pneumatic

Solenoid

Bimetallic strip

Piezoelectric

Piezoelectric

Electroactive polymer

Electroactive polymer

MEMS-comb-driveMEMS-comb-drive

Piezoelectric

Electrostatic

Electrostatic

1e–3

1e–5 1e–3

Maximum stroke (m)

0.1

1000

Ma

xim

um

ou

tpu

t fo

rce

- co

mp

ressio

n (

N)

1

Figure 12.1 “Ashby plot” relating maximum output force (in compression, N) to maximumstroke (m) for a range of materials. (Zupan et al. [3]. Reproduced with the permission of JohnWiley and Sons.)

systems engineering communities. An illustration of an Ashby plot [3] isshown in Figure 12.1.As emphasized throughout the book and specifically discussed in Chapter 9

by Liu and Broer and in Chapter 10 by Ware, shape can be a functionalproperty. It follows that shape change can be a means to introduce propertychange and variation in a system. Systems engineers and designers are wellversed in the employment of standard methods to introduce shape change:for example, compliant mechanisms in the form of various hinges and barmechanisms.Theymay be intrigued by the prospect of harnessing the potentialof light, but ultimately, structural applications have rigorous expectationsnot only for ruggedness and lifetime but also for load bearing. Considerableprogress is necessary first of all for assessing the performance of thesematerials and systems in Ashby plots, and then also for articulating the valueproposition for integrating these materials into systems. A major selling pointseems to be the ability to off-board power or passively power the mechanicalsubsystems.

�

� �

�


12.2.3 Harvesting and Harnessing Light

One of the big questions facing implementation of these materials and com-posites in systems is ultimately how will light be harvested and harnessed?Due to the nature of the azobenzene chromophore, the primary means ofinducing photomechanical responses is by irradiation with synthetic UV light.Some demonstrations have shown responsiveness to solar radiation [4–6].“Passively” powering systems with materials capable of directly convertingsolar energy into a change in shape would not be subject to the weight penaltiesof on-board lighting apparatus and the accompanying power source. Onepromising material solution to convert solar radiation into UV light is byincorporating upconverting materials that absorb light of a higher energylevel and emit light at a lower energy level [7, 8]. Unfortunately, the currentmaterials capable of this phenomenon are too inefficient, and thus, the inputenergy levels necessary to generate relevant amounts of UV or blue lightare exceedingly high. Chemists have also taken on this problem, both forthe benefit in harvesting solar radiation and for improving the biocompati-bility in potential health applications by synthesizing materials with higherwavelength absorptions [9–13]. Composite materials based on the broadspectrum absorption of carbon nanotubes, graphene, or other filler materialshave some advantages in this respect. However, these systems unabashedlyemploy a photothermal mechanism that may be difficult to decouple fromenvironmental changes in temperature unless employed in a closed system.The secondary question is what is the best way to spatially control the

mechanical response of a system?There are a number of approaches that havebeen reported – ranging from preparation of multimaterial composites todesigning the mechanical response of the material a priori [14–20]. However,especially in applications where light may not penetrate uniformly due toshading or self-folding, if light is to be used, it must be harvested but alsoharnessed. A number of papers have discussed the potential of employingoptical fibers to pipe light throughout a structure in what could be describedas a vascular-like network. To my knowledge, only the discussion and demon-strations by Kuzyk have taken steps in this direction [21–23]. The opticalengineering community, particularly those versed and skilled in fiber optics,could enjoy a rich playground with these materials and the development offiber optic networks that pipe light to dictate arbitrary local control of shapeor form.

12.2.4 Speed is Limited

For these materials to be used in control systems – for example, in shapemorphing or tactile (haptic) displays – the response time should be consid-erably improved. A misnomer is that the response time of photomechanicaleffects in these materials and composites is fast. Indeed, the fundamental

�

� �

�


t = 0 t = 80 ms

2 mm 2 mm

(a)

2.5

(b)

2.0

1.5

1.0Fo

rce

(m

N)

0.5

0.0

0 100 200

Time (ms)

Time (ms)

log

(fo

rce

)

50

1

0

–1

–2

–3100 150 200

τ = 75 ms

250 300

300 400 500

Figure 12.2 Liquid crystal elastomer loaded with noncovalently attached azobenzeneguest molecule. (a) Material was irradiated from below by a 60 ms strobe of 514 nm light.The sample deflects almost 2 mm in 80 ms. (b) Force is monitored as a function of time, withmaximum force observed after the removal of the light. (Camacho-Lopez et al. [24].Reproduced with the permission of Nature Publishing Group.)

photochemistry does occur on the picosecond timescale. However, the latencyof the response is perhaps best illustrated in Figure 12.2 by Palffy-Muhorayand coworkers, in which a 60ms irradiation of a laser induces a response[24]. However, the response of the material is latent to the illumination eventcaused by the buildup of heat (in this instance) and its ability to affect thelocal order parameter. Other examples spread throughout this book reportresponse times that are considerably slower than the fundamental absorptionevents, dictated by the crystalline or polymeric matrix.

�

� �

�


Clearly, there areways to assess fastermodes – for example, workingwith res-onant effects [6, 25, 26] or mechanical instabilities [27]. Recent work indicatesthat the assumed correlation between photochemistry and the photomechan-ical response is further convoluted [16], evident not only in the response timebut also in the restoration of the polymeric materials, in this case, back to theinitial form [14]. Such fundamental studies are critical in defining the nascentproperties of the materials examined, and importantly, the knowledge gener-ated can be leveraged to extend and enhance the response times by moleculardesign or mechanical implementation.

12.2.5 Systems Design and Implementation

Ultimately, the overall goal of this research is to realize the integration of thesematerials and composites into systems. In the limited experience I have had inthis area, it has become expediently clear that, first and foremost, the materialsmust be robust and the responses repeatable. From that basis point, relevantmodels and the accompanying design tools are key enablers to make therequisite choices that will accompany the design and implementation stage.Due to the immense tradespace, it seems clear that topology optimizationroutines to define the material properties as well as any spatial design will becritical.Furthermore, few, if any, materials are in and of themselves employed as

actuators. Think of a piston in an automobile. Chemical energy is convertedthrough combustion in a gasoline engine to drive the piston, which then turnsa crankshaft. The force is translated into motion through a system, not anyindividual material. Initial examples of photomechanical systems by Ikedause the photomechanical response of a material to drive a pulley [28]. Therecent works reporting various forms of soft robotics integrate a variety ofcomponents and subsystems to enable the desired responses [29–33]. In thisway, a system may not need large displacement. For example, a recent paperreported that a synthetic hovering bee simply uses a piezoelectric actuator andamplification to induce the large motion required for flapping [34]. Actuatorsinherently are sensitive to a control input and, in many applications, need torespond to input signals≫1Hz.

