Photonic Glasses

M

Fuxi Gan • Lei Xu editors

PHOTONIC GlflSS€S

~^< \

&s •

*.S* J™

1f^S^

N\C GV-

HCrt°

editors

Fuxi Gan .. Lei Xu Fudan University, China

p World Scientific NOOM

- sme^cs? • mm& • %nmm^ - urns tern - ;&?-> • CHWW

&

*

Published by World Scientific Publishing Co. Pte. Ltd. 5 Toh Tuck Link, Singapore 596224 USA office: 27 Warren Street, Suite 401-402, Hackensack, NJ 07601 UK office: 57 Shelton Street, Covent Garden, London WC2H 9HE

British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library.

PHOTONIC GLASSES Copyright © 2006 by World Scientific Publishing Co. Pte. Ltd. All rights reserved. This book, or parts thereof, may not be reproduced in any form or by any means, electronic or mechanical, including photocopying, recording or any information storage and retrieval system now known or to be invented, without written permission from the Publisher.

For photocopying of material in this volume, please pay a copying fee through the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA. In this case permission to photocopy is not required from the publisher.

ISBN 981-256-820-4

Printed by Mainland Press Pte Ltd

Preface

Photonics is a systematic science dealing with the photon generation and detection, as well as stimulated emission, photon frequency conversion and polarization change. The invention of laser in the middle of the last century was a great event for the development of photonics, in which photons are performed as the information and energy carriers. The photonics has been popularized recently, and indicates its rapidly escalating importance in the future. The exploration of new photonic materials, including photonic glasses, is a key issue for advance photonic devices. Inorganic glasses have been used as optical materials for a long time due to their isotropy, hence to make the large size and high optical homogeneity more easily, and high transparency over a wide spectral range from ultraviolet to infrared, as well as linear functional properties. Therefore, inorganic glasses still play essential role in photonics used as transmitting and linear functional media for photonics. Since the emergence of lasers, transition element doped glasses have become one of the most important laser materials. The laser glasses, glass fibers and waveguides are still the main groups of solid state laser materials. The nonlinear optical effects of inorganic glasses are always happened with intense electromagnetic field, where as the glasses can perform the nonlinear optical functional roles. The interaction of ultra-short laser pulse with glass media produces a series of new optical phenomena, which can be applied in fabrication of new photonic devices. Inorganic glasses have also been developed from three dimensional (bulk materials) to low dimensional (thin films and fibers), which play a significant role

V

VI

Preface

in photonic devices for optical data storage, optical communication, as well as optical processing and display. Authors of this book have been actively engaged in this field and made notable contributions to the above mentioned subjects and published many papers in both domestic and international journals. An attempt was made in this monograph to summarize the research results in photonic glasses, which have been achieved by the authors' research groups in Shanghai Institute of Optics and Fine Mechanics and Fudan University mainly in recent 15 years, and to compile all experimental information from journals and proceedings in a book. It is not quite enough to select those chapters in this book for reflecting all aspects of photonic glasses due to imbalance for our research achievements. It might be possible to ignore some important subjects or contents due to the limit of the authors' knowledge and incomplete access to published materials. We hope that our readers, especially the photonics experts, will give us their valuable suggestions in order to make correction and complements for us in the next edition. The authors of this book are grateful to the China National Natural Science foundation, Chinese Academy of Science and Chinese Ministry of Education for their continuous support to the research projects. Acknowledgements are made to the journals from which some figures have been reproduced. The editors wish to extend the gratitude to our associates who contributed to compile this book, and particularly thank Prof. Shouyun Tian, Ms. Bo Ma, Ms. Hongxia Zhao, for their assistance in preparing the manuscript.

Fuxi Gan Lei Xu Fudan University, Shanghai December, 2005

Contents

Preface

v

List of Contributors

xi

1. From Optical Glass to Photonic Glass 1.1 Introduction 1.2 Physical Fundamentals 1.3 Optical Glasses 1.4 Photonic Glasses 2. Structure and Properties of Amorphous Thin Films for Optical Data Storage 2.1 Amorphous Rare Earth-transition Metal (RE-TM) Alloy Thin Films 2.2 Amorphous Metallic and Chalcogenide Thin Films 2.3 Nonlinear Optical Amorphous Alloy Thin Films 3. New Developments in Optics and Spectroscopy of Rare Earth Ions Doped Glasses 3.1 Laser Spectroscopy of Nd 3+ and Yb 3+ High Doped Glasses 3.2 Nonlinear Luminescence of Rare-Earth (RE) Ions in Glasses 3.3 Super-luminescence of RE-doped Glass Fibers 4. Third-Order Optical Nonlinear Properties of Glasses 4.1 Measurement of Third-Order Optical Nonlinear Susceptibility of Glass 4.2 Optical Nonlinearity of Dielectric Glasses 4.3 Optical Nonlinearity of Organic-Inorganic Hybrid Glasses 4.4 Optical Nonlinearity of Nano-composite Glasses 4.5 Optical Limiting Effects

vii

1 1 2 6 18 39 40 53 64

77 77 94 110 117 118 123 131 137 146

viii

Contents

5. Second-Order Optical Nonlinear Properties of Glasses 5.1 Introduction 5.2 Second-Order Optical Nonlinearity in Silica Glasses 5.3 Second-Order Optical Nonlinearity in High Refractive Index Glasses 5.4 Applications

153 153 167

6. Glass Fibers for Optical Amplification 6.1 Brief Introduction of Optical Fiber Amplifier 6.2 Er3+-Doped Phosphate Glass Fiber Amplifiers

191 191 193

7. Glass Fibers for High Power Lasers 7.1 Introduction of Optical Fibers 7.2 Fabrication and Materials 7.3 High Power Lasers Based on Rare-Earth Ions Doped Fibers 7.4 High Power Pulsed Fiber Lasers 7.5 Recent Development and Applications of Fiber Lasers

227 227 234 239 246 254

8. Hybrid Organic-Inorganic Solid-State Dye Laser Glasses 8.1 Organic Dyes and Liquid Dye Lasers 8.2 Hybrid Solid-State Dye Laser Glasses and Preparation Techniques 8.3 Photostabilities and Photodegradation Mechanisms of Hybrid Solid-State Dye Laser Glasses 8.4 Hybrid Solid Dye Laser Glass Based on Energy Transfer Between Laser Dyes 8.5 Solid-State Dye Lasers and Parameter Optimization 8.6 DFB Laser Based on Sol-Gel Derived Organic-Inorganic Hybrid Thin Film Waveguides 8.7 Summary and Future Prospects

261 262

9. Optical Glass Waveguides 9.1 Principles of Optical Waveguides 9.2 Glass Waveguides Fabrication and Optical Properties 9.3 Organic/inorganic Hybrid Glass Waveguide Materials 9.4 Functional Glass Waveguide Devices

299 300 302 310 315

10. Glass Photosensitivity and Fiber Gratings 10.1 Glass Photosensitivity 10.2 Principles of Fiber Gratings 10.3 Fiber Grating Fabrications 10.4 Fiber Grating Applications

172 186

264 273 281 285 290 292

339 340 355 361 363

Contents

ix

11. Glass Fibers for Photonic Crystals 11.1 Light Guidance in PCF 11.2 Fabrication 11.3 Properties of PCFs and Device Applications 11.4 Non-Silica Glasses for PCFs

375 377 383 386 400

12. Functional Microstructures in Glass Induced by a Femtosecond Laser 12.1 Introduction 12.2 Micro-Structural Changes Induced by Femtosecond Lasers 12.3 Valence State Manipulation of Active Ions 12.4 Precipitation of Functional Crystals 12.5 Novel Phenomena Induced by Femtosecond Lasers

405 405 407 417 424 435

Index

445

List of Contributors

Chapters 1-4 Fuxi Gan Shanghai Institutes of Optics and Fine Mechanics, Chinese Academy of Sciences, Shanghai 201800; Department of Optical Science and Engineering, Fudan University, Shanghai 200433 Chapter 5 Liying Liu Department of Optical Science and Engineering, Fudan University, Shanghai 200433 Chapter 6 Shibin Jiang NPPhotonics, Arizona, USA Chapter 7 Qihong Lou Laboratory of Advanced Laser Technology, Shanghai Institute of Optics and Fine Mechanics, Chinese Academy of Sciences, Shanghai 201800 Chapter 8 Guodong Qian, Yu Yang Department of Materials Science, Zhejiang University, Hangzhou 310027 Chapter 9 Lei Xu Department of Optics Science and Engineering, Fudan University, Shanghai 200433 Chapter 10 Yigang Li, Guanghui Chen and Lei Xu Department of Optical Science and Engineering, Fudan University, Shanghai 200433 Chapter 11 Danping Chen Photon Craft Project Laboratory, Shanghai Institutes of Optics and Fine Mechanics, Chinese Academy of Sciences, Shanghai 201800 Chapter 12 Jianrong Qiu Department of Materials Science, Zhejiang University, Hangzhou 310027; Photon Craft Project Laboratory, Shanghai Institutes of Optics and Fine Mechanics, Chinese Academy of Sciences, Shanghai 201800 xi

Chapter 1

From Optical Glass to Photonic Glass

1.1 Introduction Optics is dealing with the propagation of electromagnetic wave and interaction of substance with electromagnetic radiation in light frequency. The former is geometric Optics and the later is molecular Optics. Optical glasses, which used for optics purposes are mainly light transmission media developed in 1880s by R&D works of Abbe and Schott in Germany. Manufacture of optical glass possessed mysterious and secret characters for a long time. After one century scientific research and development, the design of physical properties and chemical composition, as well as manufacturing technology have been put on modern scientific bases.'1"31 Optical glasses have also been developed from three dimensional (bulk materials) to low dimensional (thin films and fibers), which play an important role in modern optical and optoelectronic engineering. Photonics is dealing with the photon generation and detection, as well as stimulated emission, photon frequency conversion and polarization change. The invention of laser in the middle of last century was a great event for development of photonics, while photonics becomes more and more important in information science and engineering. The first definition of "Photonics" in 9th International Symposium on High Speed Photograph in 1970 was as following: "Photonics is a systematic science concerning the photon as the information carrier". Since photonpropagation speed (~1012 cm/s) is much higher than electron propagation speed (109 cm/s), therefore, photonic devices have very short time 1

From Optical Glass to Photonic Glass

2

response and super-high information capacity for single channel. As 21st century is the multi-media era and the tera-information era, the information capacity in Tb (1012 bits), or information density in Tb/cm2, information stream for transportation, storage, display and calculation in Tb/s, super-high frequency processing (modulation, switching, crossexchanging, coding and decoding) in T Hz (ps time response) are expected. The exploration of new photonic material, including photonic glasses, is a key issue for advance photonic devices. Here, photonic glasses with optical glasses perform different functional roles, such as laser generation, frequency conversion, light modulation and switching etc. In marketplace optics have already played significant roles in optical instruments, ophthalmic and medical industries, as well as optical engineering in space and nuclear industries. The photonics has only been popularized recently, in many ways an indication of its rapidly escalating importance in optoelectronic industries, such as optical communication, optical data storage and data processing, flat panel displays and many others. The development of photonics as well as related or overlapping fields with optics and optoelectronics, still needs to explore new materials, in which glassy or amorphous materials are the important ones. 1.2 Physical Fundamentals Optical properties of glasses are much concerning with molecular optics: light refraction, absorption, reflection and scattering based on electromagnetic radiation theory.t4] In the classic electromagnetic radiation theory the induced electrodipole moment P by electromagnetic field is P = P(1)+/>(2)+P(3) , where t* =aE, F^^l/Ijfi-E-E, I*3)=(\l6)rE-E-E.

(1.1)

l)

In which a, fi, y is the second rank, the third rank and the fourth rank polarization tensor respectively. E is the electrical field vector. In the weak electromagnetic field (classic optics) only P*1} should be considered. Therefore, the optical effect is linear. Optical properties with

3

1.2 Physical Fundamentals

classical optical glasses are linear ones and based on two main optical effects: optical dispersion and scattering. The optical functions of photonic glasses are always happened with intense electromagnetic field, where by the nonlinear optical effects are induced. In the intense electromagnetic field, the P®\ P (3) should be important, therefore the optical properties of photonic glasses are mostly nonlinear ones. 1.2.1 Light scattering The spontaneous light scattering of optical glass involves molecular vibration. Polarization tensor i* 1 ' is expanded into Taylor series. Pa> = a 0 c o s c y + (dak /dQk )Qk cosco0tcos(cot + 8k),

(1.2)

where Qk is the normal coordinate of vibration frequency cok, Sk is the phase factor. The first term is Rayleigh scattering, and second one is Raman scattering, we have studied them in optical glasses.[4] During intense coherent light (laser) irradiation it is accompanied by coherent light scattering, resulting in stimulated light scattering, e.g. stimulated Raman scattering and stimulated Brillouin scattering by P^ action. Differing from spontaneous light scattering, the incident photons are scattered by excited phonons rather than thermo-vibration phonons. The scattering light is also coherent, since the coherent light scattering is nonlinear one, we have to pay a great attention in photonic glasses. 1.2.2 Optical dispersion Optical dispersion can be described by classical electrodynamics and quantum mechanics. Absorption, reflection and refraction can be expressed by optical dispersion equation 2/,

2N ,

4nNe2 TTP m

fk{cok2-co2) 2

2 2

„ „N 2

2

^(a>k -co ) +co rk

where x is the absorption coefficient. If the light absorption can be neglected, % «1

4

From Optical Glass to Photonic Glass

•1=1+

x m

k

fk(<°k -a1) (a>k2-a>2)2+co2rk2

(1.4)

If the light interaction is not in the resonant region (<%-ft>)»71 4xHe2 ^ fk n -1 = 1 +3- ' k COk -co m

(1.5)

where 71 is the damping coefficient,^ is the oscillator strength, co* is the eigen frequency. Fig. 1.1 shows the optical dispersion curve in the resonant region and non-resonant regions, it can be divided into optical normal dispersion at non-resonant region and abnormal dispersion at resonant region. It is difficult to measure the refractive index of bulk glass at abnormal region, but for thin films it can be done by optical polarized ellipsometric method. In the normal dispersion region the refractive index of glass at definite light wavelength only depends on oscillator strength and eigen frequency.

y*-^ y ^

co

(b)

Fig. 1.1 Optical dispersion graphs near the absorption band (a) and with three absorption bands (b).

5

1.2 Physical Fundamentals

1.2.3 Optical nonlinearity In the intense electromagnetic field the dispersion curve generates a Stokes displacement and shifts toward the longer wavelength as the intensity of the field increasing. As shown in Fig. 1.2, the increment of refractive index 8n can be expressed as Sn=n2\E\\

(1.6)

where n2 is the nonlinear refractive index, which can be expressed as: n2 = {2x/n0 ) r = {2nN/n0 ) # ,

(1.7)

where y is the third order and fourth rank tensor, 6 is the nonlinear polarizability, i.e. the electric field induced change of polarizability (da/dE), a is first order and second rank polarization tensor as mentioned before.

Fig. 1.2 Stokes shift of dispersion curve.

The third order optical nonlinearity is consisted of real part Re ^3\ nonlinear refraction and imaginary ^3\ nonlinear absorption resulted from saturated absorption and multi-photon absorption.

6

From Optical Glass to Photonic Glass

^3>=[(Re^3)) +(lm^»)J •

(1-8)

Optical nonlinear refractive index n2 is related with ^ 3 ) n2 =\6xm

ln2ce,

(1.9)

where c is the light velocity and s is the dielectric susceptibility. The third order optical nonlinearity ^ 3 ) or nonlinear refractive index «2 is presented in all glassy materials. In non-resonant region the third order optical nonlinearity is caused by electronic shell (cloud) distortion, which is called electronic polarization, and nuclear core displacement, where the electronic part is dominant. We have deduced the method for calculating the nonlinear refractive index of different glasses by eigen absorption wavelength or refractive index of glass at three wavelengths,[4] based on the hypotheses of Stokes displacement of eigen absorption of glass mentioned above. The third order optical nonlinearity of glasses is described in detail in Chapter 4 of this book. As the glass is an isotropic substance, there is no second order optical nonlinear effect in glass, the second order and third rank polarization tensor p should be zero. But recent research results confirm that in intense electromagnetic field if electric charge separation could be performed, the second order optical nonlinear effects, such as light second harmonic generation etc. can be observed. By photo-voltaic model, X(2)=X0KEdc,

(1.10)

where ^ 2 ) and ^ are the second and third order optical susceptibility respectively, Eic is the electric field resulted from charge separation. In Chapter 5 of this book the second order optical nonlinearty of glasses is introduced furthermore. 1.3 Optical Glasses Modern Optical glasses are composed of following branches, which still play an important role in modern optics, optoelectronics and photonics.

7

1.3 Optical Glasses

1.3.1 Transparent optical glasses The main optical characteristics of transparent optical glasses are optical dispersion and spectral transmission. The optical dispersion characteristics of optical glasses can be expressed by «d~ua diagram, here nd is the refractive index at d spectral line and «d is called Abbe value (ud=nd-l/ nF-nc) , «F and nc are the refractive index at F and C spectral lines respectively, the n?-nc is so called mean dispersion. Different sorts of optical glasses are located in the different positions in the «d~fd diagram. The UV-transparent optical glasses are high purity oxide and fluoride glasses, and the IR-transparent glasses are heavy metal oxide and non-oxide glasses.[5"71 Fig. 1.3 shows the optical dispersion characteristics of oxide and fluoride glasses, the oxide glasses possess higher «d and Od value than that of fluoride glasses. The optical dispersion characteristics of different optical glasses in near infrared region are shown in Fig. 1.4, here °2 = n2 ~V n\ ~n3, n\, «2. «3 represent the refractive index at 1 urn, 2 um and 3um wavelength respectively. The chalcogenide glasses are located at the upper right, the oxide glasses at lower right and the fluoride glasses at lower left respectively.141 1.70

>oy i

1.65 1.60 1.55 &

1.50

/

1.45

^"W>^

II

1.40 1.35

a

1.30 1.25 0 "BeFi 110 105 100 95 90 85 80 75 70 65 60 55 50

v„ v

Fig. 1.3 H
From Optical Glass to Photonic Glass 4.0

3.0

2.0

""600

400

200

0

Vi

Fig. 1.4 «2.0~v2.o diagram of different optical glasses in near infra-red region (2nm) (A) 1 As2Se3TlTe, 2 GeAsTeSeL5, 3 AsSe4, 4 GeAsSe3, 5 GeSe4, 6 As2S3, 7 Ge3.9Sb0.iS6, 8 GeAsS3; (B) 1 Si0 2 , 2 K-9(517/615), 3 ZF-l(649/338), 4 LaK-5(673/562), 5 ZF7(805/254), 6 50ZnO-50B2O3, 7 30BaO -70B2O3, 8 BaO -P 2 0 3 ; (C) 60.48ZrF3- 31.68BaF2 -3.84GdF3 -4A1F3 -4RFX 1 YF3, 2 A1F3, 3 CdF2, 4 LiF, 5 PbF2; (D) CaF2(crystal).

1.3.1.1 Abnormal dispersion optical glasses The optical dispersion characteristics of optical glasses are important in the design of compound lenses for avoiding chromatic aberration at different wavelength range. For example, to avoid chromatic aberration at visible region an achromatic lens composed of different glasses of focal length// a n d / to have the same focal length at wavelengths of the F and C lines, the condition

*Ww+*W2«/=0

( U °)

must be satisfied. If the achromic lens design is needed in the near infrared region, different optical glasses should be composed according to Fig. 1.4. There will, however, be a residual error in the intermediate wavelength range, called second-order chromatic aberration, which can be corrected by using glasses with equal partial dispersions for different Abbe values. The corrected systems are called apochromats, which are

9

1.3 Optical Glasses

rather important in modern high precision optical engineering. Therefore, so-called abnormal dispersion glasses should be developed. 1.3.1.2 Aihermal optical glasses When a beam of light propagates through a glass, which is subjected to a thermal field or powerful light field, the optical distortions of the wavefront will appear. These optical distortions are described by the following thermo-optical coefficients. In case of no thermal stress, the thermo-optical constant W is defined as: W = p + (n-\)a,

(1.11)

where a represents the linear coefficient of thermal expansion, (1/°C), and the p= dn/dT, is the temperature coefficient of refractive index. In case of existed thermal stress rather than free expansion, the optical distortions for a glass rod, the length L is larger than the diameter D (L»D), are conveniently described by the stress thermo-optical coefficient P and the stress birefringence coefficient Q:

P = J3-

aE

(1-/0 aE Q=

2(1-//)

(c,+3c 2 ),

(cx-c2),

(1.12)

(1.13)

while for a glass disk ( D » L ) , it is described by the stress thermo-optical coefficient R and stress birefringence coefficient Q': C. -l

R = a(n-l)/u-aE v

(1.14) 2

Q' = aE

(1.15) J

where ju is the Poisson's ratio, E is the Young's modulus, c\ and c2 are the stress-optical coefficients. The athermal optical glasses are applied to avoid the optical wave-front distortion for different glass elements, their

10

From Optical Glass to Photonic Glass

thermal optical constants should approach to zero. It can be realized by composing the different glass components. 1.3.1.3 Gradient index optical glass The conventional optical glasses should possess high optical homogeneity that means the refractive index of whole glass volume must be equal. Imaging occurs by means of discrete refraction at the lens surfaces using compound lens. The gradient index optical glass or GRIN glass exhibits the refractive index gradient in axial (varying along the optical axis), radial (varying outwardly from the optical axis) or spherical (varying symmetrically from a point) by artificially making. It provides an alternative to the generation of complex aspheric lens surfaces to control optical aberrations for imaging systems.'81 For an example, in a rod or fiber having a radial refractive index gradient, light propagates in a sinusoidal fashion because the light rays passing through the center travel more slowly than those further away from the optical axis, as shown in Fig. 1.5. Therefore, GRIN glass rods and fibers are widely used in micro-optics for optical fiber communication and optical imaging instruments.

Laser diode GRIN rod lens Optical fiber

GRIN rod lens Optical fiber GRIN rod lens

Fig. 1.5 The Schematic cross section of a coupler and collimator using GRIN rod lenses with small diameter and flat surface.

Different methods have been developed to create the refractive index variation in glass. These include alkali ion exchange by immersing the glass rod in an alkali salt solution, neutron or ion implantation, chemical

1.3 Optical Glasses

11

vapor deposition, as well as ion stuffing in porous glass and sol-gel techniques. Changes in refractive index of A«>0.05 can be achieved by above methods. 1.3.2 Low loss optical glass fibers and plane wave-guides Low loss optical glass fibers still play a dominant role in optical fiber communication. Optical loss depends on light scattering loss (mainly Raleigh scattering), which proportional to l/X4, eigen absorption at UV and IR, as well as impurity absorption. Fig. 1.6 shows the theoretical intrinsic loss of silica glass fiber. The minimum total loss is around 0.15 dB/Km. Decreasing the IR eigen absorption band energy and increasing UV eigen absorption band energy the theoretical total loss will be diminished. As shown in Fig. 1.7 the total loss of fluoride glass fiber is two orders lower than that of silica glass fiber, but due to technological difficulty the ultra-low loss fluoride glass fibers for long-haul transmission are not achieved up to now.[9]

^(|im)

Fig. 1.6 Theoretical intrinsic loss of silica glass 1. Total loss 2. Rayleigh scattering loss 3. UV eigen absorption loss 4. IR eigen absorption loss 5. Structural defects loss.

Fig. 1.7 Optical loss of silica glass (a) and fluoride glass; (b) 1. IR eigen absorption loss, 2. Rayleigh scattering loss, 3. Total loss.

12

From Optical Glass to Photonic Glass

For long-haul transmission, optical dispersion compensation is important, which consists of wave-guide dispersion, material dispersion and phase modulation dispersion. Transmitting glass fibers should compensate the dispersion, polarization, optical nonlinearity and inactive losses. For fused silica fiber the minimum loss wavelength is at 1.55 um, minimum dispersion is at 1.3 um and nonzero dispersion shift optical fibers (NZ-DSF) should compensate the optical nonlinearity (stimulated Raman and Brillouin scattering, self-phase modulation, four wavemixing etc.) Optical glass waveguides, including plane light integrated waveguide, wave-guide grating and array wave-guide grating, as well as waveguide modulator and isolator, play a significant role for next generation of optical communication, which has been developed in recent years. Chapter 9 in this book is given to introduce it. 1.3.3 Optical substrate glasses with low surface roughness and large size Substrate plate glass with optical quality is used for optical application, such as glass substrates for panel displays (LCD, TFT, PDP) and for data storage (magnetic and optical disks). The manufacturing process for optical substrate glasses with low surface roughness and large size is the key technology. As an example, the requirement of surface roughness for magnetic and optical disk substrates is around 2 nm in 3.5-5 inch disk size, and for panel display the surface roughness is around 10 nm in several meters size. Table 1.1 shows the substrate size for TFT-LCD with different generations. The mirror substrate glasses with precision optical surface and large size are other important application in optical engineering. The selection of glass for mirror substrates involves many factors. For front surface mirrors where retention of the optical figure is important, the thermal stability of the environment is a major consideration. The coefficient of thermal stability k can be expressed as

13

1.3 Optical Glasses

t.-fi-

r

x

V" (1.16)

aE \^p where a — coefficient of thermal expansion (1/°C), cp—coefficient of specific heat (cal/g-°C), E — Young's modulus (Pa), P,— tensile strength (kg/cm2), p — density (g/cm3), X — coefficient of heat conductivity (cal/sec-°C). The difference of values X, cp, E, Pt and/? in different glass systems is not large; therefore the main requirement for thermal stability of the environment is to acquire the lowest possible coefficient of thermal expansion of glass. The silica glass containing TiC<2 and transparent glass ceramics (such as Zerodur by Schott and ULE by Corning) is commonly used. Table 1.2 shows the comparison of properties of several glasses. Table 1.1 Substrate size and 15" panel utilized numbers of different TFT-LCD generations. Generation

3 rd

4th

5*

6*

Substrate size (mm)

550x650

680x880

1100x1250

1500x1800

Number of 15" panel utilized Time

4x

6x

15x

30x

1996-1998

2000-2002

2002-2004

2004-

Table 1.2 The properties comparison of several glasses. Material (suppl ier)

Density (g/cm3)

Thermal expansion coefficient (10_6/K-')

Hardness (Knoop)

Stress-optical coefficient (TPa 1 )

Fused silica (various) ULE (Corning)

2.20 2.21

0.55 0.03

450 460

3.5 4.0

Zerodur (Schott)

2.53

0.10

630

3.0

Pyrex (Corning)

2.23

3.2

418

3.9

BK7 (various)

2.51

8.3

520

2.7

14

From Optical Glass to Photonic Glass

1.3.4 Optical linear functional glasses The glasses with their optical functional properties proportional to external field intensity used for optical switching, modulation and isolation, are called optical linear functional glasses, they are mainly acousto-optic glasses and magneto-optic glasses. The acousto-optic and magneto-optic effects of glass are determined by composed glass components, their atomic or ionic electron configuration, that is the electric and magnetic dipole moments of composed atoms or ions. The acousto-optic and magneto-optic effects are dependant on external (acoustic or magnetic) field intensity, not on the light intensity (usually in the weak light field), therefore, we called them as optical linear functional effects, their wavelength dependences can be created by optical dispersion theory, they are quite different in the resonant region and non-resonant region. There are many acousto-optic and magneto-optic materials. Compared to crystalline materials, glasses are attractive because they are isotropic and can be prepared in large sizes and excellent optical quality. By choosing different glass composition it can be easy to select glasses to meet the demands for application in different wavelength regions. 1.3.4.1 Acousto-optic glasses Acoustic waves create a time-varying refractive index grating in the glass via the photo-elastic effect. A light beam transmitted the glass is deflected by the grating; therefore it provides fast deflection or modulation of light. The acousto-optic glasses are widely used as deflector or modulator in optoelectronic and photonic devices. The diffraction efficiency of acousto-optic materials is important; its figure of merit M can be expressed by M =* £ . py

(1.17)

where p — density (g/cm3), n — refractive index, V — acoustic velocity (m/s),/? — photo-elastic constant (Pockels coefficient).

1.3 Optical Glasses

15

Thus an acousto-optic glass, in addition to have low losses at the acoustic and optical wavelengths, should also have a large refractive index and photo-elastic constant, as well as small sound velocity. Table 1.3 shows the properties and acousto-optic (A-O) figure of merit (M) of tellurate and sulfide glasses. Table 1.3 Properties and M (A-O) of tellurate and sulfide glasses. Glass

Tellurate

Sulphide

p(g/cm )

4.4

4.1

N V(m/s) P M

2.02 3108 0.09 3.9

2.45 4375 0.18 19.0

3

1.3.4.2 Magneto-optic glasses An ion with full-filled shells has zero orbital moment and therefore has zero permanent magnetic moment, called diamagnetic ion. Ions have unfilled inner electronic shells, such as rare earth and transition-metal ions, exhibit permanent magnetic moment. In the absence of an applied magnetic field, the average magnetic moment is zero. When a magnetic field is applied, it causes a finite current around the nucleus, resulting in a magnetic moment that is opposite to the applied field for diamagnetic ions, and magnetization results from the orientation of the permanent magnetic moment for paramagnetic ions. The magnetization M is dependent on magnetic field H M = %H,

(1.18)

where x is the macroscopic magnetic susceptibility. The diamagnetic susceptibility depends on the number of atoms per volume and the number of electrons per atom, as well as the charge distribution in the atoms. Therefore the large diamagnetic susceptibility is found for high refractive index and optical dispersion glasses, containing cations with large, easily polarized outer electronic shells, such as Te4+, Bi3+ and Pb2+, as well as anions S2", Se2".

16

From Optical Glass to Photonic Glass

The paramagnetic susceptibility is dependent on the number of paramagnetic atoms per volume and the value of permanent magnetic moment. In the rare earth group, the unoccupied 4f electron shell is shielded by outer electron shell 5 s and 5p, thus leaving the orbits of the 4f electrons practically the same in the free atom. The paramagnetic susceptibility of rare-earth ion can by calculated from the total momentum quantum number J and Lande factor g values (for an electron spin, g = 2.0). The effective magnetic moment neff is + l)f2MB,

neff=g[J(J

(1.19)

where fdB is Bohr magneton, juB = (eh/2mc)

= 0.927 x 10" 20 erg/Oe(emu). 3+

3+

3+

3+

It has been predicted that, Ce , Pr , Tb , Dy magnetic susceptibility, as shown in Fig. 1.8.

(1.20)

ions possess large

Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb

Fig. 1.8 Comparison of V ( A =600nm) with gJ(J+l) for various trivalent rare-earth ions.

