Optical switches - Materials and design

1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X Optical switches i © Woodh...

Author: Baojun Li and Soo Jin Chua

220 downloads 2076 Views 13MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form

DOWNLOAD PDF

1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

Optical switches

i © Woodhead Publishing Limited, 2010

Related titles: Handbook of advanced dielectric, piezoelectric and ferroelectric materials (ISBN 978-1-84569-186-8) This comprehensive book covers the latest developments in advanced dielectric, piezoelectric and ferroelectric materials. It presents current research from leading innovators in the field. Sections cover topics under the general headings: high-strain highperformance piezo- and ferroelectric single crystals; electric-field-induced effects and domain engineering; morphotropic phase boundary related materials and phenomena; high-power piezoelectric and microwave dielectric materials; nanoscale piezo- and ferroelectrics; piezo- and ferroelectric films; novel processing, new materials and properties. Science and technology of advanced piezoelectric materials (ISBN 978-1-84569-534-7) Science and technology of advanced piezoelectric materials provides an up-to-date comprehensive review of piezoelectrics. Written by a distinguished team of international contributors, this book focuses on the most promising piezoelectric materials and their applications. In the first part of the book, the properties of key piezoelectrics are discussed; materials include PZT, relaxor ferroelectrics, Pb-free ceramics, quartz, lithium niobate and tantalate, single crystals, polymers and composites. Part II reviews the preparation techniques of piezoelectrics during manufacturing and in the laboratory. Applications oriented materials development is discussed in Part III, focusing on actuators, high-power applications and performance under stress. Details of these and other Woodhead Publishing materials books can be obtained by: • visiting our web site at www.woodheadpublishing.com • contacting Customer Services (e-mail: [email protected]; fax: +44 (0) 1223 893694; tel.: +44 (0) 1223 891358 ext. 130; address: Woodhead Publishing Limited, Abington Hall, Granta Park, Great Abington, Cambridge CB21 6AH, UK) If you would like to receive information on forthcoming titles, please send your address details to: Francis Dodds (address, tel. and fax as above; e-mail: francis. [email protected]). Please confirm which subject areas you are interested in.

ii © Woodhead Publishing Limited, 2010

1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

Optical switches Materials and design

Edited by Baojun Li and Soo Jin Chua

iii © Woodhead Publishing Limited, 2010

1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

Published by Woodhead Publishing Limited, Abington Hall, Granta Park, Great Abington, Cambridge CB21 6AH, UK www.woodheadpublishing.com Woodhead Publishing, 525 South 4th Street #241, Philadelphia, PA19147, USA Woodhead Publishing India Private Limited, G-2, Vardaan House, 7/28 Ansari Road, Daryaganj, New Delhi – 110002, India www.woodheadpublishingindia.com First published 2010, Woodhead Publishing Limited © Woodhead Publishing Limited, 2010 The authors have asserted their moral rights. This book contains information obtained from authentic and highly regarded sources. Reprinted material is quoted with permission, and sources are indicated. Reasonable efforts have been made to publish reliable data and information, but the authors and the publisher cannot assume responsibility for the validity of all materials. Neither the authors nor the publisher, nor anyone else associated with this publication, shall be liable for any loss, damage or liability directly or indirectly caused or alleged to be caused by this book. Neither this book nor any part may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, microfilming and recording, or by any information storage or retrieval system, without permission in writing from Woodhead Publishing Limited. The consent of Woodhead Publishing Limited does not extend to copying for general distribution, for promotion, for creating new works, or for resale. Specific permission must be obtained in writing from Woodhead Publishing Limited for such copying. Trademark notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation, without intent to infringe. British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library. ISBN 978-1-84569-579-8 (print) ISBN 978-0-85709-041-6 (online) The publisher’s policy is to use permanent paper from mills that operate a sustainable forestry policy, and which has been manufactured from pulp which is processed using acid-free and elemental chlorine-free practices. Furthermore, the publisher ensures that the text paper and cover board used Have met acceptable environmental accreditation standards. Typeset by RefineCatch Limited, Bungay, Suffolk, UK Printed by TJI Digital, Padstow, Cornwall, UK

iv © Woodhead Publishing Limited, 2010

Contents

Contributor contact details 1

Introduction to optical switches

ix 1

S.J. Chua, National University of Singapore, Singapore, and B.J. Li, Sun Yat-Sen University, China

2

Electro-optical switches

B.J. Li, Sun Yat-Sen University, China

5

2.1 2.2 2.3 2.4 2.5 2.6

Introduction Theory and principles of electro-optical switches Materials and fabrication of electro-optical switches Device structures of electro-optical switches Performance and challenges References

5 6 10 12 58 59

3

Thermo-optical switches

61

L. Sirleto, G. Coppola, M. Iodice, M. Casalino, M. Gioffrè and I. Rendina, National Research Council – Institute for Microelectronics and Microsystems, Naples, Italy

3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8

Introduction Theory and principles of thermo-optic effect Materials for thermo-optical switches Device structures of thermo-optical switches Conclusions List of abbreviations List of symbols References

61 62 69 75 86 89 90 91 97

4

Magneto-optical switches

J. Tioh, R.J. Weber and M. Mina, Iowa State University, USA

4.1 4.2

Introduction History of optical communication

97 97 v

© Woodhead Publishing Limited, 2010

1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

vi

Contents

4.3 4.4 4.5 4.6 4.7 4.8 4.9 4.10

All-optical switches Magneto-optical switches Theory and principles of magneto-optical switches Material Characterization of Faraday rotation Summary Appendices References

102 104 105 116 118 129 129 132

5

MEMS-based optical switches

136

L.L.P. Wong and J.T.W. Yeow, University of Waterloo, Canada, and A.A. Goldenberg, University of Toronto, Canada

5.1 5.2 5.3 5.4 5.5 5.6 5.7 5.8 5.9

Introduction Optical systems Optical switch architectures Actuating principles of MEMS-based optical switches Materials and fabrication of MEMS-based optical switches Challenges surrounding MEMS-based optical switches Conclusions List of abbreviations References

136 136 138 144 149 153 155 155 155

6

SOA-based optical switches

158

A. Assadihaghi, H. Teimoori and T.J. Hall, University of Ottawa, Canada

6.1 6.2 6.3 6.4 6.5 6.6

Introduction SOA-based switching strategy SOA structure SOA design criteria Summary References

158 158 165 171 178 178 181

7

Switching based on optical nonlinear effects

M.P. Fok and P.R. Prucnal, Princeton University, USA

7.1 7.2 7.3 7.4 7.5 7.6

Introduction Nonlinear effects for optical switches Nonlinear devices for optical switches Structure of nonlinear-effect-based optical switches The ‘ideal’ nonlinear-effect-based optical switch? References


181 182 185 189 202 203

8

Contents

Liquid crystal optical switches

vii

206

C. Vázquez García, I. Pérez Garcilópez and P. Contreras Lallana, Universidad Carlos III, Spain, and B. Vinouze and B. Fracasso, Telecom Bretagne, France

8.1 8.2 8.3 8.4 8.5 8.6

Introduction Liquid crystal theory and principles Liquid crystal switches and applications Future trends Acknowledgments References

206 208 215 230 235 236

9

Photonic crystal all-optical switches

241

K. Asakawa, Y. Sugimoto, N. Ikeda and Y. Watanabe, National Institute for Materials Science, Japan, N. Ozaki, Wakayama University, Japan, Y. Takata, Kyocera Corporation, Japan, Y. Kitagawa, Stanley Electric Co. Ltd, Japan, S. Ohkouchi and S. Nakamura, NEC Corporation, Japan, A. Watanabe, Meijo University, Japan, and X. Wang, National Institute of Advanced Science and Technology, Japan

9.1 Introduction 9.2 Theory and principles of photonic crystal all-optical switches 9.3 Design and fabrication of advanced 2DPC waveguide for PC-SMZ 9.4 Growth and characterization of optical QDs for PC-FF 9.5 Device structures and performances of photonic crystal all-optical switches 9.6 Conclusion 9.7 Acknowledgments 9.8 References 10 Fiber, holographic, quantum optical and other types of optical switches

241 243 251 257 267 271 272 273 276

Y. Zhang and B.J. Li, Sun Yat-Sen University, China

10.1 10.2 10.3 10.4 10.5 10.6

Introduction Fiber switches Holographic switches Quantum optical switches Other switches References

276 277 294 296 305 309

11

Summary: key trends in optical switches

313

B.J. Li, Sun Yat-Sen University, China, and S.J. Chua, National University of Singapore, Singapore

Index

316


1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

viii

Contributor contact details

(* = main contact)

Chapter 1

Chapter 3

S.J. Chua Department of Electrical and Computer Engineering National University of Singapore E4-05-48, 4 Engineering Drive 3 Singapore 117567

L. Sirleto*, G. Coppola, M. Iodice, M. Casalino, M. Gioffrè and I. Rendina National Research Council – Institute for Microelectronics and Microsystems via P. Castellino 111 I-80131 Naples Italy

E-mail: [email protected]


Chapter 2 B.J. Li State Key Laboratory of Optoelectronic Materials and Technologies School of Physics and Engineering Sun Yat-Sen University Guangzhou 510275 China E-mail: [email protected]

Chapter 4 J. Tioh, R.J. Weber* and M. Mina High-speed Systems Engineering Department of Electrical and Computer Engineering Iowa State University Ames, IA 50011 USA E-mail: [email protected]; weber@ iastate.edu; mmina@engineering. iastate.edu

ix © Woodhead Publishing Limited, 2010

1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

x 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X


Chapter 5

Chapter 7

L.L.P. Wong and J.T.W. Yeow Department of Systems Design Engineering University of Waterloo 200 University Avenue West Waterloo, Ontario Canada N2L 3G1

M.P. Fok* and P.R. Prucnal Department of Electrical Engineering Princeton University Engineering Quadrangle, Olden Street Princeton, NJ 08544 USA


E-mail: [email protected]; prucnal@ princeton.edu

A.A. Goldenberg* Department of Mechanical and Industrial Engineering University of Toronto 5 King’s College Road Toronto, Ontario Canada M5S 3G8 E-mail: [email protected]

Chapter 6 A. Assadihaghi, H. Teimoori and T.J. Hall* Centre for Research in Photonics at the University of Ottawa School of Information Technology and Engineering (SITE) 800 King Edward Avenue University of Ottawa Ottawa, Ontario Canada K1N 6N5 E-mail: [email protected]

Chapter 8 C. Vázquez García*, I. Pérez Garcilópez and Pedro Contreras Lallana Grupo de Displays y Aplicaciones Fotónicas Dpto. Tecnologia Electrónica Escuela Politécnica Superior Universidad Carlos III de Madrid Av. Universidad 30 28911 Leganés Madrid Spain E-mail: [email protected]

Bruno Vinouze and Bruno Fracasso Optics Department Telecom Bretagne Brest France E-mail: [email protected]



Chapter 9

Y. Kitagawa Stanley Electric Co. Ltd. 400, Soya Hatano, 257–8555 Japan

K. Asakawa* National Institute for Materials Science 3–13, Sakura Tsukuba 305–0003 Japan E-mail: [email protected]

Y. Sugimoto and N. Ikeda National Institute for Materials Science 1–2–1, Sengen Tsukuba 305–0047 Japan E-mail: [email protected]. jp; [email protected]

Y. Watanabe National Institute for Materials Science 3–13, Sakura Tsukuba 305–0003 Japan E-mail: [email protected]

N. Ozaki Wakayama University 930, Sakaedani Wakayama 640–8510 Japan E-mail: [email protected]

xi

E-mail: yoshinori_kitagawa@stanley. co.jp

S. Ohkouchi and S. Nakamura NEC Corporation 34, Miyukigaoka Tsukuba 305–8501 Japan E-mail: [email protected]; [email protected]

A. Watanabe Meijo University, Tenshiroku Nagoya 468–8502 Japan E-mail: [email protected]

X. Wang National Institute of Advanced Industrial Science and Technology 1–1–1, Umezono Tsukuba 305–8568 Japan E-mail: [email protected]

Y. Takata KYOCERA Corporation 660–10, Shimonocho Ise 516–8510 Japan E-mail: [email protected]


1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

xii 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X


Chapter 10

Chapter 11

Y. Zhang and B.J. Li* State Key Laboratory of Optoelectronic Materials and Technologies School of Physics and Engineering Sun Yat-Sen University Guangzhou 510275 China

B.J. Li State Key Laboratory of Optoelectronic Materials and Technologies School of Physics and Engineering Sun Yat-Sen University Guangzhou 510275 China




1 Introduction to optical switches S.J. CHUA, National University of Singapore, Singapore, and B.J. LI, Sun Yat-Sen University, China Abstract: A number of technologies are used for implementation of optical switches. It ranges from simple mechanical movements to deflect the light beam to using external stimuli to change the optical properties of materials. This chapter summarizes the mechanisms used for implementation of optical switches. Key words: optical switches, thermo-optical switch, magneto-optical switch, micro-electro-mechanical (MEMS), electro-optical switch, liquid crystal optical switch.