12.2.6 Applications

12.2.6.1 Optical ElementsOnepotential application of thesematerials is in optics. Althoughnot discussedat length in this book, many related azobenzene polymeric materials have beensubject to extensive examinations relating to nonlinear optics [35] or the gen-eration of surface relief gratings [36, 37] through polarization holography. Themechanical deformations that have been the primary focus of this book have

�

� �

�


(a)

(b)

(c) (d) (e) (f)

0.9

0.8

0.7

0.6

0.5

0.4

Div

erg

en

ce

0.3

0.2

0.1

0.00 5 10 15

Power (mW)

20 25

(a) (b) (c) 70

60

50

Focal le

ngth

(m

m)

40

30

20

0.20 0.25 0.30 0.35

Power density (W/cm2)

0.40 0.45 0.50

(A)

(B)

Figure 12.3 Photoinduced optical effects in azobenzene-functionalized liquid crystalnetworks. (A) Influence of light irradiation (in power, mW) on a membrane (a, insets). Thedeformation of the material causes the passing laser beam to diverge. (B) A similar materialis used in a configuration with a liquid to adjust the focal length from 65 to 20 mm. (Figuresused with permission from the Optical Society of America (A, Ref. [38]; B, Ref. [39]).) (Zupanet al. [3]. Reproduced with the permission of John Wiley and Sons.)

been used to generate dynamic optical responses, illustrated in a couple ofrecent papers (Figure 12.3) [38, 39]. The deformation has been used to steera laser beam for lensing. Other recent papers have reported on photoinduceddeflection of patterned surfaces to repeatedly form topographical surface fea-tures that affect friction but could also be used to have surface transition fromspecular to diffuse reflection [40–43]. Although the marketplace is becomingcompetitive, the employment of photoinduced optical changes could be inte-grated in mechanical designs to control light transmission in architecture orautomotive applications.

�

� �

�


12.2.6.2 Morphing Shapes and SurfacesUsing light to reconfigure the shape or surface topography could be usefulin a range of applications, from health care to aerospace. To truly generatereconfigurable shapes or surface features, the mechanical response in thematerial must be localized. A number of approaches have been taken to date,ranging from inkjet printing of shape memory polymers to controlling theswellability of hydrogels [44]. Liquid-crystallinematerials, which are inherentlyanisotropic and can be patterned through surface alignment, are currently theleading approach to realizing morphing shapes or surfaces [15]. As evidentin the work of Terentjev and coworkers [45–48], mechanical systems can beprepared to yield haptic displays with tactile responses. Integrating crystallinematerials in multimaterial combinations or through patterning could also bea route to induce designed shapes and surface protrusions. Shape morphingsurfaces have been discussed as relevant in microfluidic devices, aerospacecontrol systems, or self-cleaning surfaces.

12.2.6.3 ActuationThemost common potential application of the materials examined to date is in“actuation.” Evident in the recent literature, it seems clear that authors defineactuation very differently. Here, we define actuation as the conversion of inputenergy into output motion in an efficient and well-controlled manner. Thus,this definition excludesmuch of thework reported relating to the photoinduceddeflection or deformation of the materials reported here. Actuation inherentlyrelates to purpose. The purpose that an actuator is designed for requires theintegration of a transducer into a system, which responds reliably and control-lably to an input stimulus. Often, the transduction is subsequently amplified byother subsystems.While the overall aim is that the efficiency of the transduction in these

materials is such that these materials could in and of themselves be consideredactuators, considerable progress must be made – particularly, the conversionefficiency, which has been calculated to be as small as 0.1% in some systems.While it is conceivable that some systems could accommodate such lowconversion efficiency, they are limited to systems that are harnessing passiveenergy such as sunlight.However, a few recent reports detail compelling opportunity space. Notably,

the work of Wiersma and coworkers report on the ability to remotely controla miniature device by localized photomechanical effects [49, 50]. Further,research at the Air Force Research Laboratory has reported on spontaneouslytwisting and rolling of films on surfaces in a seemingly perpetual motility [51].Palffy-Muhoray and coworkers [24] have also used light irradiation to triggermotion in fluids.

�

� �

�

References 401

12.3 Conclusion

While there are many challenges to the integration of these materials in thewide array of potential applications discussed here and throughout the book,the fundamental tenet of directly converting light into mechanical outputremains compelling and highly distinguished. Light is inherently unique as aninput stimulus – it can travel long distances, be easily turned on and off, andis readily available through both natural and synthetic means. Further, lightcan be readily manipulated in intensity, pulse shape, phase, direction, focus,and polarization. As this technology continues to mature and develop, theseunique attributes of light as a “smart” stimulus will further distinguish thisresearch topic from the broader topic of stimuli-responsive materials.

References

1 Council, N.R. (2014) Convergence: Facilitating Transdisciplinary Integrationof Life Sciences, Physical Sciences, Engineering, and Beyond, The NationalAcademies Press, Washington, DC, p. 152.

2 Ramos, D., Mertens, J., Calleja, M., and Tamayo, J. (2007) Sensors, 7,1757–1765.

3 Zupan, M., Ashby, M.F., and Fleck, N.A. (2002) Advanced EngineeringMaterials, 4 (12), 933–940.

4 Li, C., Liu, Y., Huang, X., and Jiang, H. (2012) Advanced Functional Materi-als, 22 (24), 5166–5174.

5 Yin, R., Xu, W., Kondo, M. et al. (2009) Journal of Materials Chemistry,19 (20), 3141–3143.

6 Serak, S., Tabiryan, N., White, T.J. et al. (2010) Soft Matter, 6, 779–783.7 Jiang, Z., Xu, M., Li, F., and Yu, Y. (2013) Journal of the American Chemical

Society, 135 (44), 16446–16453.8 Wu, W., Yao, L., Yang, T. et al. (2011) Journal of the American Chemical

Society, 133 (40), 15810–15813.9 Cheng, F., Zhang, Y., Yin, R., and Yu, Y. (2010) Journal of Materials Chem-

istry, 20 (23), 4888–4896.10 Bléger, D., Schwarz, J., Brouwer, A.M., and Hecht, S. (2012) Journal of the

American Chemical Society, 134 (51), 20597–20600.11 Yang, Y., Hughes, R.P., and Aprahamian, I. (2012) Journal of the American

Chemical Society, 134 (37), 15221–15224.12 Bushuyev, O.S., Singleton, T.A., and Barrett, C.J. (2013) Advanced Materials

(Weinheim, Germany), 25 (12), 1796–1800.13 Samanta, S., McCormick, T.M., Schmidt, S.K. et al. (2013) Chemical Com-

munications, 49 (87), 10314–10316.