The Faraday effect is the rotation of the polarization plane of the linearly polarized light propagating through the glass parallel to the applied magnetic field. It has been successfully utilized in the quick optical isolation, switching and modulation. The angle of 6 is proportional to the light path through the glass and to the magnetic field strength H, 6 = V-l-H,

(1.21)

1.3 Optical Glasses

17

where the constant V is well known as the Verdet constant depending on the magnetic properties of the glass. Table 1.4 lists the Verdet constant of several commercial magneto-optic glasses at room temperature. The wavelength dependence of constant V of the glass is connected with optical dispersion and varies at 1/A2 in non-resonant region. The temperature dependence of diamagnetic Verdet constant Vdiei is very small and paramagnetic Verdet constant varies as \/T, therefore, it increases at low temperature. Table 1.4 The Verdet constant of several magneto-optic glasses at room temperature. Type FR-5(Hoya) Pr(P0 3 ) 3 FR-4(Hoya) SF-59(Schott) Si0 2

Glass +3

Tb doped borate Meta-phosphate Ce+3 doped borate High PbO contained silicate Fused silica

F/rad • T"1 • m" -71 -39.6 -30 28.5 4.0

1.3.5 Amorphous thin films The amorphous silicon a-Si, and amorphous silicon-hydrogen alloy (aSi-H) films are used for optoelectronics, such as optical shutters in spacial modulation and optical gates in panel display. The optical disks are the key parts in optical data storage science and technology. The optical storage devices function is determined by the characteristics of optical storage media films, such as read-only memory (ROM), recordable (R), rewritable (RW) or random access memory (RAM). The optical storage media materials have specific importance to high density optical data storage.[10] There are two types of amorphous thin films for rewritable optical disk storage: amorphous rare earth-transition metal (RE-TM) alloys, such as TbFeCo, for magneto-optical storage and chalcogenide glass films, such as Ge-Sb-Te for phase change storage. Recent progress of amorphous thin films for high-density optical data storage, concerning their structure, properties and storage mechanisms, is reviewed in Chapter 2 of this book.

18

From Optical Glass to Photonic Glass

1.4 Photonic Glasses Photonic glasses include laser glasses, nonlinear functional glasses and new photonic glasses. 1.4.1 Laser glasses The first solid-state laser (ruby laser) was invented in 1960, and Nd-glass laser in 1961. Ever since then the laser glasses have been the most important photonic glasses. Due to structural disorder of glass the line broadening of optical emission and phonon spectra are much larger in glass than in crystal. Laser emission can only be obtained in rare-earth (RE) ions doped inorganic glasses. The spectroscopy and preparation technology of rare earth doped laser glasses have been summarized in a monograph.[11] For deep understanding the spectroscopic behavior of REdoped new laser glasses, in Chapter 3 of this book the new developments in optics and spectroscopy of RE ions doped glasses are introduced. In recent ten years the great achievements have been got in diode laser pumped solid-state lasers (DPSSL), which promote the development of new laser glasses. Nd3+-doped laser glasses are still applied for high power laser system, and Yb3+-doped glasses are the prospective candidate. Ho3+, Er3+, Tm3+ doped glasses and laser glass fibers are developed for optical communication and IR lasers. Intensive studies have been made on the glass and glass-ceramics for light frequency up-conversion. 1.4.1.1 Nd3^-doped laser glasses for high power and high energy laser system Nd3+-doped laser glasses have been applied for high power laser systems for a long time, and quite a lot of R&D efforts have been done on Nd3+doped phosphate laser glasses.[12'13] There are several reasons for application of Nd-doped phosphate glasses, such as high stimulated emission cross-section, high energy storage density and efficiently stored and extracted energy. The absorption band at 800 nm is matched with

19

1.4 Photonic Glasses

laser wavelength of GaAlAs diode laser; therefore it is easy to perform the DPSSL devices. The high power laser systems with shorter pulse duration (~lns) are mainly applied for laser fusion experiments, the maximum output power has been higher than 100 TW. The massive quantity of Nd3+-doped laser glasses with large size and high optical quality are required. Table 1.5 lists the application of Nd3+-doped phosphate glasses in high power laser systems for laser fusion in the different countries. After great efforts made for improving the glass preparation technology, the Nd +-doped laser glasses can be manufactured in meter-scale size of glass disks for laser amplifiers application. High glass optical homogeneity (An~106) and low Pt-inclusion concentration can be achieved. Several types of Nd3+-doped phosphate glasses, which are available for laser fusion application, such as LHG-8, LG-770 and N31, were developed by Hoya (Japan), Schott (German) and SIOFM (China) respectively. Their spectral properties are rather similar, as shown in Table 1.6. Table 1.5 The application of Nd3+-doped phosphate glasses in high power laser systems for laser fusion in the different countries. Laser system (Country)

Mark of Nd3+-doped glass

Quantity of glass disk (pieces)

Max glass disk size (mm3)

Volume of glass disk (liter)

Nova (LLNL, U.S.A) NIF (LLNL, U.S.A) Omega (U.S.A) LMJ (France) FIREX-I (Japan) Shenguang II (China) ShenguangUI (China) Proto-type ShenguangUI (China)

LHG-8,LG-750 and Q-88 LHG-8 and LG-770 LHG-8 LHG-8 and LG-770 LHG-8 N21andN31

510

629x342x43

7

3072 4752 32 48

802x457x41 448x276x30 802x457x41 800x440x40 480x220x30

15 2 15 14.1

N31

144

610x330x38

7.6

N31

1152

610x330x38

7.6

864

810x450x40

14.6

3.0

20

From Optical Glass to Photonic

Glass

Table 1.6 Spectral properties of different Nd 3+ -doped phosphate glasses for laser fusion Character

LHG-8

LG-770

N31

(Hoya, Japan)

(Schott, Germany)

(SIOFM, China)

Peak emission wavelength ^p(nm)

1054

1052

1053

Stimulated emission crosssection a p (10" 20 cm 2 )

3.6

3.9

3.9

HalflinewidthAX. efl (nm)

26.5

25.4

25.4

The high energy laser systems with longer laser pulse duration (us~ns) are used for materials processing and other military applications.[14] Recently the solid state heat capacity lasers (SSHCL) are developed for high energy lasers with repetition rate operation, it gives a new challenge for development of Nd3+-doped laser glasses. Nd3+-doped glass fibers can be used for advanced high power DPSSL devices. Fig. 1.9 shows the development of output power of single glass fiber lasers. The maximum output power of Nd3+-doped fiber lasers is approaching to several KW, the laser beam quality value M2 of advanced glass fiber DPSSL laser with KW output power can be reached to l.[15] The design of fiber configuration and preparation of glass fibers, as well as characterization of glass fiber lasers are rather important, which are introduced in Chapter 7 of this book.

£2000-1 ¥ 1800

2000W»

SL

@ 1600

1530W»

| 1400-

1300W»

*> 1200 % 1000-1 s

O

0 600 Ji 400 400 200 e on

lOlOWa

800-

800Wa 610W« 48SW« 5W

9-2W

„w

3»w

now •

w o wr / *

-*—i—t—i—1—i—•—i—•—r

1992

1994

1996

1998 2000 Year

2002

2004

2006

Fig. 1.9 Development of output power of double cladding glass single fiber lasers.

21

1.4 Photonic Glasses

1.4.1.2 Yb -doped laser glasses Diode laser pumped Yb3+ doped glass has been considered as a promising high power laser glass and has attracted much attention recently.[16'17] Glass doped with trivalent ytterbium ions (Yb3+) exhibits highly efficient emission using In GaAs laser diode as pump sources. Its long fluorescence lifetime in ms range and high quantum efficiency (>90%) due to absorption band closed to luminescence line, as well as low heat loading (<10%) are quite beneficial for high power laser application. There are only two manifolds in the Yb3+ energy level scheme and it is commonly believed that excited absorption and nonradiative energy transfer between electronic energy levels should not occur. Table 1.7 lists the spectral and physical properties of Yb3+ doped commercial glasses. The diode-pumping wavelength is around 976 nm; due to broad absorption band the pumping wavelengths at 941 nm and 915 nm are also used. Table 1.7 Spectral and physical properties of Yb -doped glasses. Laser glass

Phosphate QX/Yb

Emission bandwidth AA/nm Emission cross section 0.975 um/xl0"20cm2 Emission cross section 0.97|im /xlO"20 cm2 Emission cross sectionaem(P)0.97|im /xl0" 20 cm 2 Pump saturation intensity IsaTP0.97um /kW/ cm2 Transparency intensity I(LV.trans /kW • cm2 Yb3+ doping density Ntot/102lcm"3 Absorption length 0.97|im /mm Thermal conductivity /Wcm"1 • K"1

62 0.67 0.05 0.001 370

Refractive index n Nonlinear refractive index/xlO"16 cm2 • W"1

1.3 0.25 0.07 38 1.3 -2 2.0 0.0085 1.535 3.2

Silicate Q246/Yb 77 0.71 0.095 0.002 190 1.1 0.19 0.11 62 2.1 -1.7 3.1 0.013 1.56 3.8

Fluorophosphate 81 1.2 0.16 0.002 120 1.3 0.4 0.2 26 0.5 -1.7 1.5

-1.5

_

22

From Optical Glass to Photonic Glass

Using double cladding Yb3+-doped glass fibers, high laser efficiency up to 83% and 89% respectively can be obtained.[18'19] Fig. 1.10 shows the fiber configuration and laser amplification of double cladding Yb3+doped glass fibers. Table 1.8 lists the Polaroid double-clad Yb3+-doped glass fiber parameters. Inner cladding

Axial section

Fig. 1.10 Double cladding fiber configuration and laser application schemes. Table 1.8 Polaroid double-clad laser fiber specifications. Rectangular cladding:

Fiber device characteristics:

Dimensions:330 um±10xl70 um (circumscribed rectangle)

Outer diameter:630 um±25 um

NAat915nm:0.470±0.003

Length:50~70m (draw to draw) ±lm (without a draw)

Attenuation 1100nm< 15 dB/km

Output wavelength: 1112±5nm

Absorption/unit length 915 nm :>250 dB/km

Departure of single mode core from central axis of capillary: centration : <250 um, pointing:<10mrad

Total device absorption in single pass 915nm:>13 dB Circular Inner Core: 2

Output beam I/e NA: 0.072-0.082 Core mode field radius: 4.3 urn ~4.9 um Core attenuation 1300nm:<12 dB/km Core circularity:
Slope efficiency >60% when pumped at 915nm Minimum static bend radius:2.5 cm Temperature : Operating : 10 to 45 °C, Non operating:-40 to+65 °C

1.4 Photonic Glasses

23

1.4.1.3 Glassfibersfor laser amplification in optical fiber communication The invention of wavelength-division-multiplexing (WDM) and erbiumdoped fiber amplifier (EDFA) techniques can be considered as the most important progress in optical telecommunication in recent ten years. WDM technique can provide more number of channels. By using EDFA the light signal can be directly amplified without the conversion of light/electricity/light, the long haul optical communication can be achieved. To meet the demand of growing information capacity the dense wavelength-division-multiplexing (DWDM) is developed, the installation of broadband amplifiers is the key task for realization of DWDM. Fig. 1.11 shows the transmission curve of a typical single-mode silica glass fiber for optical communication system, there are a broad band of low loss wavelength ranging from 1200 nm to 1700 nm, and low loss window (fiber loss less than 0.3 dB/km) with 25 THz(~300 nm)(from 1450 nm to 1660 nm). 0.8

i i

0.7

!

~ 0.6

I °-5 •a ^0.4 o

!J

0.3

5=5

0.2 0.1 0 1200

1300

1400 1500 1600 1700 Wavelength(nm) Fig. 1.11 Transmission curve of a typical single-mode silica glass fiber.

Rare earth (RE) ions doped glass fibers give a series of stimulated emission in the silica glass fiber at low loss wavelength range, as show in Fig. 1.12. Table 1.9 lists the potential amplification bands of several RE ions in glass fibers from 1260 nm to 1650 nm. The main active ions for DWDM broadband amplifiers are rare earth ions Er3+, Tm3+, Ho3+ and Pr3+. High gain output has been achieved by RE-doped glass fiber

From Optical Glass to Photonic Glass

24

amplifiers, for example, in the C-band the EDFA amplifier can produced 21 dB gain and output 12 dB at wavelength of 1535 nm,[21] while in the S-band (1470-1530 nm) using Tm: ZBLAN fluoride glass fibers 10~15dB gain has been obtained by Yb3+-doped glass fiber laser pumping with wavelength of 1047 nm.[22] Due to fiber wave-guide it is easy to get laser output with single 3+ longitudinal and transverse modes, the Er , Tm , Ho,3+ and Pr ions doped fiber lasers are developed quickly in recent years for near infra-red application. In Chapter 6 the recent progress of laser amplifier glass fibers is presented.

0.9

1.1 1.3 1.5 1.7 Wavelength (y m)

1.9

Fig. 1.12 Loss characteristics of silica fiber and emission bands of some rare earth ions. Table 1.9 Potential amplification bands of some RE (rare-earth) ions in glass. Operating range(nm)

Dopant RE ion

1260-1350 1320-1400 1460-1510 1500-1600 1600-1650 1700-2015

Pr3+ Nd3+ Tm3 Er3+ Ho3+ Tm3

Transition J

H5

* M3/2

J

H4

3

F4

"* M5/2 5 3

I5-5I7 F4-+3F6

O.H.

F.H

C.H.

Band

Ref.

+ + + + +

+ + + + +

0 E S C L

24 11 22 21

+

+

U

22

O.H. = Oxide host, F.H. = Fluoride host, C.H. = chalcogenide host

23

1.4 Photonic Glasses

25

1.4.1.4 Optical passive and active glass planar wave-guides For easy making integrated photonic chips the passive and active waveguides are important for next generation of optical communication, processing and computing. Planar wave-guides are prepared on the glass surfaces or silicon substrates. Wave-guide films can be prepared by CVD, sol-gel flame hydrolysis deposition (FHD), reactive ion etching (RIE) and sputtering methods, and the wave-guide patterns are made by photolithography. In glass substrates the wave-guides are buried in glass surface by ion-exchange of alkali ions between glass and molten salts.[25] The passive wave-guides are used for star and access coupling, splitting and branching light, the transmission loss smaller than 0.5dB/cm of the passive wave-guides has been obtained. The active wave-guides are always made by deposition of rare earth doped glasses on Si substrates. The most popular application of active wave-guide is for wave-guide amplifiers and combining with optical switching and isolating wave-guides in new generation of optical communication systems.[26] There are a great deal of scientific work should be done in integration of optical passive and active wave-guides, as well as optoelectronic wave-guides to form effective and compact integrated photonic chips. In Chapter 9 the design, preparation and characterization of passive and active wave-guide are presented. 1.4.1.5 Laser glasses in infra-red (IR) region In near infra-red region besides Nd3+-doped and Yb3+ doped laser glasses, which can produce laser emissions at 0.82 um, 1.06 um and 1.33 um for Nd3+-doped glasses and 1.05-1.15 um for Yb3+ doped glasses respectively, the most important laser glasses are Er3+ and Tm3+ doped laser glasses. The most important laser wavelength of Er3+ doped laser glasses is at near IR region 1.5-1.6 |am. Upon the 980 nm excitation of laser diode (LD), Er3+is first excited to4In/2 level. The excited Er3+ ions then rapidly decay to the Ahm level mainly through a non-radiative process, emission at wavelength of 1.54 |am generated by 4\\m to 4I15/2 transition, as shown in Fig. 1.13(a). Due to the weak absorption of Er3+

26

From Optical Glass to Photonic Glass

ions at near IR region and high Er3+ ion concentration quenching, for laser diode pumping around 1 |xra, wavelength, the sensitization of Er3+ emission by co-doping Yb3+ + Er3+ is always used for erbium glass lasers. High doping Yb ions in glasses can effectively absorb the pumping light in the vicinity of 1 um, and transfer energy from excited state (2F5/2) of Yb3+ ions to meta-stable state of Er3+ ions Chvi), intense emission can be produced at wavelength 1.53 um. The strong and effective laser emission can be obtained in Er3+, Yb3+ co-doped phosphate glasses by 980 nm diode lasers pumping. Er3+ doped glass lasers are useful for medical and rangefinder applications, due to laser wavelength at 1.53 um is good for eye-safe.[27] Tm3+-doped glass fibers can generate laser emission with wavelength range 1.86-2.0 um by transition of 3£Li to 3H6 as shown in Fig. 1.13(b). With improvement of optical fiber configuration design and pumping technique by diode laser array pumping at wavelength 980 nm, the output power up to hundred watts can be achieved from double cladding Tm3+-doped silica glass fibers.[28] Ho3+-doped glass fibers can also generate laser emission with wavelength around 2 um, but it is difficult to find the pumping source. By energy transfer process as shown in Fig. 1.13 (c), the laser emission of Ho3+-doped glass fibers can be produced at 2.04 um by transition of % to 5I8. Table 1.10 summarizes the recent results of laser output from ,3+ Tm and Ho doped double cladding glass fibers. .-1

*Im-

I

i

L

1

,it,.

!

"

790nm 1.9(im

|2.ll4 M m

T?V

1

1.54pm

"^

"lis

1 Tra*

(a)

(b)

Ho" (C)

Fig. 1.13 Energy level diagram of Er , Tm and Ho in glasses (a) Emission transition of Er3+-ions in glasses pumped by 790nm and 980nm laser diode (b) Emission transition of Tm3+-ions in glasses pumped by 790nm and 980nm laser diodes (c) Emission transition of Ho3+-ions in glasses pumped by 809nm laser diode.

27

1.4 Photonic Glasses

Table 1.10 Laser output characteristics of Ho3+, Tm3+ doped double cladding glass fibers. ^s

A-p

Doping concentration

Ref.

(urn)

Slope efficiency

Cavity

(urn) Tm3+: Ho3+Silica glass

2

809

33%

Tm3+2000ppm Ho3+200ppm

F-P

29

Tm3+: Ho3+Fluoride glass

2.056

805

42%

Tm3+36000ppm Ho3+4000ppm

F-P

30

Tm3+: Al3+ Silica glass

2

805

46%

Tm3+7000ppm Al3+63000ppm

F-P

31

Tm3+: Yb3+ Silica glass

2.05

805 975

805nm, 49% 975nm, 35%

Yb3+ and Tm3+ 2-wt%

Double wavelength pumping, F-P

32

Glass fibers

Up to now the laser emission in middle infrared wavelength range (3-5 urn) has not been realized in glass media, although the coherent signal sources in this wavelength are quite important. The quantum cascade semiconductor diode lasers and optical parametric oscillation in laser crystals are the main sources developed in recent years. Several attempts to explore the laser emissions directly from RE ions doped fluoride crystals have been carried out. Table 1.11 lists the possible laser emission in 3 um~5 um by using RE ions doped halide and chalcogenide hosts.[33'34] It is obvious that the laser emissions cannot be produced in oxide hosts due to the hydroxyl (OH") group absorption and high eigen absorption in middle infra-red, it has to use non-oxide glass hosts such as halide and chalcogenide glasses. Another reason to use the non-oxide glass hosts is that with longer emission wavelength the energy gap of emission transition between upper and lower levels will be smaller, therefore the phonon-assisted non-radiative transition probability is larger, it must use low-phonon energy glass hosts. As shown in Table 1.12, the non-oxide (halide and chalcogenide) glasses possess lower phonon energy. Due to the difficulty to prepare non-oxide glass and glass fibers with high optical quality the laser emission with wavelength larger than 3 um is not easy to obtain in glass hosts. The energy transfer

28

From Optical Glass to Photonic Glass

process in RE ions co-doped glass hosts can by utilized to enhance the population of upper level and to depopulate the lower level of emission transition, thus to increase the population inversion. Table 1.11 Possible laser emissions in 3~5um by using RE doped halide or chalcogenide hosts. Wavelength(um) 3.369 3.41 3.605 3.893-3.914 4.34 4.75 5.15 5.117-5.242 4.3-4.4 3.8 4.65

Lasing ion Ho3+ Er3+ Pr3+ Ho3+ Dy3+ Er3" Nd3+ Pr3+ Dy3+ Ho3++OPO Ho3++OPO

Channel 5 S2^F2 %19-^Xll
I5-5I6

H l l / 2 - » H13/2 4

l9/2—*4Ill-2 11/2

3

F3-3H6 H i 1/2—• H u / 2

5 5

S2-5F5 S2-5F5

Operating temperature(K) 300 77 110 300 300 110 300 200 300 300 300

Table 1.12 Maximum phonon energy of various glass hosts. Glass host type

Phonon energy(cm"')

Borate Silicate Phosphate Germanate Tellurate Fluoride Chalcogenide

1400 1100 1200 900 700 500 350

1.4.1.6 Frequency up-conversion laser glasses There are very few laser emissions in visible and ultraviolet-regions by RE ions in glass, the shorter wavelength laser emissions are always obtained by frequency harmonic generation resulted from interaction of laser beam with nonlinear optical crystals. Of course, it is worth to investigate the shorter wavelength lasers generated directly from RE ions doped glass itself by frequency up-conversion of RE ions in glass. In Chapter 3 of this book the mechanism and dynamics of frequency up-

29

1.4 Photonic Glasses

conversion of RE ions in glass are introduced in detail. Due to multiphoton process the radiative emission probability of frequency upconversion is low, it has to decrease the non-radiative transition probability. As mentioned above, the low phonon energy glass hosts are necessary, and for increasing the gain length the RE ions doped glass fibers are applied. Although there have been thousand of frequency upconversion emission lines, up to now the practical frequency upconversion lasers are made by diode laser pumped RE ions doped fluoride glass fibers only. The Pr , Tm and Er ions are main doping ions, and Yb3+ ions are co-doping ions for absorption of diode laser energy in near infra-red region and for energy transferring to active RE ions. Table 1.13 lists the parameters of up-conversion laser of RE ions doped ZBLAN glass fibers.[34] Table 1.13 Laser and materials parameters of up-conversion laser of rare-earth doped ZBLAN glass fibers. D.C. (10 6 ) Pr/Yb 3000/20000 Pr/Yb 3000/20000 Pr/Yb 3000/20000 Pr/Yb 3000/20000 Pr/Yb 3000/20000

Core diameter um NA

L

Pumping light

m

nm

S.E.

W

Laser parameter nm W

5

0.2

0.51

7850-860

5.5

635

1.02

19

5

0.15

3

7840

3

635

0.44

17

3

0.3

7860

0.76

635

0.3

52

3

0.3

LD852

0.088

635

0.02

49

3.5

0.16

0.3

T850-860

1.6

491

0.16

14

Tm.1000

3

0.21

0.16

0.13

482

0.04

35

Tm.1000

3

0.21

2.5

0.8

482

0.11

30

3.5

0.19

1

2.8

482

0.11

6.6-8.6

Tm/Yb 1500/5000 Tm.1000 Er.1000 Er.1000 Er.1000 D.C. = Doping

3 0.21 5.5 2.1 0.31 6 0.5 concentration,

0.250.6 0.250.6

LD11301140 LD11301140 FL1060

3.5 D645/1064 0.6 455 0.003 _ 1.2 7970 0.9 15 543 0.05 1.5 LD971 0.15 544 0.012 12-50 1.1 T792 0.5 2700 0.09 25 L = length, S.E. = Slope efficiency (%)(light to light)

30

From Optical Glass to Photonic Glass

Due to the disorder structure of glass host, the non-radiative transition probability of RE ions is larger in glass state than that in crystalline state. For increasing the emission efficiency to create the localized order of atomic arrangement is considered, that is the reason to develop transparent glass-ceramics, in which spectral behavior is similar to that in crystal. By technical improvement the light loss (mainly scattering loss) in transparent glass-ceramics has been decreased rapidly, the more efficient frequency up-conversion emission can take place in RE ions doped glass-ceramics.[35'36] Table 1.14 shows the spectral parameters of frequency up-conversion emission of RE ions doped fluoro-oxide glass-ceramics, but the most of the results are concerned with spontaneous emission. For getting stimulated emission it must decrease the scattering loss of glass-ceramics further more. Table 1.14 Glass-ceramic hosts, doping RE ions, pumping wavelength and up-conversion emission wavelength.

Er3+, Yb3+

Up-conversion Pumping wavelength (nm) emission wavelength (nm) 800 550, 660 407, 540, 660 973 591 479 490, 545, 650 800 490 800 490, 545, 660 980 415,515,590,660,450 798 Xe lamp 420 pumping 408, 552 632.8

Er3+

970

540

Tm3+, Er3+

488 978 978 455, 470, 482 1060 1060 980 980

1550, 1650 545, 696 537, 546, 663, 696

Glass-ceramics host

Rare earth ion

Si02-PbF2

Eu3+, Er3+ Er3*, Yb3+ Pr3+ Nd 3+ ,Ho 3+ , Yb3+ Nd 3+ ,Tm 3+ ,Yb 3+ Ho3+, Yb3+ Nd3+ Nd3+

Si02-LaF3 Si02-A10i.sPbF2-CdF2 Si0 2 -Al 2 0 3 -CaOCaF2 Si0 2 -W0 3 PbF2-CdF2-BaF2 Si0 2 -Na 2 0Al203-AlF3-LaF3 Ge02-NaF-BaF2 Ge02-PbO-PbF2 Ge0 2 -W0 3 -PbF 2 Ge0 2 - Al 2 0 3 -PbF 2 PbGe02-PbF2-CdF2

Tm3+, Er3+ Tm3+, Pr3+ Tm3+ Tm3+, Yb3+ Ho3+ Er3+

450, 478 365, 650, 487 490, 650 410, 550, 525, 850

1.4 Photonic Glasses

31

1.4.2 Non-linear functional glasses Nonlinear functional glasses are of great importance to the development of optoelectronic and photonic devices, the component functions depend on material properties, as shown in Table 1.15. The nonlinear optical properties of glass are consisted of third order and second order optical nonlinearities. Table 1.15 Material properties and component functions. Component function

Material property

Optical switching Bragg grating

High third order optical nonlinear constant x<3) High photo-refraction

Modulation Frequency doubling and OPO Isolation

High x<3), X(2>> acousto-optic and electro-optic constant, High x<2' High magneto-optic constant, x(3),acousto-optic constant

1.4.2.1 Third order optical non-linearity The third order optical nonlinearity is presented in all glassy materials, the dominant contribution of third order optical nonlinearity of transparent glasses (in non-resonant region) is consisted of electronic part and nuclear part, the former is induced by electron cloud deformation, that is electronic polarization of atoms and ions, the later is due to nuclear core displacement. The electronic part is dominant and occupies 80-85% of total third order optical nonlinearity. The third-order optical nonlinearity of the transparent glass (in the non-resonant region) is always defined as nonlinear refractive index n2, its relation with the third order optical susceptibility %(3) is shown in eq. (1.9). The origin and the method of calculation for the n2 and x<3)of glass was described in detail in monograph,[4] principally the nonlinear refractive index n2 of glass depends on the polarizability of glass consisted of ions, especially anions and cations with lone electron pairs, in other words the nonlinear refractive index of glass can be evaluated by linear refractive index and optical dispersion. Fig. 1.14 shows the relationship between rcd, v d and n2 of oxide and fluoride glasses. It can be

32

From Optical Glass to Photonic Glass

estimated that the chalcogenide glasses possess the highest n2 value, 103 times higher than that of silica glass. Due to the electronic cloud (shells) deformation of glass consisted of ions the nonlinear response time of glass is quick, less than pico-second (ps). Therefore, optical glasses with high nonlinear refractive index can be applied for quick optical switching and modulation.

Abbe'iiumbci va

Fig. 1.14 Oxide and fluoride glasses on an w r v d diagram with lines of constant nonlinear refractive index coefficient w2 predicted.

The composite glasses contain semiconductor or metallic nanoparticles, which are or called quantum dots, possess high third order optical nonlinear susceptibility in resonant region, the spectral transmission can be adjusted by quantum dots size effect. The approaches for increasing optical nonlineartity of quantum dots doped glasses involve increasing particle concentration, and working at near resonant wavelength, besides the quantum confinement effect utilizing dielectric confinement effect (surface polarization effect).[37] The hybrid glasses are composed of organic molecules with large optical nonlinearty and inorganic glass as a host material, which always made by sol-gel method. The approaches for improving nonlinear optical properties of organic-inorganic hybrid glasses or gels involve selecting organic compounds with high /3), increasing solubility of doping compounds and improving thermal and optical stabilities.[37]

1.4 Photonic Glasses

33

In order to enhance the third order optical nonlinearity the recent photonic glasses with high ^ 3 ) value are composite and hybrid glasses always made by sol-gel method.[38] In Chapter 4 of this book we put special emphases on the composite and hybrid optical glasses with high third-order optical nonlinearity. 1.4.2.2 Second order optical nonlinearity The second order optical nonlinearity in glass can be induced by electric field or intense laser beam. May be due to redistribution and orientation of separated electron charge, as shown in equation (1.10), the second order optical susceptibility x(2> is proportional to the third order optical susceptibility x(3)- Chalcogenide and halide glasses and glass with chalcogenide or halide quantum dots are always used for second harmonic generation. The detail of second order optical nonlinearity of glasses by electric and optic poling can be found in Chapter 5 of this book. 1.4.3 New photonic glasses By intense UV laser or ultra-short pulsed laser such as femtosecond (fs) laser irradiation some new optical phenomena have been observed in recent years, such as photo-induced darkening, photo-induced refractive index change, photo-induced valence change of rare-earth ions, photoinduced micro-crystallization, and photo-induced long phosphorescence etc. By means of these new optical effects some new glasses have emerged in recent years. 1.4.3.1 Photonic crystal fabricated from glass and glass fiber Photonic crystal is an artificial material with periodic microstructure scale to form a photonic band gap, which can be designed by nanocomposite glass structure.'391 In order to make high performance photonic crystal, it is important to create a big dielectric index difference of periodic arrangement in wavelength scale, such as metallic particles in glass, hole glass fibers, and silica gel sphere et al. Different photonic crystals can be made by different ways, such as sol-gel method, micro-

34

From Optical Glass to Photonic Glass

fabrication in glass by fs laser and micro-pattern by ion beam and electron beam.[40] Two dimensional photonic crystals fibers with photonic band gap can be made of silica glasses and other silicate, telluride and chalcogenide glasses with large difference of refractive index of core and clad.[41] Its more popular applications are used as very narrow band filter, light deflector, and isolator etc. In Chapter 11 of this book, the design and application of photonic crystals made by glass fibers are introduced. 1.4.3.2 Photo-induced data storage glasses There are several types of new glasses can be applied for photo-induced data storage. 1.4.3.2.1 Photo-gated spectral hole-burning (PGSHB) glasses In the inhomogeneous absorption spectrum to make a spectral hole by narrow line-width laser excitation (single site excitation) is called spectral hole burning, as shown in Fig. 1.15. The number of holes can be burned in whole inhomogeneous bandwidth is determined by the ratio of inhomogeneous bandwidth (A v;) to homogenous line-width (A vH). At low temperature the homogenous line-width is very narrow, therefore the ratio is large. The photo-gated spectral hole-burning is used for frequency multiplexing high density optical data storage. Due to disorder structure the glass possesses wider inhomogeneous line width and is suitable for PGSHB material. M. Nogani has summarized the persistent spectral hole-burning of RE ions doped glasses.[42]

Fig. 1.15 Photo-gated spectral hole-burning ( PGSHB) «=Av; /AV H .AVJ: inhomogeneous line broadening; AvH : homogeneous line broadening.