Optical communication using semiconductor lasers as sources and optical fiber as the transmission medium is the only solution to handle the massive growth of both telecom and datacom traffic. A single strand of fiber offers a bandwidth of 25 000 GHz, and a cable containing about 1000 optical fibers can carry six billion simultaneous full-screen videophone conversations – one for every person on earth. With the introduction of new services such as high-definition television (HDTV) and grid computing, bandwidth demand is expected to rise. Grid computing provides on-demand access to both local and remote computational resources for storage and visualization and encourages the effective and productive use of expensive resources to simulate scientific, engineering and commercial applications. Three-dimensional (3D) movies, which are now introduced in cinemas, will soon see their emergence in the home entertainment arena. To fully realize the potential bandwidth available on these optical fibers, other components of the optical network system have to be developed, ranging from detectors to multiplexers, buffers and switches to match the transmission rate and bandwidth. This book addresses the different technologies which could be applied to switching optical signals, applications of which depend on the topology of the optical network, the switching fabric and the switching speed required. In general, a switch is concerned with the routing of message information in response to supervisory control signals. The message information could be large blocks of multiplexed traffic in the optical core network or a large number of lower bit channels to be delivered to the users in the optical access network. However, the application of an optical switch may not just be limited to the communication networks but will also be incorporated in the communication cores of a large multi-processor computer where the data rates may exceed 100 Gbit/s. With new schemes being experimented for secure communication and for computing using quantum phenomena, new architecture will be required for switches that do not interrupt the phase information of the quantum packets. 1 © Woodhead Publishing Limited, 2010

1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

2 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

Optical switches

An optical switch functions by selectively switching an optical signal delivered through an optical fiber or an integrated optical circuit to another. Several methods are available and each relies on a different physical mechanism for its operation. The various physical mechanisms used are briefly summarized below, following the order of presentation in each of the chapters in this book. 1 Decrease in refractive index due to the presence of charge carriers in the semiconductor device forming the switch. By injecting charge carriers at a material interface, the refractive index at one side of the interface can be reduced, which can cause total internal reflection (TIR) to take place when a beam travels from a high to a low refractive index media at the interface. Thus, the beam is reflected rather than transmitted across the interface, enabling the beam to be switched. Changes in refractive index of one beam path relative to another cause a phase difference between the beams which can lead to constructive or destructive interference when they arrive at the outputs of the two different arms forming the beam paths. Electro-optical switches make use of this effect in an interferometric device. 2 Change in refractive index with temperature. Refractive index of materials generally decreases with increase in temperature. Thus by incorporating this property change into an interferometric device, for example, switching can be realized. This effect is made use of in thermo-optical switches. 3 Change in polarization of light as it travels through the medium interacting with the magnetic field. The rotation of the plane of polarization, known as the Faraday effect, is proportional to the intensity of the applied magnetic field in the direction of propagation of the light beam. With a polarizer at the output, the beam can be cut off when the rotation causes the plane of polarization to be perpendicular to the transmission axis of the polarizer. This effect is made use of in magneto-optical switches. 4 When a free-space beam is deflected by a micro-mirror, the deflected beam can be made incident at a number of optical fibers by precisely controlling the deflection of the micro-mirror. Such micro-mirrors can be implemented on micro-electro-mechanical (MEMS) systems often implemented by etching the silicon surface into arrays of flat beams and membranes. The movements of micro-mirrors, for example, form the basis of MEMS-based optical switches. 5 As light propagates through an optical gain medium, its wavelength, polarization, phase and amplitude can be changed and the gating function can be performed by putting an element sensitive to the property altered by the amplifier. Such elements can be a grating polarization beam splitter while a Mach–Zehnder interferometer can act upon variations in wavelength, polarization and phase. The integration of the semiconductor optical amplifier with the gating elements forms the basis of semiconductor optical amplifier (SOA) switches. 6 With optical nonlinearities such as the Kerr effect, changes in the refractive index of a material take place in response to an applied electric field. In the


Introduction to optical switches

3

case where the electric field is due to the light itself, it is known as the optical Kerr effect or AC Kerr effect. This causes a variation in index of refraction which is proportional to the local irradiance of the light. This refractive index variation is responsible for the nonlinear optical effects of self-focusing and self-phase modulation. As the beam propagates, it experiences a phase shift due to the change in refractive index that is related to the intensity of the beam itself. Thus by applying a gating element at the output of the medium, a switching action can be implemented. 7 In liquid crystals, the orientation of the rod-like molecules causes the polarization state of a linearly polarized light transmitted through the medium to vary. Thus, if the orientation of the rod-like molecules is continuously varied from the top to the bottom of a layer by 90°, through the application of a voltage across the layer, the state of the linearly polarized light transmitted through the layer will undergo a 90° rotation with respect to that at the input. With the use of a polarizer at the output, the beam is blocked if the output polarization is perpendicular to the polarizer axis, whereas it would be transmitted when the voltage is not applied. Thus, a switching function can be performed when the input signal is distributed over several outputs, for example, and with a polarizer at each end. 8 Photonic crystals are periodic optical nanostructures, typically a hexagonal patterned array of holes in an optical slab waveguide, designed to affect the motion of photons in a similar way that periodicity of a semiconductor crystal affects the motion of electrons. By appropriate choice of hole diameter and period, specific wavelengths of light cannot propagate through the guide. Thus by eliminating a row of these holes, the light can be guided through the regions where there are no holes. By changing the refractive index of the semiconductor where the light is guided, say by a control pulse, phase shift can occur. Such a phase change can be made the basis of a switching action. 9 By physically moving sideways two fibers aligned end-on using a piezoelectric element, switching action can also be performed. When they are perfectly aligned, transmission takes place and the signal can be switched between fibers. 10 Quantum confined stark effect is the change in the quantized energy in a quantum well when an electrical field is applied across the quantum well. This results in a reduction of the transition energy between the lowest quantized energy levels of the hole and electron. The optical absorption of the quantum wells is increased for a designed wavelength with the application of the external voltage. This effect is being made used of in quantum optical switches. Each of the chapters deals with a different principle for the operation of the switch. However, they are considered in greater detail with discussions on the choice of materials, fabrication technique and treatment of the complexity of the switch design in affecting the performance and to satisfy the network topology and the switching speed. The control signal for the switches can be electrical in


1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

4 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

Optical switches

origin, viz. current or voltage, or it can be an external optical signal pulse which changes the physical properties and conditions of the switch material. In the alloptical network, the signal does not undergo any optical-to-electrical-to-optical (OEO) conversion in the transmission path from source to destination. In such a case, optical control can be achieved by the intensity of the signal beam in affecting the nonlinear optical properties, and this class of switches is important for implementation in an all-optical network. For switches, several performance criteria are specified such as switching speed, insertion loss, crosstalk, on/off ratio, power consumption and reliability. Switching speed is defined as the time it takes for the connection to be made for the signal to be transferred from the input to the output ports. It is the time it takes for the output port to see the signal after the control signal has been activated. It is a function of the delay encountered within the switch and, depending on the physical mechanism employed for the switch, can vary from nanoseconds to microseconds. Insertion loss is the amount of power loss in the signal in coupling to the output port. Crosstalk is defined as the ratio of light power in the unwanted output port to the power in the desired output port. The unwanted signal that is leaked out contributes as noise on the unintended output ports. On/off ratio is the ratio of the power in the output port when the switch is on to the power when it is switched off. In the ideal case, when the switch is off, no signal should be transmitted. As many switches are operating millions of times a second in the system, their power consumption is by no means negligible. Thus, it is important to minimize the power required to perform a switching function. Finally, for the switch to be accepted, it has to be reliable and should meet the performance parameters under a wide variety of environmental conditions. Thus, while essentially any physical mechanism can be used for making a switch, it is finally the practicality and features such as physical size, cost and stringent requirements on performance that see the switch being commercially adopted.


2 Electro-optical switches B.J. LI, Sun Yat-Sen University, China Abstract: This chapter introduces the developments of electro-optical switches and mainly focuses on Si-based semiconductor materials because of the very mature fabrication technology. The chapter first discusses theory and principles of single-mode, multi-mode interference, and plasma dispersion effect, followed by materials and fabrication of electro-optical switches. The chapter then discusses eight kinds of electro-optical switches. Finally, a brief discussion on the performance and challenges of electro-optical switches is given. Key words: electro-optical switch, plasma dispersion effect, carrier injection, Si-Ge.

2.1

Introduction

Fiber-optic communication networks are experiencing a continuing increase in demand for telephone, cable TV, digital video, data and internet services. The continuing development of fiber-optic communication networks to accommodate future demands will depend on the availability of cheap, reliable and robust components for routing, switching and detection. Among them, optical switches are essential components for 1310–1550-nm fiber-optic communications and optical networks. They can reduce the cost of the network and increase fiber transmission capacity and at the same time, distribute optical signals to different subscribers. The basic technologies for the design and production of optical switches are now in place, but there is not yet a clear winner in the area of materials. GaAs- or InP-based quaternary compound semiconductor materials are widely employed for optical switches, due mainly to their potential for integration with active devices such as lasers and photo-detectors operating at the fiber-optic windows of 1310–1550 nm. However, its fabrication technique is not compatible with the very mature Si technology and remains complex and expensive. Silica and/or glass on silicon are widely used in integrated optics. They can offer the advantage of larger cross-section waveguides and low losses, but monolithic integration with lasers and photo-detectors is difficult. Polymer-based materials are used in optical devices, but their stability needs to be improved. In this chapter, we will introduce the developments of electro-optical switches and mainly focus on Si-based semiconductor materials because of the very mature fabrication technology.

5 © Woodhead Publishing Limited, 2010

1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

6

Optical switches

2.2

Theory and principles of electro-optical switches

2.2.1 Single-mode principle Ridge waveguide is a fundamental structure for constructing optical switches. To connect a switch with a single-mode optical fiber with a core diameter of about 9 to 10 µm, the first step of the design is to get a large-scale single-mode ridge waveguide as shown in Figure 2.1. The design criteria for the single-mode ridge waveguide are as follows: (1) The numerical aperture of ridge waveguide must be matched with that of single-mode fiber (0.2–0.3); (2) The ridge waveguide must have a large cross-section, and the ridge height and ridge width should be equal to the core diameter of the singlemode fiber (9–10 µm); (3) The ridge waveguide must support single-mode. As a design example, we use Si-based materials. Figure 2.2 shows the crosssection of a ridge waveguide. It was formed by a Si-Ge layer with a refractive index of n1 grown on (100) Si substrate. The ridge width is w = 2aλ, the inner ridge height is h = 2bλ, and the etched depth of the ridge is h´ = 2b(12r)λ. In these expressions, λ is the free-space optical wavelength, and r is the fractional height of the side regions compared to the ridge center (the out-inner ratio). The three dielectric materials have refractive indices of n0, n1, and n2, respectively, at the wavelength of interest. To propagate the single-mode light in the input and the output waveguides, the lateral dimensions and the transverse dimensions must satisfy the single-mode ridge waveguide condition. Based on the single-mode ridge waveguide condition, the ratio a/b should be1

n0 n1 n2 n0

n1 n2

2.1 Cross-section view of a ridge waveguide.



7

w = 2aλ h ' = 2b(1–r)λ

n0

h = 2bλ

n1 Si-substrate

n2

2.2 Cross-section view of SiGe/Si ridge waveguide.

[2.1] [2.2] where γ0,2 = 1 for HE mode, γ0,2 = (n0,2/n1)2 for EH mode, a, b, r, and λ are the ridge width factor, ridge height factor, etching depth factor, and wavelength in free space, respectively, and n0, n1, and n2 are the refractive index of cladding layer, guiding wave layer, and substrate, respectively. It is well established that the pseudomorphic, dislocation-free SiGe alloy layers can be grown on a Si substrate, provided that their thickness is less than a critical thickness hc. This thickness is defined as the thickness above which the misfit dislocations are generated. It also depends remarkably on the Ge fraction x. For SiGe layers grown on Si (100) substrates, the function hc(x) can be expressed by2