�

� �

�


14 Ahn, S.-K., Ware, T.H., Lee, K.M. et al. (2016) Advanced Functional Materi-als, 26 (32), 5819–5826.

15 White, T.J. and Broer, D.J. (2015) Nature Materials, 14 (11), 1087–1098.16 Liu, D. and Broer, D.J. (2015) Nature Communications, 6, Article number:

8334.17 de Haan, L.T., Gimenez-Pinto, V., Konya, A. et al. (2014) Advanced Func-

tional Materials, 24 (9), 1251–1258.18 de Haan, L.T., Sanchez-Somolinos, C., Bastiaansen, C.M.W. et al. (2012)

Angewandte Chemie International Edition, 51 (50), 12469–12472.19 Ware, T.H., McConney, M.E., Wie, J.J. et al. (2015) Science, 347 (6225),

982–984.20 McConney, M.E., Martinez, A., Tondiglia, V.P. et al. (2013) Advanced Mate-

rials, 25 (41), 5880–5885.21 Dawson, N.J., Kuzyk, M.G., Neal, J. et al. (2011) Journal of the Optical Soci-

ety of America B, 28 (8), 1916–1921.22 Dawson, N.J., Kuzyk, M.G., Neal, J. et al. (2011) Journal of the Optical Soci-

ety of America B, 28 (9), 2134–2141.23 Dawson, N.J., Kuzyk, M.G., Neal, J. et al. (2010) Optics Communications,

284 (4), 991–993.24 Camacho-Lopez, M., Finkelmann, H., Palffy-Muhoray, P., and Shelley, M.

(2004) Nature Materials, 3, 307–310.25 Lee, K.M., Smith, M.L., Koerner, H. et al. (2011) Advanced Functional

Materials, 15, 2913.26 White, T.J., Tabiryan, N., Tondiglia, V.P. et al. (2008) Soft Matter, 4,

1796–1798.27 Shankar, M.R., Smith, M.L., Tondiglia, V.P. et al. (2013) Proceedings of the

National Academy of Sciences of the United States of America, 110 (47),18792–18797.

28 Yamada, M., Kondo, M., Mamiya, J.-I. et al. (2008) Angewandte ChemieInternational Edition, 47 (27), 4986–4988.

29 Ilievski, F., Mazzeo, A.D., Shepherd, R.F. et al. (2011) Angewandte ChemieInternational Edition, 50 (8), 1890–1895, S1890/1-S1890/3.

30 Mosadegh, B., Polygerinos, P., Keplinger, C. et al. (2014) Advanced Func-tional Materials, 24 (15), 2163–2170.

31 Park, S.-J., Gazzola, M., Park, K.S. et al. (2016) Science (Washington, DC,U. S.), 353 (6295), 158–162.

32 Shen, H. (2016) Nature (London, U. K.), 530 (7588), 24–26.33 Wehner, M., Truby, R.L., Fitzgerald, D.J. et al. (2016) Nature (London, UK),

536 (7617), 451–455.34 Graule, M.A., Chirarattananon, P., Fuller, S.B. et al. (2016) Science, 352

(6288), 978.

�

� �

�

References 403

35 Ikeda, T. and Zhao, Y. (2009) Smart Light-Responsive Materials:Azobenzene-Containing Polymers and Liquid Crystals, John Wiley & Sons,Inc, Hoboken, NJ, p. 514.

36 Viswanathan, N.K., Kim, D.U., Bian, S. et al. (1999) Journal of the MaterialsChemistry, 9, 1941–1955.

37 Natansohn, A. and Rochon, P. (2002) Chemical Reviews, 102 (11),4139–4175.

38 Serak, S.V., Tabiryan, N.V., White, T.J., and Bunning, T.J. (2009) OpticsExpress, 17 (18), 15736–15746.

39 Xu, S., Ren, H., Lin, Y.-J. et al. (2009) Optics Express, 17, 17590–17595.40 Kumar, K., Schenning, A.P., Broer, D.J., and Liu, D. (2016) Soft Matter,

12 (13), 3196–3201.41 Liu, D. and Broer, D.J. (2014) Langmuir, 30 (45), 13499–13509.42 Liu, D. and Broer, D.J. (2014) Soft Matter, 10 (40), 7952–7958.43 Liu, D. and Broer, D.J. (2013) Liquid Crystals Reviews, 1 (1), 20–28.44 Liu, Y., Genzer, J., and Dickey, M.D. (2016) Progress in Polymer Science, 52,

79–106.45 Marshall, J.E., Ji, Y., Torras, N. et al. (2012) Soft Matter, 8 (5), 1570–1574.46 Campo, E.M., Roig, J., Roeder, B. et al. (2011) Proceedings of the SPIE,

8107, (Nano-Opto-Mechanical Systems), 81070H-81070H-10.47 Camargo, C.J., Torras, N., Campanella, H. et al. (2011) Proceedings of the

SPIE, 8107, (Nano-Opto-Mechanical Systems), 810709-810709-7.48 Camargo, C.J., Campanella, H., Marshall, J.E. et al. (2011) Macromolecular

Rapid Communications, 32 (24), 1953–1959.49 Rogoz, M., Zeng, H., Xuan, C. et al. (2016) Advanced Optical Materials,

4 (11), 1689–1694.50 Palagi, S., Mark, A.G., Reigh, S.Y. et al. (2016) Nature Materials, 15 (6),

647–653.51 Wie, J.J., Shankar, M.R., and White, T.J. (2016) Nature Communications, 7,

13260.

�

� �

�

405

Index

aactuator devicesALCNactuation characteristics and geometrictransformations 385

arch-shaped geometries 374bidirectional actuation 374high-frequency oscillation 388mechanical designs 386photoactuation, analog control of371–373

photoinduced optical effects 399topological defects 370trans–cis isomerization 370trans–cis–trans reorientation370–371

transverse curvature 387twisted nematic strips 384, 386

compliant mechanisms 369, 370digital photomechanical actuator

373–374binary actuators 374–380bistable helical mechanism 384–386multistable implementations380–384

LCNcross-linked polymer network 370monodomain sample 375, 377multiple actuation modes 370photoresponsive liquid crystalpolymers 370

photo-fixed strains 371

photomechanical mechanisms andmachines 387–388

photoresponsive polymers 371soft actuators 370

ALCN see azobenzene-functionalized liquidcrystalline polymers networks(ALCN)

amorphous polymers 24azobenzene-functionalized linear

polyimidesbackbone rigidity 128–132photomechanical response andcrystallinity 126–128

coordinated molecular motions 120cross-linked amorphous polyimidesazobenzene cross-linker 136, 137,139

dichroic absorbance 136photoinduced bending 136, 137RQ and SQ 139

exponential decay function 120free volume 120–122glass transition temperature 120Kuhn length scale 120melt and solid states, molecular

structures in 122molecular rigidity 120persistence length 120second-order glass–rubber transition

119sub-Tg segmental mobility 142

anodic aluminum oxide (AAO) 262


�

� �

�

406 Index

anomalous photovoltaic (APV) effect278–279

anthracenes 52anticlasticity, nematic cantilevers 96Ashby plotsmaximum output force 395Young’s modulus and density 128

atomic force microscopy (AFM) 118, 236azobenzene 117amorphous and semicrystalline