35

1.4 Photonic Glasses

Low valence rare earth ions, such as Eu2+, Sm2+, Ce3+ doped glasses can be used as storage media. By shorter wavelength light excitation, the photo-ionization process is as follows: Eu2+or Sm2+ Eu2+or Sm2+ <-

hvx

hv2

>Eu3+or Sm3+ + e -Eu3+orSm3+ +e

The ionized electron trapped in the defect center, this is the write process. By another longer wavelength light excitation the photoreaction can be back, this is erase process. The low valence RE ions can be obtained in glass by strong reduction condition in preparation or by strong energy beam irradiation. Fig. 1.16 shows the spectral burning hole spectra for 5D0-7F0 transition of Eu3+ ion in aluminosilicate glass at liquid herium temperature (4.2 K).t43] Recent publication has shown that the spectral hole burning can be achieved at liquid nitrogen temperature (77 K) in chalcogenide glasses.'441

CO

c

ffi

I 0.95

-15 -10 -5 0 5 10 Laser frequency offset (GHz)

15

20

Fig. 1.16 Hole spectra for the 5D0-7F0 transition of Eu3+ ion in (75-x)Si02.xA103/2. 25NaOi/2.1.0EuO3/2(x=10,15,25,35) glasses at 4,2K. In all the spectra, the burning wavelength is 576.0nm, and the burning time is 300s.

1.4.3.2.2 Three dimension (3D) optical data storage of photochromic glass It is well known that the darkening and fading upon exposure to shorter wavelength visible light can be performed in photo-chromic glasses contained Ag (CI, Br) particles, and chalcogenide glasses, but the

36

From Optical Glass to Photonic Glass

darkening or fading process is too slow. J These glasses can be used only in write-once memory, such as holographic memory. The photodarkening has been observed in common glasses by fs laser irradiation.[46] When the fs laser beam is focused into glass, thus the three dimensional recording can be performed.'471 Fs pulse laser can also induce the color change at focused spot in RE ions or Ag, Au ions containing glasses by ion valence change.'48'491 1.4.3.3 Micro-structural fabrication of glasses The micro-optical elements inside glasses, such as 3D optical waveguides, nano-gratings, and binary lenses etc. are very important for integrated photonic devices, which can be fabricated in doped and undoped glasses by fs laser irradiation. In Chapter 12 the mechanism of interaction of glass with fs laser beam and micro-structural fabrication of optical elements inside glasses are introduced in details. Conclusion After one-century development of glass property and technology optical glasses still play an important role in modern optical and optoelectronic engineering. They have been developed from 3 dimensional (bulk material) to low dimensional (thin film and fiber). The optical quality (homogeneity and loss) of modern optical glass is even higher than that of classical optical glasses, so the technology improvement and innovation are the great concerns. To explore new optical properties of glass and to meet the demands of recent photonic devices are the urgent tasks for development of photonic glasses. Much attention should be paid to the physical mechanisms of new optical phenomena in glass and to the high performances of new photonic glass products, as well as nanotechnology of glass fabrication.

References

37

References 1. Fuxi Gan, Optical Glasses (in Chinese), Science Press, Beijing, First Edition in 1964, revised edition: Vol. 1,2,1982; Vol. 3,1985. 2. Fuxi Gan, Change Rules and Calculation of Physical Properties of Silicate Glasses (in Chinese), Science Press, Beijing, 1966. 3. Fuxi Gan, The Calculation of Physical Properties and Design of Chemical Composition of Inorganic Glasses (in Chinese), Shanghai Scientific and Technical Publishers, Shanghai, 1981. 4. Fuxi Gan, Optical and Spectroscopic Properties of Glasses, Spring-Velag Publisher, Berlin, Heidelberg, 1992. 5. Fuxi Gan, "Nonoxide Glasses for Infrared Optics", High Performance Optics, Proc. Ernst. Abbe Conf., (1987) 129-147. 6. Fuxi Gan, "New Glass-forming systems and their practical application", Invited paper on 15th International Congress on Glass, Leningrad, 1989; Journal of Noncrystalline Solids, 123 (1990) 385-399. 7. Fuxi Gan, "Optical properties of fluoride glasses: a review", invited paper on 9th International Symposium on Non-oxide Glasses. Journal of Non-crystalline Solids, 184(1995)9-20. 8. Jerzy Zarzychi ed.,"Glasses and Amorphous Materials in Materials Science and Technology", Sovie 19 (1991) 649-650. 9. Fuxi Gan, "Long Wavelength (2~5um) Infrared Optical Fiber Communication," (in Chinese) Shandong Scientific Publishing, Jinan, 1993. 10. Fuxi Gan, "Advanced thin films for optical storage", invited paper of Int. Conf. on Thin Film Physics And Application., Proc. SPIE, 1519 (1991) 530-538. 11. Fuxi Gan, Laser Materials, World Scientific Publishing Co. Ltd., Singapore, 1995. 12. J. Campbell and T. Suratwala, J. Noncryst. Solids, 263/264 (2000) 318-342. 13. J. Murray, Proc. SPIE, 3492 (1998) 1-10. 14. A. Parker, S&TR, No.4(2002) 19-21; No.10 (2002)8-9. 15. Laser Focus World, Aug, 2004, 36. 16. D. W. Hughes, J. R. M. Bar, J. Phys. (D), 25 (1992) 563-585. 17. J. F. Chanteloup, G. Bourdet, et al, Proc. SPIE. 5707 (2003 ) 12. 18. N.CherkaMp/?/. Phys. 27 (1999) 365. 19. L. Goldberg, R.P. Koplow, D.A.V. Kliner Opt. Lett. 24 (1999) 673. 20. S.Tanabe, Proc. SPIE.506 (2003)34. 21. Bor-chyuan Hwang, Shibin Jiang, Tao Luo, et. al, Proc. SPIE. 5061 (2003) 41, IEEE. Photon. Tech. Lett. 13 (2001) 197. 22. T. Kasamatsu, Y.Yano. T.Ono, IEEE. Photon. Tech. Lett. 13 (2001) 31. 23. Taehoonlee Jong , Jong Heo, Je Park, et.al.Proc. SPIE. 5061 (2003) 70. 24. E. Ishikawa, H. Yanagita, et al, Technical Digest, Optical Amplifiers and Their Applications, (1998) 216.

38

From Optical Glass to Photonic Glass

25. M. F. Grant, Critical Review of Optical Science and Technology. CR 53 (1994) 5580. 26. E. Lallier, Appl. Opt. 31 (1992) 5276. 27. W. Koechner, Solid State Laser Engineering, chapt.7, Springer-Verlag, Berlin, 1996. 28. W. A. Clarkson, N. P. Barnes and P. W. Turner, Opt. Lett. 27 (2002) 1989-1991. 29. C. Ghisler, W. Luthy, H. P. Weber, Optics communication, 132 (1996) 474-478. 30. S. D. Jackson, Electronics Letters, 37 (2001) 821-822. 31. S. D. Jackson, S. Mossman, Appl. Opt. 42 (2003) 2702-2707. 32. S. D. Jackson, Opt. Lett., 28 (2003) 2192-2194. 33. T. T. Basiev, Yu. V. Orlovski, B. L. Galagan, et. al, Laser Physics. 12 (2002) 859877. 34. M. J. Weber, Handbook of Laser, CRC, 2001. 35. Y. Wang, J. Ohwaki, Appl. Phys. Lett. 63 (1993) 3268-3271. 36. V. Lar, I. Iparragnirre, Jon Azkargorta et al., Opt. Mater, 25 (2004) 201-208. 37. Fuxi Gan, Natural Science Progress in China, 9 (1999) 289-295. 38. Fuxi Gan, Sol & Gel Science. 13 (1998) 559-563. 39. J.D.Joanopoulos, R.Z.-Meade and J.N.Wim, Photonic crystals, Princeton University Press, 1995, pp.137. 40. J.C. Knight, T.A.Broeog, J.A.Birks and P.ST.J.Russell, Science. 282 (1998) 1476. 41. T.M. Monro, Y.D.West, D.W.Hewak et.al., Electron. Lett.36 (2000) 1998-2000 42. Masyauki Nogami, Proc. SPIE. 5061 (2003) 164-168. 43. K. Fujita, K.Tanaka, K.Hirao and N.Soga, Optics Lett.23 (1998) 543. 44. Woon Jin Chung, Jong Heo, Animesh Jha, J. Noncryst. Solid., 326&327 (2003) 292295. 45. S.Sakka, Photochromic Glass, in: New Glass Handbook, Asakura, 1991, pp.324-328. 46. Fuxi Gan, Proc. First Intern Workshop, "Glass and the Photonics Revolution", Glass Sci. Tech., 75C1 (2002) 63-73. 47. Xiongwei Jiang, Congshan Zhu, Fuxi Gan, Proc. SPIE. 4085 (2001) 216-219. 48. J.Qiu, K.Miura, K.Nouchi et.al., Solid State Comm. 113 (2000) 341-344. 49. H. Zeng, J. Qiu, X. Jiang, C. Zhu and F. Gan, J. Cryst. Growth, 262 (2004) 255.

Chapter 2

Structure and Properties of Amorphous Thin Films for Optical Data Storage

With the astonishing progresses of the computer and network applications, as well as audio-video (A/V) appliances the information volume has been greatly increased. 21st century is the tera-bit (1012 bits) information century. The processing, transfer and storage of the tera-bit information must be based on the characteristics of super-high density and super-high transfer rate. In recent years, optical data storage has been a very rapid developing field in response to the ever-growing demands. The optical nano-recording is arising now for huge information storage. The amorphous thin films have played an important role in optoelectronic and photonic devices, especially in optical data storage owing to easy fabrication and integration. There is no grain boundary in amorphous thin films, it is easy to get high quality thin films with very fine surface roughness, suitable for nano-storage with low-level noise. The magneto-optical (M-O) recording and phase change (P-C) recording are still the main approaches for high-density optical data storage, especially for rewritable storage. Both the recording media are amorphous; they are amorphous rare earth-transition metal (RE-TM) alloys and amorphous chalcogenide alloys. Some amorphous metallic and semi-metallic alloy thin films are also used for storage media and optical nonlinear mask (ONM) media. The structures of amorphous thin film materials exhibit quite complex and complicated features. To understand the structural relation with their properties and to interpret their storage mechanism are very important. The fundamentals of optics, as well as magneto-optical and optical recording properties of glass thin films have been introduced in a monograph before.[1] In this chapter the 39

40

Structure and Properties of Amorphous Thin Films for Optical Data Storage

recent experimental results of the structure and properties of RE-TM, P-C and ONM thin films are presented. 2.1 Amorphous Rare Earth-transition Metal (RE-TM) Alloy Thin Films 2.1.1 Principles of magneto-optical storage In magneto-optical (M-O) recording the information is stored in the form of thermally induced magnetic domains and read-out is accomplished by sensing the polarization change (optical Kerr effect) resulted in the optical beam. For high density M-O recording, it is required that the film has an axis of easy magnetization perpendicular to the film plane. The following condition should be satisfied.[2] Ku>4nMs,

(2.1)

where Ku is magnetic anisotropy for perpendicular magnetization and Ms is saturated magnetization intensity. To stabilize the written bit (magnetic domain) it requires 4nMJ[V (2d) -1/ (/+ (3d/2h)) ]
(2.2)

where He— coercivity, d— diameter of written domain, /z-thickness of the film, /c— haracteristic length of medium [\-aJ4nMs; GW—energy density of domain wall]. When the written domain diameter is much larger than the film thickness, the minimum domain size is given by the following equation. d = aj2MsHc.

(2.3)

Hence, to stabilize the perpendicular magnetic domain and to minimize its size, the values of Hc and Ku should be high. For optical recording, high signal-to-noise ratio (S/N) is needed. The S/N ratio can be expressed simply as S/N=ARPosm20k,

(2.4)

where R is the reflectivity of the multilayer film, P0 is the laser power, Ok is the Kerr rotation angle and A is the constant of the sensing system. We propose an evaluation factor F, thus for Kerr angles less than 5° F=(PoR)m0k,

(2.5)

2.1 Amorphous Rare Earth-transition Metal (RE-TM) Alloy Thin Films

41

while from equation 2.4 it would be seen that a high S/N ratio could be obtained by increasing the laser power. This strategy is limited by that a higher laser power could lead to a rise in temperature at the beam spot, and therefore to a decrease in the Kerr angle. High values of Ok (and R) are equally necessary for good M-0 recording performance. 2.1.2 Magnetic and magneto-optical properties of single layer and multilayer amorphous RE-TM films There are several approaches for increasing magneto-optical storage density, such as shortening recording laser wavelength, using magnetic super-resolution, magnetic and optical hybrid recording (or heat assisted magnetic recording). The different new amorphous RE-TM films should be developed to meet the specific requirements for increasing the storage density. 2.1.2.1 RE-TM thin films for short wavelength recording In order to increase the recording density from present levels, use of shorter wavelength (-400 nm) light is required. This approach is expected to dominate over the next generation M-O recording. An optical Kerr rotation angle (fy) is the most important requirement for M-O media. Fig. 2.1 illustrates the wavelength dependence of Ok for amorphous binary and ternary RE-TM films. It can be seen from the Kerr rotation spectra that Ok value decreases with shortening the light wavelength for the most of heavy rare earth and transition metal (HRETM) alloys. It is well known that the TbFeCo alloy films are popularly used in M-O recording, but its Ok is small at short wavelength. TbFeCo films are not suitable for high-density optical data storage with shorter wavelength laser recording. The development of new media that show large Kerr rotation at short wavelengths is being eagerly expected. The amorphous RE-TM alloys with light RE (LRE) elements, such as Nd, Pr, Ce, Sm can increase Ok at short wavelength, however, as the magnetization of (Nd, Ce, Pr, Sm) -TM films is larger than those of Tb-Fe-Co, it is difficult to prepare films with perpendicular magnetization. Early as in 1984,

42

Structure and Properties of Amorphous Thin Films for Optical Data Storage

Z.Y.Lee et al. prepared amorphous SmCo thin films with perpendicular dominant magnetic anisotropy for the first time.[3] In 1993 K.Chen prepared crystalline SmCo films with excellent magnetic anisotropy.m S.Takei et al. developed SmCo/Cu thin film structure and obtained SmCo thin films with dominant perpendicular anisotropy successfully, but the films finally transferred into crystalline.[5] Although all above researches discussed how to induce perpendicular anisotropy in SmCo thin films, but the easy axis of all SmCo thin films were only partially normal to the thin film plane, that is, the magnetic anisotropics of these SmCo films were not completely perpendicular to the thin film plane. Therefore, there should be a lot of research work on SmCo recording media ahead.

»' 300

400 500 600 Wavelength(nni)

700

800

Fig. 2.1 Wavelength dependence of Kerr rotation angle of binary and ternary amorphous RE-TM alloys 1. Gd26 (Fe8iCoi9)74; 2. Tb2i(Fe5oCo5o)79; 3 (Gd5oTb5o) 28.6Co7i.4; 4. GdTbFe; 5. Gd7Tb,3Fe80; 6. TbDyFe; 7. GdTbDyFe; 8. GdFeBi; 9. Gd26Fe74; 10 (Gd26Fe74)84Sn16; 11. Gd28(Fe90Co10)68Bi4; 12. Nd19Co81; 13. Ce21Co79.

It has been found that artificially superstructured multi-layer films (MLFs) have an axis of easy magnetization perpendicular to the film plane, and display strong interfacial magnetic anisotropy. Examples include PrGd/FeCo and Nd/FeCo superstructure multi-layer films.[6] The influence of binary layer period on Kerr rotation spectra of PrGd/FeCo MLFs is shown in Fig. 2.2. The optimized bilayer period is about 1 nm. The spectra of figure of merit F for the bilayer periodic films of NdGd/FeCo and PrGd/FeCo and the single layer film of TbFeCo are compared, as given in Fig. 2.3. The enhancement of F value for the bilayer periodic films is obvious from the spectra.

2.1 Amorphous Rare Earth-transition Metal (RE-TM) Alloy Thin Films

^0.50 DC Of

43

0.50

Pr35Gd65/Fe89Coii

Period(A)

w0.45

o A

4.0 8.0

1"0.40 «0.35 ©

>-0.30 * 0 . 2 5 400

600 800 wavelength(nm)

Fig. 2.2 Influence of bilayer period on Kerr rotation spectra of PrGd/FeCo MLFs.

350

550 Wavelength(nm)

750

Fig. 2.3 Spectra of figure of merit F for the three films NdGd/FeCo, PrGd/FeCo and TbFeCo. O modulated multilayer Pr35Gd65( 1 nm)/Fe89Con( 1 nm). • modulated multi-layer Nd36Gd64 (1 nm)/Fe89Con( 1 nm). • Single layer TbFeCo.

2.1.2.2 RE-TM thinfilmsfor magnetic super-resolution For high density M-0 data storage the magnetic alloy induced superresolution (MSR) technology is applied, which includes center aperture detection (CAD), magnetic amplifying magneto-optical system (MAMMOS) and domain wall displacements (DWD). MSR technology can make us to read the magnetic domain, which is smaller than laser spot.[7] Fig. 2.4 shows the schematic diagram of magnetic superresolution with central aperture detection (CAD-MSR) mechanism. The magnetic layer configurations are exchange-coupled double-layer (ECDL) or exchange-coupled multi-layer (ECML) consisted of the readout layer, and recording layer or the intermediate (switching) layer. In addition, the function of readout layer is to replicate magnetic domain of the recording layer by magnetic coupling or the switching layer by exchange catena force. The readout layer with RE-rich possesses high magnetization and

44

Structure and Properties ofAmorphous Thin Films for Optical Data Storage

low coercivity to make magnetic coupling easy and the recording layer with TM-rich possesses high coercivity to guarantee the perpendicular recording. Tables 2.1, 2.2 listed the materials composition and magnetic properties of ECDL and ECML for MSR.[8]

(b) Fig. 2.4 Schematic diagram of CAD-MSR (a) ECDL (b) ECML. Table 2.1 Magneto-optical materials and properties of ECDLforMRS Layer Readout Recording

Layer

Material

r Q (°c)

-*comp V W

Nd8Gd27(Fe75Co25)7o TbI9(Fe85Co,5)8i

306 250

130
Hc (KOe) <0.1 8

Table 2.2 Magneto-optical materials and properties of ECML for MSR Material Thickness Tc (°C) Hc (KOe)

Readout Switching Recording

Gd27(Fe25Co25)67 GdFe Tbj^FegsCo^g

30 10 40

-300 -140 -250

in plane -2 4

2.1 Amorphous Rare Earth-transition Metal (RE-TM) Alloy Thin Films

45

The magnetostatic coupling effect is used for MSR. Fig. 2.5 shows the temperature dependence of the Kerr rotation angle of readout layer in multilayer film. The Kerr rotation angle increases with increasing temperature, the magnetization direction of readout layer changed from in plane to perpendicular in the narrow temperature range. The sharp change is important to obtain an excellent MSR performance.191 0.7

j? 0.6 •a o.s e 0.4 o

•

/

: /

0.3 fc

0.2 i

X 0.1

•

0.0 1-~-u

-

/

J I

40

•

1

•

1

•

80 120 Temperature/'C

, 1

160

1

1

200

Fig. 2.5 Temperature dependence of the Kerr rotation angle of the readout layer in the GdFeCo/AlN/TbFeCo magnetostatically coupled multilayer film.

It is reported that the high density up to 11 Gbit/in2 has been achieved with CAD-MSR by using blue ray (406 nm) and objective lens with numerical aperture (NA) of 0.6. Combining blue ray recording and MAMMOS detection storage density up to 50 Gbit/in2 was obtained.[10] 2.1.2.3 RE-TM thin films for optical and magnetic hybrid storage The hybrid recording is a combination of the M-0 writing, which can form perpendicular magnetic domain by thermo-magnetic effect, with the magnetic flux detection, which enables high-sensitivity and high resolution by magneto-resistive sensor.[11] In perpendicular M-0 writing, the track width is determined by thermal profile, it can overcome the super-paramagnetic limitation in longitudinal magnetic recording. For further reducing the thermo-magnetic written domain size, the near-field optical approach can be adopted. In magneto-resistive detection the bit length is dependent on domain magnetic field profile through the transition parameters. The signal detection sensitivity can be increased by using giant magneto-resistive (GMR) head, thus can overcome the

46

Structure and Properties of Amorphous Thin Films for Optical Data Storage

weak signal detection of Kerr rotation angle in M-0 storage. Therefore, narrowing the track width and bit length from hybrid recording can increase the area data storage density. For perpendicular M-O recording the media should have high coercivity and low magnetization, but it is not suitable for magnetic flux detection. Therefore dual layer construction should be used for hybrid recording, the M-0 writing layer is with high He and low Ms, it exchange-couples a readout layer (or capping layer) with high Ms that can generate the magnetic flux detectable by GMR sensor. The schematic diagram is shown in Fig. 2.6. Capping layer GdFeCo T ^ S ' C TC=230'C

W I + I

Recording layer TbFeCo T ^ - 6 5 ' C TC=245'C

I Room temperature

Laser Spot

«t

** i

H

_•

+. GdFeCo

-4

I TbFeCo

T=1Q0 *C

Fig. 2.6 Schematic diagram of film configuration for hybrid storage.

Different proposals of ECDL have been given for hybrid recording. H.Saga et al. presented the film configuration, as shown in Table 2.3, for hybrid recording.[12] C.C. Lin et al. prepared TbFeCo double layer thin film; one layer is TM-rich for magnetic flux detection readout and the other layer is RE-rich for thermo-magnetic writing as well as data storage. The saturation magnetization of the readout layer reaches 370 emu/cm3 while the coercivity of the memory layer is 4.6 KOe.[13] Based on the experimental results of SmCo system, exchange coupled double

2.1 Amorphous Rare Earth-transition Metal (RE-TM) Alloy Thin Films

47

layer thin films were prepared as hybrid recording media. The Co-rich SmTbCo used for readout and RE-rich SmTbCo for recording. Table 2.4 lists the magnetic properties of each layer. Although the coercivity is very low, the magnetization of the readout layer is larger than 300 emu/cm3, which is adequate for GMR head readout. The recording layer is characterized by its high coercivity to ensure the data thermally stable.[14] It is worth to point out that hybrid recording technology requires its media behaving high Ku, large Ms at room temperature and proper temperature dependence of its magnetic properties. Among all these requirements, high Kn is especially important for higher area density recording toward 1 Tbits/in2. Crystalline FePt and SmCo thin films have larger Ku value than that of amorphous RE-TM alloy films.[15] For increasing Ku value to study the origin of the magnetic anisotropy in amorphous RE-TM thin films is quite necessary. Table 2.3 Film structure of hybrid recording disk Layer

Materials

Thickness (nm)

Ms( emu/cm 3 )

Surface protective layer

SiN

20

—

Readout layer, high Ms

TbDyFeCo

75

200-300

Writing layer, high He

TbFeCo

40

< 50

Substrate protective layer

SiN

60

Substrate

Polycarbonate

Table 2.4 Magnetic properties of the readout layer and memory layer of SmTbCo exchange coupled double layer thin film Ms (emu/cm 3 )

Mr (emu/cm 3 )

Hc (kO e )

Tc (°C)

7/comp (°C)

Readout layer

Sm 665 Tbi2.3 5 Co 81 (20nm)

340

330

L85

230

150

Memory layer

Sm^Tbz.ieCoss^ (20nm)

115

110

5.8

300


48

Structure and Properties of Amorphous Thin Films for Optical Data Storage

2.1.2 Microstructure of amorphous RE-TM alloy films and the structural origin of magnetization and magnetic anisotropy In RE-TM inter-metallic compounds and amorphous alloys with the heavy rare earths (Gd to Lu) the spin (S) is anti-parallel to the transition metal TM moment (a). Following Hund's rule J-L+S, so J is also antiparallel to the TM magnetization. With the light rare earths (b) S is also anti-parallel to the TM moment. However, Hund's rule for the light rare earths is J=L-S and, since L is larger than S, the net rare earth moment J is opposite to S and therefore in the same direction as the TM atomic moment as shown in Fig. 2.7.[16] The net magnetic moment is (2.6)

Mnet=Mve-Mm TM

RE

f

*

•\ f (A) Fig. 2.7 A simple vector model diagram showing the exchange coupling of the rare-earth elements and the magnetic transition metals.

It can be divided into two groups: ferrimagnets and ferromagnets according to electronic configuration (light RE and heavy RE). The magnetization depends on magnetic coupling between two atoms. It seems only to have atomic short-range order and no requirement for long-range order.

2.1 Amorphous Rare Earth-transition Metal (RE-TM) Alloy Thin Films

49

For high-density magnetic and magneto-optical recording, the magnetic domain axis should be perpendicular to film plane, recording media should have large magnetic anisotropy Ku and coercivity Hc. There have been several points of view concerning possible mechanisms, related to magnetic anisotropy, such as magnetic striction,[17] growth induced structural anisotropy,[18] dipolar effects[19] etc, but the most important one is the structural origins, which rely on the long range order, middle range order or short range order of the film structure. Taking the magnetic garnet film (BiAl) DyIG prepared by sol-gel method as an example, the as deposited film is amorphous (<625°C). During annealing the film is crystallized and long range order is performed, Fig. 2.8 shows the XRD patterns of (BiAl) DyIG film annealed at different temperature. From the Kerr loops of (BiAl) DyIG film, as shown in Fig. 2.9, the dominated effect of Hcmd Ku appears at

i»

(c)650°C


t.J

«f

... 10

I 20

3. T ~

.

1 30

I

sff -33

s

.

I

40

.

-

I

50

60

70

2 0 Meg

C-a

§

(e)700*C

(d)675t;

f

i 10

i 20

.

iSji

4*

<

1

i 30

,

1 40

,

2 6 /deg

1

, 50

1

, 60

1 70

10

<....!..

i„

20

30

.i

X

•

X

40 50 2 0/deg

i

JL

.

60

X.

70

Fig. 2.8 XRD patterns of (BiAl)DylG film annealed at different temperatures.

temperature higher than 625 °C. It is obviously shown that the anisotropy is contributed by long-range order of the film structure^201

50

Structure and Properties of Amorphous Thin Films for Optical Data Storage

-1

0 H/kOe

1

, _

{d)675iC

:

\

i

-3

•

i

•

\

"X\ 1

-1

.

1

.

1

.

1

.

1

0 H/kOe

Fig. 2.9 The Kerr loops of (BiAl)DylG film annealed at different temperatures.

Amorphous TM-RE films are popularly used in magnetic and magneto-optical perpendicular recording, the structural relation of the perpendicular magnetic anisotropy is an important problem. V.G. Hrris et.al obtained evidence for structural anisotropy in amorphous TbFe films using X-ray-absorption fine structure (EXAFS) measurements.[21] They found that Fe-Fe and Tb-Tb pairs correlations are greater in plane and Tb-Fe correlations are greater while perpendicular to the film plane, as shown in Table 2.5. It means that atomic arrangement Fe and Tb to film plane is anisotropic, which induced the magnetic anisotropy. Upon annealing at 300 °C the measured structure anisotropy disappears and the magnetic anisotropy decreases to a level consistent with magneto-elastic interaction between the film and substrate. Fig. 2.10 shows the Fourier transformed EXAFS data for Fe at room temperature (a) and heat treated at300°Cforlhr(b).

2.1 Amorphous Rare Earth-transition Metal (RE-TM) Alloy Thin Films

51

Table 2.5 Calculated coordination numbers (CNs) deduced from modeling of Fe and Tb EXAFS data for Tb0.26Fe0.74 sample. ||: electric field vector oriented in the plane of the sample; 1: electric field vector oriented near perpendicular to the plane of the sample.

Fe-Fe Fe-Fe Fe-Tb Tb-Fe Tb-Fe Tb-Tb

R(A)

CN"

CN 1

CN"-CNX

2.46±0.005 2.64±0.01 3.00±0.02 2.84±0.005 3.04±0.01 3.5±0.06

4.94±0.09 1.5±0.16 1.35±0.25 4.46±0.18 2.60±0.16 3.5±1

4.4±0.15 1.57±0.15 1.85±0.25 4.27±0.15 3.4±0.18 2.5±1

0.54 -0.07 -0.55 0.19 -0.80 1

It can be seen that the structural anisotropy in Fig. 2.10(a) has vanished by the annealing as in Fig. 2.10(b). But they did not give the atomic order range and image. Should the middle range order remain in amorphous RE-TM films?

0.8 0.6

'

(a>

-

A

J*V~ « H

j !

T =300 "C. 1 hr

-

• I1 11

F e EXAFS

;,. *-J.,i., •

*.*±*.*+*.i'....„

E in-plane

Ell

0.0

Fe EXAFS: Tb0MFeOM

-

0.4 0.2

•

E out-of-plane

EX 0

1 2 3 4 5 6 7 Radial Coordinate (A)

8

Fig. 2.10 Fourier-transformed Fe XAFS data for a Tbo.26Feo.74 alloy film at room temperature (a) and heat treated at 300 °C for 1 hr.

We use the magnetic effect in scanning tunnel microscope (STM) to investigate the structure of amorphous RE-TM thin films. The tunnel current of STM can be modulated by the external magnetic field. The transverse of longitudinal magnetic field is applied between the tip and the sample, the tunnel current is correlated with the strength and the direction of the external field. The experimental arrangement of transverse magnetic effect in STM and the tunnel current modulation

52

Structure and Properties of Amorphous Thin Films for Optical Data Storage

with the alternating transverse magnetic field are shown in Fig. 2.11 and Fig. 2.12, respectively.p2]

Voltage source

-t,

Fig. 2.11 Experimental arrangement of transverse magnetic effect in STM.

The STM images of sputtered amorphous TbFeCo thin film are illustrated in Fig. 2.13. By our experimental results, sputtered amorphous. TbFeCo thin films are not real amorphous, they have cluster structure (10~30 nm). In the cluster there exist nano-scale fine crystallites in 2.2-2.8 nm size. The atomic lattice spacing of nano-crystallites is about 0.2 nm. These nano-crystallite, orienting randomly (amorphous structure in atomic scale) may be contributed to the origin of perpendicular magnetic anisotropy.[23] to

ac tunnel current 0

|

> |-20 3 "3

-40

(a)

\

« — J J k

-u—

100

200

Magnetic &M (10~*T) (a)

ac magnetic field (b)

Fig. 2.12 Z voltage of piezotube depending on transverse magnetic field (a) and alternating tunnel current signal generated by an alternating magnetic field (b).