[2.3]

where a(x) ≈ 0.554 nm is the mean bulk lattice constant of SiGe, b is the Burger’s vector modulus, ν is the Poisson’s ratio, and fm(x) is the substrate-alloy misfit parameter. For SiGe alloys, the lattice constant can approximately be related to the lattice constant of Si and Ge according to Vegard’s rule for a (100) Si substrate with fm(x) = 0.04117x. To reduce the misfit dislocation of SiGe-Si interface, the thickness of strained SiGe alloy layer, which is grown on the Si (100) substrate, must be less than hc. For the SiGe optical waveguide, the Ge content must be less than 15%. According to the calculation, the optimum Ge content is x = 0.03 to 0.05 (Figure 2.3). Here we choose x = 0.04. According to Eq. (2.3), for x = 0.04, the hc = 6.5 µm. Therefore, the thickness of the ridge waveguide was chosen to be 2.5 µm. Figure 2.4 shows the


1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

8

0.5

0.4

0.3

NA

1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

Optical switches

0.2

0.1

0.0 0.00

0.02

0.04

0.06

0.08

0.10

Ge content, x 2.3 Numerical aperture of a SiGe/Si waveguide with Ge mole fraction, x.

critical value of the ratio a/b as a function of b for λ = 1.3 and 1.55 µm, respectively, using the factors r = 0.5 and 0.8 as parameters3 using Eq. (2.1). In Figure 2.4, we observe that the HE 00 and EH 00 modes are essentially identical. To simplify the design, only the HE 00 mode is discussed here. For a thickness of 2.5 µm and r = 0.5, an etched depth of 1.2 µm and a width of 3 µm were chosen, respectively.

2.2.2 Multi-mode interference principle From the viewpoint of integration, a small size is desirable. The main advantages of optical devices based on the multi-mode interference (MMI) effect are low loss, compact size, and large fabrication tolerances. They are quite easy to design and fabricate. Because of the excellent properties of MMI devices, optical switches were demonstrated based on MMI effect. The operation of the optical MMI switch is based on the self-imaging principle. Self-imaging is a property of a multi-mode waveguide by which an input field is reproduced in single or multiple images at periodic intervals along the propagation direction of the waveguide. In the MMI switch, a multi-mode waveguide is designed to support a large number of modes. According to the self-imaging theory4, when an input light beam is coupled into the multi-mode waveguide from a single-mode input waveguide, the input optical field will be reproduced in single or multiple images at periodic intervals along the propagation direction. In general, taking the Goos-Hähnchen shifts into account, an effective width We of the multi-mode waveguide can be expressed as:



9 [2.4]

where σ = 0 for the TE mode and σ = 1 for the TM mode, λ0 is a free-space wavelength; nr and nc are effective refractive indices of ridge waveguide and cladding layer, respectively; WM is the width of the multi-mode waveguide. (a)

8

Rib width/Rib height, a/b

7

HE 00

l = 1.3 mm

6

EH00

Ge0.04 Si0.96 /Si

5 4

2

r = 0.5

1 0

Multi-mode region

r = 0.8

3

single-mode region 0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

Rib height factor, b (b)

8

Rib width/Rib height, a/b

7

HE 00

l = 1.55 mm

6

EH00

Ge0.04 Si0.96 /Si

5 4 r = 0.8

3 2

r = 0.5

1 0

Multi-mode region

single-mode region 0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

Rib height factor, b 2.4 Critical a/b versus b for SiGe ridge waveguide shown in Fig. 2: (a) λ = 1.3 µm and (b) λ = 1.55 µm.


4.0

1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

Optical switches

By defining Lπ as the beat length of the two lowest-order modes:

[2.5]

where β0 and β1 are the propagation constants of the fundamental and the first-order lateral modes, respectively. When the width WM and length L of the multi-mode waveguide satisfy the condition: L = p (3Lπ) p = 0, 1, 2, …

[2.6]

the input light field will be repeated and single image can be obtained. For the multi-mode waveguide with a length of z = 3Lπ, according to the partial index-modulation principle for the multi-mode waveguide5, when a π phase shift is introduced at around the position z = 3L π /2, a transformation between the even and odd modes will take place. Thus after a further propagation of z = 3L π /2, the input optical signal will be outputted from another corresponding output waveguide.

2.2.3 Plasma dispersion effect The principle of the plasma dispersion effect is that the refractive index of materials is related to its carrier concentration6. As an example, for SiGe/Si materials, when the Ge composition is low (x < 20%), the free-carrier plasma dispersion effect in Si12xGex will lead to a variation of refractive index, which can be described by the relation as follows: ∆n = – (q2λ2/8π2c2nε0)·[(∆Ne/mce*) + (∆Nh/mch*)]

[2.7]

where q is the electron charge, ε0 is the permittivity of free space, n is the refractive index of Si12xGex, λ is the wavelength, c is the light velocity, ∆Ne and ∆Nh are the concentration changes of electrons and holes, respectively, mce* and mch* are the conductivity effective masses of electrons and holes of Si1-xGex and could be given by

2.3

mce* = mce*(Si) (12x) + mce*(Ge) x

[2.8]

m* ch = m* ch(Si) (12x) + m* ch(Ge) x

[2.9]

Materials and fabrication of electro-optical switches

Materials that can be used to fabricate optical waveguide switches are LiNbO3, III-V compound conductors, polymers, and Si-based materials. To achieve good reliability and highly monolithic integration with Si-based chips, SiGe material was used to fabricate multi-functional photonic devices because of low propagation loss (< 0.5 dB/cm) in the wavelength region of λ = 1.3–1.55 µm. SiGe epitaxy has



11

the advantage that its fabrication techniques are compatible with large-scale Si integration. Due to greater electron mobility and smaller bandgap in germanium, remarkable improvements in the performance of devices can be achieved, with virtually no changes in existing all-silicon designs. SiGe is also relatively easy to fabricate using existing silicon facilities. For Si-based materials, they were usually grown by molecular beam epitaxy (MBE) or UHV-chemical vapor deposition (CVD). By using an MBE, two waveguide switches are described as follows. One contains a lateral p+-n junction (Fig. 2.5) and the other contains a vertical p+-n junction (Fig. 2.6). p-SiGe

p-SiGe

p-Si substrate

p-Si substrate

(a)

(b)

SiO2

SiO2 p-SiGe

n+

p-Si substrate

p-Si substrate

(c)

(d)

SiO2

SiO2

Al electrode

+ p

n+

p-Si substrate

+ p

n+

p-Si substrate

(e)

(f)

2.5 Fabrication procedure: (a) SiGe material growth; (b) dry etching to form ridge waveguide; (c) deposition of SiO2; (d) phosphorus ion implantation to from n+ carrier injection regions; (e) boron ion implantation to form p+ collector; and (f) sputtering deposition of aluminum films for ohmic contacts. SiO2

32.2 μm

5 nm, n+-Si cap, 1 × 1018 cm–3

1.0 μm 2.5 μm

p-SiGe, 2 × 1016 cm–3

5 nm, p-Si cladding, 2 × 1016 cm–3 p+-Si sub, 2 × 1018 cm–3

2.6 A schematic diagram of a waveguide cross-section view of an optical switch.


1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

12 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

Optical switches

2.3.1 Fabrication of a lateral p+-n junction switch The sample was made in the stages as shown in Figure 2.5. First, SiGe materials were grown by MBE. A 50-nm p-type Si buffer and a lightly doped p-type SiGe (Ge content at ~4%) layer with the thickness of 2.6 µm was grown by molecular beam epitaxy on a p-Si(100) substrate. The boron doped-concentrations in the buffer layer and waveguide layer are approx. 5 × 1016 cm23. The substrate temperature during the growth was kept at 600°C. Second stage is the fabrication of the device. The device was fabricated in a 3-µm manufacturing process line. The ridge waveguides were formed by reactive ion etching (RIE) technique with etching rate of 250–300 Å min21. A 550-nm thickness SiO2 film was deposited at 400°C by plasma-enhanced chemical vapor deposition (PECVD) on the top of the sample surface to serve as the ion implantation mask and the surface passivation layer. The n+ carrier injection regions of the switch were realized using phosphorus ion implantation with energy of 60 keV and a dose of 5 × 1015 cm22 (Fig. 2.5d). The p+ collector was formed by boron ion implantation with energy of 80 keV and the same dose. The n+ and p+ ohmic contacts were formed by sputter deposition of aluminum films with thickness of 2.0 µm and followed by alloying at 440°C for 30 min in N2 ambient. The last stage is end polishing. The wafer was cut and the input and output facets of the waveguides were polished by a mechanical method in order to couple the incident light from a single-mode optical fiber.

2.3.2 Fabrication of a vertical p+-n junction switch A 5-nm p-type Si lower cladding layer with a concentration of about 2 × 1016 cm23 was grown by UHV-CVD (900°C) on a p+-type Si(100) substrate. This is followed by the growth of a 2.5-µm p-type strained Si0.96Ge0.04 core waveguide layer with a concentration of 2 × 1016 cm23 (Figure 2.6). On the top of the sample, an abrupt n+-p junction was formed by growing a 5-nm n+-type Si cap layer with a concentration of 1 × 1018 cm23. The whole device was fabricated using siliconoptical bench technology. Waveguides were developed based on lithography and plasma etching of silicon. The waveguide with an abrupt n+-p junction was formed by removing other n+-Si cap layer. The device surface was next passivated by a SiO2 layer. The n+ and p+ ohmic contact electrodes were deposited by evaporating 1.0-µm thick aluminum layer followed by alloying at 450°C. The chips were diced and the input/output facets of the switch were mechanically polished in order to have easy coupling of the incident light from a single-mode fiber.

2.4

Device structures of electro-optical switches

2.4.1 1 × 1 switch 1 × 1 optical switch is usually a 1 × 1 optical modulator. It can be fabricated in III–Vs materials, Si(Ge) materials, LiNbO3, or polymers. Figure 2.7 shows a 1 × 1



13

p+

n+

n+

n-Si Si-sub.

SiO2

2.7 Schematic diagram of 1 × 1 optical switch/modulator in SOI.

optical switch fabricated in silicon-on-insulator (SOI)7. A single-mode rib waveguide with a width of W is composed of an n-type silicon guided-wave layer on a SiO2 layer. An abrupt p+-n junction is formed below the top surface of the ridge waveguide to inject the carriers into the waveguide. If the p+-n junction of the waveguide is forward-biased when guided-mode optical signals are end-coupled into the rib waveguide, a large number of carriers will be injected into the guided-wave layer of the waveguide, and the refractive index in the waveguide will decrease because of the plasma dispersion effect, which can make the guided-mode convert into the radiation mode of the substrate and the cover. This causes a lot of guided-mode energy to be lost and absorbed in the rib waveguide, which will cause the rib waveguide to cut off, resulting in the so-called waveguide ‘vanishing.’ Consequently, there is no output light in the waveguide, and thus switching is achieved. For the fabrication, SOI material is used. This is produced by cleaning followed by oxidizing the substrate wafer (the SiO2 is 400–500 nm thick), bonding at a high temperature (in O2, at 1200°C, for 2 h), and thinning (grinding precisely rear face down to 20 µm), and then, polishing to a thickness of 6 µm with r.m.s. roughness > 2

[8.3]

where d is the cell thickness, ∆n is the LC birefringence and λ is the light wavelength. LC molecules reorient if voltages greater than the threshold voltage, Vth, are applied to the electrodes. When a sufficient voltage value, named switching voltage Vsw > Vth, is applied (ON state), LC molecules align parallel to the electric field and the polarization rotation disappears. In most display applications, two crossed polarizers are placed on the outside of the substrates, with the transmissive axis of each polarizer parallel to the rubbing direction of each alignment layer. The basic operation of a TN device working in this mode, known as normally white (NW) mode, is roughly depicted in Fig. 8.4(a). When no voltage is applied


Liquid crystal optical switches (a) LC between crossed polarizers

211

(b) LC between parallel polarizers

Input light Polarizer

LC cell

Analyzer Output

8.4 Representation of a TN cell operation: (a) crossed polarizers and (b) parallel polarizers.

(OFF state) the incident light is transmitted. In the ON state, the light is blocked by the output polarizer and the device appears dark. From an electrical point of view, a LC cell acts, basically, as a capacitor with a non-ideal dielectric material (the LC). The electrical equivalent circuit (EEC) of the LC device can be obtained using an experimental procedure based on the impedance spectroscopy technique (Barsoukov and Macdonald, 2005). This method consists in measuring the complex impedance (magnitude and phase) of the device and fitting the EEC components from these impedance measurements (Pena et al., 2002; Pérez et al., 2007). Results for a TN device, filled with LC mixture E7 from Merck (Darmstadt, Germany), with an active area of 1 cm2 and a thickness of 5 µm, are summarized in Fig. 8.5. The EEC consists in a

–20

1E+005

–40

1E+004

–60 –80

1E+003

TN cell

RS

RP CLC

–100 1E+002 1E+002 1E+003 1E+004 1E+005 1E+006 Frequency (Hz)

(a)

4

RS = 120 W RP = 0.95 MW

Capacitance (nF)

Impedance magnitude Impedance phase

Impedance phase (º)

Impedance magnitude (W)

1E+006

3 2 1 Vsw

Vth

CLC = 0.8 nF

0

(b)

1

2 3 4 Applied voltage (Vrms)

(c)

8.5 (a) Impedance (magnitude and phase) measurement of TN cell in the OFF state, from 100 Hz to 1 MHz, (b) EEC in the OFF state, and (c) capacitance variation as a function of applied voltage (threshold and switching voltages can be estimated from this).