LCEs (see liquid crystal elastomers(LCEs))

liquid-crystalline monomers 118nematic order, guest dye rods 82–83nonlinear optics 398photochromism 37, 38electron transfer, photocontrol of 40free-volume requirement 39geometrical changes in 39hydrogen-bonding networks 40macroscopic motion 40–41molecular-scale motions 39–40nanometer-thick command surfaces of38

nanoscale force 39osmotic pressure pumps 40

photoinduced deformationdye hydrophobic fibers 4polyimide films, photogeneratedstresses 5

polymer chains 4–5polymer monolayers 8, 9, 25trans–cis isomerization 3–4

azobenzene-functionalized liquidcrystalline polymers networks(ALCN)

actuation characteristics and geometrictransformations 385

analog photomechanical actuators371–373

arch-shaped geometries 374bidirectional actuation 374cantilever geometry 164

cross-linker concentration 48film geometry 164flexural–torsional response 168high-frequency oscillation 388laminated films, rotational motion

44–46mechanical designs 386monodomain films, bending of 14, 16photo-driven oscillation of 46–47,

167–168photoinduced optical effects 399photoinduced phase transition 11–13polarized blue-green light, irradiation

with 13, 163, 167polydomain film, direction-selective

bending of 14–16, 161–162shape memory, all-optical control of

163surface-mediated photoalignment 13,

14thermal and photoinduced twisting of

168, 169topological defects 370trans–cis isomerization 162, 370trans–cis–trans reorientation 13, 15,

163, 370–371transverse curvature 387twisted nematic strips 384, 386

bBeer–Lambert law 124, 125, 328bend 394azobenzene, guest dye rods 82, 83differential response with depth 80–81saddle, spontaneous distortions 95, 96splay–bend director field 92–94twisted nematic cantilevers 92, 93,

95–963,3′,4,4′-benzophenonetetracarboxylic

dianhydride (BTDA) 129–132𝛽-cyclodextrin–polyethylene glycol

(𝛽-CD-PEG) gel 60, 61bimorph structure 240–2413,3′,4,4′-biphenyltetracarboxylic

dianhydride (BPDA) 129–131bistable helical mechanism 384–386block copolymers 20–21

�

� �

�

Index 407

Boltzmann’s constant 120Boyer–Beaman rule 124Brewster angle microscopy 8, 9bulk photovoltaic effectAPV effect 278–279current source model 279–282experimental setup 278, 279illumination responses 277voltage source model 282–285

ccamouflage 1cantilever geometry 164carbon black (CB) 208–210carbon nanotubes (CNTs) 19–20, 24anisotropic polymer chains 191compressed-exponential law 190Debye relaxation 186, 187, 189Euler buckling instability 189kinetic equation 189near-infrared illumination 187photoinduced polaron excitations 190stimulated response 186tube orientational ordering 187

Cartesian director pattern 103–104ceramic preparation method effectgrain size effect 290–291processing method 290surface/geometry dependence 291–294

chlorophyllin 7chrysophenine (CHP) 4, 117circadian rhythm plants 1circular director pattern 103–104CNTs see carbon nanotubes (CNTs)command surface 14, 38, 134complementary metal-oxide-semiconductor

(CMOS) 224converse piezoelectric effect 6, 276Coulombic interaction 4, 24covalent adaptive networks 171cross-link density 145–147cross-linked amorphous polyimidesazobenzene cross-linker 136, 137, 139dichroic absorbance 136photoinduced bending 136, 137RQ and SQ 139

cross-linked liquid crystal polymers(CLCPs) see liquid crystal polymernetworks (LCNs)

current source model 279–282

ddensity grating 64diacrylate liquid crystal monomers 158,

159diarylethene (DAE) 3, 21, 22, 37, 38, 238,

254, 261, 263dielectrics 282digital photomechanical actuator 373–374binary actuators 374–380bistable helical mechanism 384–386multistable implementations 380–384

4-dimethylamino-4’-nitroazobenzene(DANAB) chromophore 129, 130

dimethylformamide (DMF) 263displacement amplification mechanism

294–295dynamic mechanical analyzer (DMA) 143

eelastomeric liquid-crystalline polymer

networks see liquid crystalelastomers (LCEs)

energy band-gap model 280

fferroelectricity 6, 80, 91figure of merit (FOM) 276–277, 288–289film geometry 164Finkelmann method 10, 11, 156, 157, 165Finke–Watzky model 236Flory’s rubber elasticity theory 145fluorophore 56, 57Förster resonance energy transfer (FRET)

56, 574-chloro-cinnamic acid (4Cl-CA) 241Fourier transform infrared (FT-IR)

spectroscopy 134freely jointed chains (FJC) 120free-radical polymerization 157–159free-volume creation 79, 80free-volume fluctuation model 121fulgide 3

�

� �

�

408 Index

gGaussian curvature 81, 94, 95, 97–99,

101–102, 387Cartesian and circular director patterns

103–104cylinder 97discontinuities–advanced origami,

director fields with 107–110sphere 97spherical spindles 103–106

glass transition temperature 120, 154,159

glassy liquid-crystalline polymer networkssee liquid crystal polymer networks(LCNs)

gold nanorods 333grain size effect 290–291graphenefilms, thermal conductivity of 214nanopositioning system 219nanotube LC elastomers 203photomechanical actuators 201–203,

224polymer compositesCB/PDMS 208–210CNT/PDMS 206–209, 212GNP/PDMS 206–213, 219GO/PDMS 208–210plain PDMS 206–209, 211–212

sp2 bonds, nanocarbons 214graphene nanoribbons (GNRs) 205graphite intercalation compounds (GICs)

205, 206graphite oxide (GO) 205–210

hhairy ball theorem 105hemi-stadia director patterns 101–102high-pressure spectroscopy 39Hooke’s law 214hot-drawing polymer processing 133–137hydrosilylation reaction 154, 156, 157, 165

iindium tin oxide (ITO) 314interpenetrating polymer network (IPN)

146, 147

jJohnson–Mehl–Avrami–Kolmogorov

(JMAK) model 236

kKevlar 154Kohlrausch–Williams–Watts (KWW)

function 200Kuhn length scale 120

lLambert–Beer law 89Langmuir–Blodgett (LB) films 60Langmuir technique 8lanthanum-modified lead zirconate titanate

(PLZT) ceramics see piezoelectricceramics

layer-by-layer (LBL) assemblies 60LCEs see liquid crystal elastomers (LCEs)LCNs see liquid crystal polymer networks

(LCNs)LCPs see liquid crystal polymers (LCPs)lightadvantages 159–160applicationsactuation 400morphing shapes and surfaces 400optical elements 398–399

convergence 394harvesting and harnessing 396light-energy transducers 160mechanics implementations 394–395robotic manipulator 372, 373speed 396–398systems design and implementation 398twisted nematic ALCN, bending of 371,

372light-emitting diodes (LEDs) 215, 218,

219, 321light-energy transducers 160linear extrapolation method 279linearly polarized light (LPL) 13–14, 16,