2.2 Amorphous Metallic and Chalcogenide Thin Films

(a)

53

(b)

(c) Fig. 2.13 STM image of sputtered amorphous TbFeCo thin film, (a) Cluster microstracture image with 109.5nmxl09.5nm scan area, (b) nanocrystallite image with 10.7nmx 10.7nm scan area, and (c) with 5.9nm><5.9nm scan area, (d) lattice image of monocrystals image with 1.88nmx 1.88nm scan area.

2.2 Amorphous Metallic and Chalcogenide Thin Films Phase change phenomena can be applied for optical data storage, electrical probe recording, nonvolatile semiconductor random access, and fast optical switching. The reversible phase change (crystalline oamorphous) of alloy films, irradiated by a short pulse laser, has been used for recordable and rewritable optical data storage. Due to the difference in reflectivity between the amorphous and crystalline states of the film, the readout signal can be detected by the difference of reflectivity. Among the amorphous alloy films, the chalcogenide alloys such as Ge-Sb-Te and In-Sb-Te and metallic alloys, such as Cu-Al and Cu-Ge were found to be the most suitable media for rewritable and recordable phase change recording. Heat treatment of materials with a laser beam has developed in recent years. According to theoretical estimations, heating and cooling rates up to 1014oC /s can be achieved

54

Structure and Properties of Amorphous Thin Films for Optical Data Storage

by very short (ps or fs) laser pulses, this is much faster than that can be achieved by common methods (~107oC/s).[24] Therefore, the phase transition of the laser treated materials is under extra-non-equilibrium conditions. Using ultra-short laser pulse and with quick crystallization, the high recording and erasing process can be achieved, but it needs a deep understanding of the mechanism and the kinetics of the phase change under such non-equilibrium conditions.[25] 2.2.1 Thermal dynamics of phase change media For phase change storage media in the writing the crystalline state transforms to amorphous, therefore, the lifetime of data storage is related to the thermal stability of the amorphous state of the P-C media, which should possess a larger crystallization activation energy, AEa. For highspeed erase, the amorphous state should be rapidly crystallized. This requirement is in contradiction with choosing the chemical composition of P-C media. F.Jiang and M.Okuda proposed a concept for erasable P-C films.[26] Elements with Tg/Tm<\ are found to be favorable for acceleration of erase speed of recording films, that means greater crystallization speed, while those with larger Tg are found to be useful for improvement of stability of the recording domain in the amorphous state at room temperature. Here, Tm is the melting temperature and Tg is the glass transition temperature (temperature at viscosity of 1012 Pa/s). Therefore, elements with Tg/Tm«l, and larger Tg, can be used to increase both the erase speed and stability. Table 2.6 lists the Tg and TJTm of some elements. Table 2.6 Glass transition temperatures, Tg , and ratio of transition temperature to melting temperature (Tm) Tg ITm, of some elements Element

Sb

rg(k)

180 0.2

TsITm

Ag 250 0.2

Cu 298 0.22

Co 445 0.25

Pb 152 0.25

Te 285 0.39

Ge 750 0.62

Table 2.7 shows the activation energy of crystallization (AEa) of some chalcogenide films. It can be seen that the doping effect of Cu, Co for increasing crystallization activation energy is obvious due to larger

55

2.2 Amorphous Metallic and Chalcogenide Thin Films

Ts, but the Tg/Tm of Cu and Co elements is comparably low, it is favorable to accelerate the crystallization speed. Table 2.7 Crystallization activation energies of some amorphous chalcogenide films Composition AEa(eV) AEaClO'^Jmor 1 )

Ins8Se42 1.20 1.92

In58Se23Pb19

In58Se23PbiiCu8

In58Se23PbiiCo8

1.61 2.58

2.90 4.65

3.64 5.83

Crystallization of the eutectic materials in binary and pseudo-binary systems possesses quick behavior, it is reasonable to be used for phase change media, because it is well known that the shorter crystallization time corresponds to smaller recording mark. As shown in Table 2.6, Sb possesses lower 7g/7rn value, therefore for the materials of eutectic composition (GaSb, GeSb, GaSbSn, AglnSbTe and InSbTe) Sb-rich recording media have been utilized for rapid crystallization, the composition and recording speed of eutectic materials are summarized in Tables 2.8, 2.9 respectively.1271 The crystallization time of eutectic materials is shown in Table 2.10. This table shows that the eutectic materials of GaSb and InSb have very fast crystallization time (7 ns and 15 ns) as compared with the stoichiometric materials of Ge2Sb2Te5 and GeiSb2Te4. Table 2.8 Composition for the eutectic materials of phase change memory Component GaSb GeSb GelnSbTe InSbTe InSb

Eutectic Composition Gan.6Sb88.4 Ge17Sb83 Ge2In7Sb80Te11 In9Sb77.6Te13.4 In3i7Sb68.3

Sb quantity (%) 88.4 83 80 77.6 68.3

Table 2.9 Recording speed of eutectic materials System

Eutectic Composition

Ag-In-Sb-Te Ge-In-Sb-Te In-Sb-Te Sb

Ag5In5Sb6oTe3o Ge2in7Sb8oTei, In9Sb77.6Te13.4 Sb

Recording Speed (m/s) 25 35 50 80

56

Structure and Properties of Amorphous Thin Films for Optical Data Storage Table 2.10 Crystallization Time of Eutectic Materials Materials GeTe Ge2Sb2Te5 Ge!Sb2Te4 AglnSbTe InSb GaSb Te

Crystallization Time 30ns 25ns 25ns 30-100ns 15ns 7ns 200ns

References 28 29 30 29 31 31 31

2.2.2 Meta-stable phase formation during laser irradiation Amorphous chalcogenide and metallic alloy materials crystallize into meta-stable phase at first during heat annealing and laser irradiation. Meta-stable phase are most likely a simple crystal structure: cubic or FCC. Table 2.11 shows the meta-stable phase observed in chalogenide SbTe-Ge, Ag-In-Ge, Sn-Sb-Ge and metallic alloys Cu-Al, Cu-Ag, Al-Ag, Cu-Ge thin films, during heat treatment and laser irradiation. Table 2.11 Meta-stable phase in semiconductor and metallic alloy films Semiconductor Metallic alloy

InSb, In2SbTe2, GeTe, GeSb2Te4, TeGeSnAg Cu20Al50, Cu80Ge20, CugoAgzo, Cu23Ag77, Ali8Ag77, Al18Ag8: Al58Ag42

E. EjJ

t

AG,

\ A

AGJ

>|U'

\\EJ

/ \

f4 %L/ \ 1 ^CP^Z_\^/

MA

MJ

Mj

ME

TRANSITION COORDINATE

Fig. 2.14 Two Kinetic Path for the transition equilibrium state. AG: Free Energy. Ei: Kinette Barrier.MA: Amorphous State. (i=l,2,3 ) ME: Equilibrium State. M^ Meta-stable State.

2.2 Amorphous Metallic and Chalcogenide Thin Films

57

Crystallization process prefers to undergo a step-by-step sequence. Fig. 2.14 shows a schematic diagram of two kinetic paths for the transition from a quenched amorphous state (Ma) to the equilibrium state (Me) directly and through several meta-stable states (Mj, i=l,2,3) respectively. The experimental results of step phase transition of Cu-Al (78A1 22Cu) alloy film shows that the amorphous phase goes through disordered fee phase, ordered fee phase to the equilibrium phase (Al2Cu + Al) during annealing as shown in Fig. 2.15. ] B.Laine el al.p3] investigated the isothermal crystallization process in Ge2Sb2Te5 and GeiSb2Te4 thin films using X-ray and optical transmission measurements, they found the presence of a metastable phase with the crystalline Ge!Sb2Te4 fee composition due to local fluctuation in the film composition during sputtering preparation. The results of isokinetic measurements have shown that the crystallization temperatures of Ge2Sb2Te5 and GeiSb2Te4 amorphous films are about 135 °C and 125 °C respectively, and the crystallization temperature of GeiSb2Te4 film is about 112°C. The meta-stable states can be also found in laser-irradiated microregion (written domain). Using less than ljas laser pulse irradiation, no equilibrium hexagonal phase appeared in GeSb2Te4 film, but only the meta-stable FCC phase was formed. Fig. 2.16 shows TEM micrograph and ED pattern for laser induced crystallize region of GeSb2Te4 film. It indicates only FCC phase existed. The stable hexagonal phase is expected to appear when using laser pulse longer than lus. [34] These observations are not surprising since a meta-stable crystalline phase which forms from the amorphous phase are favored kinetically. It takes shorter time and less energy from amorphous state to metastable state, so it is suitable for quick writing and erasing in optical recording. However, a potential shortcoming is associated with the use of a meta-stable of crystalline phase. That is, some stable crystals may form after repeated irradiation, and this could introduce noise as well as alter the optimum writing laser power. In addition, if the writing laser power were insufficient to melt the crystal, it would result in gradual degradation of the contrast.

58

Structure and Properties of Amorphous Thin Films for Optical Data Storage CuAlfttM, 22CuJ amorphous phase (as-deposited state)

its

AlCu S +A1 phases [equilibrium slate)

Fig. 2.15 Step phase transition state to amorphous from of Cu-Al alloy film.

Fig. 2.16 TEM image and ED pattern for the laser crystallized region of Ge-TeSb film, irradiated by a 100ns, 9.6 mW He-Ne laser pulse: (a) TEM image (15400x); (b) ED pattern (camera const.: 34.384 mm A).

2.2.3 Structure and morphology change of laser induced phase change in micro-region During laser irradiation the morphology of written domain (irradiated micro-region) depends on laser pulse duration and laser power. Fig. 2.17 shows a series of TEM morphologies and electron diffraction patterns of the irradiated spots obtained by a 17mw laser with various laser pulse durations in evaporated Al7iCu29 film (initially an fee sold solution).[35] Changes in morphology and structure can be seen. In Fig. 2.17a, a small faint mark can just be distinguished, indicating the 100 ns pulse is just below the threshold of mark making. An increase in the pulse duration to 200 ns results in the formation of an amorphous spot, plus a damage hole in the center (Fig. 2.17b). The high-resolution images of the structural state, both initially and after irradiation, are shown in Fig. 2.18a and Fig. 2.18b respectively, clearly showing a structure change from a fine crystalline state to an amorphous structure. Fig. 2.17c and 2.17d show other spots produced by a 500 ns and 1000 ns pulse respectively, characterized by a significant crystallized zone and a large damage hole.

2.2 Amorphous Metallic and Chalcogenide Thin Films

The structure is determined from the diffraction patterns as Al and Al2Cu (0) mixture phases; this is the equilibrium state according to the phase diagram.

? T ; -I

Fig. 2.17 Electron micrographs and diffraction patterns of irradiated spots in evaporated Al71Cu29film(initially an fee solid solution) by a 17 mW laser with varying pulse length (a) 100 ns pulse; (b) 200ns pulse; (c) 500ns pulse; (d) 1000 ns pulse.

Fig. 2.18 High resolution image of (a) uncooled evaporated film, showing fine crystallites, about 4 nm in size, distributed in the amorphous matrix (the structure is determined by the diffraction analysis as an fee phase with a=0.396 nm). Some {111} and {200} fringes are distinguishable; (b) amorphous spot (as in Fig. 17(b)) near the damage edge, no crystalline fringe detectable.

Fig. 2.19 (a) (b) (c) and (d) shows the high magnification TEM micrographs and ED patterns for the laser-crystallized bit irradiated by 200 ns, 16 mw laser pulse in GeSb2Te4 RF-sputtered film. Fig. 2.19 (a) is

59

60

Structure and Properties of Amorphous Thin Films for Optical Data Storage

the bright field TEM image. We can expect that the maximum temperature of the central area during the laser pulse irradiation is above the melting point, while that of the surrounding area is below the melting point, thus causing two distinct morphologies; Fig. 2.19(b) is the dark field TEM image, showing the crystal grains in the bit; Fig, 2.19(c) shows the ED pattern of the bit, the analysis shows that the bit is in a polycrystalline state with FCC structure, but it can be expected that if the cooling rate is high, the central part of the bit will be in an amorphous state after the laser pulse irradiation; Fig. 2.19(d) gives the single crystal ED mark for the biggest crystal shown in Fig. 2 J 9(b). An analysis shows that the single crystal ED spots are the reflections from a <211> oriented crystal of the FCC structure, the corresponding indices are shown in Fig. 2.19(d). Therefore, the 200 ns laser pulse irradiation on GeSb2Te4 thin film still produces a FCC structure with increased grain size due to the increased laser power-deposition.[25]

(a)(15400x)

(b)

(c)

(d)

Fig, 2.19 Shows the high magnification TEM images and ED patterns for the lasercrystallized bit, irradiated by a 20Qns, 16 mW laser pulse in GeSb/I^ film, (a): Bright field TEM image, (b): Dark field TEM image, (c): ED pattern of the bit, (d): Single crystal ED spot for the biggest crystal shown in (b).

2.2 Amorphous Metallic and Chalcogenide Thin Films

61

2.2.4 Kinetics of phase change by laser irradiation A simple thermal melting and re-crystallization model for laser induced phase change can be established. An equation treated by Bartholom, which is based on a one-dimensional heat flow model, can be used.[36] It is assumed to fit a thin film case. Fig. 2.20(a) is a calculated temperature profile of the Al7iCu29film in response to the various laser pulses. The increase in peak temperature and widening of the heat region with the increasing pulse duration is obvious. The temperature reaches over 1400 °C after short duration beyond 200 ns. This may explain the center damage of the irradiated spot as shown in Fig. 2.17 (c) that is, the high peak temperature in the center can lead to not only melting of the materials, but also yield violent splashing, hence high tension or recoil pressure as well. The size of the hole may roughly represents this violently splashed region, while the spot can be regarded as the irradiation-affected area.

(c)

Radial distance turn)

Fig. 2.20 Computer-simulated temperature rise profile with radial distance of Al7iCu29 film by a 17mW laser with various pulse irradiations. The peak tem-perature for 500 and 1000ns cases is indicated (a). The cooling process imme-diately after irradiation of the 200 and 1000 ns pulse cases, respectively (b), (c).

62

Structure and Properties of Amorphous Thin Films for Optical Data Storage

The simulated cooling process of the 200 and 1000 ns irradiated cases is shown in Fig. 2.20(b) and 2.20(c). According to this, the cooling down period is up to 10 ns in both cases, in respective of their different initial temperatures, hence a cooling rate of about 10 llo C/s is estimated. With this extreme cooling rate, an amorphous phase would be favored to form, if the materials had been melted. As seen from Fig. 2.20(a) the temperature rise with radius forms a continuous curve, from room temperature to the peak temperature; a certain region must exist where the elevated temperature is between the transition temperature and the melting temperature. The crystallization or re-crystallization can take place in this region. Therefore, the amorphous and crystalline phases can be formed in the different parts of the same irradiated spot. The similar phenomena have been observed in GeSb2Te4 thin film, as shown in Fig. 2.19. 2.2.5 fs laser induced phase change Femtosecond laser-induced phase change in films has been studied widely in recent years. This interest is focused on the quest for developing a microscopic theory of phase change and on the chance to apply the femtosecond laser induced effects in ultra-fast optical switching of optical signal processing and in read-write optical data storage.[37] Phenomena for phase change have been explored in the ultrashort pulse duration regime for a wide set of group IV semiconductors such as silicon, germanium and for binary compound semiconductors such as GaAs, GeSb or InSb.138"401 Femto-secend (fs) laser pulse has less thermal effect and can avoid the materials thermal distortion, it makes a possibility to form recording mark with very clear mark edge, leading to a substantial decrease of the mask size and hence to increase the storage capacity. Ohta et al. pointed out that the short pulse duration of 120fs recording on the phase-change media indicates a high data rate capability of more than 1 Tbit/s.[41] The kinetic phase change process by fs laser irradiation should be very fast, thus the high data rate can be achieved in P-C optical recording. Experiments have shown that fs laser irradiation can induce complete

2.2 Amorphous Metallic and Chalcogenide Thin Films

63

crystallization of amorphous semiconductor alloy film. Fig. 2.21 shows the spectral reflectivity curve and X-ray diffraction pattern of amorphous film a-Ge2Sb2Te5, crystalline film c-Ge2Sb2Te5 annealed at 200 °C and crystalline film f-Ge2Sb2Te5 induced by fs laser pulse (120 fs).[42] There is significant difference of reflectivity between amorphous and crystalline states both for heat-annealed and fs laser induced films. The similar experimental results have been achieved in amorphous Sb-rich AglnSbTe films.[43] By comparison of the X-ray diffraction (XRD) results of crystallized products of the annealed and fs laser induced

500 600 700 Wanvelength / nm

800

(a)

400

500 600 700 Wanvelength / nm

800

(b) Fig. 2.21 Spectral reflectivity (a) and X-ray diffraction pattern of semiconductor Ge2Sb2Te3 films.

64

Structure and Properties of Amorphous Thin Films for Optical Data Storage

Ge2Sb2Te5 and Sb rich AglnSbTe films, both types of XRD are essentially the same in spite of the different cooling rates. Concerning the physical process of fs-laser induced phase change, different points of view have been presented. While at femtosecond laser irradiation on a sample, the energy deposition time is much shorter than the phase change time. In addition, femtosecond laser pulse has much higher peak power, it can create small highly excited plasma before energy transfers from the excited carriers to the lattice. The surrounding atoms are still cold. After several picoseconds, energy from excited electrons transfers to thermal motion of the lattice by the emission of optical phonons. When the free carriers and the lattice come to an equilibrium temperature and the excess free carriers have been removed.[44] It is well known that the some semiconductor compounds (GaSb,GaInSb) show the explosive crystallization.[45] This explosive generation of crystallites caused by the highly exothermic transition and electron-hole contributed bond weakening. Therefore, the solid state-phase change can take place on the film surface. Solkolowshi-Tinten et al.[46] have evidenced that upon lOOfs laser pulse irradiation the final crystalline state is reached after a melting process occurred in the ps time scale. The cooling rates caused by femto -second laser in films are more than 1012 K/s and therefore the formation of amorphous phases may be expected after laser pulse ended. The recrystallization of the film is through the melted state. It is worth to study the physical process of crystallization of chalcogenide thin films induced by fs laser irradiation further more. 2.3 Nonlinear Optical Amorphous Alloy Thin Films Amorphous alloy thin films with large optical nonlinearity, such as third order optical nonlinearity or thermo-optical nonlinearity can induce transient nonlinear refractive index and transmittance change. Due to self-focusing effect, a smaller beam aperture can be obtained as the light beam through nonlinear optical thin film (mask layer). Recently the super-resolution near-field structure (Super-RENS) has played an important role in optical nano-storage. The multilayer thin film structure

2.3 Nonlinear Optical Amorphous Alloy Thin Films

65

is shown in Fig. 2.22 (1) the additional optical mask layer is applied. Fig. 2.22 (2) shows the principle of the optical mask super-resolution. Protection layer Optical mask layer protection layer Recording layer

(1)

light spot ffective aperture

(2)

Fig. 2.22 schematic diagram of Super-RENS. (1) Multi-layer structure of Super-RENS. (2) Principle of the optical mask super-resolution. (a)Conventional readout, (b) Optical mask super-resolution readout.

In conventional readout process of disk system, the optical recorded pits cannot be read out when the spatial frequency is higher than 2NA/A, where A is the laser wavelength, NA is the numerical aperture of optical head. To the contrary, by using the optical mask super-resolution, pits with a spatial frequency higher than 2NA/X can be read out because the effective aperture is restricted within a crescent-shaped area. A film structure of optical disk based on the self-masking superresolution developed by J. Tominaga et al.[47] This is called superresolution near field structure (Super-RENS), because there is a dynamic nearfield "probe" in it, which can be adjusted by changing the laser intensity. The typical structure of a super-RENS disk is "polycarbonate substrate/protection layer (170 nm) /mask layer (15 nm) /protection layer (20-60 nm) /optical recording layer (15 nm) / protection layer (20 nm) ". The mask thin film plays the role of a near-field aperture (see Fig. 2.23). The properties of optical mask layer materials are rather important to

66

Structure and Properties ofAmorphous Thin Films for Optical Data Storage

perform super-resolution effect, the main requirements are: fast response under laser irradiation, high sensitivity with low readout power, large spectra change with high contrast, high stability over a large number of readout cycles. The photo-mode mask super-resolution may be a promising approach and some amorphous thin films with high optical nonlinearity show strong potential for achieving super-resolution. Laser beam

0~6C 20~60nm

t

Marks

Aperture

Generated optical-near -field Fig. 2.23 Principle of the aperture-type super-resolution nearfield structure (SuperRENS).

2.3.1 Third-order optical nonlinear amorphous mask films. The amorphous thin films with high third-order nonlinear index and small linear and nonlinear losses are necessary for observing selffocusing (SF) effect in a very short interaction length with a low power laser. Self-focusing (SF) due to the nonlinear optical Kerr effect is a direct method to obtain smaller optical spot. Fig. 2.24 shows the optical Kerr effect of a chalcogenide amorphous film induced by fs laser, the time response is fast only in hundred fs range.f48] Arsenic trisulfide (As2S3) glass is a material with very large thirdorder non-linear susceptibility when the pump beam wavelength is contained in the Urbach tail region. It has been shown that the nonlinear

2.3 Nonlinear Optical Amorphous Alloy Thin Films

67

refractive index n2 (8.65x10 m/W) of the As2S3 glass induced by 633 nm He-Ne laser, whose photon energy falls into the Urbach tail region, is 8 orders of magnitude larger than n2 measured using near infrared light as a pump source (2.5xl0"18 m2/W at 1.06 um) [49]. The self-focusing effect in As2S3 glass has been studied by Song and his coworkers using a near-field scanning optical microscope and the first image that shows an actual self-focused area was obtained (see Fig. 2.25).[49] 4-|

3

3

Realx(3,ofAs2Se} Amplitude: 4.42

n

C

* ! 04

""%


>

V

2-

DC " 3 -4

-1000 -750 -500 -250

250

500

750

1000

500

750 1000

delayed Time(fs) 4 34.

2

Imagine V" of As2Sej Amplitude: 1.56

1 M C

0 -1 -2 -3

-1000 -750 -560 -250

' 250

Delayed Time(fs) Fig. 2.24 Femto-secend heterodyne detection of optical Kerr effect on GeAsSSe amorphous thin film, (a) real and (b) imaginary part of x(3)-

68

Structure and Properties of Amorphous Thin Films for Optical Data Storage

- 3 - 2 - 1

»

1

2

3

OisiH!!ct{micromtter) (c) Fig. 2.25 Optical images at 1.6 mW: (a) the image was taken at the glass side, where the As2S3 film was not coated, (b) image of the output beam at the As2S3 film surface, (c) the line scan images taken along the directions specified by the corresponding arrows in (a) and (b). The downward arrows in (c) indicate distortions during self-focusing.

Because of the unusually large nonlinear refractive index («2=2.64xlO"10m2/W at 690 nm) thefilamentswith minimum size of less than 0.3 nm (the bright spots at the bottom of Fig. 2.25 (b)) were observed in a 1.6 mW beam propagating through the 1.7 urn thin film at wavelength of 690nm and NA of 0.1. The spot size is approximately equal to the wavelength of light in the film (0.265 um) and is far beyond the diffraction limit of the optical test system (A,/2NA=0.69/(2x0.1)= 3.45 nm, full width at half maximum). In fact, the photo-induced effects in As2S3 glass are very complex. The graded index (GRIN) structure induced by photo-darkening effect and geometrical lens structure caused by the photo-expansion effect are concomitant with the nonlinear optical self-focusing. According to Song's analysis, the beam size reduction mainly attributes to the GRIN and SF effect. The GRIN structure effect

2.3 Nonlinear Optical Amorphous Alloy Thin Films

69

can contribute to beam reduction in initial stages of the SF effect but exhibits negligible effects for smaller self-focused beams.[49] Besides the focusing effect, super-resolution spot can be obtained through semiconductor-doped glass layer with nonlinear absorptionsaturation. T. Nagase et al. reported the transmitted beam narrowing by a CdSSe-doped glass,[50] which is usually used as color filter glass. Fig. 2.26 shows the calculated beam intensity profile transmitted through the saturation absorption materials A and B (different CdSSe-doped glass). The inset figure shows the absorption-saturation phenomenon of them. The full width at half maximum of the transmitted light is reduced by 5% in sample A and by 25% in sample B, compared to that of the incident light. This suggests that the storage density will be enhanced by about 10% and 80% by using these materials in optical recording media.

5 10 Relative position (arb. units) Fig. 2.26 Calculated beam intensity profile transmitted through the CdSSe-doped glass, the inset figure shows the transmittance of different samples as a function of the power density of the pumping beam.

This material also showed a nanosecond-order response in transmittance change. The trigger time was less than 10ns and the transmittance change reached about 30%. Better super-resolution effect can be expected with more pronounced absorption-saturation phenomenon by suppressing the

70

Structure and Properties of Amorphous Thin Films for Optical Data Storage

de-excitation of the excited electrons. The de-excitation probability can be made small if the semiconductor materials are made of hyper-fine particles and quantum size effects appear.[50] 2.3.2 Optic-Thermal nonlinear amorphous mask films The optical mask layer materials usually include the glass thin films of Sb, TiSb2Te3 et al. For the Sb film, the refractive index n changed with temperature can be approximately expressed as [51] n (7) =3.11 T<864K n (T) =3.11+0.0252 (T-864) 864K923K. The refractive index distribution of Sb thin film in the spot can be illustrated with Fig. 2.27 (a).

ttttf

Incident light

ttft

Fig. 2.27 (a) Refractive index distribution induced by strong optic-thermal nonlinearity of Sb thin films (b) Simulated physical model.

For homogeneous incident plane wave, the corresponding near-field optical profile is simulated by finite-difference-time-domain (FDTD) method, and the simulated physical model is depicted in Fig. 2.27 (b).[52]

71

2.3 Nonlinear Optical Amorphous Alloy Thin Films

The time response for optic-thermal effect in amorphous Sb film is around several nano-seconds (ns). Fig. 2.28 shows the reflectivity change during the pico-second laser pulse irradiation. By the Super-RENS the shortening optical recording pit (spot) size is obviously, Fig. 2.29 shows the comparison of static recording obtained on single layer Ge2Sb2Te5 thin film and double layer with Sb mask thin film at the same recording condition.[53]

-IS, — ^ — , — , — , — J — , — , — , — , — , — , — , — , — , —

•2

0

2

4

« >

10

12

14

Tuw/ns Laser «ils« width: 35ps

Fig. 2.28 Structural transformation of Sb thin films under picosecond laser pulse 1. 110J/m2 2.115 J/m2 3.120 J/m2 4.125 J/m2 5.130 J/m2.

(a)

(b)

Fig. 2.29 Comparison of recording pits obtained on Ge2Sb2Te5 thin film (a) Recording pits obtained in single layer Ge2Sb2Te5 thin film (b) Recording pits obtained by Sb-superRENS. Laser wavelength 650 nm, pulse duration 100 ns, laser power 15mW.

72

Structure and Properties of Amorphous Thin Films for Optical Data Storage

Recently, Liu et al.m found that thin bismuth films have unusually giant third-order optical nonlinearity at low laser intensity, which is ~105 times larger than that of composites containing Bi nano-particles.[55] The giant nonlinearity of Bi thin films may be useful for the energy concentration of the laser beam and the size reduction of recording marks. In comparison with Sb, Bi has a much lower melting point and it can reduce recording and readout powers. The static optical recording properties of super-RENS with Bi mask layer (Bi-super-RENS) were studied and compared with those of the Sb-super-RENS.[56] Recording marks were directly observed by scanning electron microscopy (SEM) and optical microscopy with a CCD camera. Fig. 2.30 shows the optical microscopy images and corresponding SEM images of the recording marks on (a) single-layered Ge2Sb2Te5 film, (b) Sb-super-RENS and (c)

Fig. 2.30 The Optical microscopy and corresponding SEM images of the recording marks on (a) single-layered Ge2Sb2Te5 film, (b) Sb-super-RENS and (c) Bi-Super-RENS with power of 14mW and laser pulse width of 100 ns using 650 nm semiconductor laser.

References

73

Bi-Super-RENS[56] with power of 14 mW and laser pulse width of 100 ns using 650 nm semiconductor laser (Optical microscopy images are on the left and SEM images are on the right). We can see that the Bi film has the similar functions as the Sb film in respect of concentrating the energy into the center of the laser beam. The static recording mechanisms of the Sb and Bi mask layer in the super-RENS may be understandable in the similar way. However, for Bi, dn/dT is only in the order of lO^/K,1281 the number is so small that the thermal-lens model could not simply account for the large nonlinearity of Bi films. Conclusion Recent experimental and theoretical research results have shown that the amorphous thin films still play an important role in the development of high density optical data storage, specially in next generation data storage-optical nano-storage. The quick change of properties in picosecond range and structural transformation in atomic scale should be studied further more. The research on fs laser interaction with amorphous films and optical storage behaviors in optical near field will provide us more new information concerning the materials structure and property understanding, as well as promote the progress of optical data storage with ultra-high storage density and data rate.

References l.Fuxi Gan, Optical and Spectroscopic Properties of Glass, Springer-Verlag, 1992, 253-275. 2. Fuxi Gan, Rare earth alloy and oxide thin films for optical storage, In: V.G. Kumar Das ed. Main Group Elements and their Compounds, Narosa Publishing House, 1996, 63-76. 3. Z Y Lee, T Numata, S Inoluchi, Y Sakurai, IEEE Trans Magn MAG-20 (5) (1984) : 1335.