5

212

Optical switches

voltage-dependent capacitor (CLC ) with series (RS) and parallel (RP) resistors. The variation of CLC with the voltage applied to the cell electrodes is due to the dielectric permittivity modifications linked to the molecular reorientations. Therefore, the threshold (usually 1–2 V) and switching (usually 3–5 V) voltages can be derived from this electrical modeling. At low voltages (below Vth) the capacitance is constant. A nonlinear variation of CLC is obtained if the voltage is increased. Finally, the device capacitance remains almost constant for voltages greater than Vsw. Power consumption (nW) and response time (usually in the 20–30 ms range) of the TN device can be also estimated from EEC simulation. The optical transmission factor, T, for polarized monochromatic light incident on a 90° TN cell sandwiched between parallel polarizers in the ON state (see Fig. 8.4(b)) can be calculated using Jones matrices (Gooch and Tarry, 1975):

sin 2 i 1+ u 2 2 T = To 1+ u 2

(

)

[8.4]

where To is ideal maximum transmission and u = 2∆nd/λ. – –– The minimum transmission occurs when u = √3, √15 … Since a smaller u gives – a smaller cell gap and a faster response speed, usually u = √3 is used in practical devices (Yang and Wu, 2006). Note that contrast is maximized for a λ value and optical transmission varies for other wavelengths. Wavelength dependence must be reduced by optimizing the fabrication parameters. Surface-stabilized ferroelectric liquid crystal (SSFLC) cells SSFLC cells are the most widely used devices based on FLC. In these devices, the FLC material is sandwiched between two substrates separated by a very thin LC layer (1–2 µm). Although a variety of molecular orientations have been applied in SSFLC (Lagerwall, 1999), the bi-stable bookshelf layer structure is the most employed (see Fig. 8.6). This material exhibits a macroscopic spontaneous polarization Ps, which must be stabilized by the surfaces in order to prevent the natural helix formation (Clark and Lagerwall, 1980). The electro-optical effect is a rotation of the smectic cone driven by coupling between the polarization and the electric field. This structure has response time of a few microseconds as well as a memory effect (bi-stability). In the presence of an electric field, the molecular orientation changes and the device remains in this state until a reverse polarity voltage is applied. An electrical modeling (Moore and Travis, 1999; Rep and Prins, 1999) of these devices allows obtaining switching voltage (a few volts), power consumption (nW) and response time (a few microseconds) in practical devices, as a function of fabrication parameters. When an SSFLC device is placed between crossed polarizers, with one of them parallel to the molecular axis of one of the stables states, one of the two states will



213

P –V

θ

θ

+V P

8.6 Bi-stable bookshelf SSFLC cell.

be black. Optical transmission for the other state can be calculated, using the Jones calculus, and is given by: [8.5] T = To sin 2 ( 4 ) i sin 2 nd where To is the maximum possible transmission in the device (i.e., transmission between parallel polarizers), θ is the cone angle of SmC* material (optimal value is 22.5°) and d is the thickness of the device (d = λ/2∆n for optimal working). As happens in TN cells, the device performance is optimal for a given λ, optical transmission for other wavelengths changes abruptly compared with TN cells. TN and SSFLC cells are just some of the possible configurations used in developing optical switches, other examples such as using PDLCs will be shown in section 8.3. Liquid crystal spatial light modulators (SLMs) A SLM is a device that modulates in one or two dimensions an optical beam in amplitude, phase or polarization, using the birefringence properties of the LC cell. Most two-dimensional (2D) LC SLMs are driven electrically; but optically addressed analog light valves are also proposed (Moddel et al., 1989). The spatial structure of an electrically addressed SLM is shown in Fig. 8.7. The pixel pitch p is defined as the center-to-center spacing between adjacent pixels. The interpixel gap i describes the edge-to-edge spacing between adjacent pixels. Assuming square-shaped pixels, the geometrical fill factor F is defined as the ratio (p/i)2 and this parameter puts an upper bound to the SLM optical efficiency. Most widespread LC SLMs are transmissive panels that consist of a LC layer aligned between two glass sheets, with a control circuitry added using the thin film transistor (TFT) technology. Such displays are rather large and used for laptop computers, TV sets and head-up displays. Major drawbacks are: rather large pixels, a moderate ‘fill factor’ (95%). The main applications for LCOS SLMs include high-definition rear-projection TV sets, portable video-projectors, wavefront control, adaptive optics, beam-steering for optical tweezers (Hossack et al., 2003), optical switching matrices or wavelengthselective switches (WSS) (Baxter et al., 2006). In that case, it is necessary to adapt the LCOS cell characteristics to optical fibers transmission constraints (Heggarty et al., 2003).

8.3

Liquid crystal switches and applications

Different types of switches can be distinguished, depending on the physical mechanisms used to steer the light with LCs, such as reflection, wave-guiding, polarization management or beam-steering (planar or volume). Some of them will be summarized in the following sections, with special emphasis on their switching parameters and applications.

8.3.1 Optical crystal switching architectures Broadly speaking, optical space-switches can be implemented according to two basic architectures: broadcast-and-select (BS) and space-routing (SR). Using BS, the light information from an input channel is split to the entire output channel array through an intermediate blocking stage, which then selects the desired output channel (Fig. 8.9). Using LC modulators, the blocking stage is implemented by amplitude modulators that must exhibit high contrast ratio to avoid crosstalk. Although strictly non-blocking, this scheme suffers from: (1) a large complexity of the intermediate stage when the number of channel increases (considering a 2D input array involves a 4D selecting mask) and (2) a very poor power budget which means that optical amplification is mandatory in this case. The SR scheme (Fig. 8.10) is generally more adapted to LC switching. It consists in driving the information from an input to an output channel using SR intermediate elements, usually arranged in different stages. This scheme limits the IL, but may be subject to connection blockings if the number of switching elements is limited, as for the Benes or Banyan topologies (Yu et al., 2006). Using LC devices to implement SR switches by individual light steering can be performed in two ways: a multi-stage planar topology using arrays of 2 × 2 binary polarization switches (Hogari et al., 1991) or single- and dual-stage schemes in which SR is performed by beam-steering in free-space (Fukushima et al., 1991). This will be illustrated in the next subsections (8.3.2–8.3.4).


216

Optical switches Broadcast

On

Combine

Off A

A

Off Off

B

Off

C

On

C

Off

B

On Input channels

Off

Output channels

Select

8.9 Broadcast-and-select switching architecture. Connecting N inputs to N outputs requires 2N couplers and N 2 ON/OFF selectors.

8.3.2 Switches based on polarization management The polarization rotation (PolRot) is the first configuration used in LC switches (Wagner and Cheng, 1980) because the basic principles are widely used in the TN displays for tens of years. This switch is based on the change of polarization state of the incident light when applying an electric field over the LC cell (see Fig. 8.4). The change of polarization with a TN cell in combination with spatial polarization selective calcite crystals or polarization beam splitters (PBS) allows optical space-switching. In order to make the device polarization insensitive and to minimize losses, the polarization diversity method is used by treating each polarization mode in parallel. The input signal is decomposed into its TE and TM components, which are separately recombined at the switch output. A schematic of this generic polarization management LC optical switch can be seen in Fig. 8.11. A simple example of the previous description is the 1 × 2 LC switch structure shown in Fig. 8.12. The principle of operation is as follows (McAdams and



217

2×2 switch 1 Bar state Input 2 channels Cross state

3

2 Output channels

1

3

8.10 Space-routing switching architecture, illustrated here by the so-called ‘Crossbar’ scheme (strictly non-blocking). Connecting N inputs to N outputs requires N 2 elementary 2 × 2 switching points.

LC

Inputs

Polarization separation

Polarization handling

Recombination

Outputs

8.11 Block diagram of a generic polarization management LC switch.


Optical switches Port 3

POF

218

L3 s

PS L1 Port 1

Port 2

s

POF

POF p+s

s P–s

p

p

L2

PBS

8.12 Scheme of a 1 × 2 LC optical switch. POF: polymer optical fibers; PS: TN-LC cell; PBS: polarization beam splitter; P: polarizer; L: focusing/collimating lens.

Goodman, 1990; Vázquez et al., 2003): the first polarizer, P, at the input manipulates the polarization states to the desired one at the output. The liquid crystal, denoted as polarization switch PS, can shift the polarization of the input light depending on the voltage applied to it. The linear polarization of the input light is shifted 90° after passing through the TN-LC cell when no bias voltage is applied. This light passes though the PBS to output port 2. By applying a voltage greater than Vsw to the TN-LC cell, LC molecules align parallel to the electric field, and the polarization rotation does not occur. This light in the ON state of the switch is reflected by the PBS to the other output or port 3. Switch transmission is controlled by the voltage applied to the LC cells; lower voltages induce less polarization shifts. Then, these switches can also operate as variable optical attenuators (VOAs). By applying a voltage Vth < V < Vsw a VOA can be implemented, splitting the input signal at both outputs with a variable ratio depending on the applied voltage and consequently on the manipulation of the stage of light polarization by the LC. Performance of polarization management based liquid crystal switches There are different configurations based on the previous basic principles depending on the type and the number of elements used and the switch capacity. Evolution of the state-of-the-art showing different implementations and their characteristics are shown in Table 8.1. In these switches, in addition to LC cells, polarizing beam splitters and calcite plates, optional elements are: mirrors, half wave plates, quarter wave plates, halfangle prisms, right-angle prisms, beam displacement prisms, total internal



TN–LC

(Wagner and Cheng, 1980) (Soref, 1981)

(Riza and Yuan, 1998) (Riza and Yuan, 1999) (Vázquez et al., 2003) (Riza and Madamopoulus, 2005) (Lallana et al., 2006)

1550

650–850

—

650–850

NLC

TN–LC

NLC

2×2

1300

FLC

6×6

820(670)

3×1

—

1×2

2×2

2×2

1×4

633

1300

2×2

—

2×2 2×2

632.8

633

1×2

Type

633

λ (nm)

FLC

(Soref and TN–LC McMahon, 1982) (McAdams et al., NLC–FLC 1990) (McAdams and FLC Goodman, 1990) (Grimes et al., FLC 1991) (Fujii, 1993) TN–LC

TN–LC

LC cell

Contribution

MM

—

MM

SM

SM

SM

MM

—

—

None

MM

MM

Fiber

223

220

222

240

234.1

243.3

221.6

220

232

227

220

CT (dB)

3

2

7

6.76

6.94

2.2

11.1

3.5

1.4

3

20–5 ms

—

ms

35.3µs

35.3µs

—

150µs

50µs

250µs

—

50/150ms

—

0.41 2.5

Turn–on/D. time

IL (dB)

Table 8.1 State-of-the-art and performance parameters of RotPol LC switches.