136, 163liquid crystal elastomers (LCEs) 43–45,

68, 80azobenzene-functionalized LCE 387a +1 azimuthal defect 168–170

�

� �

�

Index 409

guest additives, actuation stress 166photoinduced contraction 160trans–cis isomerization 162trans–cis–trans reorientation 163

cross-linked 82free radical/cationic photopolymerization

157–159molecular configurations of 155monodomain LCEs, reversible uniaxial

shape change 118nanotube LC elastomer compositesactuation kinetics 200commercial development 204Debye relaxation 201dichroic nanotubes 204KWW functions 200optical-to-mechanical conversionfactors 201

polarization microscopy 205self-assembled nanotubes 202strain-dependent energy conversion205

thin-film transistors 202, 204used actuators 203viscoelastic PDMS matrix 203

nematic order (see nematic elastomers)notional properties 155pendent/in-chain LC-forming rods 82photoinduced order reduction 82polysiloxanes 10Finkelmann method 156, 157, 165hydrosilylation reaction 154, 156,157, 165

side chain/main chain 156, 157, 165volume and shape change 84–85

liquid-crystalline (LC) glassesdense cross-linking 85immobile component rods 85nematic glassesirradiation 85photocantilever 92, 93polydomain glass, curl direction 90

optothermal Poisson ratio 85perpendicular response 85relative volume change 85

liquid-crystalline (LC) solids 81

liquid crystal polymer networks (LCNs)68

azobenzene-functionalized LCNcantilever geometry 164cross-linker concentration 48film geometry 164flexural–torsional response 168laminated films, rotational motion44–46

monodomain films, bending of 14, 16photo-driven oscillation of 46–47,167–168

photoinduced phase transition 11–13polarized blue-green light, irradiationwith 13, 163, 167

polydomain film, direction-selectivebending of 14–16, 161–162

shape memory, all-optical control of163

surface-mediated photoalignment 13,14

thermal and photoinduced twisting of168, 169

trans–cis isomerization 162trans–cis–trans reorientation 13, 15,163

CNTs 20, 24contraction and extension 10, 12, 14–15cross-linked polymer network 370Finkelmann method 10, 11free radical/cationic photopolymerization

157–159glass transition temperature 154in situ polymerization 10–11(meth)acrylate photopolymerization

166molecular configurations of 155monodomain sample 375, 377multiple actuation modes 370notional properties 155photoresponsive liquid crystal polymers

370splay and twisted nematic geometry 168three-dimensional motions of 15, 17–19

liquid crystal polymers (LCPs) 10advantages of 118anisotropic steric repulsions 132

�

� �

�

410 Index

liquid crystal polymers (LCPs) (contd.)melt and solid states, molecular

structures in 122molecular configurations of 155notional properties 155photoinduced actuation 43–49Vectra/Kevlar 154

lower critical solution temperature (LCST)345

low-molecular-weight glasses 23, 24LPL see linearly polarized light (LPL)

mMaier–Saupe model 132melting temperature (Tm) 123–124mesoporous silica nanocrystals 56, 57metal–organic frameworks (MOFs) 65–68Michael addition reaction 59micromechatronics 299micro-propelling robot 297–299micro-walking machine 295–296morphotropic phase boundary (MPB) 286multiwalled carbon nanotube (MWCNT)

19, 182, 209multiwalled nanotubes (MWNTs) 180,

181, 183

nnanopositioningaccuracy and resolution 223–224dual actuators 213GNP/PDMS actuator fabrication and

characterization 216–219light-driven actuators 214nanopositioner system integration

219–221photothermal nanopositioners, kinetics

of 221–222polymeric chains 213principle GNP/elastomer photothermal

actuation 214, 215, 218single-axis photothermal

nanopositioning system 215, 218useful vs. maximum displacement

222–223nanotube liquid crystalsand highly oriented nanotubes

anodisc alumina filter 194dispersion and fabrication process194

elastomer composites 192, 193isotropic-nematic transition 195Lambert–Beer law 196macroscopic and long-range ordering192

nucleated nematic islands 195PDMS 192polarization microscopy 196schlieren textures 197SWNT 1922-D FFT analysis 196

LCEactuation kinetics 200commercial development 204Debye relaxation 201dichroic nanotubes 204KWW functions 200optical-to-mechanical conversionfactors 201

polarization microscopy 205self-assembled nanotubes 202strain-dependent energy conversion205

thin-film transistors 202, 204used actuators 203viscoelastic PDMS matrix 203

naphthalene diimide (NDI) 257, 259nematic elastomersazobenzene, guest dye rods 82, 83chain elongation and spontaneous

distortion 83–84films/cantilevers, illumination effects oncurvature 87dynamical intensity and dyepopulations 88–90

intensity 86–87Lambert–Beer limit 86modified Lambert–Beer law 86photon absorption 86photostationary dye populations andmechanical response 87–88

photostrain 87stress 87thermal recovery time 86

�

� �

�

Index 411

mean-square polymer chain size, ratio of82–83

photomechanics vs. thermal mechanics91–92

polydomain photoelastomers 90–919-anthracene carboxylic acid (9AC) 241,

254–2579-methylanthracene (9MA) 240, 250–2549-tert-butylanthracene ester (9TBAE)

237, 250, 251, 254nonelectronic photoresponsive receivers 2nonisometric origami 110, 111nonradiative deactivation process 6

oone-way shape memorycross-linking segments 331definition 327photochemical mechanismsamorphous polyimides andliquid-crystalline acrylic networks341

azobenzene derivatives 340–342coumarin moieties 337cross-links 336light-responsive shape memorypolymers 338–339

molecular photoisomerization 340photo-healable polymers 343photo-origami 340rearrangement mechanisms 340reversible binding groups 337stimuli-responsive behaviors 336

photothermal behaviorabsorbing fillers 333, 335, 336bending recovery 332elastomeric matrix 335gold nanorods 333icosahedrons 335Nafion 332photoresponsive SMP foams 331SMP 331–3363D shape transformations 333–335UV light 335

switching segments 329, 331thermally responsive materials 330

Onsager model 132

Onsager steric ordering mechanism 82optical communication 2optoactuation 79organic crystals 21, 22, 24oxy-4,4′-di(phthalic anhydride) (OPDA)

129–132

pPauli’s exclusion principle 120PDMS see polydimethylsiloxane (PDMS)perfluoronaphthalene 21, 22perylene diimide (PDI) 257photoacoustic tomography 2photochemical responses 23–24photochromic crystalshistory and backgroundbending mechanism 237crystal packing effects 234crystal photochemistry 236DAE derivative 238, 239organic nano-and microcrystals 237photoreactive crystals 240reversible photochromic reactions235

single-crystal X-ray analysis 236UV light irradiation 236

interfacing molecular crystals 262–263intermolecular photochemical reactionsirreversible solid-state [4+4]photodimerization reactions250–254

nonequilibrium charge distributionand electronic heating 257–259

[2+2] photodimerization 248–250reversible [4+4] photodimerizationsystems 254–257

intramolecular photochemical reactionsphotodissociation 247–248photoisomerization 244–247ring-opening/closing reactions242–244