74

Structure and Properties ofAmorphous Thin Films for Optical Data Storage

4. K.Chen, H.Hegde, S.U.Jen, and F.J.Cadieu, J.Appl. Phys., 73 (1993) 5923. 5. S.Takei, A.Morisako, M.Matsumoto, J. Magn. Mang.Mater., 272-276 (2004) 1703. 6. X. Y. Yu, H. Watable, S. Iwata, S. Tsunashima and M. Ychiyama, SPIE Proceedings on Third International Symposium on Optical Storage, ed., Fuxi Gan, Yunnan, China, 2053(1993)2. 7. M.Kaneko, K.Aratani, M.Ohta, Jpn. J.Appl.Phys, 31 (1992) 568-575. 8. Xianging Wang,Ming Fang,Defang Shen,Fuxi Gan, Thin Solid Films,489 (2005) 181-185. 9. Yuepin Zhang, Xianying Wang, Zuoyi Li et al, J.Mater. Sci. 39 (2004) 2233-236. 10. Y.Tanaka, M.Shinodea, K.Yamaguchi, et al, Proc. SPIE, 4090 (2000) 246. 11. T. Ransch, Data Storage, No.6 (2001) 16-22. 12. H.Saga, H.Nemoto, H.Sweda, et al. Jpn. J. Appl. Phys. Part 1,38 (1999) 1839-1840. 13. C.C.Lin, C.H.Lai, B.M. Chen, and H.P.Shieh, IEEE Trans. Magn., 37 (2001) 1399. 14. F.Jin, Z.Y.Li, Z.X.Huang, et al, J.Magn.Soc.Jpn.,2% (2004) 208. 15. Fuxi Gan, J. Cer. Soc. Japan, Suppl. 112 (2004) 755-760. 16. Richard I. Gambina, Takao Suzuki, Magneto-optical Recording Materials, IEE Press, New York, 1999,33-35. 17. S. N. Cheng, M.H. Kryder, and M.C.A. Mathur, IEEE Trans. Magn. 25 (1989) 4018 18. F. Hellman and E.M. Gyorgy, Phys.Rev. Lett. 68 (1992) 1391. 19. H.Fu, M.Mansuripur and P. Meystre, Phys. Rev, Lett, 66 (1991) 1086. 20. Yue-Pin Zhang, Xian-Ying Wang, De-Fang Shen, Fu-xi Gan, Chin. Phys. Lett. 20 (2003) 746-748. 21. V. G. Harris, K. D. Alyesworth, B. N. Dos, W. T. Elam and N. C. Koon, Phys. Rew Lett. 28 (1992)1939-1942. 22. Zhanghua Wu and Fuxi Gan, Chinese Science Bulletin, 38 (1993) 1941-1943. 23. Zhanghau Wu and Fuxi Gan, J. Magnetics Soc. Japan, Supplement, 17 (1993) 68-71. 24. P.L. Lin, R.Yan, N.Bloembergen, Appl. Phys. Lett. 34 (1979) 844. 25. Fuxi Gan, J. Non-Cryst. Solids, 256-257 (1999) 176-182. 26. F.Jiang and M.Okuda, Jpn. J. Appl. Phys. 30 (.1991) 97. 27. M.Okuda, H. Inaba and S. Usuda, Proc. SPIE, 5060 (2003) 145; Proc. MRS, 803 (2004)HH5-18. 28. M.Chen, et al, Appl.Phys.Lett, 49 (1986) 502. 29. G.F.Zhou, Mater Sci. Eng, A304-306 (2001) 73. 30. G.H.Zhou, Jan. J.Appl.Phys, 38 (1999) 1625. 31. D.J.Gravesteijin, et al, Philips Tech.Rep. 44 (1989) 250. 32. Fuxi Gan and Haibo Yan, Proceedings of Thin Films and Beam-Solid Interactions, ed. L.Hung, North-Holland (1991) 445-450. 33. B. Laine, C. Rivera-Rodringues, E. Morales-Sanchez, et al., JNon-Cryst.Solids, 345 & 346 (2004) 173-177. 34. Fuxi Gan, Sonsheng Xue and Zhengxiu Fan, Annaln der Physik 1 (1992) 391-398 35. H. Yan, and F. Gan, J.Materials Science, 28 (1993) 5382-5386. 36. BJ.Bartholom, J.Appl. Phys, 64 (1988) 3815.

References

75

37. A.Rousse, G.Rischel, S.Fourmaux, I.Uschmann, S.Sebban, G.Grillon, Ph.Balcou, Nature, 410 (2001)65. 38. J.Siegel, J.Solis, C.N.Afonso, Appl.Phys.Lett. 75(1999) 1071. 39. J.Solis, C.N.Afonso, S.C.W.Hyde, N.P.Barry, P.M.W.French, Phys. Rev. Lett. 76 (1996)2519. 40. I.L.Shumay, U.Hfer, Phys.Rev.B 53 (1996) 15878. 41. T.Ohta, M Birukawa, H.Yamamoto, K Hirao, Journal of Magnetism and Magnetic Materials.242-245 (2002) 108. 42. Guangjun Zhang, Donghong Gu, Fuxi Gan, et al.Jhin Solid Films, 474 (2005) 169172. 43. G. Zhang, D. Gu, X.Jiang and Fuxi Gan ,. Appl. Phys. A 80 (2005) 1039-1043. 44. Li Huang, J.Paul. Callan, Eli N.Glezer, Eric Mazur, Phys. Rev. Lett. 80 (1998) 185 45. V.Aleksandrov, Inorganic Materials, 28 (1992) 544. 46. K.Sokolowski-Tinten, J.Solis, J.Bialkowshi, J.Siegel, C.N.Afonso, D.von der Linde, Phys RevLett.Sl (1998) 3679. 47. J. Tominaga, T. Nakano, et.al., Appl.Phys.Lett.J3 (1998) 2078-2080. 48. Qiming Liu, Jun Mi, Shixiong Qian and Fuxi Gan, Chinese Phys. Lett. 19 (2002) 575577. 49. K.B.Song, J.Lee, J.H.Kim et al., Phys.Rev.Lett., 85 (2000) 3842-3845. 50. T.Nagase, S Ashida and K.Ichihara, Jpn.J.Appl.Phys. Part 1, 38 (1999) 1665-1668. 51. J.Wei and F. Gan, Appl. Phys. Lett. 82 (2003) 2607-2609. 52. J.Wei and F. Gan, Opt. Commun. 219 (2003) 261-269. 53. Jingsong Wei, Fei Zhou, Yang Wang, Fuxi Gan, J. Appl. Phys. 97 (2005) 0173102. 54. D.R.Liu, et.al., Opt.Lett., 11 (2002) 1549-1551. 55. Z.Pan et.al, Opt.Mater., 4 (1995) 675-684. 56. F.Zhang, W.D.Xu, Y. Wang and Fuxi Gan, Jpn.J.Appl.Phys., 43 (2004) 780-780.

Chapter 3

New Developments in Optics and Spectroscopy of Rare Earth Ions Doped Glasses

3.1 Laser Spectroscopy of Nd and Yb High Doped Glasses As mentioned above the Nd3+-doped glasses have been playing a significant role in high power and high energy output solid state laser systems,1-1'2] and the Yb3+-doped laser glasses are the prospective candidates in the future.[3] Since the application of diode laser pumping the high emission cross section and high concentration doping are required, it is worth to study the spectroscopic characteristics and concentration quenching behaviors of Nd3+ or Yd3+-high doped glasses in detail. The laser selective excitation and time-resolved spectroscopic measurements are two powerful tools to study the doped ion site structure and the energy transfer dynamics between the doped ions. 3.1.1 Laser spectroscopy ofNd3+-high doped phosphate glasses It is known that knowledge of the spectroscopic properties of Nd3+ions is of fundamental importance for manipulating laser action. These properties include spectral absorption, peak wavelength and line-width, branching ratio, emission cross-section, fluorescence lifetime, quantum efficiency, and fluorescence quenching process. The relationship between these parameters and with the chemical composition of the glass host is also of particular interest for developing new laser glasses. Above mentioned spectral research results of Nd-doped glasses have been introduced and discussed in a monograph before.[4] 77

78 New Developments in Optics and Spectroscopy of Rare Earth Ions Doped Glasses

The phosphate glass hosts are suitable for high doping Nd + ions, therefore, it is particular interest to study the laser spectral properties of Nd3+ high doped phosphate glasses.[5] The spectroscopic and laser properties of Nd3+ ions are strongly affected by the local structure at its sites and distribution of the doped Nd3+ ions in glass matrix. For a glass laser, the effects of site-to-site variations, in the large-signal gain regime, cause hole burning and reduced energy output. Therefore, the knowledge of the laser selective site spectroscopic properties for laser ions in glass is important for selecting the optimum host glass and operation conditions. 3.1.1.1 Absorption and emission properties The absorption spectrum of a phosphate glass sample (with 2.2wt% Nd203) consists of a large number of lines, as shown in Fig. 3.1, attributed to transitions occurring within 4f shell of the Nd ions. Data from this spectrum can be used to calculate the intensity parameters, branching ratios and radiative transition probabilities of fluorescence transitions of 4F3/2 —• 4Ij manifolds and so on, according to the JO theory.[6'7] The calculated J-0 parameters are listed in Table 3.1. Q2 is a measure of the covalence degree related to the established rare-earth ionligand bonds. Moreover, the Q2 parameter also reflects the non-symmetry of the Nd3+ local environment; the lower the Q2 value is the more centrosymmetrical the ion site is and the more ionic its chemical bonds with the ligands. The obtained value of Q2 in phosphate glass is 3.9, which is lower than that in silicate and borate glasses, indicates that the larger asymmetry is connected with their chain structure compared to silicate and borate glasses. The 4F3/2 —> \n , 4In/2 and 4I3/2 steady-state fluorescence spectra at room temperature is measured for the phosphate glass with 2.2wt% of Nd3+ by exciting the sample with an Ar laser (Fig. 3.2). The emission spectrum consists of three large and non-symmetric bands centered near 897, 1054 and 1325 nm, and they are in-homogeneously broadened due to site-to-site variation in the local ligand field.

3.1 Laser Spectroscopy ofNds+ and Yb3+ High Doped Glasses

79

Table 3.1 Judd-Ofelt calculation results a2 a, Q6 Transition 4F3/2 —• 4 I9/2 4 In/2 4 I3/2

3.9 4.6 5.3 Branching ratio, /? 0.431 0.478 0.09

0 30 £ 0.25 s a

1o.20 !> SO

§ 0.15 | 0.10 : a. O 0.05

ft

ML Jl

300

wl

400

, .

l/w

500 600 700 800 Wavelength >7nm

900 1000

Fig. 3.1 Absorption spectrum of the neodymium-doped phosphate glass (with 2.2wt% Nd 2 0 3 ). 1UU

.

80 jS

e

».M)

Jl

.

u

-8 •^40 £• xn §20 ts

.

•

A ' A \ \

i i 1

V

0 -20

800

A.

i

1000 1200 Wavelength Wnm

1400

1600

Fig. 3.2 Room-temperature fluorescence spectrum of Nd3+ in phosphate glass (with 2.2wt% Nd 2 0 3 ).

80 New Developments in Optics and Spectroscopy of Rare Earth Ions Doped Glasses

Table 3.2 lists the emission properties of Nd +-doped phosphate glasses,A,p is the emission peak wavelength, AA,eff is the effective emission line-width, a p is the stimulated emission cross-section, n is the refractive index, xexP is the measured lifetime. Table 3.2 Room-temperature emission properties of Nd3+ (2.2wt% of Nd 2 0 3 ) in phosphate glass A,p/nm 4

4

F3/2-* I,i/2 1054

AA.eff/nm 25

o-p/l(r20cm2 4.0

n 1.52

Texp fa 330

The 4F3/2 —• 4In/2 emission cross-section in Nd3+-doped phosphate glasses is much higher than those obtained in other glass systems.[8'9'10] The larger cross section is mainly due to the narrower effective emission line-width. In phosphate glasses, there is a systematic decrease in the effective line-width with decreasing charge and increasing size of the glass-modifying cations. Therefore the introduction of K 2 0 and BaO modifiers has the effect of reducing the emission bandwidth. 3.1.1.2 Non-radiative transition It is well known that the emission efficiency Cne) is determined by the ratio of radiative transition probability Pr and non-radiative transition probability Pnon, it can be expressed by the ratio of calculated lifetime rc (1/P r ) and measured lifetime Tm (1/P„0„) rie=PrIPnon=*J*c-

(3-1)

The non-radiative transition probability is mainly composed of nonradiative transition by the interaction between active ions itself and with un-active ions, as well as with glass host. The last one is glass host phonon assisted non-radiative transition probability, which is determined by glass phonon energy and the energy gap between upper and lower levels of the transition (9040 cm"1, 4 F 3/2 —%m for Nd3+). Owing to the high vibration frequencies of the phosphate hosts (1121cm"1), this process is more pronounced in phosphate glass than in other oxide glasses. It can be measured by luminescence decay change. The decays of the 4F3/2 —>• 4In/2 transition are measured with a narrow band tunable

3.1 Laser Spectroscopy ofNd3* and Yb3+ High Doped Glasses

81

dye laser exciting the sample at the 4I9/2 -» 4G5/2 absorption band (575 nm). The measured fluorescence lifetimes at room temperature of seven Nd-doped phosphate glass samples with different Nd3+ concentrations are displayed in Fig. 3.3. It can be observed that the lifetime x decreases linearly with Nd3+ ions increasing. The rate of concentration quenching in phosphate glass is lower than that in silicate glass, because the interactions of Nd-Nd ions are weaker subjected to the chain structure of phosphate glass. 500 r

q t-

s a

•

400 •

, 300 •

1

200 •

S §

100 •

u

E

0

Nd ions concentration n/1010cm~3 Fig. 3.3 Lifetime of the 4F3/2 state as a function of the neodymium ion.

The energy transfer processes in case of higher Nd3+ concentrations are of cross-relaxation between Nd3+ ions, the non-radiative transition probability is proportional to Nd3+ ion volume concentration square.[11] The interaction of Nd3+ ions with OH" anion is a particular case for phosphate glasses. Due to the hydrophilicity the content of OH" groups in phosphate glasses is considerably higher than that in silicate, borate and fluoro-phosphate glasses. By dehydration treatment during the glass melting the hydroxyl group absorption of tested glass samples is low, less than 1.0 cm"1 at 3000 cm"1 wavelength. The phonon assisted energy transfer from the 4F3/2 level to OH group plays an important role in phosphate glasses and reduces the 4F3/2 state lifetime. 3.1.1.3 Site-dependent effects Taking advantage of the tunability and narrow bandwidth of the Ti:

82 New Developments in Optics and Spectroscopy of Rare Earth Ions Doped Glasses

sapphire ring laser as an excitation source for the 4I9/2 -» 4F3/2 transition, the emission spectra of the 4F3/2 —> 4In/2 transition at different excitation wavelengths along the 4I9/2 -» 4F3/2 absorption band of the Nd3+-doped phosphate glasses can be obtained. Fig. 3.4 shows the steady-state emission spectra obtained at different excitation wavelengths measured at 77K. None of the laser-excited fluorescence spectra show the extreme line narrowing in spite of the good resolution of narrow bandwidth of Ti: sapphire ring laser. Instead, the line-widths of the 4F3/2 —» 4lm transition are much broader than the homogeneous widths. This is attributed to a residual inhomogeneous broadening, which occurs for non-resonant fluorescence, because the excitation energy of ions in physically distinct sites may accidentally coincide while the location of other levels remains different.[12] However, the characteristics of site-selection of spectra are still exhibited. As it can be observed, the spectrum slightly narrows and red-shift occurs as the excitation wavelength increases, and the shape of the emission band changes as the excitation wavelength varies. 5

§4 .o

Is & 1 2 'e a. o

877mn

—

875iun 873nm870ran-

1000

1020

1040

1060

1080

1100

Wavelength X/nm Fig. 3.4 Steady-state emission spectra of the 4 F 3/2 -> 4 I n / 2 transition in phosphate glass with 2.2wt% of Nd 2 0 3 for different excitation wavelengths. Measurements were performed at 77K.

The fluorescence decay for the 4 F 3/2 -> 4In/2 emission is also measured as a function of the excitation wavelength along the lm -> G5/2 absorption band at 6K by using a dye laser. The variations in the time x with excitation wavelength are shown in Fig. 3.5, which displays that longer decay time occurs at shorter excitation wavelengths and vice versa, shorter decay time occurs at longer excitation wavelengths.

3.1 Laser Spectroscopy ofNd3+ and Yb3+ High Doped Glasses

83

567 568 569 570 571 572 573 574 575 576 Wavelength A/nm Fig. 3.5 Lifetime t of the 4 F 3/2 state as a function of excitation wavelength for the phosphate glass with 2.2wt% of Nd 2 0 3 . Lifetime were obtained at 6K and collecting the fluorescence at the emission peak of the F3/2 -* I11/2 transition.

These results suggest that the Nd3+ ions in phosphate glasses have a different nearest neighbour coordination. At the shorter wavelength excitations we should observe the emissions from Nd3+ ions with higher field strength of ligand Dq, whereas at the longer wavelength excitations the emission from lower field strength of ligand Dq sites becomes dominant. According to non-radiative transition theory,[13] the ions with larger Dq values have a larger energy gap between the excited state and ground state, higher radiative transition probability and so possess shorter emission wavelength, lower non-radiative decay rates and larger fluorescence lifetime. The ions with smaller Dq values have reverse properties. Therefore, the peaks of emission wavelengths shift to high energy and longer lifetime with excitation energy increasing. The distribution of lifetime as a function of excitation energy shows us a picture of the type of local coordination around Nd ions. However, mpared to other Nd3+-doped glasses system,1141 the smaller site-to-site variation in the fluorescence lifetime indicates that the phosphate glass has smaller site-to-site variations in local field and a comparatively low degree of inhomogeneous broadening, so it has a relatively higher emission cross-section.

84 New Developments in Optics and Spectroscopy of Rare Earth Ions Doped Glasses

3.1.2 Laser spectroscopy ofYb

high doped glasses

Yb3+ ions possess wide absorption band (Av~18 nm), it is easy to couple InGaAs diode laser pumping without strict temperature control. The laser wavelength is close to the absorption band, and the quantum efficiency should be high (>90%), thus the heat effect is low. The emission lifetime is high, this allows more energy to be stored at high power pumping. There are only two manifolds in the Yb3+ energy level scheme, it is commonly believed that excited absorption and non-radiative energy transfer between electronic energy levels should not occur. Yb3+-doped glasses have been considered as a promising high power laser glasses and have attracted much attentions. The spectroscopic properties and laser parameters of Yb3+ ions in different glasses have been studied before,[15~20] the laser selective excitation and time resolved spectra of high doping Yb3+ phosphate glasses are reported recently.[21] The summary of the these research results is presented in this section. 3.1.2.1 Absorption and emission spectra The absorption and emission spectra of Yb3+ doped phosphate (PNK), telluorogermanate (GTN), niobosilicate (SN), borate (BL), fluorophosphates (FP) and fluoride (FL) glasses are given in Fig. 3.6 and Fig. 3.7. The absorption and emission bands of Yb3+ in different glass hosts are listed in Table 3.3. Table 3.3 Absorption and emission bands of Ytterbium doped phosphate, borate, silicate and tellurate glasses, fluorophosphate and fluoride Glass Phosphate Borate Silicate Tellurate Fluorophossphate Fluoride

Absorption band (nm)

h

h.

^3

942 940 939 932 930 930

I I I 956 950 /

971 973 974 975 970 970

Emission band (nm) X2 X3 A-i 1006 972 984 974 988 1010 994 1017 975 995 1018 975 / 1005 970 970 / 1000

3.1 Laser Spectroscopy ofNd3+ and Yb High Doped Glasses

900

850

950 1000 Wavctength/nm

1050

900

1100

85

950 1000 1050 Wavelength/nm

1100

- Absorption ' Emission

= 20 V ~ IK

•£»«

g 1" c 12

it <

2

e-

850

900

950 1000 1050 Wavelenglh/nm

900

950 1000 Wavelength/nm

1050

1100

Fig. 3.6 Absorption and emission spectra of Yb3+-doped glasses (a) PNK; (b) GTN; (c) SN; (d) BL. ill

I J';

9 8

"1 7

1.0

;.!

:k

£S (0

!

.-— 1 L/~\

h

2

:

1

-^

0 850

900

1

o1—1 |0.6 •

*'l\ ! i

?4

•'••• Emission

~s if 0.8 -

V\

'•g 0.4

£

1\ \ I \ \

10.2 •-*-—

950 1000 1050 1100 Wavelength (nm) (a)

u

n n

I *

850

900

950 1000 W a v e l e n g t h Wnm

J>

1050

1100

(b) 3+

Fig. 3.7 Absorption and emission cross-section of Yb -doped glasses (a) FP; (b) FL.

86 New Developments in Optics and Spectroscopy of Rare Earth Ions Doped Glasses

3.1.2.2 Absorption of high Yb + doped phosphate glasses at low temperature Due to low concentration quenching effect the high Yb3+ ions doped phosphate glasses are noted as good laser media for thin slab and fiber lasers. The molar composition of studied Yb3+ doped phosphate glass is (60-65) P 2 0 5 : (4-8) B 2 0 3 : (5-10) A1203: (10-15)K2O: (5-10) BaO: (0-2) La 2 0 3 : (0-2)Nb2O5: (4-6)Yb203. Fig. 3.8 shows the absorption spectra of Yb3+ in phosphate glass at different temperature (8K-300K). Because lattice vibrations are weakly coupled to the 4f electrons of the rare earths, it is the usual case that only zero-phonon lines are observed in rare earth spectra. There are four distinct peaks at low temperature, which correspond to the lowest sub-level of 2F7/2 to 2F5/2 transitions. The width of these lines remains essentially constant, which indicates that these lines show strong inhomogeneous broadening. The position of absorption peak of 2F7/2 (1) — 2F5/2 (1) transition does not shift, while the other peaks shift to shorter wavelength with the temperature increasing. The positions of the peaks at different temperatures are listed in Table 3.4. Table 3.4 The peak-position of the four absorption bands at different temperature

8K 80K 150K 200K 300K

First peak/nm

Second peak/nm Third peak/nm

918.2 918.0 917.6 Dispersion Dispersion

933.0 932.6 931.9 930.8 930.7

959.0 958.1 957.2 956.7 Dispersion

Fourth peak/nm 975.2 975.2 975.2 975.2 975.2

According to crystal-field theory, the maximum of allowed splitting number of J=5/2 level is three; therefore, the absorption spectrum should be resolved into three broad bands. However, the absorption spectrum consists of four bands. This is the fourth weaker peak besides the other three stronger peaks. Apparently, a small portion of Yb3+ ions is found in the sites distinctly different from those occupied by Yb3+ ions, which give rise to the principal spectra. The energy level diagram of Yb3+ in phosphate glass can be determined by absorption spectra at low temperature, as well as emission spectra, and is shown in Fig. 3.9.

3.1 Laser Spectroscopy ofNd3+ and Yb3+ High Doped Glasses

87

I06I6 10460 10269 2

: 800

900

1000

—

Wavelength/hm

_,_:: j

3+

F.„

1

570 372 150

7/2

Fig. 3.8 Absorption spectra of Yb in phosphate glass at different temperature. Fig. 3.9 Energy level structures of ytterbium ion in phosphate glass.

3.1.2.3 Spectroscopic parameters Due to overlap of emission and absorption bands of Yb3+, the emission cross-section measurement error is mostly attributed to the radiation trapping effect. We have already proposed the determination of emission cross-sections of Yb3+ in glasses by the reciprocity method Okm(A)

= Ohbs(A)

exp

An

(Ki-hcV) kT

(3.2)

where oab(A,) and aem(k) are the absorption and emission cross-sections at wavelength A. respectively. Zu Zu represent the partition functions for lower and upper levels. At the high temperature limit, the ratio of ZXIZ^ simply becomes the degeneracy weighing of the two states; Ez\ is the zero-line energy, which is defined as the energy separation between the lowest components of Stark split levels of the upper and lower states, h is the Planck constant, c is the light velocity, and Tis the temperature (K). The spectroscopic parameters can be calculated by FuchbauerLadenburger equations.[19] Table 3.5 gives spectroscopic properties such as peak absorption cross-section (op) and emission cross-section, which are determined from the absorption and emission spectra shown in Figs. 3.6, 3.7. The spectroscopic properties of the Yb-doped laser glasses

88 New Developments in Optics and Spectroscopy of Rare Earth Ions Doped Glasses

developed recently are listed in Table 3.6 to compare them with above glasses. The ADY, LY and PN are phosphate laser glasses which were reported to have high emission cross-sections and were recently developed by HOYA Corporation/231 QX is phosphate laser glass which was reported to have high heat capacity and the largest laser output power so far and was developed by KIGRE Company in USA,[24] and FP is fluorophosphates laser glass which was reported to have femtosecond ultra-fast pulse laser output and was developed by JENA University and the BONN-MAX Institute.[25] Table 3.5 Spectroscopic properties of Yb3+-doped PNK, GTN, SN, BL, FP glasses. Glass PNK GTN SN BL FP FL

Xp/nm 972.0 975.0 974.2 974.0 980.0 975.0

0D/pm2 0.68 1.45 1.52 0.98 0.62 0.65

0-e.r/pm2 1.09 2.29 1.92 1.93 0.68 0.65

tf/ms 2.0 0.90 1.00 0.90 1.50 1.77

Wan 1019 1030 1026 1016 1005 998

tf-Oen,/

2.07 2.06 1.92 1.74 1.02 1.15

Table 3.6 Spectroscopic properties of Yb3+-doped ADY, LY, PN, QX glasses. Glass ADY LY PN QX

XJnm 971.0 970.5 973.0 970.5

oypm 2 0.60 0.55 1.00 0.5

aem/pm2 1.03 0.80 1.35 0.70

T(/ms 1.58 1.68 1.09 2.00

A.em/nm 0.020 1.028 1.019 1.010

TfOem/ms-pm2 1.63 1.35 1.48 1.40

From Tables 3.5, 3.6 it is shown that PNK has higher emission cross section and longer fluorescence lifetime than ADY, LY and FP, and has a much higher emission cross section than QX, and has the same lifetime as QX. Although the emission cross-section of PNK is lower than that of PN, its lifetime is much longer than that of PN. In addition, in all glasses, PNK has the largest if, that is, the highest extraction efficiency. GTN has the highest emission cross section and a shorter lifetime and its is nearly equal to that of PNK; SN and BL have a nearly equal emission cross section, but the former has a longer fluorescence lifetime, and thus larger C e m *T,[ 1 9 l

3.1 Laser Spectroscopy ofNd3+ and Yb3+ High Doped Glasses

89

3.1.2.4 Laser selective excited fluorescence line narrowing (FLN) spectra Fluorescence line narrowing (FLN) spectroscopy is used for studying the local environment around the Yb3+ ions. The FLN spectra were measured at 77K under Ti: sapphire laser excitation and the 2F5/2-2F7/2 fluorescence spectra were recorded after excitation at each absorption peak according to the absorption spectra. Fig. 3.10 shows the fluorescence spectra with the excitation wavelengths corresponding to three absorption peaks (918 nm, 933 nm and 958 nm) at 77 K. The fluorescence spectrum excited by 975 nm absorption peak is not shown, because the 975 nm wavelength overlaps with the fluorescence spectra, the positions of the emission peaks are identical for excitation wavelength at 933 nm and 958 nm, and located at 975 nm, 1002 nm, 1013 nm and 1026 nm corresponding to the transitions from the lowest sub-level of 2F5/2 to the components of the 2 F7/2 level, according to energy level diagram shown in Fig. 3.9. In the case of 918 nm excitation wavelength, there is a redshift at the shortest wavelength emission band, while the other three emission bands have no obvious shift. This behavior indicates that the Yb3+ ions possibly occupy two groups of sites. The absorption peak of 918 nm probably arises from another type of site.[21] 4 3 3

•5 2 ^. in

c * •a c

1 i

0 1000

1050

1100

1150

Wavelength /nm Fig. 3.10 Fluorescence spectra of Yb3+ in phosphate glass at 77K with excitation wavelength at 918nm, 933nm and 958nm, respectively.

90 New Developments in Optics and Spectroscopy of Rare Earth Ions Doped Glasses

Additionally, we measured the fluorescence spectra at 77K with the excitation wavelength shifted inside the inhomogeneous absorption band. Fig. 3.11 and 3.12 show the site-selective fluorescence spectra with the variation of excitation wavelength inside 933 nm and 958 nm absorption lines respectively. It can be seen that similar fluorescence spectra occur with continuously changing the excitation wavelength within the absorption bands 933 nm and 958 nm respectively. It indicates that the site structure of Yb3+ ions in phosphate glass is uniform. We can still find some spectacular things, the shortest emission wavelength (2F5/2(1)2 F7/2 (1)) is sharper than the other three emission band and its location is much more sensitive to the excitation wavelength. It shifted systematically to longer wavelength as the excitation wavelength decreasing, which revealed the sited dependent behaviours of Yb3+ ions in glass attributed to distortion of the local environment of Yb3+ ions. However, comparing to the other rare earth ions, as Nd3+ ions shown above in Fig. 3.4, the variation of the fluorescence wavelength with the excitation frequency in Yb3+-doped glasses is smaller. 8 6 -5 4 ' B

2

0 950

1000

1050

1100

1150

Emission Wavelength /nm Fig. 3.11 Fluorescence of Yb3+ in phosphate glass at 77K with changing wavelength within the absorption line at 933nm.

3.1 Laser Spectroscopy ofNd3+ and Yb3+ High Doped Glasses

91

990 1020 1050 1080 E m i s s o n W a v e l e n g t h inm Fig. 3.12 Fluorescence of Yb3+ in phosphate glass at 77K with changing wavelength within the absorption line at 958nm.

3.1.2.5 Yb ions concentration effect on luminescence Table 3.7 shows the Yb3+ ion concentration effect on measured luminescence lifetime. It can be seen that the concentration quenching is very weak in Yb3+ doped phosphate and fluorophosphates glasses, it is pronounced in borate glasses. The Yb3+ ion concentration effect on luminescence in different glass hosts is similar to that of Nd +-doped glasses.1111 The optimum concentration of Yb3+ ions in borate, niobosilicate, tellurate, phosphate, fluorophosphates and fluoride glasses appears in the range of (3-5)xl0 20 , (5-7) xlO20, (3-8) xlO20, (5-11) xlO20, (7-15) xio 20 and >5xl020/cm3 respectively. The concentration quenching is attributed to the energy transfer between Yb3+ ions and cooperative upconversion. [21'26] Table 3.7 Concentration effect of Yb 2 0 3 on measured lifetime in different glasses Glass system 60TeO210La2O3-30ZnO 50P2O3-10Nb2O5-20K2O-20BaO 40SiO2-20Nb2O5-20BaO20SrO 53ZrF4-17BaF2-4LaF3-3 A1F3-10PbF2-13LiF 50B2O3-10La2O3-40ZnO 15Ba(P03)2-75 (35A1F3-15YF3-50MF2)

xm (ms) Tm= Tm= Tm= Tm= Tm= *m=

Refe-

Yb 2 0 3 %wt 3 0.81 1.60 1.0 2.4 1.15 1.6

5 0.71 2.0 0.9 2.5 1.10 1.8

7 0.60 1.90 0.7 / 0.76 1.8

9 0.64 1.85 0.64

11 0.60 1.72 0.60

/

/

0.39 0.3 2.0 2.0

rence 17 17 17 26 27 16

92 New Developments in Optics and Spectroscopy of Rare Earth Ions Doped Glasses

Due to more homogeneous site structure of Yb ions in phosphate glass, the cross-relaxation loss between Yb -Yb ions should be small; therefore, the concentration quenching effect should be weak. It can be confirmed by experiment of concentration dependence of luminescence lifetime of Yb3+-doped phosphate glass. As shown in Fig. 3.13 a constant up to 5mol% of Yb 2 0 3 can be kept, which is higher than Nd3+-doped phosphate glass in comparison of Fig. 3.13(a) with Fig. 3.13(b).