3V

—

8V

—

—

—

—

—

15Vrms

6V

5V

2.5V

Control voltage

2FO–Circulator, 2 PBS, 2 LC, 2 TIR, 2 BDP 2 PBS, 4 L, 6LC, 1 P

4PBS, 2M, 4LC, 2AP, 2HWP, 2QWP, 1LB 1PBS, 2LC, 2P, 1M, 1HWP, 1QWP, 1AP 1PBS, 1LC, 1P

2 PBS, 2 AP, 5 LC, 2 BR

6FLC, 6 GL

2 LC, 2 HWP, 3 Calcites 2 NLC 2 SS–FLC, 2 M, 4 HIEP 4 FLC, 4 PBS

4PBS, 2 LLC, 7 AP

2PBS, 2AP, 1LLC

Elements


AP: BDP: BR: FLC: GL: HWP: HIEP: LB: LLC: M: MM: NLC: P: PBS: QWP: SM: TIR: TN–LC:

1 neglecting

650–850

808

NLC

NLC/FLC

2×2

Dual 3 × 2 —

MM 236.2

220 2.5

reflections, expected up to 1.2dB with MM fibers and GRIN–rod lenses Right–Angle Prism Beam Displacing Prism Birefringent Crystal Ferroelectric Liquid Crystal GRIN Lens Half Wave Plate High Index Equilateral Prism Leakage Block Large LC Cells Mirror MultiMode Nematic Liquid Crystal Polariser Polarizing Beam Splitter Quarter Wave Plate SingleMode Total Internal Reflection Prism Twist Nematic Liquid Crystal

(Lallana et al., 2007) (Yang et al., 2008) 60.6µs 35 µs

13,5ms ±15transient ±5V hold

5V 4 PBS, 2 HWP, 4 QWP, 4 M

3 PBS, 8 L, 6LC


221

reflection prisms, birefringent crystals (see Table 8.1). Most of these are based on free-space optics bulk elements, using lenses for coupling light in optical fibers and only a few of them use fiber-optic devices (Sumriddetchkajorn and Riza, 2000). Better response times are obtained when FLC cells are used, near 35 µs (Riza and Yuan, 1998; Yang et al., 2008). Standard NLC cells exhibit higher response times, not less than 20 ms (Lallana et al., 2006). A reduced response time can be obtained in combining NLC cells and the transient nematic effect (TNE); 60 µs is reported in Yang et al. (2008). NLC cells have a slower response time than FLC, but NLC can operate in a wider wavelength range, because the FLC cell thickness, d, determines the wavelength in which the polarization shift is 90°. On the other hand, NLC cell thickness at first or second minimum can be optimized for a multiband operation fulfilling Mauguin’s regime in order to obtain the polarization switch. A broadband 3 × 1 reconfigurable optical multiplexer, from 650 to 850 nm, is presented in Lallana et al. (2006). With such TNLC, a 3 × 2 multiplexer based on graded index plastic optical fibers (GI-POF) can operate from 850 to 1300 nm (Lallana et al., 2007). Other TNLC systems give good uniformity in the C-band (1530–1560 nm) (Sumriddetchkajorn and Riza, 2000). Switch IL depends on the structure of the device. Simpler switches can manage only one polarization; thus, higher ILs are expected, 3.5 dB in McAdams and Goodman (1990). More complex switches using polarization diversity management (see Fig. 8.11) exhibit low IL, 1.4 dB (McAdams et al., 1990). Optical switches with IL less than 1 dB, crosstalk from 220 to 245 dB, low PDL (around 0.1–0.2 dB) and low PMD have been developed (see Table 8.1). Although not specifically reported in the papers, LC switches have very low power consumption, in nWs.

8.3.3 Liquid crystal materials for amplitude and phase modulation An engineer designing a LC switch aims at specific electro-optical performances (contrast, diffraction efficiency, transmission, voltages, speed) compatible with transmission parameter requirements (IL, crosstalk, PDL, PMD, response time, temperature dependence, power consumption, bit rate tolerance). These perform ances are linked to the molecules’ chemical properties designed by chemists (refractive indexes, elastic constants, dielectric anisotropy and temperature range). LC manufacturers continue to synthesize hundreds of new molecules for displays or specific applications. The relatively slow speed of the LC switches, as already mentioned, could, however, limit their practical application in telecoms. The critical performances of the PolRot cells are contrast and bandwidth. The TN contrast is determined by the minima of Gooch and Tarry’s law (Gooch and Tarry, 1975) as previously reported in equation [8.4]. The first minimum solution


222

Optical switches 1000000 100000

Contrast

10000 1000 100 10 1

3.5

3.6

3.7

3.8

3.9

4

4.1

LC thickness (mm)

8.13 LC cell contrast optimization for 1.5 µm wavelength (0.25 ∆n LC need a 3.8 ± .1 µm thickness for reaching 30 dB contrast in all the C band).

– is 2∆nd/λ = √3. The LC thickness management is demanding, a 3.8 µm ± .1 µm cell thickness for a 30 dB contrast at 1.5 µm wavelength (Fig. 8.13). Then, the 40 nm bandwidth is fully compatible with the C band for telecom applications. So, a clean cell assembly technology combined with the wide catalogue of nematic mixtures covers all the needs of PolRot switches (Pain et al., 1997). An interesting approach consists of using a PDLC as a variable switch between optical fibers (Lallana et al., 2008). The PDLC is composed of a polymer matrix with many LC droplets having a radius of the same size as the wavelength (Fig. 8.14(b)). Inside each droplet, the nematic is uniformly aligned; but from droplet to droplet, the nematic directors are randomly aligned and polarization is independent. Without voltage, the structure scatters light; the switch is in the OFF state. When voltage is applied, nematic molecules align parallel to the electric field, and the structure becomes transparent because the refractive index of the polymer is close to the LC refractive index (n0); so the switch is in the ON state. A common PDLC mixture uses 80% by weight TL205 LC with 20% of PN393 monomer (from Merck). The contrast adaptation to the wavelength is performed by adjusting the size of the LC droplets by: ∆n.a/λ = 0.3

[8.6]

where a is the droplet radius (Bosc et al., 1996). During the UV polymerization of the monomer, a higher power leads to smaller droplet radius. For the phase modulation used in beam deflection gratings, the main issues are diffraction efficiency and response time. Phase modulation can be easily realized with nematic LC in a parallel alignment configuration with a 5° pre-tilt standard



(a) parallel nematic

223

(b) polymer dispersed LC

8.14 Other LC structures suitable for phase or amplitude switching: (a) parallel alignment nematic LC and (b) PDLC.

polyimide material (Fig. 8.14(a)). The parallel structure must be adjusted to reach the desired phase shift given by: ∆φ = 2π ∆n d/λ

[8.7]

A 2π phase shift at 1.5 µm wavelength is achieved with a 7.5 µm LC cell with a 0.2 ∆n. The diffraction efficiency reaches 40%, the theoretical limit of a twophase level grating. A reflective configuration allows dividing the response time by 4, because it is proportional to the viscosity, inversely proportional to the dielectric anisotropy and the square of the electric field. Note that LC birefringence – whatever the material – decreases with increasing wavelength according to Cauchy law nλ = n∞ + b/λ2 (b, specific for each mixture). For telecom wavelengths, the birefringence ∆n typically decreases around 20% with respect to the visible range value. The LC ∆n must be as high as possible (as an example, 0.29 for BL009 from Merck). The elastic splay constant (K11) must be as low as possible. Response time is relatively short (typically 20 ms) due to saturation voltages used there. FLC devices also display bi-level amplitude or phase gratings (Fig. 8.6). The main advantage is a symmetrical and fast response time (typically 10–100 µs). Because of a ∆n lower than 0.15, the FLC (like SCE13 from Merck) usually exhibits a diffraction efficiency in the range of 20%. An unstable LC alignment with time is observed due to thick cell for telecom wavelengths. Special fiber arrays with non-uniform pitch to prevent the coupling of the parasitic diffraction orders were proposed both in one and two dimensions (Fracasso et al., 2001; Letort et al., 2008). The gray scale capability of nematic LCs allows producing multi-phase gratings which lead to high diffraction efficiencies (Klaus et al., 1996). For a blazed grating (saw-tooth profile) with a number of phase levels N, we have a theoretical efficiency (Goodman and Silvestri, 1970), given by:

η (N) = [sinc(1/N)]2


[8.8]

224

Optical switches

with sinc(x) = sin(πx)/πx. This gives η = 81% for four phase levels and η = 91% with N = 6. As a consequence, numerous gray levels are needed to achieve the smooth angle adaptations at each new routing configuration as also seen on DIMOS simulation (Fig. 8.15). Experimentally measured diffraction efficiency reaches 80% for six gray shades at 1.55 µm wavelength (Wolffer et al., 2000). The initial long response times can be reduced by a factor of 100 by using an over driving addressing scheme down to 50 ms rise time and 2 ms decay time (Tan et al., 2000). The reflective configuration in LCOS divides the rise time by a factor of 4. NLCs are easily implemented in LCOS due to their stable planar alignment (Lelah et al., 2001).

8.15 Simulation of NLC blazed grating with multi-electrode structure (DIMOS simulation).

To summarize, LC materials allow making amplitude as well as phase modulation switches. Ferroelectrics could be used in binary gratings, being potentially fast. Nematics are mainly used for PolRot switches and high diffraction efficiency blazed gratings. The wide range of nematic LCs is suitable to design devices from visible to IR wavelengths. Nevertheless, this technology, based on organic materials, still has to be accepted by telecom professionals who are used to manipulate mineral materials and components.

8.3.4 Switches based on beam deflection To overcome the capacity bottleneck inherent in planar switching architectures, free-space optics using programmable 2D beam-steering elements appear as the unique solution to implement large-capacity optical switches, with strictly nonblocking capabilities. The main application aimed here is transparent switching cores of OXCs for WDM networks (Smith et al., 1993). Two generic switching architectures between single-mode fiber arrays are the single-deflector 1 × N



225

correlation setup (O’Brien et al., 1994) and the N × N dual-deflector scheme (Wolffer et al., 2000), depicted in Fig. 8.16. The light from a 2D fiber array is collimated by a 2D micro-lens array, and each collimated beamlet is locally deflected and coupled towards the proper output using a second deflector/lens combination. Although such large capacities as 256 × 256 channels with 1 dB average IL are obtained using reflective 3D micro-mirror arrays (Neilson et al., 2004), the liquid crystal SLM technology used in the refractive or diffractive type shows promising advantages such as: (1) no mechanical motion and hence high angle repeatability and stability, (2) low controlling voltages (a few volts) and (3) high reliability. The first option is refractive LC elements in the form of scanning Fresnel lenses (Sato et al., 1985) or LC micro-prism arrays (Hirabayashi et al., 1995). In the latter case, the steering structure is a homogeneously aligned nematic LC cell in which a micro-prism array (250 µm pitch) is deposited on one glass electrode. The local LC refractive index changes continuously with the voltage applied and an incident optical beam is refracted (and hence deflected) according to SnellDescartes’s law. High transmittance (95%) and high deflection angle are obtained at low driving voltage (2.8 V), but the steering is one-dimensional, the response time is very slow (~1 s) and the device is polarization dependent. 2D beamsteering could be obtained by crossing such 1D structures, to the detriment, however, of steering efficiency.

Micro-lens array 1

Input fiber array

Deflector array 1

Micro-lens array 2

Deflector array 2

Output fiber array

8.16 Two-stage N × N beam-steering switching architecture scheme, which is strictly non-blocking. The setup can be easily folded (and hence made more compact) using reflective deflector arrays.