mechanical actioncomplete transformation and crystalreconfiguration 241–242

partial reaction and bimorphformation 240–241

�

� �

�

412 Index

photochromic crystals (contd.)molecular crystals, reaction dynamics in

260–261new photomechanical materials

261–262organic photomechanical materials 234

photochromismdefinition 37in solid state (see solid-state

photochromism)photo-driven relay 295photoelectrical processes 6, 24photoinduced shape programminganisotropic stimulus 328, 329Beer–Lambert law 328light 328one-way shape memory (see one-way

shape memory)two-way shape memory (see two-way

shape memory)photomechanical crystalsamorphous phase-mediated azobenzene

isomerization 53, 54anthracenes 52cocrystal, photomechanical bending in

53, 54diarylethene crystals 50–51DR1 54salicylidenephenylethylamine molecules

52–53photophone 1–2, 276, 297photopolymerization 157–159photosalient crystals 50, 246photo-softening effect 48photostrictive effect 6ceramic preparation method effectgrain size effect 290–291processing method 290surface/geometry dependence291–294

experimental setup 278figures of merit 276–277, 288–289materials considerations 289–290

photothermal processes 6, 7, 24photoviscosity effect 117photovoltaic effect 6bulk photovoltaic effect 276

APV effect 278–279current source model 279–282experimental setup 278, 279photovoltaic current, illuminationresponses 277–278

voltage source model 282–285dopant research 287–288ferroelectric/piezoelectric materials 276light polarization direction effect

285–286photostrictive effect 276PLZT composition research 286–287

piezoelectric ceramicsphotostrictive device applicationsdisplacement amplification mechanism294–295

micro-propelling robot 297–299micro-walking machine 295–296photo-driven relay 295photophones 297

photostrictive effect (see photostrictiveeffect)

photovoltaic effect (see photovoltaiceffect)

Poisson ratio 85, 348polyacrylamide gels 6–7polydimethylsiloxane (PDMS)MWNTs 180nanocomposite photomechanical

actuators, CNTs 202nanotube liquid crystal elastomer

composite 192–193nanotubes and carbon black 189plain elastomer 197–198prestrains 180, 183, 187SWNT LC films 1922-D nanomaterial (graphene)–polymer

compositesCNT 206–209, 212GNP 206–213GO 208–210plain polymer 206–209, 211–212

polyetherimide (PEI) film 129, 130poly(isopropyl methacrylate) (PIPMA)

142polymer gel 2, 4, 24

�

� �

�

Index 413

polymeric liquid crystals (PLCs) see liquidcrystal polymers (LCPs)

polymeric materialsabsorption processes 124–126amorphous state (see amorphous

polymers)annealing and aging 138–141azobenzene, photoisomerization of 117cross-link density 145–147macromolecules 119molecular alignment, hot-drawing

process 133–137molecular weight 119monomer, dimer, and trimer 119oligomers 119semicrystalline state (see semicrystalline

polymers)spirobenzopyran, photoisomerization of

118sub-Tg segmental mobility 142–145

polymer nanocompositesCNT polymer composites, relaxation ofanisotropic polymer chains 191compressed-exponential law 186, 190Debye relaxation 187, 189Euler buckling instability 189kinetic equation 189, 190light-off relaxation 187, 189, 190light-on compressed-exponentialresponse 190

near-infrared illumination 187normalized stress vs. time 187, 188photoinduced polaron excitations190

stimulated response 186, 187tube orientational ordering 187

highly oriented nanotubes (see nanotubeliquid crystals)

nanopositioningaccuracy and resolution 223–224dual actuators 213GNP/PDMS actuator fabrication andcharacterization 216–219

light-driven actuators 214nanopositioner system integration219–221

photothermal nanopositioners,kinetics of 221–222

polymeric chains 213principle GNP/elastomerphotothermal actuation 214, 215,218

single-axis photothermalnanopositioning system 215, 218

useful vs. maximum displacement222–223

nanotube LC elastomer (see nanotubeliquid crystals)

oriented nanotube composites 197–200photomechanical actuationcarbon fiber technologies 186continuum elasticity and nanotubemorphology 184

exerted actuation stress, magnitude of182, 183

IR stimulations 181–183local and macroscopic strains 185MWNTs 180, 181, 183nonradiative photon decay 183orientational bias 181, 183PDMS network 180–181photomechanical testing apparatus180, 181

resonant undulation 186scattering reflexes 180, 181stress vs. pre-strain data 181–182Young’s modulus 181, 184

2-D nanomaterial (graphene)–polymercomposites

CB/PDMS 208–210CNT/PDMS 206–209, 212GNP/PDMS 206–213GO/PDMS 208–210plain PDMS 206–209, 211–212

poly(meth)acrylate liquid-crystallinepolymer networks 166–170

polymethacrylic acid (PMAA) 117poly(methyl acrylate) (PMA) 142poly(n-isopropylacrylamide) (PNIPAM) 7gold nanoshells 347iron oxide nanoparticles 348malachite green 345

�

� �

�

414 Index

poly(n-isopropylacrylamide) (PNIPAM)(contd.)

swelling–deswelling carbon nanotubehybrid gels 345

thermoresponsive gels 345polysiloxanes 10Finkelmann method 156, 157, 165hydrosilylation reaction 154, 156, 157,

165side chain/main chain 156, 157, 165

polyvinylidene fluoride (PVDF) 205popping crystals 50profilometer 293proportional–integral–derivative (PID)

control loop 218pulse-width modulation (PWM) 220pyromellitic dianhydride (PMDA)

126–132

rRaman spectroscopy 134, 248rapid quenching (RQ) 139–141reactive mesogens (RMs) 157, 304reversible bends 256, 263, 351robotic manipulator 372, 373

ssalicylidenephenylethylamine 52schlieren textures 197selenium 2self-assembled monolayers (SAMs) 60semicrystalline polymers 24azobenzene-functionalized linear


azobenzene moieties, absorption kineticsof 126

Boyer–Beaman rule 124cantilever bending 123melt and solid states, molecular

structures in 122melting temperature 123–124specific volume 121, 122WLF theory, Boltzmann TTS 121

shape memory polymers (SMPs)

absorbing materials 334allyl sulfide-containing networks 340amorphous and semicrystal 333aneurisms, treatment of 331biomedical devices 336carbon nanotubes 331–333chromophore, photoisomerization of

336dye-dope 331gold nanorods 333infrared light 331optical absorption 331photoisomerization, two-way SMPs 343photoresponsive behavior 329photostable cross-link 337polymeric materials 329programmed absorption 333reversible exchange/addition reactions

336thermally responsive materials 329, 330

simple melt spinning method 59single-molecule force spectroscopy 39single-walled carbon nanotube (SWCNT)

20single-wall nanotube (SWNT) 192, 194,

204slow quenching (SQ) 139–141SMPs see shape memory polymers (SMPs)solid-state photochromismartificial molecular machines 54–55acrylate-type azobenzene monomers59

definition 55glass surface, dual-responsivemicropump on 60, 61

hydrogen-bonded cross-linked fibers59

Langmuir–Blodgett films 60layer-by-layer assemblies 60mesoporous nanocrystals 55–57poly(para-phenylene) backbone 56,58, 59