Yb phosphate

•

*

«

a

•

2

Nd ioiis concertration/xl(J!0/cm (a)

4

6

8

10

12

14

16

Yb^ions concentration/ xlO^/cm3 (b)

Fig. 3.13 Lifetime as a function of the RE ion concentration, in phosphate glass of (a) Nd3+ ion (b) Yb3+ ion.

3.1.2.6 Laser performance parameters The laser performance parameters of Yb3+-doped laser materials are characterized by /?min, which is defined as the minimum fraction of Yb3+ ions that must be excited to balance the gain exactly with the ground state absorption at laser wavelength, and 7sat which is the pumping saturation intensity, that is to acquire, as the ground state depletion may be accomplished by available laser diodes, which can pump all Yb3+ ions into the excited state. Im\a, which is defined as the minimum pumping intensity, is a measure of the case of pumping the laser material to overcome the threshold power in the absence of other losses. With smaller /?min, 7sat and 7m;n values, the better laser performance can be obtained.

3.1 Laser Spectroscopy ofNd3+ and Yb3+ High Doped Glasses

93

The main parameters describing laser performance of laser glasses include minimum pumping intensity (/min), saturating pumping intensity (7sat) and minimum fraction of excitations G#min). The relationship between Imin, Isat and /tan, is as follows: min _ Pmin=

A'min

(3.3)

sat '

Oabs(Ao)

L Z\ = <\ + exp Oem(Xo) + Oabs(Xo) { Zl

(Ezi - hcAo'1

kT

|

,

(3.4)

he

hat =

(3.5)

[ApCTabs(Ap)Tf\

he

f,

Zi

'

[(Ezi - hcAo'111

,~ ^

where h, k, c are the Planck constant, Boltzmann constant and light velocity, respectively. A,em, ^ and T are the secondary emission wavelength (laser wavelength), peak absorption wavelength (pumping wavelength) and temperature (K), respectively. The calculated 7min, 7sat and/?min, of different glasses are shown in Table 3.8 and Table 3.9. Table 3.8 Laser performance parameters of Yb3+-doped PNK, GTN, SN, BL, F P glasses Glass PNK GTN SN BL FP

oyo-em-T(/pm4-ms 1.48 2.99 2.92 1.70 0.56

Amn(%) 0.0744 0.0351 0.0423 0.0147 0.123

/sal/KW-cm'2 7.93 8.27 7.10 12.23 16.1

0.59 0.29 0.30 0.51 2.45

Table 3.9 Laser performance parameters of Yb3+-dbped ADY, LY, PN, QX glasses Glass ADY LY PN OX

Op-cwTf/pm -ms 0.90 0.74 1.47 0.70

Anin(%)

Aa,/KW-cm-2

U/KW-cm" 2

0.0984 0.1670 0.0596 0.1946

11.38 11.68 9.90 10.79

1.12 1.95 0.59 2.10

94

New Developments in Optics and Spectroscopy of Rare Earth Ions Doped Glasses

The gain coefficient (G) is related to N*x0emand N* is the inversion population numbers, which is proportional to apXTf. Therefore, G is in proportion to apxTfxGem. Furthermore, the stored energy and extraction efficiency are in proportion to the product of the absorption cross section and fluorescence lifetime, and the product of emission cross section and fluorescence lifetime, respectively. From Tables 3.8, 3.9, it can be seen that GTN and SN glasses have the smaller 7min, 7sat and /?min, BL has the higher /sat and lower, 7min and fi^n than PN, PNK has larger /?min and smaller 7sat than PN. Therefore, BL has some advantages over PN, and PNK is similar to PN for laser emission. Due to relatively high gain coefficient and low saturating pumping intensity, as well as weak concentration effect the phosphate glass is a good glass host for high doping Yb3+ ions. 3.2 Nonlinear Luminescence of Rare-Earth (RE) Ions in Glasses The nonlinear luminescence is a nonlinear emission process, that is, luminescence intensity is no longer linear with excitation power, but most commonly proportional to the square or cube of excitation power. It always takes place by two-photon or multi-photon process. The frequency up-conversion fluorescence of rare-earth ions in glasses is one of the nonlinear luminescence examples. To generate short wavelength laser by frequency up-conversion process is important for widening laser application. Because the transition probability or the stimulated emission cross-section of two-photon or multi-photon transitions is rather low, it should decrease the non-radiative transition probability and increase gain length. The low phonon glass hosts, such as non-oxide glasses, and glass fibers are always used for frequency up-conversion, such as laser generation. There are several mechanisms of frequency up-conversion: excited state absorption, cross energy transfer between ions, cooperative luminescence, avalanche luminescence etc. Fig. 3.14 illustrates the general energy level schemes of four frequency up-conversion processes mentioned above.

3.2 Nonlinear Luminescence of Rare-Earth (RE) Ions in Glasses

95

*2

1

R, ion 1

ion2 (b)

(a) i k

! \s3 i

\ -

<1>

ion 1
il

(d)

Fig. 3.14 General energy level schemes related to frequency up-conversion process (a) excited state absorption, (b) cross relaxation,(c) cooperative luminescence, (d) photon avalanche.

3.2.1 Excited state absorption The up-conversion process by excited state absorption (ESA) is the simplest one, the process happen in single ions. The energy level scheme is shown in Fig. 3.14(a).Taking the UV up-conversion luminescence of Nd3+ and Er3+ ions at low concentration (~lmol %) in fluorozirconate glass (ZBLAN) as an example,[28] Fig. 3.15 shows the up-conversion luminescence spectra of Nd3+ and Er3+ doped ZBLAN glasses in the region of 350-500 nm pumped by Ar+ laser with 514.5 nm. The pumping efficiency of Ar+ laser is mainly dependent on the wavelength matching with absorption band of Nd3+ and Er3+ respectively. The UV upconversion fluorescence intensity against pump power intensity is shown in Fig. 3.16, the slope value is 2.05 for Nd3+ and 1.65 for Er3+, it means the two-photon ESA process, which is illustrated in Fig. 3.17. The UV up-conversion process is as follows. Absorbing one photon, 3+ Nd is excited from ground state to 2G9/2, and relaxes to 4F3/2 (about 0.4ms), excited ions stay at this level, and jump to 2Ls/2 level by absorbing a second photon, then relax to 4D3/2, from which the upconversion luminescence is emitted. This is the sequential absorption process by a single ion, i.e. the excited state absorption (ESA) process.

96

New Developments in Optics and Spectroscopy of Rare Earth Ions Doped Glasses

The related rate equation has been described before.[29] When the doping concentrations increased, the interaction between ions becomes stronger, it means that the energy transfer process is enhanced. It has been mentioned that the concentration quenching of up-conversion UVluminescence was observed by the cross-relaxation process.[30] Er3+

Nd3+ a

450

500

A (nm) Fig. 3.15 Up-conversion spectra of Nd 3+ and Er3+pumped by argon ion laser with 514.5 nm.

slope=1.65 (Nd3+)

Pump Power (mW) Fig. 3.16 Up-conversion fluorescence intensity against pump power monitored at 382.8nm (For Nd3+) and 410.nm (for Er3+ ions) respectively.

3.2 Nonlinear Luminescence of Rare-Earth (RE) Ions in Glasses

97

'C,,*

*f ;3/1

3fi'S,/.-

«F,5 / 3 -

K.r1'

Fig. 3.17 Up-conversion process of Nd3+ and Er3+ pumped at 514.5nm. Wavy arrows indicate non-radiative process. Lines are drawn as guides for the eye.

3.2.2 Cross-energy transfer between activated ions The energy level scheme for cross relaxation between activated ions itself is shown in Fig. 3.14(b). The up-conversion emission of Nd3+ and Er3+ ions at higher concentration (>lmol %) in ZBLAN fluoride glass excited at about 800 nm is an example to illustrate the energy transfer process between activated ions. The fluorescence and excitation spectra of up-conversion emission of Nd3+ and Er3+ in ZBLAN glass are shown in Fig. 3.18(1) and 3.18(2) respectively. The dependence of up-conversion emission intensity against pump power intensity of Nd3+ and Er3+ in ZBLAN glasses is shown in Fig. 3.19.[29] According to slope values they are both two-photon processes. Taking Nd3+ ions as an example, the upconversion emission in Nd3+ doped glass excited at about 790 nm

98

New Developments in Optics and Spectroscopy of Rare Earth Ions Doped Glasses

belongs to the type of dipole-dipole interaction between two Nd3+ ions in their excited state, this interaction causes one ion to be promoted to a higher excited state (4G7/2,2G9/2 etc), while the other ion drops to a lower energy state (4I]3/2, 4In/2, 4I°/2),[28] see Fig. 3.20(1). These energy transfer processes may be phonon-assisted. The rate equations described this energy transfer processes were shown before.[29] Our analysis shows that the energy transfer processes were the principal processes in these upconversion emissions.

500 520 5J0 560 580 600

780

800

Wavelength

(1)

820 (nm)

520 540 560 580 780 800 820 840 wavelength (nm)

Fig. 3.18 Fluorescence (a) and excitation spectra (b) of up-conversion emission of Nd and Er3+ in ZBLAN glass: (1) Nd3+ (lmol %) (2) Er3+ 1, (X 10) 1 mol% Er; 2, (X4) 1 mol% Er, 0.5 mol% Cr; 3, 5 mol% Er. (a) fluorescence; (b) excitation. 100

- 50

= " >> a

Intens o

/•

:

550 nm /

o 5 •o c a o r, c u

530 nm slope 1 .85

A

jlope

1.89

0

ta

i

, '

.

i

,

5

,

,

, . i

10

30

Excitation Intensity (a.u.) (1)

1 Excitation

5 10 Intensity

30 (a.u.)

(2)

Fig. 3.19 The dependence of up-conversion emission intensity against pumping intensity of Nd3+ and Er3+ ions ZBLAN glasses: (l)Nd3+, (2) Er3+.

3.2 Nonlinear Luminescence of Rare-Earth (RE) Ions in Glasses

ioa 1

too 2

ioa 1

ion 2

ion I

iee 2

ton 1

99

i&n 2

d)

\

V

a b c

|S

ftn Q

Sft

:s (in/! ion 2 (2)

Fig. 3.20 Energy levels and excitation processes of Er3+ (1) and Nd3+ (2) in ZBLAN glass.

3.2.3 Cooperative luminescence The energy level scheme for cooperative luminescence is shown in Fig. 1.13(c). Yb3+ doped glass is a good example to demonstrate cooperative luminescence. With 4f-13 electronic configuration the energy level diagram of Yb3+ is very simple, in the range of visible to near IR, there are only two energy levels (ground state 2F7/2 and excited state 2F5/2). The luminescence owing to the transition 2F7/2—>-2F5/2 is in the vicinity of 1 um and the excitation wavelength is around 900 nm. When the excitation light intensity increases, it is easy to observe the green luminescence with a peak wavelength at 480 nm.13'1 Fig 3.21 gives the visible fluorescence of Yb3+ doped sample, which is excited by a Ti: sapphire laser at wavelength of 919.6 nm. The fluorescence wavelength range extends approximately from 460 nm to 490 nm, with a peak wavelength

100 New Developments in Optics and Spectroscopy of Rare Earth Ions Doped Glasses

at 480 nm. It can be found that there is a wavelength correspondence between cooperative and one photon fluorescence of Yb3+, which is X cooP=V2, where A,C00p is the wavelength of cooperative emission and X; is the wavelength of the of 2F5/2—>2F7/2 transition in Yb3+. In view of cooperative theory, a photon is produced by an ion pair resulted from the interaction of two Yb3+ ions. Therefore, we have the relationship hvcoop= hv! +hvj, where I and j denote the Stark levels of 2F5/2 and 2F7/2 respectively, in terms of conservation of energy. From this, in the case of a low broadening approximation the above wavelength relationship can be deduced easily in consideration of the approximation of hvi +hvj. 100

r

is

50 -

460

470

480 490 500 Wavelength (nm)

Fig. 3.21 Cooperative luminescence of Yb3+ ions in ZBLAN glass. Excitation wavelength: 919.6nm.

Rate equations can give a relationship between excitation intensity and cooperative luminescence intensity. These are as follows: rl,

2

dt = o<J>Mn ,^

at

'12"1 n "0 , C„n, n

^ 1 2 ' ' l "0

'

(3.7)

(3.8)

T0

iV = n0 + nx + 2n2.

(3.9)

In the case of continuous-wave excitation, we find a solution for n2 as

3.2 Nonlinear Luminescence of Rare-Earth (RE) Ions in Glasses

101

Cn{zxo^NfN

(3.10)

(l + r,o)3 In the low pumping saturation approximation, that is r1oO»/.

(3.11)

We reduce equation (3.10) to /i 2 *r 2 C 1 2 (T,oa>JV) 2 ^,

(3-12) +

where Cu is the cooperative possibility; N is the Y b molar concentration: a is the absorption cross section for pumping light; ii is the experimental lifetime of the 2F5/2 level;
slope-1,86

1L

3

4

5 6 P (mW)

7

8

Fig. 3.22 .Evolution of ytterbium cooperative fluorescence power with pumping power. Pumping wavelength: 919.6 nm.

3.2.4 Avalanche luminescence Among non-linear luminescence, avalanche luminescence is of specific characteristic, which is based on cross relaxation between two ions and

102 New Developments in Optics and Spectroscopy of Rare Earth Ions Doped Glasses

absorption of pump photon from an excited state. Fig. 3.14(d) shows the energy level scheme of photon avalanche process. It has been observed in doped Er3+ and Ho3+ ZBLA fluoride glasses and glass fibers early.[32'33] The mechanisms of avalanche processes include: side band absorption, ion pair absorption, thermal effects and absorption caused by Raman scattering. In Er3+ doped ZBLAN glass, side band absorption plays a major role, because all other factor are negligible.1331 Strong green emission at 579 nm has been achieved in Er3+ doped fluoride glass fiber by using 640 nm pumping source. A green light has been observed in Ho3+ doped ZBLAN glasses (with 2.5mol%) when the laser beam wavelength is tuned around 585 nm. Fig. 3.23 gives the green fluorescence spectrum. The first peak wavelength is located at 543 nm, and the second peak at 552 nm. Another visible fluorescence has also been detected in the range of 630670 nm. Fig. 3.24 shows this fluorescence with a peak wavelength of about 660 nm.

540

550

560

Wavelength (nm)

Fig. 3.23 Green luminescence observed in Ho3+ doped ZBLAN glass under excitation by an R6G dye laser operating at 588.lnm. (The first peak wavelength is located at 543nm and the second peak wavelength is 552nm).

6A0 650 660 W a v e l e n g t h (nm)

Fig. 3.24 Red fluorescence observed in Ho3+ doped ZBLAN glass under excitation by an R6G laser operating at 582.8 nm.

A plot of fluorescence power versus pump power is shown in Fig. 3.25. It suggests that a two-step transition should account for the green fluorescence because the slopes are greater than unity although depending on the excitation wavelength. Considering sample 3, Fig. 3.25 shows that slope is 1.66 for an excitation wavelength of 588.1 nm

3.2 Nonlinear Luminescence of Rare-Earth (RE) Ions in Glasses

103

whereas it is 1.87 for an excitation wavelength of 582.8 nm. For these two excitation wavelengths, the interaction cross sections are different and this can account for the difference in slope (1.87 at 582.8 nm and 1.66 at 588.1 nm). There are no energy levels of Tb3+, Eu3+ and Sm3+ ions, which can match the pumping photon energy from the ground state, so these ions do not absorb 588 nm pumping light and it is impossible for them to up-convert by absorbing two photons. Secondly, it is impossible for them to compare the green fluorescence of Ho3+ ions with that of Er3+ ions (the up-conversion emission spectra are shown in Fig. 3.18 (2)). It is easy to compare the differences between both fluorescence spectra. For the fluorescence resulting from the transition of 4S3/2—*2Ii5/2 of Er3+ ions the peak wavelength is 550 nm, but for Ho3+ ions, it is 543 nm. The difference is not due to the response of the detection system. So, the green fluorescence recorded must result from the transition of ( 5 F 4 , 5 S 2 )^ 5 I 8 ofHo 3+ ions.

100 80 • s l o p e ~ l .66 (3J$F 60

/

40 20 10

3

/

/

i

4

s l o p e - 2 . 6 1 (4*)

i

i

t i i i

6 8 P (mW)

i

10 (b) 3+

Fig. 3.25 Dependence of fluorescence intensity of different Ho doped ZBLAN glass samples on pumping power with excitation wavelength of 588.1 nm and 582.8nm: (a) excitation wavelength: 588.lnm; (b) excitation wavelength: 582.8nm. Sample 2: 2.5 mol% Ho, Sample 3: 5 mol% Ho, Sample 4: 1 mol% Ho + 4.5 mol% Tm.

Considering the energy level diagram of Ho ions and a pump photon energy of 17000 cm"1, we can deduce the cross-relaxation process as follows (Fig. 3.26). Ions in states 5 F 3 , 5 F 4 and 5I4 interact with ions in the ground state, which results in the cress-relaxations of 5F3—> 5I7=> 5 I 8 —> 5 F5, 5 F 4 -> 5I7 => % -» 5I4 and 5I4-+5I7 => 5I8-^5l6- Among these

104 New Developments in Optics and Spectroscopy of Rare Earth Ions Doped Glasses

transitions, the last one contributes largely to the build-up of the population at 5I7 owing to the non-radiative decay from 5I6 to 5I7. It is also important for the first and second cross-relaxations to provide 5I4 with enough populations.[34]

Fig. 3.26 Avalanche process in Ho3+ ions. The cross-relaxation between 5I4—>5I7 and 5 Ig—»5I6. The population at 5I7 is obtained by non-radiative decay from 5I6 and 5I7. A description of other cross-relaxations is giver in the text: 2, green luminescence at 543nm and 552nm;2, red luminescence ranging from 630nm to 670nm.

The lifetime of energy level 5I7 is very long (about 12 ms), so a large population can be maintained in this level. In terms of upper absorption energy levels, a Ho3+ ion can be excited from 5I7 to higher excited states by absorbing a photon which has a wavelength of 477 nm (—>3F7), 584 nm (->5Fi), 632 nm (^ 5 F 2 ), 739 nm (^ 5 F 4 ) or 956 nm (^ 5 F 5 ). Avalanche up-conversion cannot happen at these excitation wavelengths due to ground-state absorption. If a two-photon absorption occurs under 584 nm excitation (that is % —• 5F4, 5F3 and 5I7 —• 5Fi as shown in Fig. 3.26), the green fluorescence with a peak wavelength of 540 nm (that is (5F4,5S2)—> % ) , and a red fluorescence of 650 nm in a visible waveband should be observed (that is 5F5 —• 5I8). An avalanche luminescence of Ho3+ ions in ZBLAN glass may appear. The slope for sample 3 excited

3.2 Nonlinear Luminescence of Rare-Earth (RE) Ions in Glasses

105

by 582.8 nm and 588.lnm is almost equal and approaches 2, which indicates the two-photon process and suggests that excitation wavelength has little influence on the slope. The excitation spectrum of the green fluorescence demonstrates that it is not a phonon-assisted up-conversion process (Fig. 3.27). In the case of a multi-phonon process, the closer the exciting wavelength to the centre of an absorption band, the more intense the absorption will be, due to the lower number of phonons involved. The intensity of green fluorescence, however, changes from low to high and then to low as the pumping wavelength scans from 582 nm to 595 nm with a constant output power of 25 mW. To sum up, we think that the observed luminescence must result from photon avalanche up-conversion. 100.

i50

°580V0~"5«5.'b 590.0 Wavelength (nm)

595.0

Fig. 3.27. Excitation spectrum of green fluorescence. The pumping power is kept to 25Mw and the wavelength changes from 580nm to 595nm.

In Ho3+ doped ZBLAN glass, thermal effects and Raman scattering are negligible because of the high lying first excited level of Ho3+. Only pair absorption and side band absorption are possible. Spectral measurements and analysis suggest that the photon avalanche is due to excited state absorption from the 5I7 state of Ho3+. Several cross-relaxation processes can populate this meta-stable state. Absorption by pair levels and multiphoton-induced side bands initiates the photon avalanche process. M.F. Joubert gave a review of photon avalanche up-conversion in rare earth ions doped crystals and fluoride glass fibers and mentioned the state of the art concerning the up-conversion of diode pumped solid state lasers.[35]

106 New Developments in Optics and Spectroscopy of Rare Earth Ions Doped Glasses

3.2.5 Energy transfer between rare earth ions in up-conversion luminescence Great attentions have been paid to the energy transfer between rare earth (RE) ions in up-conversion luminescence, because some active RE ions have weak absorption cross-section at pumping wavelength or low quantum efficiency due to multi-photon process. Several useful pairs of RE ions have been utilized for practical applications. 3.2.5.1 Energy transfer in Yb /Er

co-doped glasses

The low absorption cross-section of Er3+ ions limits the pump efficiency. Ytterbium ions, a two energy level system, exhibit not only a large absorption cross-section, but also a broad absorption band between 800 and 1100 nm, which can be seen from Fig. 3.28. Furthermore, the large spectral overlap between Yb3+ emission (2F7/2—»• 2F5/2) and Er3+ (4Iis/2—*• 4 In/2) results in an efficient resonant energy transfer from Yb3+ to Er3+ in Yb3+/Er3+ co-doped system. The energy transfer effect is obvious. The intense luminescence of Yb3+ /Er3+ co-doped has been observed by 966 nm laser diode pumping, the peak wavelengths at 549 nm and 665 nm belong to two-photon up-conversion and the peak wavelengths of emission at 406 nm and 381 nm are three-photon up-conversion of Er3+ ions.[36] j 1

~E

Er Yb

2 6 -

cross

10 4 -!

.1 2 " |

--" \ \

< 0 800

900

1000

1100

1200

Wavelegth/nm Fig. 3.28 Absorption cross section spectrum for Yb3+ and Er at 970nm band.

3.2 Nonlinear Luminescence of Rare-Earth (RE) Ions in Glasses

107

The transition processes in Yb3+ and Er3+ and energy transfer between Yb3+ and Er3+ are quite complicated. Fig. 3.29 illustrates a simplified energy level diagram of the Yb3+/Er3+doped laser glass pumped at 970 nm. The main processes involved are the following by referring to the number in Fig. 3.29. (1, 2) are the radiative absorption and emission of a pump photon from 2F7/2 and 2F5/2 levels of Yb3+, respectively. (3) is the energy transfer from Yb3+ to Er3+, Yb(2F5/2)+Er(4I15/2HYb(2F7/2)+Er(4I11/2), that acts as indirect pumping of Er3+ ions. (4) Ground state absorption (GSA) of Er3+ ions. (5) The erbium nonradiative decay from the 4In/2 level to 4 IB/ 2 upper laser level. (6,7) The stimulate emission and absorption between 4Ii3/2 upper and 4Ii5/2 ground Er laser level. (8) Cooperative up-conversion process between two excited Er3+ ions that promotes one ion to the %/2 upper level, while the second ion decays to the ground level. The first ion then relaxes, through a fast non-radiative decay, to its original 4Ii3/2 state so that overall effect is the loss one, 4In/2, excited ion. (9) Further energy transfer (ET) from Yb3+ to Er3+, which will cause to the ESA of Er3+ ions.[37]
? J 3 ^

rr

11/2

j

Si

6

i

3/4

Nj MJ/2

Yb 3

Er' T

1'

t Eru

Fig. 3.29 The energy level diagram of the erbium-ytterbium system. The solid-line arrows refer to radiative phenomena; the dash-line arrows refer to Er-Yb energy transfer process. ,3+

3.2.5.2 Energy transfer in Tm /Ho co-doped glass From Fig. 3.25 it can be seen that the slope of sample 4 co-doped with

108 New Developments in Optics and Spectroscopy of Rare Earth Ions Doped Glasses

Ho3+ and Tm3+ differs from those of samples 2 and 3. Energy transfer occurs from Tm3+ to Ho3+ ions. Fig. 3.30 gives a pairs of resonant energy transfer between Tm3+ and F£o3+. They include the following processes: (i) Ho (5F5) => Tm (3F2) -> Tm (3H4) (ii) Tm: 3H4 - • 3H6 ^ H o : 5I7 -> 5F4 (iii)Tm: 3H4 -» 3 F 4 =>Tm: 3H6 -» 3 F 4 and Tm: 3 F 4 =>Ho: (5I7). Clearly, populations at 3F5 and 5I4 of holmium ions are moved to 3Hi of thulium ions by energy transfers between Tm and Ho ions. In addition, for sample 4 the avalanche 'threshold' is about 3 mW, while for sample 2 it is only about a fraction of a milliwatt. It also indicates that the losses of population looping increase owing to the decrease of populations of 3F5 and 5I4. As the slope efficiency of up-conversion process changes to 2.6, it can be considered as a three-photon process. % D

F4

5

F5

0= - ^ \

:3F23F3

J

3

F4

_3H*

I8. 3+

Ho

Tm3+

Fig. 3.30 Energy transfers between Tm 3 + ions and Ho 3 + ions.

3.2.5.3 Energy transfer in Nd3+/Yb3+/Tb3+ co-doped glass Up-conversion luminescence of a ZrF4-based fluoride glass co-doped with Nd , Yb and Tb was examined under 800 nm excitation by Qiu et al.[3S] Fig. 3.31 shows the up-conversion luminescence excited by 800 nm laser light of Nd/Yb/ Tb co-doped glass and normal fluorescence excited by 350 nm light of Tb3+ doped glass. The up-conversion luminescence around 490, 545, 580 and 624 nm, which originate from the 5D4 level of Tb3+ ions, were observed. The schematic diagram of up-

3.2 Nonlinear Luminescence of Rare-Earth (RE) Ions in Glasses

109

conversion mechanism, as shown in Fig. 3.32 has been proposed. The conclusions are as follows: (1) only Nd3+ is excited by 800 nm light, (2) the phonon relaxation, (3) the energy of Nd3+ transfers to Yb3+ and (4) the Tb3+: 5D4 level is populated by the co-operative energy transfer from two Yb3+ ions. •T

-

-r 1 ' i—r—f—r— | . Ujwwnversion luminescence (X „= 800nm) Normal fluorescence ( X ^^SOnm)

• •?">

-J\

'

£

-

t

JFf

A jp-

_ .

380 400 420 440 Wavelength (nm)

Jp*

T P~

.... i . . . . i . , . . J iW./ 400

450

500

-"•

U \600A .650

550

Wavelength (nm)

Fig. 3.31 Up-conversion luminescence spectrum of 50ZrF4-29BaF2-lNdF3-10YbF2-10TbF3 glass under 800nm excitation (solid line) and normal fluorescence spectrum of 60ZrF4-30BaF2-5TbF3- 5La F3 glass under 325nm excitation (broken line). xlO'

I

'D,

" *D„ S G 4

24

e JO E?

A

3

D4\

•• "

" 4<WO,n

16

4>

a W

2

= 1

4f7

^

I8

\ -l-m

m

s-

Nd3"*

4

JJl

- 7I;,IM).l.2l • 'F, •If, •7F,

*i,

'

H»*IV

% V

ii

Yb)*

2f

"

Tb"

• 1=6

2

'

Yb-1*

Nd-1*

Fig. 3.32 Schematic diagram of up-conversion mechanism under 800nm excitation in 50ZrF 4 -29BaF 2 lNdF 3 10YbF 2 10TbF 3 (ZBNdYbTb) glass.

110 New Developments in Optics and Spectroscopy of Rare Earth Ions Doped Glasses

For development of visible solid-state lasers by frequency upconversion, the most practical approaches are using RE co-doped nonoxide glass fibers and pumped by near IR laser diodes. The possible upconversion laser designs are summarized in Chapter 1 (1.4.1.6). 3.3 Super-luminescence of RE-doped Glass Fibers Super-fluorescence is an amplified spontaneous emission (ASE) under extremely strong excitation. In this process the spectral width of the emission is narrowed, but the line shape still keep smooth and a large output can be achieved. For getting the high gain the RE-doped fluoride glass fibers are always used. 3.3.1 Super-luminescence in RE-doped silica glass fibers To produce the super-luminescence in glass fibers the experiment is arranged as follows. A laser beam was coupled into the fiber core by an achromatic microscopic objective (the magnifying power 10x, and the N.A. 0.3). An Ar-ion laser and CW tunable Ti: sapphire laser was used as pumping source. The output signal light from the fiber was dispersed by a GDM-1000 monochromater and detected by a near-infrared photomultiplier tube GDB-411. There is no resonant cavity in the experiment for double pass experiment, only a reflective mirror put at single pass output side.[39'40] The schematic diagram of measuring arrangement is shown in Fig. 3.33.

(a) ' Power meter

Ar Laser

Recorder

PMT

GDM-1000 Monochromator

Fig. 3.33 The experimental configuration for fiber super-fluorescence.

3.3 Super-luminescence ofRE-doped Glass Fibers

111

The fiber we used in this experiment was a heavily doped one. Its length was 3m. The fiber core was 5um in diameter with NA=0.15. Super-fluorescence characteristics were investigated in single-pass and double-pass configurations. The changes of the spectral profile with increasing pumping power of a Nd3+-doped silica glass fiber is shown in Fig. 3.34. No obvious threshold effect was observed in this process. When the pumping was weak, the peak of the spectrum was at 1080 nm with spectral width 28 nm, but the intensity at 1088 nm increased faster with increasing pump power than that at 1080 nm, and soon it became the most intense. In this process, the spectra narrowing are very obvious. The narrowest spectral width we have obtained in this experiment is 3 nm when about 40 mW of pump power was absorbed. No further narrowing has been observed when the pumping power increased after this. The maximum outputs we achieved were 0.32 mW and 6 mW in the single-pass and double-pass configurations respectively.'41'421

1080 1130 W a v e l e n g t h (ran)

Fig. 3.34 Fluorescence spectrum narrowed with the increase of pump power of a Nd3+doped silica glass fiber. 1, 5mW; 2, lOmW; 3, 15mW; 4, 20mW; 5, 40mW.