226

Optical switches

More flexible and truly 2D beam deflectors are provided by 2D LC SLMs displaying dynamic diffraction gratings, or more generally holographic optical elements (Fracasso et al., 1990; Fukushima et al., 1991; O’Brien et al., 1991) that can be configured for a wide variety of routing topologies. Since the response time of LC devices is related to the LC layer thickness, the prism function of the refractive case is replaced by a blazed grating (saw-tooth profile) with a phase amplitude of 2π. Figure 8.17 depicts two effects of a pixellated SLM: (1) the phase modulation distribution is approximated by a staircase grating profile with phase values φk = 2kπ/N and 0 ≤ k ≤ N–1 and (2) a spatial quantization of the grating periods (Px, Py) by the 2D pixel grid. Nematic LC cells allow quasi-linear phase ramps (i.e., very large N values) (Wolffer et al., 2000) but the 2π phase modulation depth is polarization dependent and exhibits rather slow reconfiguration times (a few tens of milliseconds). In contrast, SSFLC cells (Clark and Lagerwall, 1980) provide binary phase states (N = 2) with a purely polarization-insensitive scheme (Warr and Mears, 1995) and a fast reconfigurable time of a few tens of microseconds. From the so-called grating equation, a monochromatic plane wave with wavelength λ under normal incidence on a 2D SLM will be deflected at angles (θx, θy) such that:

sin ym = m Ny p

[8.9]

sin xm = m Nx p

[8.10]

and

where m is the diffraction order, Nx and Ny denote the number of pixels per period in the x and y dimensions and p is the pixel pitch. The first diffraction order Phase (rad)

Grating period Px

fN–1 = 2p fk x (m)

f0 = 0

Py

Pixel electrode (a)

Px (b)

8.17 (a) 1D quantized phase profile of a blazed grating (N = 4 phase levels). (b) Spatial quantization of a 2D grating displayed on a pixellated LC SLM. Px and Py are the grating period lengths and correspond to multiple values of the pixel pitch.



227

(m = 1) is commonly used for beam-steering with a diffraction efficiency η as given in equation [8.8] with respect to the number of phase levels. We obtain a theoretical value η = 41% for two phase levels (FLC case) and η = 95% with N = 8. The maximum deflection angle is obtained for Nx = 2, yielding from equation [8.9] θmax = arcsin(λ/2p) which, for a typical pixel pitch p = 10 µm and λ = 1550 nm, gives θmax = 0.1 rad (5.7 degrees). In addition, steering resolutions lower than the micro-radian can be obtained using either pseudo-periodic diffraction gratings (Fracasso et al., 2003) or phase offsets (Engström, 2008). This great steering flexibility by monitoring the optical powers coupled in the output fibers permits adaptive beam positioning for possible mechanical tolerances in the packaging structure (Johansson et al., 2002). As an illustration, a 8 × 8 beam-steering spaceswitch using high-resolution transmissive nematic LC cell arrays (parallel-aligned) is presented in Wolffer et al. (2000). The device operates at 1550 nm using the twostage architecture shown in Fig. 8.16, embedded in a polarization diversity scheme. The basic LC cell is 9 µm thick and individual deflectors are made up with 309 electrodes over a 1.4 mm width, leading to a pixel pitch as small as 4.5 µm. The deflection efficiency of the two cell stages varies between 1 and 4 dB and the average measured fiber-to-fiber IL is 9 dB, with a PDL of 0.5 dB and an average crosstalk level of 243 dB. Finally, BER characterizations performed on switched connections through the device show that it is optically transparent at 10 Gbit/s.

8.3.5 Wavelength-selective switches (WSS) Recently, the emergence of devices such as wavelength blockers (Vasilyev et al., 2003) and WSS (Rhee et al., 2001) has made the design of efficient and integrated ROADM or OXC nodes possible. This solution is shown to be more flexible than using optical space-switching matrices in association with multiplexers and demultiplexer stages. Basically, WSS is an all-optical subsystem that can be viewed as an integrated 1 × N OXC. Its purpose is to switch any incoming wavelength from its input fiber to any of N output fiber ports. Figure 8.18 shows the particular switching configuration of a schematic WSS. Individual power attenuations of the routed channels can also be performed at that stage to compensate for possible power imbalance of the input WDM multiplex. A number of approaches to implementing WSS have been demonstrated (Bonenfant and Loyd, 2004), but the most mature technological solutions to date are MEMS arrays (Marom et al., 2005) and LCOS modulators (Baxter et al., 2006). LC-WSS based on polarization switching The generic optical scheme of a WSS is based on the structure of an ultra-short optical pulse shaper (Heritage et al., 1985), with micro-beam displacement between a pair of high-resolution wavelength dispersing diffraction gratings. Figure 8.19 shows the scheme for the first proposed solution, involving beam shifting by


228

Optical switches λ1

λ8

O1 λ1

WSS

2 3 4 5 6 7 8

O2 Input port

O3

O4

Output ports

8.18 Schematic of a 1 × 4 wavelength selective switch operating on an input wavelength multiplexed with eight channels.

polarization rotation (Patel and Silberberg, 1995). The input fiber polychromatic channels (λ1, …, λN) are collimated, then angularly dispersed by the first grating and the beamlets are individually focussed by the first lens as separate spots on the LC pixels of a nematic twisted array. This element is followed by a polarizationselective deflective element like a Wollaston prism or a calcite plate. Without any polarization rotation (LC cell ON), the wavelength channels are recollimated by Positive Positive LC Polarization rotator array Polarization lens 2 lens 1 deflective plate

Diffraction grating 1

λ1

λ1

λ2

λ2

f

Collimator

Diffraction grating 2

f Output 1

Input fiber 1

Output 2

8.19 Beam-shifting LC-WSS structure. For clarity, the polarization diversity apparatus is not represented. The gratings are placed at the front and rear focal planes of the first and second lens, respectively.



229

the second lens and recombined by the second grating into a single beam to the output fiber 1. When selected by the LC pixel (LC cell OFF), the wavelength channels are polarization rotated, then deflected off by the birefringent plate and recombined by the second grating into the second output fiber. The intermediate spatial shift is determined as a function of the lens focal length and the distance between the output fibers. In addition, a polarization diversity setup is used to make the device polarization independent, which involves a duplication of the number of beams crossing the optical setup. To achieve low CT and PDL values, the beamlets for the two polarizations of a given optical channel should experience the same polarization rotation and path length. To do this, one of the two eigen-polarizations of an input channel is halfwave-retarded, so that both beams share the same polarization and experience the same loss by crossing through the same optical components. First demonstration (Patel and Silberberg, 1995) on an eightwavelength switch with 4 nm separation exhibited 225 dB crosstalk and IL of about 10 dB. Further improvement of the architecture has led to devices operating on 80 channels with 5 dB IL and crosstalk values lower than 235 dB. System demonstrations have shown that this first WSS solution could be used to build ROADM structures operating at 10 and 40 Gbit/s. WSS based on LCOS beam-steering The configuration of Fig. 8.19 can be generalized to 1 × N wavelength switches, using beam-steering elements instead of polarization beam shifting elements. Figure 8.20 shows the generic and compact WSS configuration using a 4f imaging Diffraction grating

Input/output fiber array (1) (2) (4)

2

1

(3)

2

Beam deflector (LC or MEMS)

(5)

1 2

Fourier mirror

8.20 Compact and generic WSS architecture using beam steering at the intermediate plane.


230

Optical switches

setup, where the same bulk diffraction grating is shared for both the demultiplexing and multiplexing around a dynamic beam-steering element, including MEMS, LCD and LCOS devices (Baxter et al., 2006). The wavelengths are dispersed along one dimension, and the orthogonal axis is used for port selection and possible amplitude control by beam shifts around the output fiber core center. The input light from a given fiber of the array is collimated and reflected by the concave mirror and then angularly dispersed by the diffraction grating in the horizontal plane. A second reflection on the mirror focuses each wavelength channel λi to a given area of the beam-steering SLM, which then reflects the channel back with a vertical angle θi (channel dependent). The path for the channel is then retraced through the device and the light is recoupled to a particular port of the vertical fiber array, depending on the initial θi value. Two-dimensional LCOS phase modulators can be employed as high-resolution, flexible and motionless reflective beam deflectors. In that particular case, the SLM area is divided into N vertical stripes corresponding to N input wavelength channels. Each pixellated stripe is configured to display a diffraction grating that does not interfere with the other stripes. Using 2D gratings, the diffraction efficiency can be adjusted to provide attenuation or power splitting functions (multi-cast). Using this technique with nematic LCOS SLMs and a polarization diversity setup (Baxter et al., 2006), a 1 × 9 WSS operating on 100 wavelengths and a 0.4 nm grid is demonstrated. Low IL ( 1, the switch is able to route an input optical signal to any one of the N output fibers. 1 × N fiber switches with relatively large N values can be realized by cascading multiple stages of 1 × 2 switches, as demonstrated in the work of Ford and DiGiovanni8. In other words, a 1 × 2 fiber switch can be considered as an elementary device to construct 1 × N fiber switches. Therefore, as examples, here Input fiber

Output fibers

10.1 Scheme of l × N fiber moving type switch.


1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

278 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

Optical switches

we mainly focus on several kinds of 1 × 2 moving-fiber type switches. An N × M (M = 1, 2, 3 …) switch with multiple movable input fibers and multiple output fibers may also be built based on the concept of l × N fiber switch. Fiber switch with electrostatic actuation As mentioned above, the actuation for the mechanical force on the movable input fiber can be generated in various ways. In principle, electromagnetic and piezoelectric forces do scale better, but are relatively difficult to fabricate. Electrostatic actuation is a simple way to move fibers. It can be generated by magnet acting on a magnetic pipe which circumvents the fiber16. Figure 10.2 shows schematically the structures of a typical 1 × 2 fiber switch of moving-fiber type with electrostatic actuation as reported in the work of Herding et al.16. To move the fibers directly, an electrostatic field is built up between a metalized flexible fiber and a fixed electrode which is also shown in Fig. 10.2.

Fiber 3

Actuation chamber

Substrate Fiber 2

Fiber 1 Metal layer

Insulation Bottom-electrode

Bottom-layer

10.2 Schematic for the structures of the 1 × 2 fiber switch and the planar electrode design16.


Fiber, holographic, quantum optical and other switches

279

As shown in Fig. 10.2, the fiber switch is realized by a simple actuation chamber with suitable cavities for the fibers and appropriate electrodes for electrostatic actuation of the metalized fibers. Simultaneous application of the actuation voltage to the input fiber and the respective output fiber will pull the input fiber to the same wall of the actuation chamber as the output fiber, where they settle in a perfect optical alignment. The design uses planar electrodes integrated into the bottom surface of the actuation chamber. This concept is easy to realize and has the benefit of pulling the fiber not only to the side wall but also to the bottom of the actuation chamber. In addition, this design will provide a fiber alignment in two degrees of freedom and good optical coupling. To switch light between two optical fibers, it is necessary for the two fibers to overlap at least half of its diameter. For standard 125 µm fibers used here, the minimum travel range is 62.5 µm. In order to minimize the crosstalk between the two output fibers, a greater distance between the output fibers, i.e. a greater travel distance might be required. The main geometrical misalignments are summarized in Fig. 10.3. The extrinsic losses are related to a poor fiber-to-fiber alignment. The main sources of extrinsic losses are the longitudinal, lateral and angular misalignments. As the fibers are aligned by the specially formed electrodes to a common micromachined stop (see Fig. 10.2), the lateral and angular mismatch between the input and the output fibers are negligible. The optical coupling efficiency will be mainly determined by the longitudinal alignment. The specified maximal 2 dB insertion loss is achieved up to a longitudinal distance of 80 µm between the faces of the input and the output fibers for core guided light propagation and extended ray intensity profile. The crosstalk damping is mainly determined by the longitudinal distance between the input and the output fibers and by the lateral distance between the two output fibers. To ensure the specified coupling loss of less than 2 dB and to avoid optical degradation of the switch due to the intrinsic losses, the fibers used as input and output fiber pieces (pigtails) in a switch should be taken from the same piece of fiber, which will reduce the effect of different NA values of the coupled fibers. According to simulation results, the following design values for the fiber switch are used: actuation stroke of 65 µm (no additional separation of the two output fibers), length of movable fiber 30 mm and curved electrodes. As a result, the driving voltage as low as 40 V can be used. The specified crosstalk is achieved even with imperfect coupling into the input fiber as long as the longitudinal mismatch is less than 100 µm. The switching speed was measured to be 7 ms; the switching frequency is limited by the fiber mass to be less than 130 Hz. This fiber

x (a)

Q

y (b)

(c)

10.3 Different types of alignment errors: (a) longitudinal, (b) lateral and (c) angular misalignments.


1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

280 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

Optical switches

switch can be made out of a dry film photo resist TM100 in a simple one-step fabrication process. Fiber switch with comb-drive actuation 1 × 2 fiber switches can also be built by using various mechanical actuators. Among the reported types of designs, the one with comb-drive actuation shows its advantages in terms of endurance for long-term applications, switching time, requirement in structure length/thickness and convenience in fabrication process. Therefore, here we introduce a 1 × 2 fiber switch with comb-drive actuation reported in the work of Yang et al.17 for illustration. Figure 10.4 schematically depicts the structure of the 1 × 2 fiber switch. The device consists of a suspended fiber-holding table, folded flexures and two comb-drives. All the components can be monolithically fabricated using the dual-thickness SOI process. In order to achieve a displacement of about 125 µm, which is required for the fiber-switching design, the suspension beams are very flexible to produce a relatively large displacement under a reasonable driving voltage. Also, the inter-digital comb fingers should move only in the desired lateral direction with a minimal sideward movement. Doing so requires a large stiffness ratio between the sideward spring constant and the lateral spring constant. Also, stoppers are used for precise radial alignment to ensure a low butt-coupling loss between the input fiber and the output fiber. Based on the constraint of fabrication capacity for the large device thickness (around 100 µm) of suspended structures with a driving voltage of 100 V, the optimal device dimensions are listed in Table 10.1.