SAMs 60simple melt spinning method 59

azobenzene chromophoreselectron transfer, photocontrol of 40free-volume requirement 39

�

� �

�

Index 415

geometrical changes in 39hydrogen-bonding networks 40macroscopic motion 40–41molecular-scale motions 39–40nanometer-thick command surfaces of38

nanoscale force 39osmotic pressure pumps 40

matrix effects 38metal–organic frameworks 65–68photomechanical effectsamorphous azo polymers 42–43definition 41LCNs and LCEs 43–49, 68photosalient/photomechanical crystals49–54

small-molecule switches 37–38surface mass transport and phase change

effectsazobenzene crystals 64, 65glass surface, single crystals 64–66surface relief grating 62–64

spirobenzopyran 5, 118spiropyrans 3, 37, 38splay–bend director field 92–94square pyramids 110, 111stimuli-induced deformation 2–3stimuli-responsive materials 171surface relief gratings (SRGs) 62–64, 118surface topographycurved shells, creation of 100definition 303liquid crystal (LC) networkschiral-nematic network 305, 306,309, 310, 318–322

copolymerized azobenzene molecule304, 307, 308

cross-linked LC networks 304diacrylates 305, 306formation of 305, 306initially highly ordered state, surfaceactuation 307–309

inkjet-print cilia, photoinducedbending 307, 308

isotropic films 309, 311light intensity ratios 321

line-patterned coatings, directorpattern 311–315

monoacrylates 305–307photopolymerized LC networks,advantage of 304

polymerization 304random director pattern 315–318reactive mesogens 304splayed LC network 305–308thermal expansion 305twisted 305, 306uniaxial alignment 305volumetric relaxation 320–321

nanoscopic effects 303nature, dynamic topographic structures

in 303static patterns 303

SWNT see single-wall nanotube (SWNT)

tthermal decay 80thermosalient crystals 50thiol–ene polymerization 145thiol–yne polymerization 145time–temperature superposition (TTS)

121triphenyl methane leuco dyes 6–7twisted nematic cantilevers 92, 93, 95–96two-photon excitation (TPE) 56, 57two-photon photoexcitation (2PE) methods

255–256two-step cross-linking method see

Finkelmann methodtwo-way shape memoryanisotropy 344definition 328photochemical reactionsazobenzene 353–354dissociation reactions 353irradiation 355liquid-crystalline materials 354–356molecular isomerization 353organic crystalline materials 357photoexpansion/photocontraction356

realizable strains 357ring-opening reactions 353, 356, 357

�

� �

�

416 Index

two-way shape memory (contd.)semicrystalline polyimides 354spriropyran 356tensile and bending reversible shapechanges 355

photothermal stimulusambient temperature 352gold nanoshells 347graphene oxide 346light stimulus, removal of 344liquid crystal polymers 350, 351optical patterning methodologies 352order–disorder transitions 350, 351PNIPAM 345–348polymer–silicon microcantilever 352reconfigurable two-way shape memorymaterials 348

self-folding origami-like structures346

spatially patterned hydrogel 348, 349swelling and deswelling process 348,350

thermoresponsive gels 345, 347

piezoelectric ceramics 343shape memory alloys 343

uUV–vis spectroscopy 139, 320

vvacuum filtration technique 192Vectra 154voltage source model 282–285

wWarner theory 253Weigert effect 13, 163Weigert’s effect 126wide-angle X-ray diffraction (WAXD) 133Williams–Landel–Ferry (WLF) theory

121, 138

yYoung’s modulus 87, 128–129

�

� �

�

(a) (b) (c) (d)

(e) (f) (g) (h)

(j) (k) (l)

(n) (p)

(v)(t)

(i)

(m)

(q) (r)

(u) (w)

(x) (y) (z)

(s)

(o)

Figure 1.11 Various three-dimensional motions of CLCP systems induced by light.(a) Oscillation [75]. (b) Swimming [74]. (c) Light-driven plastic motor [73]. (d) Inchwormwalk [76]. (e) Robotic arm [76]. (f ) Manipulation of an object [77]. (g) Actuation throughtissues [78]. (h) Gripper [128]. (i) Crawling up [128]. (j) Adaptive liquid lens [80]. (k) Localizedactuation [81]. (l) Tactile device [82]. (m) Heliotropism [83]. (n) Microparticle [138].(o) Artificial cilia [85]. (p) Pillar array [86]. (q) Size-changeable pores [87]. (r) Fiber [88].(s) Micropump [89]. (t) Snap-through [90]. (u) Deformation into cone [114]. (v) Accordionfolding [115]. (w) Checkerboard pattern [115]. (x) Photoswitchable stripes [116]. (y) Dynamic3D finger print [79]. (z) Winding of spring [118].


�

�

��

“Off” “Activated”

“On”

Ar+ off

Ar+ off

E⊥n

E⊥n

E||n

Figure 2.3 The optical protocol for activating the light-powered oscillation of a cantilever. The nematic director (n) is positioned parallel to thelong axis of the polymer cantilever of dimension 5 mm × 1 mm × 50 mm. When exposed to light polarized orthogonal to n (E⊥n), bending occurstoward the laser source. Cycling the Ar+ laser from E⊥n to E‖n can turn oscillation “on,” while blocking the Ar+ or returning the polarization of thelaser beam to E⊥n turns the oscillation “off.” (White et al. [51]. Reproduced with the permission of Royal Society of Chemistry.)

�

�

��

RO N RO RON N

NNOR OR OR

3a: R = C12H25

3b: R = C6H13

2a: R = C12H25

2b: R = C6H13

2c: R = C10H21

1a: R = C12H25

1b: R = C6H13

(a)

(b) (c)

N

Figure 2.13 (a) Structures and (b) photographs of the crystalline powders of azobenzenes utilized in the study. (c) The same powders afterirradiation with 365 nm light for 30 min at 100 mW/cm2. (Norikane et al. [186]. Reproduced with the permission of American Chemical Society.)

�

� �

�

+1

–1

(a)(b)

(e)

(f)(g)

(d)

(c)

Figure 3.14 Arrays and networks of defects for topography and topology. (a) An array ofpyramids rises from a flat sheet decorated with a director field consisting of concentricsquares. These are discrete forms of m = +1 defects, but now the square pyramids intowhich they rise (b) cannot relax to cones. The ±1 defects are identified by two labelledexamples in red. Rising pyramids, similar to these, were used to lift heavy loads [43]. (c) Tworegions of uniform director can be welded together by a grain boundary of (discrete) ±1/2defects (circled in blue and red); after [52]. (d) Upon weak deformation, a transverselyshrunk, still planar region in contact with a ridged region remains. (e) Upon strongeractuation, both regions are turned into parts of a faceted bottle in order to reduce theoverall bend energy along the ridges. (f ) An array of + 1

2and −1 defects, where neighboring

+ 1

2s’ cores are connected by slits (heavy lines) that are as yet unopened. (g) Contraction

along the directors around the defects leads to the opening of slits while remainingplanar–a topological rather than topographical change; taken from Ref. [44] whereexperimental realizations are shown. (Ware et al. [43]. Reproduced with the permission ofThe American Association for the Advancement of Science.)