The Er +-doped silica glass fiber used in this experiment is 4m long and with a core diameter of 6(am. The doping concentration is 1000 PPM. Fig. 3.35 shows the fluorescence spectrum and super-fluorescence spectra of an Er-doped silica fiber at different pump levels. The

112 New Developments in Optics and Spectroscopy of Rare Earth Ions Doped Glasses

fluorescence spectrum was measured in a short fiber and it has a peak at 1530 nm and a shoulder peak at 1550 nm. Curves 2-5 in Fig. 3.35 are the super-fluorescence spectra of the fiber of 4m measured at different WP, which exhibit the change of spectral profile with pump intensity. This change was for similar reasons to those indicated in the last section. The maximum output is 0.28 mW in the single-pass configuration while WP=290 mW. In the double-pass configuration, 3.2 mW output was achieved with Wp=240 mW. The corresponding conversion efficiencies were 0.31% and 4.1%. Although these values will increase with the pump power, they are too small compared with the ideal values 33.8%and 16.9% in the case that the quantum efficiency is unity. We regard this as the effect of ESA, a sizeable fraction of the absorbed pump power is absorbed from the excited state and not contributed to signal photon emission.[41] 3.3.2 Evolution of super-luminescence in Nd3+-dopoed fluoride glass fibers Nd3+-doped ZBLAN glass fibers with a core diameter of 5-1Oum were prepared. The inactive loss at emission wavelength is about 20-30 dB/km. An evolving process of the spectral shapes from fluorescence to laser emission of the ZBLAN fiber pumped by the Ar-ion laser is shown in Fig. 3.36. As a comparison, Fig. 3.36(a) gives a fluorescence spectrum of a bulk glass sample. Its peak is situated at 9544 cm"1, and the line width (FWHM) is about 60 cm"1 (-176 nm). Fig. 3.36(b) is the fluorescence spectrum of the fiber at 7.5 mW of injected pump power; the fluorescence peak locates at 9549 cm"1, and the line-width becomes 127 cm"1, which is a little narrower than that of the bulk sample. When pump power is doubled, the line-width narrows to 114 cm"1, and another peak at 9499 cm"1 appears (Fig. 3.36(c)). At 30 mW, there are three well resolved peaks located at 9515, 9480 and 9450 cm"1, respectively (Fig. 3.36 (d)), and the line-width drops to 108 cm"1 (-120 nm). But the peak at 9549 cm"1 disappears. So with increasing the pump intensity, the linewidth narrows progressively. At the same time not only the splitting of

3.3 Super-luminescence ofRE-doped Glass Fibers

113

fluorescence but also the change of these peak locations has been observed.[43]

9800

1500 1620 1540 1560 1530 1600 Wavelength (ran)

Fig. 3.35 Fluorescence spectrum (1) and Super-fluorescence spectra (2-5) of Erdoped silica fiber at different pump power: 2, 40mW; 3, 85mW;4, 120mW; 5, 180mW.

9500 9200 lfavenumber(cm"')

Fig. 3.36 The developing process from fluorescence to laser emission pumped by 514.5nm of Ar-ion laser, a: bulk sample, b: 7.5mW, c: 15mW d: 30mW, e: lasing close to the threshold.

Fig. 3.37 shows the development of fluorescence spectra of doublepass pumped by Ti: sapphire laser, in which there is a clear saturation

100T3

T025 1050 1075 Wavelength(nm)

1100

Fig. 3.37 Typical spectra of fluorescence for double-pass at different injected pump power levels pumped by 800nm,l: the line-width 11.6nm (injected pump power 7.0mw), 2: the line-width 8.6nm (injected pump power 9.2mw), 3. the line-width 8.2nm (injected pump power ll.Omw), 4. the line-width 7.7nm (injected pump power 14.4mw), 5. the line-width 7.5nm (injected pump power 20.0mw).

114 New Developments in Optics and Spectroscopy of Rare Earth Ions Doped Glasses

range. In this regime, the line-width maintains about 7.5 nm although the pump power increases. The line-widths of output spectra change progressively. For the single-pass, the line-width decreases from 16.2 nm (injected pump power 7 mW) to 7.6 nm (24 mW). For the double-pass, it decreases from 11.6 nm (7 mW) to 7.5 nm. No clear saturation, however, is recorded for single-pass due to the limit of the pump power. From the change of line-widths, it can be seen that there is a super-fluorescence phenomenon.

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18.

J Murray, Proceedings of SPIE.3492 (1998)1-10. J Campbell and T Suratwala, J. Non-Cryst. Solids, 263/264 (2000) 318-345. W.F.Krupke, IEEE J.Select. Topics Quantum Electron., 6 (2000) 1287-1298. Fuxi Gan, Optical Spectroscopy of laser Materials, In: Laser Materials, World Scientific, (1995)1-102. Yanli Mao, Zhanrong Sun, Xiuli Jiang, Peizhen Deng and Fuxi Gan, Chinese Physics, 11 (2002) 613-618. B R Judd, Phys. Rev. 127 (1962) 750. G S Ofelt, J. Chem. Phys. 37 (1983) 511. M Ajroud, M Haouari, Ouada Ben, H Maaref, A H Brenier, and C Garapon, J.Phys.: Condens. Matter 12 (2000) 3181. V Mehta, G Aka, A L Dawar and A Mansingh, Opt. Mater. 12 (1999) 53. G A Kumar, P R Biju, C Venugopal and N V Unnikrishnan, J. Non-Cryst. Solids 111 (1997)47. Fuxi Gan, Optical and spectroscopic properties of transition elements in glass. In Optical and spectroscopic properties of glass. Springer Verlag. 1992,176-200. M J Weber, J A Paisner, S S Sussman, W M Yen, L A Riseberg and C Brecher, J.Lumin. 12/13 (1976) 729. F Auzel, Luminescence of Inorganic Solids, New York: Plenum, (1978) 67. C Brecher, L A Risberg and M J Weber, Phys. Rev. B18 (1978) 5799. Hongbin Yin, Peizhen Deng, Fuxi Gan, J.Noncryst. Solids. 210 (1997) 243-248. Hongbing Yin, Peizhen Deng and Fuxi Gan, J.Noncryst.Solids. 210 (1997) 243-248. Peizhen Deng, Chun Jiang, Hongbin Yin and Fuxi Gan, Proc. SPIE. 3416 (1998) 99103. Hongbin Yin, Peizhen Deng, Fuxi Gan, Junzhou Zhang, Proceedings of 17th Intern. Congress on Glass, Beijing. 17 (1995) 240-243.

References

115

19. Chun Jiang, Peizhen Deng, Guosong Huang, Fuxi Gan, Science in China, 42 (1999) 616. 20. Chun Jiang, Fuxi Gan, Peizhen Deng, Materials letters, 41 (1999) 209. 21. Peizhen Deng, Yanli Mao and Fuxi Gan, Proceedings of 7,h Intern. Otto Schott Colloquium, C.Russel, G. Volksch, eds. Glastech. Ber. Glass Sci. Techn. 75 (2000) 203-209. 22. Hongbing Yin, Peizhen Deng, Junzhou Zhang, Fuxi Gan, Materails Lett. 30 (1997) 29-33. 23. Xuelu Zou, H. Toratani, Phys. Rev. B, 52 (1995) 15889. 24. R.Koch, U.Griebner, H.Schonnagel, etal, Optical Communication, 134 (1997) 175. 25. V.Petrov, U.Griebner, D.Ehrt, et al, Optics Letters, 22 (1997) 19. 26. Junjie Zhang, Long Zhang, Changhong Qi, Fengying Lin, Hafeng Hu, Chinese J. of lasers, A26 (1999) 739-744(in Chinese). 27. Chung Jiang, Fuxi Gan, Junzhou Zhang, Peizhen Deng and Guosong Huang, J. solid state chemistry 144 (1999) 449-453. 28. Fuxi Gan, Jie Wang, Yihong Chen, J.Noncryst. Solids, 213&214 (1997) 261-265. 29. Fuxi Gan and Yihong Chen, Opt. Mater.l (1993) 45-52. 30. R.Piramidowicz, P.Witonski, M.Klimczak, M.Malinowski, Opt .Mater, 28 (2006) 152-156. 31. Fuxi Gan, Jie Wang, Yihong Chen, Pure Appl. Opt. 5 (1996) 855-862. 32. Yihong Chen, Jie Wang, Fuxi Gan, Proc XVII International Congress on Glass,! (1995)15-21. 33. Y. Chen and F. Auzel, Electron. Lett. 30 (1995) 1602. 34. Y.H.Chen, G.K.Liu, J.V.Beitz, Fuxi Gan, Jie Wang, 10th International Sympoium on Non-oxide Glasses, June 19-22, Corning, NY (1996). 35. Marie-France Joubert, Opt. Mater. 11 (1999) 181-203. 36. Liyan Zhang, Jianghu Yang, Lili Hu et.al. Chin. Phys. Lett. 20 (2003) 1344-1346. 37. P. Lapota, S. Taccheo, S. Longhi et. al., Optical Mater. 11 (1999) 269-288. 38. J.Qiu, M Shojiya, R. Kanno, Y.Kawamoto, Opt. Mater. 13 (1999) 319-325. 39. Fuxi Gan, Chemical Express, 6 (1992) 933-948. 40. Yihong Chen, Ruihua Cheng and Fuxi Gan, Chinese Science Bulletin 67 (1992) 556559. 41. Fuxi Gan and Yihong Chen, Pure Appl. Opt. 2 (1993) 359-365 42. Yihong Chen, Ruihua Cheng and Fuxi Gan, Chinese Journal of Infrared and Millimeter Waves, 11 (1992) 415-419. 43. Jie Wang, Fuxi Gan, Changhong Qi, J.Noncryst. Solids 184 (1995) 235-239.

Chapter 4

Third-Order Optical Nonlinear Properties of Glasses

The nonlinear optical effects of glass have recently become a focus of increasing scientific and technological interest. All glassy materials possess third-order optical nonlinearity, which is remained a subject of considerable theoretical and experimental efforts stimulated by the need of materials understanding in applications. In the past ten years of development of high power solid state laser devices the self-focusing and laser induced damage were the serious obstacles for application of laser glasses, a great progress has been achieved in development of new laser glasses with low figures of third-order optical nonlinear susceptibility. The optical wavefront distortion in optical glass fibers should be prevented for long haul optical fiber communication, some new approaches to diminish or compensate the nonlinear optical effects in glasses fibers have been explored. Recently on the contrary the strong third-order nonlinear optical effects in glasses have come into a hot spot of many researchers because of their potential applications in photonic devices, such as switching, modulation, optical signal processing etc. It is worth to understand deeply the mechanism and glass composition relationship of the third-order optical nonlinearity of glasses. In the previous monograph we have introduced the third-order nonlinear optical effects, such as laser induced self-focusing and damage, and optical blooming etc. in glasses, much concerning on dielectric bulk glasses.[I] Therefore, we put more emphases in this chapter on the thirdorder optical nonlinearity of dielectric glass, thin films, semiconductor and metallic particles doped composite glasses, and organic and inorganic hybrid glasses. 117

118

Third-Order Optical Nonlinear Properties of Glasses

4.1 Measurement of Third-Order Optical Nonlinear Susceptibility of Glass According to nonlinear optical effects several methods have been developed for measuring the third-order optical nonlinear susceptibility of different bulk glasses and glass thin films. In optical and laser glasses the third-order optical nonlinearity is always expressed as nonlinear refractive index n2(E), that is the refractive index change (<5n) by external electromagnetic field (E) or light intensity (I) Sn =n2-\E\

= n'2-I,

(4.1)

where E is the intensity of electric field (V/m), / is the intensity of light beam (W/cm2). The units of n2 and ri2 are in esu and cm2/W, respectively. The relation between nonlinear refractive index n2(E) with the thirdorder optical nonlinear susceptibility x(3) in non-resonant region is » 2 (£) = ^

( 3

>,

(4.2)

"o

where «o is linear refractive index. The measurements of nonlinear refractive index of transparent dielectric bulk glasses, such as by self-focusing damage, self-induced polarization change and laser-induced birefringence, were described before,[1] here we introduce several advanced methods for measuring the / (3) or n2(E) in bulk glasses and glass thin films, commonly used recently. 4.1.1 Degenerate four-wave mixing (DFWM) DFWM technique can provide information of the magnitude, speed and origin of/(3), therefore, it is popularly used in the measurement of/(3). The degenerate four-wave mixing process involves the two strong pump beams Ki and K2 and a week probe beam K3 injected to the pump wave. The three beams interact with the nonlinear glass sample to produce the phase conjugate beam K4. All four waves are at the same optical frequency and / ( 3 ) values of the sample can be determined by the intensity ratio of the phase conjugate beam K4 to the probe beam K3.

4.1 Measurement of Third-Order Optical Nonlinear Susceptibility of Glass Gi

Gi

Aperture

Z7 7h

119

Sample

7\i (b)

(a)

Fig. 4.1 (a) Experimental setup of 3D DFWM; (b) wave vector diagram of the threedimensional configuration.

The experimental setup of forward DFWM is shown in Fig. 4.1(a). The pump source is a mode-locked and frequency-doubled Nd: YAG laser with 35 ps pulse-width, 532 nm wavelength, and 10 Hz pulse repetition rate. The low repetition rate is used to avoid thermal effects. The input beam is split into three beams Ki, K2 and K3 with nearly equal energy by use of reflecting beam splitters (BSi, BS2, and BS3), then focused on a plot of the sample by a lens (L) with 20cm focal length. The beam geometry is shown in Fig. 4.1(b), where K4 is signal beam generated by input beams through the sample. The angles between the three beams K b K2 and K3 are about 2°. K2 and K3 can be controlled by stepping motors, so that the arrival time at the sample can be varied. When Ki, K2 and K3 are overlapped spatially and temporally in the sample, the generated signal beam K4 is detected through slits, recorded by a photodiode, and then analyzed by a boxcar and computer.[2] The third-order nonlinear optical coefficient %(3) can be obtained by comparing the measured signals for the sample with that for carbon disulfide as reference under the same experimental condition according to the following formula:'31 0)

Z

=

2 Lr_f n \ aL expyccL 12)

VMr J 1/2

L

1 - exp(- aL)

(3)

(4.3)

here 74 is the intensity of phase conjugate beam K4, L the sample path length, n the linear refractive index, and a the linear absorption coefficient of the sample at 532 nm. The subscript "r" refers to carbon

120

Third-Order Optical Nonlinear Properties of Glasses

disulfide (CS2). To convert relative to absolute susceptibilities we use X(3W=6-8x10"13

e s u for cs

2-

4.1.2 Z-scan method Z-scan method is based on nonlinear effects of self-focusing and defocusing by intense light beam in glass. As shown in Fig. 4.2(a) the transmittance of a nonlinear glass is measured through a finite aperture placed in the far field as a function of the sample position (z) with respect to the focal plane. The z-scan method can be used to determine the nonlinear refractive index n, and also nonlinear absorption (3, both are the origin of the third-order optical nonlinearity, thus the third-order optical susceptibility %(3)can be expressed as:

z(3) =

(Re^3>)2+(lm^)2

Rezm=2n02s0cy,

1/2

(4.4)

(4.5)

2

Im%^=P

2n

(4.6)

here «o is the linear refractive index, SQ is the permittivity of free space, c is the velocity of light, y is the coefficient of nonlinear refraction and /?is the coefficient of nonlinear absorption. The optical nonlinearity is composed of electronic part (short time response) and thermal parts (long time response), which can also be measured by z-scan method using laser beams with different pulse duration.

Incident beam It

Fig.4.2 (a) Z-scan schematic diagram, (b) the experimental setup of Z-scan. A, aperture; BSj (i=l,2), beam splitters; L; (i=l,2), lens; Dj (i=l,2,3), detectors.

4.1 Measurement of Third-Order Optical Nonlinear Susceptibility of Glass

121

The experimental setup is shown in Fig. 4.2(b). A Q-switched Nd: YAG laser with 1060 nm wavelength and 15ns pulse duration and 1Hz repetition rate was used as the excitation light source. After spatial filtering, each pulse had a maximum energy of several millijoules and was used in the z-scan measurements. A laser energy ration meter (Laser precision Corp., Model Rj-7200) was used to detect the output signal. This technique relies on the fact that the intensity of a focused laser beam varies along the axis of a lens and becomes a maximum at the focus. By moving the sample through the focus, the nonlinear refraction and the intensity dependent nonlinear absorption can be measured as the changes of the transmittance through the sample. The sample itself acts as a thin lens with varying focal length as it moves through the focal plane. Such a device has been used to make sensitive simultaneous measurements of the sign and the magnitude of the nonlinear refractive index y, and the two-photon absorption coefficient /?, as well as the variation of refractive index per unit photo-excited carrier density aY, and the free carrier absorption cross section aat» for semiconductor materials.[4'5] It is easy to measure the normalized peak (maximum) transmittance Tp and valley (minimum) transmittance Tm the difference between them is a quantity ATpw, which is connected with A
(4.7)

AQ>0=KLeff-yI0,

(4.8)

where L e ^(l-e"V')/ao; cto and L—the linear absorption coefficient and length of sample, 1$ —the light intensity at z=0, y—the coefficient of nonlinear refraction. For |A=0.406(l-s)0'25, s is the aperture transmittance. At central-small-aperture diaphragm (5=0), there are Ar po «0.64|A
(4.9)

The changes of the intensity and the phase of a light in glass are related with the coefficient of nonlinear ft (for two-photon absorption) q(z,r,t) = AaI(z,r,t)Leff,

(4.10)

122

Third-Order Optical Nonlinear Properties of Glasses

, I(z,r,t)e~aL Ie(z,r,t)= y

'/

T(z,S = 1) = X [ 7 ° ( ^3S/ 2 r > (when?o
g 0 (z,0) = Aa/ 0 L

(4.11)

(4.12) (4.13)

M + (s/z0) where 7T[z,5=l)—the transmittance integration at open-hole. 4.1.3 Optical Kerr effect Due to low signal intensity of optical Kerr effect (OKE) the real and imaginary parts of the third-order optical nonlinearity of glass cannot be separated normally. By using of femtosecond optical heterodyne technique (OHD) the real and imaginary parts of complex third-order optical nonlinearity can be effectively separated and their values and signs can be determined.[6] The OHD-OKE experimental set-up is show n in Fig. 4.3. A modelock Ti: sapphire laser pumped by the Millennia Vs laser which provides 4.1W of green (532 nm) light, is operated at 82MHz and has an average output power of 680 mW, a wavelength of 805 nm and a pulse duration of about 80fs. The output beam is divided into a pump and a probe beam by a beam-splitter. The probe beam passed through a polarizer PI that is directed at 45° with respect to that of the pump beam. The pump and probe beams, with the average power 20 mW and 2 mW, respectively, are focused into the sample, and their con-focus length is about 200 (am. A polarization analyser is placed behind the sample at the crossed polarization direction to the input polarizer. A quarter-wave plate is inserted between the input polarizer and the focusing lens to measure the real part of %(3) in the OHD-OKE experiments, and without quarter-wave plate for probing x(3) imaginary part. The optical axis of quarter-wave plate is also directed at 45° to the polarization direction of the pump beam.[7]

4.2 Optical Nonlinearity of Dielectric Glass

123

For measuring the third-order nonlinear susceptibility of glass samples CS2 was used as the reference sample. R2 represents a correction factor for attenuation of beams in the sample due to linear absorption. The third-order susceptibility of the sample can be calculated by comparing with the reference >l/2

( * %

=

h,R(i)

R

ref

ref "2.RU)

V

re

fJ

ref

R,

tt%

(')

ref

(4.14)

The superscript ref denotes the reference, and \% )Rre is the effective third-order susceptibility of the reference for which we assume no imaginary part.

Fig. 4.3 Schematic diagram of OHD-OKE set-up. LS: Ti: Sapphire laser, BS: Beam splitter, M: Mirror, Ch: Chopper, P: Polarizer, S: Sample, D: Detector, ODL: Optical delay line, L: lens.

Besides the x(3) measurement methods mentioned above, the spectrally-resolved two-beam coupling (SRTBC), interferometry, third harmoric generation and pump-probe method are also used. 4.2 Optical Nonlinearity of Dielectric Glass [8] The nonlinear refractive index of dielectric glass in non-resonant region is consisted of electronic and nuclear parts, the former is a polarizability due to the distortion of electronic cloud (shell) of consisted ions in glass, and the nuclear contribution is induced by nuclear core displacement,

124

Third-Order Optical Nonlinear Properties of Glasses

which can be determined by Raman scattering cross-section. It has been estimated that the nuclear contribution is about 15% for inorganic glasses.[9] 4.2.1 Optical nonlinearity of transparent oxide glasses On the basis of dispersion theory, the polarizability of glass can be described as

A

r~-C-Z

2

—

(4.15)

2 '

m k cok -co where m and c are the mass and the charge of electron, respectively.^ is the oscillator strength corresponding to the inherent frequencyffifcand a> is the frequency of incident beam. For representation of the dispersion curve of glass in the visible region, it is sufficient to take account of only two inherent frequencies, i.e. the electron transition frequency ooi in the ultra-violet region and the vibration frequency of glass structural network ©2 in the infrared region. Therefore, e

r=—

/i

CQX

•+ •

CO

-0)

f2

(4.16)

-CO,

m The change in the polarizability of glass under the actions of an external field is due to the change of co\, oh, f\ and/;. The change in the energy of inherent electron transition of ions, induced by the actions of powerful light, gives rise to changes in the polarizability of glass. For transparent oxide glasses the change of UV eigen absorption plays a more important role than that in IR region. The change of cu\ results in a shift of the dispersion curve only, which could either increase or decrease the refractive index. In case where both f\ and co\ values change the shape of the dispersion curve will also vary. During the Stokes shift the oscillator strength f\ changes very small the simplified dispersion equation can be expressed as follows: 1/2

con

-1 = / co0

-co

(4.17)

125

4.2 OpticalNonlinearity of Dielectric Glass [8]

where f ~{nl~l), follows:

when
1

1

1-*

(4.18)

W-ij'H&k'i 1.1 1.0 0.9 1

0.8 0.7 0.6

>=*=

0.5

•

2

0.05* 0.04

^

^

0.03

—J*

5 * " ^ 4~"""~

0.02 ,1

,., . . J

2.0

.1

3.0

i

A.O

1 . .

1

5,0

1

^hO

,.

, 1

•

,. 1

KO

itf ivM)

Fig. 4.4 Optical dispersion curves of inorganic glasses, 1. Fused silica, 2. ZrF/t-Based fluoride glass. 3.As2S3,4.GeSe4, 5.AsSe4

Therefore, the ultra-violet inherent wavelength ks can be found by plotting ! / ( « / - l j v s ( l - 1 " 2 ) . Fig. 4.4 shows this relationship for inorganic glasses, the deviation from straight line in infrared region is due to the resonant absorption of structural lattice vibration. The ensuring straight line can be extrapolated to zero to give Xs. On the basis of some simplification, the Ju, value can also be obtained from the next equation with two linear refractive index values: (o t 2 -l) 1 _

K

(n22-\) ^2

(4.19)

Fig. 4.5 shows the relation between n2(E) and Xs2 of optical glasses. Thus we can empirically derive several formulas for calculating nonlinear refractive index nt{E) of transparent glasses in the non-resonant region. There also exist some simple relationship between nonlinear refractive index «z(.E) and dispersion coefficient v (Abbe value). We can

126

Third-Order Optical Nonlinear Properties of Glasses

also estimate ni{E) value from v value. 150

a


_

IOO-

50

"\V

ZFi(^SF7)/^

-

UK 6 SiOo

X

(Wo;,

L_

-J.... , 1

,, 1

2

?.J 10l0 (cm ) Fig. 4.5 Relationship between n2(E) and ^.s2 of optical glasses

n2 = - 3 9 x 10

14

+ 29 x 10 ~* xA, 2 (esu), (Oxide glasses)

(4.20)

n2 = - l l x l O " 1 4 + 1 2 x l 0 " 4 x/l, 2 (esu), (Fluoride glasses)

(4.21)

vd at 68-90, n2 = 25.24 - 0.23t^(l0~ 14 esw),

(4.22)

vd at 25-68, n2=32\0/ud

-34.67(l0" 1 4 era).

(4.23)

Table 4.1 gives the comparison between the calculated and measured values of «2(£) of glasses. The calculation error is about 15%. Table 4.1 Comparison between calculated and measured values of n2(£) Type of glasses

Optical glasses

Laser glasses

Glass mark ZF-7 BaF-2 QK-3 N0312 N0812 N1012 N2120 N2420

Cal. Value of n2(E) (IP' 13 esu) 9.0 3.6 1.2 1.7 1.9 1.7 1.5 1.4

Exp. Value of n2 (IP"13 esu) 6.4+1.1 4.1+0.4 1.3+0.4 1.9+0.4 1.6+0.5 1.6+0.5 1.3+0.3 1.2

127

4.2 Optical Nonlinearity of Dielectric Glass

4.2.2 Optical nonlinearity of chalcogenide glasses and glass thin films Chalcogenides, included sulfides, selenides and tellurides, can form amorphous state in a wide range of composition. Due to anions, such as S2", Se2" and Te2", with lone electron pairs the third-order optical susceptibility x<3) m the non-resonant region was found to be several orders higher in chalcogenide glasses than this one of Si0 2 glass.110'11] The nonlinear response time of chalcogenide glasses is in the ps range, and the chalcogenide glasses and films are well transparent in the middle infrared region (<20 urn), therefore they are quite suitable for ultrafast switching and modulation in IR photonics devices.[12'13] Table 4.2 lists the x(3) value of several chalcogenide glasses with x(3) value of silica glass as comparison. It can be seen that some measured values are spread; the variation in these results may be caused by fluctuation of glass composition and quality, measurement technique and conditions. Table 4.2 The x(3) and n2 values of several chalcogenide glasses

Glass type Si0 2 As2S3

GeS3, GeS2 GeSe4 Ge-As-S-Se Ge1538AS30.77S48.47Se5.39

As-S-Se As4oS3oSe3o Ge^As^Sego Ge16As32Se52

Refractive index (X, um) 1.45(0.59) 2.48(1.06), 2.53(2)

Nonlinear refractive Nonlinear index n 2 ' susceptibility x<3) Reference 16 2 (10" cm /w) (X, um•) (10"13esu) (X, urn ) 3.1(1.06) 400(1.303),40(1.3), 250(1.00),80(1.55), 200(1.55)

800(1.06) 2.41(10.6) 35(1.55), 70(1.55) 2.47(10.6) 175(1.55) 190(1.55) 1020(1.06) 270(1.55)

0.28(1.06) 14.8(0.675), 22(1.9), 17.4(1.06), 120(1.319) 10(1.9)

14 14, 15,16, 17,

14, 15 15 16,17 16,17 15

Due to high polarizability of chalcogenide glasses the nonlinear absorption should be high, we have to measure the real part and imaginary part of third-order susceptibility of chalcogenide glasses. As mentioned above the OHD-OKE method can effectively separate the real

128

Third-Order Optical Nonlinear Properties of Glasses

and imaginary parts of complex third-order optical nonlinearity and determine their values and signs, thus the relaxation dynamics can be discussed. The diagrams of OHD-OKE signal intensity with delayed time of GeSe2 and Ge2oAs25Se35 are shown in Fig. 4.6 and Fig. 4.7 respectively. The nonlinear properties of amorphous chalcogenide films of As2Se3, GeSe2, Ge2oAs25Se55 and GeioAs4oS3oSe20 measured by OHDOKE are summarized in Table 4.3. Imaginary yF1 of GeSe2 Amplitude: 1.24

Real x"31 of GeSe. Amplitude: 3.64

0

-1000 -750 -500 -250

250

500

750 1000

-1000 -750 -500 -250

Delayed Time(fs)

0

250 500 750 1000

Delayed time(fs)

Fig. 4.6 The OHD-OKE signals of amorphous GeSe2 film (Real and imaginary parts) at two different heterodyne angles,-2° and 2°. Table 4.3 Nonlinear properties of amorphous chalcogenide films (X (3) )R

Sample

(X (3) )I

Response ]

(10' esu) (10"13esu) time (fs) ' 56 -20 <200 As2Se3 88 <200 -30 GeSe2 66 <200 -24 Ge20As25Ses5 <200 -40 Gei0As4oS3oSe2o 90

l

Real^'ofGe^As^Se^ Amplitude: 5.52 A I \2°

I

17

" „ (10-10esu) m2 /W) 5.18 3.09 7.4114 4.42 5.5586 3.32 7.5798 4.52

13

Imaginary V31 of

(1

Reference

°

m2/W) 0.2514 0.3944 0.3155 0.5299

7 18 19 20

Ge^s^^

Amplitude: 1.99

J

- 4 '

•

'

'

-1000 -750 -500 -250

^t

0

250 500 750 1000

Delayed Time(fs)

1

•

'

'

-1000 -750 -500 -250

0

250 500 750 1000

Delayed timeffs)

Fig. 4.7 The OHD-OKE signals of amorphous Ge2oAs25Se55 film (Real and imaginary parts) at two different heterodyne angles, -2° and 2°.

4.2 Optical Nonlinearity of Dielectric Glass

129

The major contribution to third-order susceptibility in the nonresonant region in dielectric glass media comes from one-photon resonant process, but when the frequency of the incident light approaches to the eigen absorption frequency of glass media the two-photon process is accompanied. By using fs laser beam in OHD-OKE measurement the two-photon and three-photon absorptions should be happened, therefore, y and /lvalues are higher than that of weak light intensity measurement. In amorphous GeSe2 and Ge2oAs25 Se55 films, the chalcogen elements (Se atoms) are only two-fold coordinated and possess lone pair electrons which are normally non-bonding. The non-bonding electrons lie at the top of the valence band and hence are preferentially excited by light and produce some short-life free electrons plasma and band filling effects. It is just the forming of the short-life electrons that make amorphous GeSe2 and Ge2oAs25Se55 films show a very fast response time within 200 fs, as shown in Figs. 4.6 and 4.7. The structure and properties of chalcogenide glasses and glass films can be changed by light exposure and heat treatment. Fig. 4.8 shows the spectral dependences of x(3) values of As40S6o films. After irradiation by a high pressure mercury lamp or heat treatment at 160°C for lh, the %{3) values are all increased.[21] It shows that the structure of chalcogenide glasses is not stable and the x(3> value depends on structural compactness.

8.0 7.0

I 6° o x

5.0

%

4.0 3.0 2.0 600

800 1000 1200 1400 wavelength (nm)

1600

Fig. 4.8 Spectral dependences of x(3) values of As40S60 films: (1) fresh-evaporated film, (2) exposed film and (3) annealed film.