Folded-flexure

Hbeam

Output fibers Channel 1

Channel 2

Input fiber

Fiber holder

Stopper Lfiber Comb-drive actuator Fixed end of fiber

Fiber holding table

10.4 Schematic of the 1 × 2 fiber switch with comb-drive actuation17.



281

Table 10.1 Optimal device dimensions of the 1 × 2 fiber switch17 Parameters

Values

Device thickness Gap between comb fingers Number of pairs of comb fingers Folded-beam length Folded-beam width Input fiber length

100 µm 5 µm 1000 2557 µm 15 µm 14.2 mm

Without the application of electrostatic voltage, the holding table and the input fiber are in the middle position (neutral position). When the left comb-drive actuator is actuated (100 V), the holding table is driven to the left and touches the left stopper; then the input fiber aligns with the output fiber (channel 1), as shown in Fig. 10.5(a). When the right comb-drive actuator is actuated with 100 V (and the left comb-drive is not actuated), the holding table is driven to the right and touches the right stopper; then the input fiber aligns with the output fiber (channel 2), as shown in Fig. 10.5(b). Note that the pictures in the right-bottom corner of Fig. 10.5(a) and (b) are the CCD photographs of the device in the corresponding status. Accordingly, 1 × 2 optical switching can be achieved by alternating the actuation of the comb-drive actuators. Since the thickness of the device is designed to be 100 µm and the depth of the fiber mounting groove is 65 µm, bulk micromachining technology must be utilized to fabricate the device. The gap between comb fingers is 5 µm and the maximum trench aspect ratio is 20. For fabricating suspended microstructures with high aspect ratios, the SOI process is one of the most popular approaches because of its simplicity. In order to reduce the complexity of fabrication and assembly, a dual-thickness SOI process is proposed to fabricate the suspended structure with layers of two different thicknesses. This process can monolithically create most of the parts of the device (folded-flexures, comb-drive actuators, stoppers, fiber holder and fiber-holding table) without the assembly process. The thicker layer serves as the structure layer and the thinner layer serves as the fiber mounting grooves. Figure 10.6(a) and (b) show the scanning electron micrograph (SEM) and the optical photograph of the device structures, respectively. Figure 10.7 shows the measured switching behaviors of the voltage output signals of channels 1 and 2. As shown in Fig. 10.7, the input fiber, which originally aligns with the output fiber of channel 1, switches to (align with) the output fiber of channel 2. The oscillation of the channel 2 signal arises from the bouncing effect when the fiber-holding table and the stopper collide with each other. This bouncing effect disappears within 2.5 ms. The total switching time (including the bouncing effect) is about 3.5 ms. The measured average insertion losses for channels 1 and 2 are about 0.92 dB and 0.89 dB, respectively. The deviations for the two channels are both 0.06 dB.


1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

282 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

Optical switches

10.5 Schematic views and CCD pictures of the switch at different switching positions: (a) the left comb-drive is actuated and (b) the right comb-drive is actuated17.

Fiber switch with piezoelectric actuation The most remarkable merit of 1 × 2 fiber switches with piezoelectric actuation is the very simple device structure18, which may greatly benefit mass production. The form of a 1 × 2 fiber switch with piezoelectric actuation is shown in Fig. 10.8. The input fiber is bonded inside the piezoelectric tube and the two output fibers are fixed closely together on the other side. The piezoelectric tube deflects when a step voltage is applied to the electrodes on the surface of the tube, as shown in Fig. 10.9. In this case, the deflection of the tube causes signals from the input fiber to be directed into the desired output fibers. The input fiber movement needs to be


Fiber, holographic, quantum optical and other switches (a)

20 kV

X30

500 µm

(b)

10.6 (a) SEM image and (b) optical photograph of the device structures17.


283 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

Optical switches 3 2.5 Output voltage (Volt)

284

2 Channel 1

1.5

Channel 2

1 0.5 0 –5

–4

–3

–2

–1

0

1

2

3

4

5

Time (ms)

10.7 Measured switching behaviors. The transient curves are the measured photo-diode output voltages of channels 1 and 217.

Input fiber

Piezoelectric tube

Bending of the piezo tube

Output fibers

Wires for applying signals

10.8 Scheme of a 1 × 2 fiber switch with piezoelectric actuation18.

Input fiber movement

10.9 Deflection of the piezoelectric tube when a step voltage is applied18.


Fiber, holographic, quantum optical and other switches Lp

Holder

Piezoelectric tube

Lg

Graphite rod

285 Lf

Input fiber

10.10 Construction of the input fiber18.

at least 62.5 µm (half the diameter of an optical fiber) in order to perform switching. The exact amount of deflection depends on the space between the output fibers. Figure 10.10 shows the construction of the input fiber. To achieve more flexible movement of the input fiber, the fiber protrudes beyond the piezoelectric tube. A thin graphite rod is bonded with part of the input fiber to control the bending of the fiber. In Fig. 10.10, Lp is the length of the piezoelectric tube, Lg is the length of the part of the input fiber, to which the graphite rod is bonded on its surface, and Lf is the length of free input optical fiber. When a voltage is applied on the piezoelectric tube, it deflects and actuates the input fiber. Once the input fiber hits the v-grooves at the output fibers (see Fig. 10.11), it stops and aligns with one of the output fibers. Accordingly, 1 × 2 optical switching can be realized by alternating the applied voltage on the piezoelectric tube. The piezoelectric tubes used in the switch were fabricated using the electrophoretic deposition process (EDP). EDP is a relatively low cost, fast and easy process to form ceramic tubes. Tubes of various dimensions can be fabricated using EDP. The cross-section of piezoelectric tube with two sectored outer electrodes for applying voltage is shown in Fig. 10.12.

Output fibers

0.16mm

v-grooves

10.11 Cross-section of the output fibers and v-grooves18.


1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

286 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

Optical switches +

Outer electrode

Inner electrode

–

Piezoelectric tube

10.12 Cross-section of piezoelectric tube18.

The latest piezoelectric-based optical fiber switch prototype is shown in Fig. 10.13. The switch has a length of 110 mm. The fibers in the latest prototype are aligned using translation stages. Different components are then bonded together using low shrinkage adhesive. A compact electronic control unit is integrated into the switch. Fig. 10.14(a) and (b) show that the input fiber is being switched into different outputs. The space between the input and output fibers is 25 µm. Figure 10.15 is a graph showing the change in insertion loss of the switch by adjusting the applied voltage on the piezoelectric tube. The graph shows that as the voltage increases to around 65 V, the insertion loss drops to around 2 dB. This

10.13 Latest prototype of the 1 × 2 fiber switches with piezoelectric actuation18.



287

(a)

(b)

10.14 Input fiber movement: (a) input aligned to the top fiber and (b) input aligned to the bottom fiber18.

is the voltage when the input fiber hits the v-grooves. Further increase of the voltage causes a decrease in the angular misalignment between the fibers and hence a drop in insertion loss. The insertion loss is the lowest (0.88 dB) when the applied voltage is 115 V. At this point, the angular misalignment between the


1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

288

20 18 16 Insertion loss (dB)

1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

Optical switches

14 12 10 8 6 4 2 0 0

20

40

60

80

100

120

140

Applied voltage (V)

10.15 Insertion loss of the optical switch against applied voltage on the piezoelectric tube18.

input and output fibers is minimized. Increasing the applied voltage further causes a rise in angular misalignment between the fibers. This can be seen in the graph as the applied voltage is increased from 115 V, the insertion loss rises from 1.05 to 2 dB. The average insertion loss of the switch is 1 dB and the average crosstalk is 245 dB. The switching time of the prototype can be controlled by adjusting the output capacitance of the power supply circuit. Lowering the output capacitance of the power supply increases the rise time of the step voltage and vice versa. When ceramic capacitors of 19.6 nF are connected to the output of the power supply, the step voltage from the power supply has a fast rise time of 0.8 ms. The input fiber is actuated too fast and causes vibration as it hits the v-grooves. Figure 10.16 is a graph showing the change in switching time of the optical switch versus the output capacitance. Based on this result, the output capacitance is adjusted to around 260 nF in order to achieve a switching time of less than 5 ms.

10.2.2 Component moving type Figure 10.17 schematically depicts a generic principle of a l × N fiber switch with component moving type. The movable component usually consists of a lens and a moving mirror. A gradient index lens can be used and the mirror can rotate to reflect light from the input onto one of the N outputs. In this section, two typical l × N fiber switches of the component moving type will be introduced for illustration. In the work of Peter et al.14, a l × N fiber switch using a conventional lens and an adaptive mirror to correct for the aberrations is proposed and studied. Figure 10.18 shows the switch composed of a fiber bundle, a lens and a mirror. The merit of this switch is its power coupling efficiency. The limitations, mainly due to aberrations and misalignments of the optical components, are overcome using a



289

25

20 Switching time (ms)

15

10

5

0 0

100

200

300

400

500

600

Output capacitance (nF)

10.16 Switching time of the switch against output capacitance of the power supply18.

Output fibers

Input fiber Movable component

10.17 Scheme of l × N fiber switch with component moving type.

Fiber bundle

Lens

Mirror

α

f

f

10.18 Scheme of l × N fiber switch composed of a fiber bundle, a lens and a mirror14.


1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

290 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

Optical switches

micromachined deformable mirror. The source fiber is imaged (4f system) onto one of the receiver fibers by moving the lens laterally. The switch serves two functions. First, it images a single-mode source fiber onto another single-mode receiver fiber (coupling function). Second, it deflects the beam to address one of the receiver fibers (switching function). In this fiber switch, an adaptive membrane mirror fabricated by bulk silicon micromachining is used as a deformable mirror, as shown in Fig. 10.19. Electrostatic deflection is generated by 37 electrodes arranged under the membrane in a hexagonal array. The maximum applied voltage is around 190 V. Since the membrane can only be deflected toward the electrodes, a bias voltage of about 141 V is applied to the membrane to achieve bidirectional movement. The deflection produced corresponds to a concave mirror with a focal length of about 2 m. The initial setup is made to compensate this slight defocusing. The optimization of the shape of the membrane for maximum power coupling efficiency of each connection is obtained with the help of an evolutionary algorithm. With good repeatability of the deformation of the membrane, the optimized deformation can be recalled for the specified connection to the receiver fiber. The measured crosstalk is less than 30 dB. The total loss due to the optical elements is estimated to be 22%, by considering two interfaces between the receiver fibers and air with 4% Fresnel losses each, 1% transmission loss inside the achromat and 0.3% of reflectance at both interfaces and a reflectivity of 87% for the mirror. Fluctuations have been measured which lead to deformation fluctuations of 0.1 µm at the center of the membrane at the maximum voltage of 190 V. In the work of Duparré et al.15, a l × 4 fiber switch for multimode fibers with 600 µm core diameters is proposed and studied. A microlens array (MLA) telescope allows for variable and fast beam deflection when the positions of the

Si chip

Al-coated SiN membrane

Control electrodes

V1

V2

V3

V4

Vb

10.19 Schematic for the micromachined deformable membrane mirror. The membrane has a diameter of 15 mm and a thickness of less than 1 µm. The surface of the membrane is coated with a 0.2-µm-thick reflective aluminum layer. Its active area has a diameter of 12 mm14.