�

� �

�

Rapidly quenched (RQ) Slowly quenched (SQ)

Po

ten

tia

l e

ne

rgy

Configuration space

150

180

210

240

270

300

330

0

30

60

90

1201.0

1.0

1.0

1.0

0.9

0.80.8

0.8

0.9

0.9

0.9

150

180

210

240

270

300

330

0

30

60

90

1201.0

1.0

1.0

1.0

0.9

0.80.8

0.8

0.9

0.9

0.9

RQ SQ

(a)

(b)

Figure 4.8 (a) Schematic description of the potential energy landscape for glasses withdifferent configuration spaces. The left image describes the rapidly quenched (RQ) samplehaving larger free volume. The right image illustrates the slowly quenched (SQ) sample withdense environment. (b) Polar plots of the normalized absorption value at 𝜆 = 355 nm for thephysically aged azobenzene-containing polyimides (left, RQ; right, SQ) at different lightirradiation conditions: (•) Before irradiation, ( ) after irradiation with linearly polarized442 nm light polarized along the y-direction (90–270∘ axis), or ( ) along the x-direction(0–180∘ axis), and ( ) 4 days after irradiation with linearly polarized 442 nm light polarizedalong the x-direction (0–180∘ axis). (c) Photomechanical bending is monitored with acantilever geometry (5 mm × 1 mm × 20 μm) upon 100 mW/cm2 intensity of 𝜆 = 442 nmlight linearly polarized along the x-direction. Effect of different physical aging conditions iscontrasted by monitoring RQ (i–iii) and SQ (iv–vi) samples. The (i, iv) inset images showcantilevers before light irradiation and after 2 h of irradiation with polarized 442 nm light(parallel to the long axis of the cantilever (E‖x)) shown in (ii, v). The (iii, vi) images arecaptured after 72 h of dark relaxation after the light irradiation. (d) Summarizedphotomechanical bending response of azobenzene-functionalized polyimide cantilevers forRQ ( ) and SQ ( ) during 2 h of continuous irradiation 100 mW/cm2 intensity of 𝜆 = 442 nmlight linearly polarized to x-axis followed by subsequent dark relaxation. (Lee et al. [55].Reproduced with the permission of American Chemical Society.)

�

� �

�

SQ

RQ

00

5

10

15

20

25

30

35

1 2 3 4 5 110 120

Time (h)On Off

Bendin

g a

ngle

(°)

RQ

SQ

E||x

E||x OFF

OFF

72 h

72 h

(iv) (v) (vi)

(i) (ii) (iii)

(c)

(d)


�

� �

�

366 nm >540 nm

>540 nm

>540 nm

>540 nm366 nm

366 nm

366 nm

–45°

–90°

–135°

0°

Figure 5.5 All-optical control of bending direction and flattening in anazobenzene-functionalized liquid-crystalline polymer network, in the polydomainorientation. The film in this experiment was heated above the glass transition temperature.Irradiation with linearly polarized UV light in the orientations inset into the images dictatesthe direction of the bending. Irradiation with light >540 nm restores the films to the flatcondition. (Yu et al. [55]. Reproduced with the permission of Nature Publishing Group.)

�

� �

�

(a)

(c) 1 2 3

AP

5 mm

4 5

(d)

(e)

(f)

(b)

Figure 5.9 Azo-LCE compositions were prepared by aza-Michael addition reactionsincluding 2azo. Samples were prepared with +1 azimuthal defects subsumed in the centerof square films of 5 × 5 mm with 50 μ thickness. (a) Illustration of the director profiledescribed by a +1 azimuthal defect. (b) A representative photograph of a +1 azimuthaldefect within a azo-LCE taken between cross-polarizers. (c–f ) The five azo-LCE films wereplaced on a white surface and subjected to 365 nm irradiation of 100 mW/cm2 for 15 min.Photographs were taken to measure the relative deflection of the materials (c) duringexposure, (d) 5 s after exposure, (e) 2 min after exposure, and (f ) after 532 nm exposure(∼50 mW/cm2 for 10 min). (Ahn et al. [64]. Reproduced with the permission of John Wileyand Sons.)

�

�

��

(a)

(b)

(c)

Figure 9.6 Line-patterned coatings with locally different director profiles. (a) A coating with alternating stripes chiral-nematic order next toisotropic order deforms from a flat state (b) to a deformed state 9c) under exposure to UV light. (b, c) Interference microscopic images takenbefore and during exposure, which show a modulation depth of around 10% relative to the initial coating thickness. (d) A coating with alternatingstripes of planar chiral-nematic order next to homeotropic order. (e) The cross section of the coating measured by interference microscopemeasure prior to UV exposure. The small corrugations, enlarged in the inset, originate from an imprinted ITO pattern. (f ) The same film underexposure to UV light. The planar chiral-nematic area expands relative to the homeotropic area with a modulation depth of around 20%.

�

�

��

20.2

0.150.1

0.050

1.5

1

0.5

0

1.5

1

0.5

0

0 100 200

200 400 600 800 1000 12000

300 400 500 600 700 800 900 1000 1100 1200

0 100 200 300 400 500 600 700 800 900 1000 1100 1200

Height (μm)2

Height (μm)

(d)Position (μm)

(f)

Position (μm)

(e)


�

�

��

Homeotropic

Planar

(a) (b)hν

Pf

(b)

UV

Δ

0.6

0.4

0.2

0

−0.2

−0.4

−0.6

He

igh

t (μ

m)

0.6

0.4

0.2

0

−0.2

−0.4

−0.6

He

igh

t (μ

m)

0 01 02 03 04 50 0 01 02 03 04 50

Length(μm) Length(μm)

Figure 9.7 Randomly patterned coatings with locally different director patterns. (a) A fingerprint pattern, which forms in a chiral-nematic coatingwith the helix axes aligned parallel to the substrate surface. (b) The fingerprint expands at the locations where the rod-like molecular units areoriented parallel to the surface and shrink at the positions where they are aligned perpendicular, resulting in a modulation depth of around 20%of the initial coating thickness. (c) An illustration of a coating with a polydomain pattern. (d) Here also, the domains with (close to) planarorientation expand, whereas the domains with (close to) homeotropic orientation shrink.

�

�

��

(c)(d)

hνΔ

0.6

0.4

0.2

000 200 4000 200 300100 400

−0.2

−0.4

−0.6 Length (μm)Length (μm)

389 μm292 μm 389 μm

292 μm

μm2

1.5

1

0.5

0

Heig

ht (μ

m)

0.6

0.4

0.2

−0.2

−0.4

−0.6

Heig

ht (μ

m)


Photomechanical Materials, Composites, and Systems: Wireless Transduction of Light into Work

Documents

Transcript of Photomechanical Materials, Composites, and Systems: Wireless Transduction of Light into Work