291

cylindrical MLAs relative to one another are altered by specially designed piezomechanical actuators. Standard achromats are used for collimation of light emitted by the input multimode fiber and for focusing of the deflected light onto a linear array of output multimode fibers. Figures 10.20 and 10.21 show the configurations of the decenterable cylindrical MLA telescopes for beam deflection and the l × 4 fiber switch, respectively. A multimode fiber with core radius Rin and numerical aperture NA 1 emits a light bundle. Bulk collimating and focusing optics with focal lengths FCOL and FFOC image the input fiber end face onto the output fiber end faces. After collimation of the light by the first collimating optics the bundle has a diameter of approximately A = 2NA 1/FCOL and a residual divergence (half-angle) of θmax = arctan (Rin/FCOL ). Beam deflection is achieved with two MLAs with pitches p1 and p2 (equal in value for this type of application) and focal lengths f1 and f2. The first MLA focuses the collimated beam into a large number of sub-foci and the second MLA re-collimates each beam. These microoptical elements permit fast beam deflection because even small displacements result in large deflection angles as a result of the short focal lengths of the microlenses. Deflection angle α, as a function of the focal lengths of the lens in the second MLA, f2, and the relative positions of the two MLAs, r0, is given by Collimator

Cylindrical lens arrays

Multimode fiber

Deflected beam

10.20 Scheme of a 2D micro-optical scanner based on MLA telescopes15. MLA

r0

p 2 Rin

Input fiber

A 2R

out

FCOL

f1 f2

Output fibers

FFOC

10.21 Scheme of the l × 4 fiber switch that uses a MLA telescope for beam deflection15.


1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

292 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

Optical switches

α = arctan(r0 /f2). The maximum angle of deflection is determined by αmax = arctan(p2/2f2) because r0max = p2/2. The displacement of the spot in the image plane deflected from the optical axis, v = FFOC tan(α), can then be given by v = FFOC r0/f2. The overall magnification of the imaged fiber end face is Mnet = f1/ f2(FFOC /FCOL ). For application of MLA telescopes in commercial devices such as multimode fiber switches, stability and repeatability are of major importance. Tolerance analysis shows that the decentration for beam deflection needs to be accurate on a submicrometer scale. Additionally, rotation of the two MLAs that build up a telescope about the optical axis led to a strong decrease in transfer efficiency. Therefore, the l × 4 fiber switch applies a newly developed piezo-based actuator for highly accurate and highly parallel decentration of the two MLAs for beam deflection, as shown in Fig. 10.22. Figure 10.23 shows the completely assembled prototype of the l × 4 fiber switch for multimode fibers with 600 µm core diameters without the control electronics. The overall system dimensions are 15 cm × 10 cm × 13 cm with an overall weight of the optomechanics of 3 kg. Switching among channels can be achieved by the variable beam deflector by digitally memorizing the values of voltage for the largest possible transfer efficiency for each channel and using a PC to assign a channel number to this voltage value. Figure 10.24 shows the measured switching behavior of the l × 4 fiber switch for switching among channels with different amounts of separation. The signal curves for the switching states channel Direction of translation

Frames holding the substrates of the cylindrical microlens arrays

Solid-state hinges

Space determined for implementation of piezostacks (2 × 2)

Two actuators acting in opposite directions

10.22 Mechanical setup for highly parallel decentration of cylindrical MLA telescopes15.



293

10.23 Overall view of the prototype (without the driving electronics for the actuators)15.

0.4

Electronic control signal Optical output signal

0.3

0.3

0.2

0.2

Voltage (V)

Voltage (V)

0.4

0.1 0.0

–0.1


0.1 0.0

–0.1 0

2

4

6 Time (ms)

8

10

0

1

(a)

3 Time (ms)

4

5

(b) 0.4

0.4


0.3

0.3 Electronic control signal Optical output signal

0.2

Voltage (V)

Voltage (V)

2

0.1 0.0

0.2 0.1 0.0

0

2

4

6 Time (ms)

8

10

–0.1

0

2

(c)

4 6 Time (ms)

8

10

(d)

10.24 Activation and deactivation behavior of the four channels: (a) activation of channel 4 when one is switching from channel 1 (7 ms), (b) deactivation of channel 1 when one is switching to channel 4 (3 ms), (c) activation of channel 2 when one is switching from channel 3 (9 ms) and (d) deactivation of channel 2 when one is switching to channel 3 (9 ms)15.


1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

294 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

Optical switches

ON and channel OFF are shown. The ramps for switching between farther separated channels (Fig. 24(a) and (b)) are steeper than for channels that are closer to each other (Fig. 24(c) and (d)) because of the behavior of the control electronics. So switching off to farther separated channels is faster (3 ms) than switching to closer channels (9 ms). Switching on from farther separated channels takes place in 7 ms and from closer channels takes 9 ms. These switching times for the thick fibers are attractive for practical use. The measured insertion loss is 1.5 dB, and the experimental values of crosstalks are 234.2 to 244.7 dB.

10.3 Holographic switches Holography is a method to generate optical images. Instead of recording the image of the object as conventional photography does, holography provides the means to record the object wave itself. This wave is recorded in such a way that a subsequent illumination of the record serves to reconstruct the original object wave. The hologram consists of a complex distribution of clear and opaque areas corresponding to the recorded interference fringes. If the hologram is illuminated under certain conditions, a duplicate of the original reference wave can be obtained from the light transmitted through the hologram. Since a hologram can serve as a controllable, or programmable, diffraction grating for switching purposes, a number of approaches to holographic-optical switches have been reported20–25. Generally, holographic switches are capable of multi-dimensional steering of light using electrical control/activation of a holographic system. One of the advantages of holographic switches is that they involve no moving parts26. The functionalities of holographic switches may include space switching, wavelength selection, wavelength drift compensation, channel monitoring, variable optical attenuation, dynamic reconfiguration and parallelism (multiple interconnections may be handled simultaneously, which is also useful for multicasting). In the following paragraphs, we will introduce some typical approaches among the large number of reported approaches. l × N holographic switches can be formed based on the use of liquid-crystal spatial light modulators (LCSLMs) as the hologram recording media. Figure 10.25 shows a switching system where a LCSLM is used as a programmable Positive lens (achromat doublet) Programmable beam steerer

Single-mode fiber array

Nx

Ny f

f

10.25 Configuration of the holographic switch with a LCSLM as a programmable beam steerer25.



295

beam steerer25. Here, many advantages are combined within the same device. First, the operation of diffractive liquid-crystal devices with no moving parts provides stable, accurate and reproducible switching maps, as well as a very simple addressing and supervision scheme. Then, it is shown both qualitatively and quantitatively that diffractive (or holographic) structures enable the extension of the device functionalities to multicast switching and variable-wavelength-band selection. Average insertion loss for the 14-channel device is 6 dB (in protection mode). Constant improvement on the LCSLM – in terms of mirror quality and liquid-crystal cell parameter control – as well as upgrades of the optomechanical alignment should rapidly lead to insertion loss values less than 4 dB. An N × N holographic switch can be built using a single spatial light modulator (SLM). However, the fan-out to the output fibers limits scalability and performance. A well-known solution to this problem is to build a holographic switch using two arrays of sub-holograms, as shown in Fig. 10.2623. This architecture ensures that the output beams are collinear with the axes of output fibers. Although the beamsteering angle has a wavelength dependence, in a double hologram system the ‘complementary’ nature of the holograms significantly reduces the overall wavelength dependence of the switch. This approach also has the added advantage of improving the crosstalk characteristics of the switch as the second hologram array tends to reject unwanted diffraction orders from the output channels. Finally, it should be noted that the insertion loss for such a switch architecture remains constant as the size of the system is scaled up. Thus, large switching fabrics scaling to many hundreds of channels are possible using this technology. Such a 3 × 3 switch was designed and constructed as part of the ROSES project23. The target performance figures for the switch are presented in Table 10.2. Table 10.2 Summary of 3 × 3 switch target performances23 Insertion loss Noise isolation (all paths) Back reflection Switching time Bandwidth

40 dB >40 dB 1 ms 30 nm

FLC-SLM #1

FLC-SLM #2 ∆h

φ h Input fiber array

z

10.26 Two hologram N × N holographic switch architecture23.


Output fiber array

1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

296 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 20 1 2 3 4 5 6 7 8 9 30 1 2 3 4 5 6 7 8 9 40 1 2 43X

Optical switches

(a)

Polarization selective A hologram

B Upper substrate

(b) (c)

Substrate mode hologram

Liquid crystal cell

Lower substrate Substrate mode hologram

10.27 Planar integrated polarization-insensitive 2 × 2 holographic switch: (a) fan-in stage, (b) switching and (c) fan-out stage21.

In the work of Moreau et al.21, a polarization-insensitive 2 × 2 holographic switch is realized by implementing multistage network in a planar configuration. The switch is a multilayered structure composed of a reflective polarizing hologram sandwiched between two planar substrates. Signals are coupled in and out of the switch by diffraction on substrate-mode holograms and their optical paths are controlled by total internal reflection in a liquid-crystal cell. Holographicoptical elements are recorded in high-index modulation photopolymer. Figure 10.27 schematically shows the polarization-insensitive 2 × 2 holographic switch proposed in the work of Moreau et al.21 Two or more parallel paths can be achieved in the same planar component by fixing the upper-substrate thickness to a fractional part of the lower-substrate thickness. In the present case, the uppersubstrate thickness is half of the lower one. When the spatial separation of A and B inputs is 2 mm, the signal polarization components are distributed in two parallel polarization-dependent switches. The first switch processes the TM component of signal A and the TE component of signal B, while the second one processes the TE component of signal A with the TM component of signal B. The switch symmetry ensures the correct fan-out of each signal. The measured crosstalk is 5.5% (in the worst case) and the switching time is 20 ms, which remains excessive for high-rate applications, but is sufficient to demonstrate the exchange operation and its applications in network protection and restoration services.

10.4 Quantum optical switches Quantum well (QW)-/quantum dot (QD)-based optical switches may be regarded as a special group relative to the nine main categories of optical switches. However, this group has been greatly expanded especially in recent years27–43. The most outstanding advantage of the QW-/QD-based switches is their ultra-high switching speed (response

Optical switches - Materials and design

Optical Sensors and Switches

Optical Materials

Optical Sensors and Switches (Molecular and Supramolecular Photochemistry)

Molecular Switches

Handbook of optical materials

Geometrical optics and optical design

Geometrical optics and optical design

Optical filter design and analysis

Molecular Switches

Optical System Design

Optical Filter Design

Optical System Design

Optical System Design

Colour and The Optical Properties of Materials

Supramolecular Organization and Materials Design

Supramolecular organization and materials design

Supramolecular Organization and Materials Design

Composite Materials Design and Applications

Modern Optical Engineering: The Design of Optical Systems (Optical and Electro-Optical Engineering Series)

Multiwavelength optical networks: architectures, design, and control

The art and science of optical design

Optical System Design, 1st Edition

Design Issues in Optical Processing

Optical components based on high index materials

Product Design and Testing of Polymeric Materials

Liquid Crystals: Materials Design and Self-assembly

Asphalt Materials and Mix Design Manual

Fiber-Reinforced Composites Materials Manufacturing and Design

Nano and Microstructural Design of Advanced Materials

Optical and Electronic Properties of Fullerenes and Fullerene-Based Materials

Optical switches - Materials and design

Optical Sensors and Switches

Optical Materials

Optical Sensors and Switches (Molecular and Supramolecular Photochemistry)

Molecular Switches

Handbook of optical materials

Geometrical optics and optical design

Geometrical optics and optical design

Optical filter design and analysis

Molecular Switches

Optical System Design

Optical Filter Design

Optical System Design

Optical System Design

Colour and The Optical Properties of Materials

Supramolecular Organization and Materials Design

Supramolecular organization and materials design

Supramolecular Organization and Materials Design

Composite Materials Design and Applications

Modern Optical Engineering: The Design of Optical Systems (Optical and Electro-Optical Engineering Series)

Multiwavelength optical networks: architectures, design, and control

The art and science of optical design

Optical System Design, 1st Edition

Design Issues in Optical Processing

Optical components based on high index materials

Product Design and Testing of Polymeric Materials

Liquid Crystals: Materials Design and Self-assembly

Asphalt Materials and Mix Design Manual

Fiber-Reinforced Composites Materials Manufacturing and Design

Nano and Microstructural Design of Advanced Materials

Optical and Electronic Properties of Fullerenes and Fullerene-Based Materials

Recommend Documents