130

Third-Order Optical Nonlinear Properties of Glasses

During light exposure and heat treatment the photodarkening and photobleaching in amorphous chalcogenides are more pronounced effects. With bandgap light illumination of wavelength of 514.5 nm from an argon ion laser, photobleaching was observed in amorphous GeS2 thin films and the optical transmittance edge shifted to shorter wavelength as shown in Fig. 4.9. Photobleaching in well-annealed films is reversible and could be erased by annealing near the glass-transition temperature but it was irreversible in un-annealed films.[22] The photodarkening effect in amorphous AS2S3 films was observed. The optical absorption edge shifts to a lower energy after illumination at the bandgap light of 514.5 nm wavelengths by an argon laser (Fig. 4.10). The shift in well-annealed films can be recovered by annealing at 180°C for lh, but in un-annealed films it is irreversible.'231 500

b 0

800

1000

12i

Wavelength.nm

Fig. 4.9 Optical transmittance edge of aGeS2 thin films, (a) As-deposited film, (b) illuminated film, (c) annealed film, (d) illuminated after annealing.

E (eV)

Fig. 4.10 Optical absorption edge of a-As2S3 films with a being the absorption coefficient (cm 1 ) and hu the photon energy: (a) as-deposited film; (b) annealed film; (c) film illuminated after annealing; (d) illuminated film.

The reversible photobleaching and photodarkening show that the lone pair electrons play an important role. Because in amorphous chalcogenide films, the S and Se atoms are only two-fold coordinated and possess a lone pair of electrons which are normally non-bonding, but can undergo light-induced reactions to produce structural defects of three-fold or singly coordinated S or Se atoms. The sites associated with

4.3 Optical Nonlinearity of Organic-Inorganic Hybrid Glasses

131

the non-bonding electrons lie at the top of the valence band and hence are preferentially excited by illumination to induce the structure change. In case of light illumination, inner core electrons can be excited, i.e. generation of inner core holes. The inner core holes can be immediately filled by outer electrons with Auger process which could induce more holes in upper states (bonding and lone pair states), since one Auger process creates two holes (vacancy cascade process). In this situation, bond-breaking or ionization of atoms is easy to occur, leading to a change in local structural order in the amorphous network.[24] The two irreversible photo-induced effects belong to local structural order change, which resulted in crystallization in micro-region, thus induced the irreversible refractive index change. 4.3 Optical Nonlinearity of Organic-Inorganic Hybrid Glasses Although some organic compounds possess high ability of optical nonlinear absorption and refraction, the thermal and light instabilities are the main problems for the application of organic materials. To solve these problems, the research on doping active organic phase into inorganic host matrix to prepare hybrid materials has been intensified in recent years.1251 The main inorganic matrices are Si02, A1203, Si02-Ti02 based glasses prepared by sol-gel method. The silica gel and glass are the most popular ones, therefore, the organic and inorganic hybrids are called ormosils (organic materials in silica host). 4.3.1 Non-linear optical susceptibility of organic material Most of the third-order nonlinear organic materials are conjugated polymers and dyes. Fig. 4.11 shows the molecular formula and %(3) values of some conjugated polymers. It can be seen that the x(3) values of organic compounds is much higher than that of inorganic glasses. The 2D and 3D octupolar structures open new perspectives for the third-order nonlinear optical media. High nonlinearity value and improved transparency as compared to the corresponding dipolar structures can be expected. Fig. 4.12 shows the molecular structures of typical 2D and 3D compounds.[26]

132

Third-Order Optical Nonlinear Properties of Glasses

R—C—C=C—C=C — C ==C—C=C— C = C — C—R

x'h' 8.5 X 10"'°

PolydiacetylewfPDA) O R=—CH,—O

-

-// %

CH,

O

Para-toluene sulphonate (PTS) JT"'=2.5X10""eiu Polybenzoxazole (PBO)

-oaro v

-?w«10-'°eSu

Polybenzothiazole (PBT) CH,—CH,\

/-y

* " ' - 1 0 ' - 10" esu

CH, — C H / \ = / 4-(N-diethylamino)-4-nitrostylbene (DEANS)

Fig. 4.11 Structural formula and x(3) values of some organic compounds. Guanidinium analogs Central acceptor NH,

N(CH,)j ' ' Crystal vio

NH,

^NHj

Central donor CN

CN'

-N(CH,),

N(CH,):

CN

TATB analogs NO,

f

HJN^^A^-NHJ

N0fV-N0, NH: 0

*

*

X J% I ^ X ^ ^

CN

' ^ ^

CN'' ""CN

A

.^!'V A

Fig. 4.12 Structural formula of 2D (Guanidinium analogs and TATB analogs) and 3D organic compounds.

4.3 Optical Nonlinearity of Organic-Inorganic Hybrid Glasses

133

Another group of organic materials are organo-metallic compounds, such as metallo-azodyes and metallo-phthylocyanine etc. Their molecular structures are shown in Fig. 4.13. The organo-metallic compounds such as phthalocyanine , which possess not only high thirdorder nonlinearity, but also nonlinear optical absorption (two-photon), thus can be used as optical limiter. The more prominent bio-molecular materials emerging in the field of nonlinear optics are the photochromic protein, such as the bacteriorhodopsin (Br). M-Phthalocyanine

M-Azo

"4.

>

m/

Pi

i O ~ \ - J > = 0 L» •o lfrCo,Hi,Zh,

R 3 '\*s= i '

" • * = • -R,

Fig. 4.13 Molecular structure of metallo-azo and metallo-phthylocyanine dyes.

4.3.2 Method of Preparation The nonlinear organic compounds are easy to dope into organic polymers, but the main problem is their thermal and light instabilities. The optical nonlinear hybrid materials prepared by the sol-gel process have drawn more attention in recent years due to its moderate temperature involved in preparation, its homogeneity in molecular level, and its increased thermal and chemical stability. Sol-gel process has been thought to be an effective method to prepare this kind of non-linear optical organic materials. The properties of organic-inorganic hybrid materials not only depend on the properties of organic and inorganic parts, but also on the interaction between two parts. There are the two types of doping: 1. Organic molecules embedded in inorganic matrix, there is only weak

134

Third-Order Optical Nonlinear Properties of Glasses

interaction by hydrogen bond, or Vander Waal's bond and electrostatic interaction; 2,Organic molecules bonded in inorganic chains or networks by covalent bond. The type of doping is dependant on preparation process. For preparation of dye-doped ormosils, Si(OC2H5)4(TESO), (CH3)2Si(OC2H5)2(DMDEOS), CH2(0)CH-CH2-0(CH2)3Si(OCH3)3 (GPTMOS) and CH3Si(OCH3)3(MTMOS) were used as the starting materials, in which ethanol (C2H5OH) mixed with HCL solution to form the sol. Dye dissolved in chloroethane was dropped into the above mentioned mixture to form the precursor sol. By using spin coating or dip coating the ormosils thin films can be prepared. The precursor sol is poured into dry box to form the dye-doped gels after gelation. In this preparation process the dye molecules are embedded into the holes of porous silica glass. The flowchart for preparation of phthalocyanine dye PPcVO doped film is shown in Fig. 4.14. In type 2 doping the nonlinear organic (NLO) chromophores are functionalized with trialkoxysilanes, Si(OR)3 by side chain or main chain combination with in-situ synthesis to form R'Si(OR)3 precursor sol (R' is NLO chromophor). Taking an example, Fig. 4.15 shows the molecular structure of NLO chromophore, aminosulphone (ASF) with trialkoxysilane S(OR)3 by main chain combination. GPTMS+ C2H50H

TEOS+C2H50H

1

t

Stirring 4

0.04NHCI+C2H5OHN

PPcVO Stirring for 3 hours Sol

''

Spin-coalinj

Dye-doped sol-gel

•

Drying from R:T. to 120°C ov srday

Films

Fig. 4.14 Flowchart for preparation of phthalocyanine PPcVO doped ormosil film.

4.3 Optical Nonlinearity of Organic-Inorganic Hybrid Glasses

135

O

(l-lO)jSi(CH;).vHSCOO(CH2)2- ] * ~ \ 0 / " ~

N

- "~"

A >N N

O

,CH.Cll20CONH-(CH2)jSiiOEt)j

N ~(0}~ @ > ~ N'v N

VV

CH2CH;OCONH-{CH2)3Si(Oni)\

Fig. 4.15 NLO chromophores, ASF fiinctionalized with trialkoxysilanes

4.3.3 Third-order optical nonlinearity of organic-inorganic hybrid glasses Table 4.4 shows the nonlinear refractive index n2 of a series of dye in TiOM (ormosil-Ti02). The data listed above were measured by z-scan method using doubled frequency of Nd:YAG laser with 0.5 us pulse at wavelength of 532 nm (near resonant).[27] Fig. 4.16 shows the measured nonlinear refractive index values of pyrylium G, fluorescein and acridine orange are extremely high. The acridine in heavy metal fluoride glass has been shown to exhibit saturable absorption in a ns time scale.[28] Table 4.4 Nonlinear refractive index of some dye doped ormosil-Ti02 Dye

Concentration 5xl0 -4 2xl0"5 2xl0' 5 2xl0"5 4xl0 -5 3xl(r 5

DODCI MG Pyrylium G Pyrylium G Fluorescein Acridine orange

ynm 591.0 629.6 577.4 550.0 480.6 491.0

n2xlO"H/esu 9.21 13.13 12900 3523 12191 3489

&

I I I 1

0.50 0

o o

0.45

o

00

A 0


0

o° o 0

0

0.40

2

0.35 -

0°

0°

0.10 -2

i/cra

Fig. 4.16 Measured z-scan of the pyrylium-G-doped ormosil.

136

Third-Order Optical Nonlinear Properties of Glasses

Phthalocyanine compounds have good physical properties, that is, a thermal and light stability. It is easy to get a thin film by spin coating of molecularly doped polymer or ormosil. Table 4.5 shows the third-order susceptibility and absorption peak of metallo-phthalocyanine compound films at a wavelength of 1060 nm.[29'30] The %(3) value of vanadylphthalocyanine is larger than that of the others. Fig. 4.17(a) shows the measured z-scan of Pr4VOPc-ormosil film with thickness of 80 nm. The shape of pike-valley in the curve demonstrates the negative value of nonlinear refractive index (defocusing). The n2-value is around 10"11 esu at wavelength of 532 nm and pulse duration of 100 ps.[31] With 1.064u,m 38 ps laser pulses, the nonlinear refractive index n2, TPA, coefficient fi and the third-order nonlinear susceptibility x(3) of Palladium-octaisopentyloxy-phthalocyanine (octa-PdPc) doped ormosil were obtained to be l.lxlO"12 esu, 2.5xlO"19 m2/W, 33 cm/GW, and 1.6xl0"13 esu, respectively.[32] Fig. 4.17(b) shows the z-scan curve of Octa-PdPc doped gel. Table 4.5 Third-order susceptibilities and absorption peaks of PcM thin films. X (3) (xl0- 12 esu) 190 80 15 6.0 4.6

Compound PcVO PcSn PcCo PcH2 PcNi

Absorption peak (nm) 736,821 723,878 617,684 634,700(s) 627,675(s)

(a)

1.15

•

(b) •

1.10

1.05

1.00

1*-

«•

•

- • "#

M

&

ta-

• •

li.

• 0.95

0.90

1

,B

•

• • .«

ot.

•

OT.

-

0R">

t

O

©

tP

«B-

i

aa. o»-

&o

'

" ' ' ' •

*

**$?>

\

V %

*

•

•

0.0

-2.5 z/cm

Him)

Fig. 4.17 Measured z-scan of the Pr4 VOPc-doped ormosil (a) and octa-PdPc-doped gel. (b)

4.4 Optical Nonlinearity ofNano-composite Glasses

137

4.4 Optical Nonlinearity ofNano-composite Glasses Nano-composite glass means the nano-particles dispersed in inorganic glass matrix. There are two kinds of nano-particles doped glasses: the semiconductor dots doped glasses with quantum confinement effects and the metallic dots doped glasses with dielectric confinement effects. A comprehensive review of nonlinear optical properties of semiconductordoped and metal-doped glasses has been reviewed before.[331 Here we present the new progress in this field. There are emerged new groups of nano-composite glasses. They are semiconductor micro-crystallites with high %(3) value, such as CdTe, PbSe and Bi2S3, and fullerenes (C6o) nano-particles and semiconductor oxide, such as ZnO, Sn0 2 , Fe 2 0 3 etc, nano-particles doped composite glasses, which were developed by us recently, the experimental results are shown in this section. 4.4.1 Semiconductor micro-crystallites doped nano-composite glasses The optical properties of semiconductor micro-crystallites are very different from that of the bulk one. Due to the quantum confinement effect the conduction and valence bands of bulk materials split into a series of discrete energy levels, and the surface effects also become increasingly important.[34] The quantum size effects can be classified into two categories depending on the ratio of micro-crystallite radius y to Bohr exciton radius y^. If the radius of the micro-crystallites is close to or smaller than Bohr radius of exciton (y<jB), it belongs to electron and hole individual confinement, which is called the strong confinement, and can be observed in chalcogenide micro-crystallites, such as CdS, CdSe etc. If the radius of the micro-crystallites is larger than the Bohr radius of exciton (y>yB), all the excitons take place to confinement, which is called weak quantum size confinement. The halide semiconductor, such as CuCl, CuBr, belongs to this category. The third-order optical nonlinearity of semiconductor microcrystallite doped glasses can be divided into two cases: wavelength of optical nonlinear effects in the resonant region and the non-resonant region in dependence of the ratio of incident light wavelength ^ n to eigen

138

Third-Order Optical Nonlinear Properties of Glasses

absorption. In resonant region the optical nonhnearity is originated from enharmonic electronic structure of different electronic energy levels, the optical nonlinear effect is large, but the response time is slow and the signal absorption loss is high, sometimes it is called active optical nonlinear process. In non-resonant region the third-order optical nonhnearity is induced by the distortion of electronic cloud (or shell), the nonlinear effect is smaller than that of the former, but the response time is quick and the signal loss is low, it can be used as passive optical nonlinear process. 4.4.1.1 Chalcogenide semiconductor micro-crystallites Since Jain and Lind (1983) firstly reported on the high nonlinear optical properties of CdSSe-doped silicate glasses,[35] there have been many papers on the optical nonhnearity of semiconductor microcrystallines in glass.[36"38] The approaches for increasing optical nonhnearity of semiconductor nano-composite materials are as follows: increasing particle concentration, working at near resonant wavelength and adding dielectric confinement effect besides quantum size confinement effect. (l)For increasing particle concentration the sol-gel method has been adopted. As an example, under near resonant condition the ^ 3 ) of 2% CdS doped glass was estimated to be 1.5xl0"12esu (390 nm). With increasing the doping concentration up to 8%wt of CdS by sol-gel process the ^ 3 ) value rises to 7xl0"10 esu at room temperature measured by DFWM method.[39] (2)Semiconductor micro-crystallites with higher optical nonhnearity, such as CdTe, PbeSe and Bi2S3 doped in glasses have been developed. The CdTe micro-crystallites doped glasses were prepared by laser evaporation method.[40] The third-order nonlinear susceptibility ^ and decay time v of the thick glass films (thickness is around 1-1.5um) were measured by DFWM method. The ^ value depends on the light wavelength, as shown in Fig. 4.18. It can be seen that near the eigen absorption the ^ value increases rapidly. The x value was estimated to be shorter than 10 ps.

139

4.4 Optical Nonlinearity ofNano-composite Glasses

Recently the PbSe micro-crystallites doped glasses have been studies in detail.[41] Te host glass composition is borosilicate (Na20-PbO-ZnOB 2 0 3 -Si0 2 ). The absorption spectra of glass samples with different heat treatments are shown in Fig. 4.19. The spectra do not show well resolved band structures but only the exciton peaks or so-called exciton shoulders at a wavelength less than the absorption band edge. The exciton shoulder and the absorption band edge shift to a higher-energy region (blue shift)

700 Wavelength (ran)

Fig. 4.18 x<3> values measured on CdTe micro-crystallites doped glass sample, plotted as a function of wavelength. Linear absorption spectrum is also presented.

1300 1700 Wavelenglh (nm)

Fig. 4.19 Absorption spectra of samples 1,2,3,4 and 5.

as the average size of the PbSe micro-crystallite decreases. The blue shift can be attributed to the quantum size effect of the carriers (electron and hole) and the exciton shoulder is tentatively assigned to a transition from the highest hole sub-band to the lowest electron sub-band (Is-Is transition). The energies at the exciton shoulders are summarized in Table 4.6. Table 4.6 Heat treatment of samples and the experimental results at room temperature Sample

Heat treatment temperature(°C)

1 2 3 4 5

460 480 500 520 540

Average radius (nm) 1.6+0. 3 2.510.3 4.010.4 6.510.3 7.3+0.4

Band gap (eV)

%(3)(esu) ( + 20%)

0.730 0.654 0.591 0.565 0.552

1.46xl0"9 1.15xl0~9 1.30xl0"9 LlOxlO" 9 0.75x10"9

140

Third-Order Optical Nonlinear Properties of Glasses

The optical nonlinearities of PbSe micro-crystallites doped glass were measured by the single beam Z-scan technique. The irradiation source is a mode-locked Nd-doped yttrium aluminium garnet laser operating at 1 Hz with 50ps pulse width and 1.06um wavelength at room temperature. Fig. 4.20(a) gives the Z-scan curve with a collecting aperture for sample 1. The normalized far-field transmittance for a small aperture behaved as a function of sample position relative to the focal plane of the lens shows a decreasing and then increasing intensity while moving through the plane, which indicates a positive intensity-dependent

/

" -5

-4

-3

-2

-1

0 1 z(cm>

2

3

4

5

-.5

j -4

1 -3

1 -1

^

1 -1

^

1 1 0 1 z(cm)

^

1 2

1 3

(b) 1 4 5

Fig. 4.20 Single-beam Z-scan curves of sample 1. The peak intensity of the incident beam is I0=0.6GWcm"2; the linear transmittance of the collecting aperture is (a) S=0.4 and (b) S=l.

refractive index y. All the samples displayed similar behaviours. Fig. 4.20(b) gives the Z-scan curve without the collecting aperture for sample 1, which shows that the normalized transmittance has a peak at the focus (z=0). This curve indicates that there is a saturable absorption in the sample 1. The estimated values of x(3> for different samples are shown in Table 4.6. As shown in Fig. 4.19, the exciton shoulder was clearly observed, which indicates the presence of exciton effect in the present system, which would enhance the third-order optical nonlinearity. Thus, the larger optical nonlinearity may arise from the exciton resonance effect.

4.4 Optical NonHnearity ofNano-composite Glasses

141

The response time of optical nonHnearity for sample 1 was also measured by pump-probe time-resolved Z scan (Fig. 4.21), which indicates that the response time is much faster than the 5 Ops pulse duration. These observations are of the results of electronic third-order nonHnearity for the samples of PbSe micro-crystallites doped in glass. The larger third-order susceptibility %0) (about 10"9 esu) and shorter response time (less than 50ps) for such material were estimated. These experimental results suggest that PbSe micro-crystallites doped in glass constitute a promising material for applications in photonics devices. 8c 1.05 to

I

1-0

I 0.95 m a I 0.90 T3

a 1 0.85

•

'

I

I

I

I

L

30

40

CO

E 2 0.80 - 5 0 - 4 0 -30 - 2 0 - 1 0 10 20 Probe delay time (ps) z

50

Fig. 4.21 Normalized probe transmittance versus probe delay for sample 1. A pulse duration of 50ps full width at half-maximum is calculated from the fit indicated by the solid curve.

(3)Optical nonHnearity can be enhanced by dielectric confinement effect, which is s surface polarization effect induced by trapped state and atomic vacancy defect. This effect depends on the dielectric constant ratio (ei/e2) of micro-particle and surrounding. The influence of surface coating and heat treatment on the third-order optical nonHnearity of Bi2S3 nano-particles with benzene sulfonic acid sodium salt (DBS) doped ormosil has been studied.[42] Fig. 4.22 shows the absorption spectra of Bi2S3 doped ormosils (sample 1) dried at nitrogen atmosphere and heated at 150°C for 2 hrs. The obvious red-shift can be observed, it demonstrates the dielectric confinement effect, because the Bi2S3 particles are coated with organic DBS doped layer during heat treatment. Fig. 4.23 shows the z-scan curves of sample I and sample II. It can be seen that the self-focusing effect of both two samples and only exists the nonlinear refraction, while the nonlinear absorption is very small,

142

Third-Order Optical Nonlinear Properties of Glasses

Saltio-

\

ij»1.0-

s.

\

\

\ \

06-

*0

«00

^ ^ ^ '

600

700

900

60O

WBwtength(rm)

Fig. 4.22 Absorption spectra of Bi2S3 doped glasses sample I dried at RT(dashed line). Sample II heated at 150°C for two hours in nitrogen gas atmosphere (dot line).

l-lai

£ 1-°"

°

o

» 0 9-

1

o

-

^v„

o

^08
o

«j 0.7-

-

/

O 2

<*%

oaoo0^1^^

0.605-, -50

-40

-30

-20

-10

10

z (mm)

z(mm)

z (mm)

z (mm)

(i)

(ii)

20

30

40

50

(b) Fig. 4.23 Z-scan curves of sample (I) and sample (II), (a) close hole, (b) open hole.

4.4 Optical Nonlinearity ofNano-composite Glasses

143

because the peak and valley of Z-scan curves are in symmetry. The %{' value of samples I and II are 5.8><10"13 esu and 3.7><10"12 esu respectively. Due to low concentration of Bi2S3 in ormosil, the x(3) value is not large. The %(3) value of sample II is one order larger than that of the former, these results show that the thermal treatment increased the dielectric confinement effect, thus increased the x(3) value of sample II. The n2 value of Bi2S3 quantum dots gives 3.3xlO"9 esu near the resonant region.[43] 4.4.1.2 Oxide semiconductor micro-crystallites Some oxides, such as ZnO, Sn0 2 , Fe 2 0 3 , Bi 2 0 3 belong to wide gap semiconductors. Their nonlinear optical properties in microcrystalline state are quite different with that of bulk material due to quantum confinement, dielectric confinement and surface defects effects. Recently great attentions have been paid to the optical nonlinearities of those oxide nano-particles. Nonlinear optical properties of Fe 2 0 3 nano-particles were investigated by the signal-beam Z-scan technique. With nanosecond laser pulses, an off-resonant nonlinear refractive index up to 10"12 cm2/W has been reported in Fe 2 0 3 nano-particles.[44] The nonlinearity is due to the free charge carriers generated by two-photon absorption. It was also found that the n2 value is significantly enhanced in the coated Fe 2 0 3 nano-particles with respect to the bare particles due to dielectric confinement effect. Because the relaxation time of the surface states is generally long (on the order of ~10"9, ns), it should be possible to observe the nonlinear behavior with low-power laser irradiation. With Ar+ and Ne-He lasers the reported largest effective non-linear coefficient, «2= -8.07xlO"7cm2/W, was obtained. The dielectric confinement effect can be also observed in Bi 2 0 3 nanoparticles. Table 4.7 shows the comparative experimental results of x(3) values of nano-particle Bi 2 0 3 coated by stearic acid (ST), bare particle Bi 2 0 3 and bulk Bi203.[45] Between the difference of %(3) is about one order of magnitude. This shows a strong evidence of the enhancement of optical nonlinearity by dielectric confinement effect.

144

Third-Order Optical Nonlinear Properties of Glasses

Table 4.7 Experimental results of the z-scan measurements for bare Bi 2 0 3 nano-particles in hydrosol and coated ones in sol (±20%) Sample Band-edge (eV) P(cm/GW) Rex(3)(esu) Imx(3)(esu) y,(3)(esu)

Coated Bi 2 0 3 Hi 0.23 3.47X10"11 -2.61X1011 3.48X1Q-"

Bare Bi 2 0 3 3^00 0 3.47xl0"12 0 3.47* 10'12

BulkBi 2 0 3 2JW 0 4.2* 10"13 0 4.2xlQ-13

The investigation indicates that Fe 2 0 3 , Bi 2 0 3 and ln 2 0 3 [46] nanoparticles are promising materials for potential application in new photonic devices. Searching the methods for doping these nano-particles in glass media or uniformly crystallizing from glass media are essential challenges for development of new nano-composite glasses. 4.4.2 Fullerene (C60) doped nano-composite glasses New composites of fullerene/Si02-Ti02 systems have been synthesized by selecting suitable solvents and special sol-gel processes.'47' C6oNH2(CH2)3Si(OC2H5)3, a novel fullerene-silane compound, in which fiillerenes have high reactivity to primary and secondary amines through N-H addition reaction across the C=C bands of fiillerenes, and the Si-OR group can be hydrolyzed and subsequently condensed into a gel was synthesized. The maximum reaction ratio of C6o/Si can be reached to 1.6xl0"3. A good mechanical strength of the compound was obtained by leading some organic modified precursors such as 3Glcidoxypropltrimethoxysilane (KH550), Diethoxydimethylsine. The nonlinear index of refraction (n2) of sol and gel was obtained in the range of 5-6xl0"12 esu, corresponding to x(3)=l-5~2x10"11esu (at 1.06|nm) measured by z-scan method.[48] The approximate magnitude of n2 shows that fiillerenes are located in a similar environment, which are surround by amino-group. 4.4.3 Metallic particles doped nano-composite glasses The nonlinear optical properties of metallic particles doped nanocomposite glasses are induced by local surface plasma resonance at the

145

4.4 Optical Nonlinearity ofNano-composite Glasses

interface between metallic particle and glass matrix. Therefore, the nonlinear optical effects are mainly in the resonant region, and affected either by an electric-field enhancement effect (dielectric confinement) or by a quantum confinement effect. The commonly used metallic particles are Au, Ag and Cu, which are doped into glass media by melting, implantation, sputtering and sol-gel process. The % value of metallic particle itself is high. For example, the %(3) value is around (2-4) xlO"9 esu for silver, ~10"6 esu for copper and ~8 x 10"9 esu for gold particles,[49,50] thus the estimated value of x(3) of metallic particles doped glasses is in the range of 10"8~10"9esu for Au- or Ag-doped glasses and 10"6~10"7 esu for Cu-doped glasses in dependant of incident light wavelength. In practical application it should consider both %(3) value and absorption coefficient a at incident light wavelength, thus the materials figure of merit F is defined as F = z(3)/cc.

(4.24)

By using a conventional melting method the maximum amount of gold particle content is limited to less than 0.01%vol. High concentration of gold particles can be obtained by the sol-gel synthesis and the thin films prepared by dip-coating method. Table 4.8 shows the x(3) values of Au micro-crystallites doped glasses at ns and ps laser pulse duration.[51] Xps(3) values of an order of 10"8esu were obtained, while xps(3) were 20 to 40% XPs(3)- It should be noted that %ps'3) purely reflects the electronic effect, while %ps(3) includes both the electronic nonlinearity and the thermooptic effect from laser-induced heating. Therefore, 60 to 80% of the Xps(3) is due to the thermo-optical contribution. Table 4.8 The Optical characteristics and third-order nonlinear optical susceptibility of gold microcrystal-doped silica films heat-treated under different conditions NQ

#1 #2 #3

P>no-8esirt *• '

Absorbanceat 532nm(cm )

Peak position (nm)

FWHM of peak (nm)

Ans

0.18 0.14 0.12

530 551 557

45 64 78

1.38 1.68 1.34

y (3) / y (3) *PS ^ns 0.42 0.38 0.19

146

Third-Order Optical Nonlinear Properties of Glasses

4.5 Optical Limiting Effects The devices to control the light intensity are called optical limiters. The passive optical limiters are the devices that control the intensity of light by utilizing the optical nonlinear effects of materials. The passive optical limiting effects can be performed by reverse saturated absorption (RSA), two-photon absorption (TPA), absorption of free carriers, nonlinear refraction, photo-induced scattering and photo-refraction etc. Some optical limiting effects, as shown below, are mainly induced by the thirdorder optical nonlinearities. 4.5.1 Measurement of optical limiting effects The schematic diagram of a optical limiting set up is shown in Fig. 4.24. A Q-switched Nd:YAG laser with a 532 nm wavelength, 10ns pulse duration and 1HZ repetition rate was used for the optical limiting measurements. The laser beam was focused onto the sample by a lens with 180 nm focal length. A small part of the input beam was picked off with a beam splitter (BS) and measured by an energy meter (DI) to monitor the input energy. The laser transmission was measured with the energy meter (D2). For identifying the type of optical limiting effect, spontaneous or accumulative, it can be performed by changing the laser pulse duration, from pulsed to CW lasers. Sample

Lens

BS

—^

IV

Nd:YAG

•0-

Probe 2

IChs 532m

Probe 1

Fig. 4.24 Optical limiting measurement.

4.5.2 Optical limiting effects ofC6o containing ormosil New hybrid organic-inorganic composites of C6o-NH2(CH2)3Si(OC2H5)3 (C6o-3-Aminopropyltriethoxysilane,C60-KH550), C6o-NH2(CH2)3 Si (OC2H5)3

4.5 Optical Limiting Effects

147

-CH2OCHCH2O(CH2)3Si(OCH3)3(3-Glcidoxypropltrimethoxysilane,KH560), andC6o-NH2(CH2)3Si(OC2H5)3-(CH3)2Si(OCH3)2 (Diethoxydimethylsilane, DTDS) have been synthesized with sol-gel process. The precursors and composition of the samples are listed in Table 4.9.[52] Optical limiting effect of C60-KH550 ormosil is shown in Fig. 4.25. By increasing the incident laser intensity, the increase of transmitted laser energy tends to saturation gradually. This is caused by the reverse saturation characteristics of the fullerene in the gel. No laser damage phenomenon in the sample was observed under microscope when the incident laser intensity raised up to 5J/cm2. At low incidence, the transmittance of C60-KH550 was about 51% and 21%. Along with increasing the fluence in, the fluence out was limited to about 0.85J/cm2 and 0.4J/cm2 respectively.'531 The optical limiting parameters are governed by the absorption of Cgo and the matrix, the reflection of the face and the scattering losses of the sample etc. The absorption at 530 nm of undoped (blank) KH550 gel is very low; the optical limiting effect is mainly due to the nonlinear absorption of C6o- The absorption at 530 nm is on the edge of the broadband absorption of C6o, the peak absorption wavelength is located at 280 nm in ultraviolet region. The optical limiting effects of all samples, mentioned in Table 4.9, are shown in Fig. 4.26.[54] Table 4.9. The precurors and compositions of the samples. Component KH560 0.04NHCL KH550 C60/Si Gel thickness Nomol C2H5QH DTDS molar ratio (cm) Al 3.5 1.2 1 1.63x10-4 0.10 A2 3.5 1.5 1 3.26x10-4 0.12 Bl 1 3.0 4.0 0.37 4.48x10-4 0.115 B2 1 3.2 1.8 0.25 2.0x10-4 0.11 B3 1 3.5 1.5 0.14 0.81x10-4 0.189 B4 1 3.0 1.7 0.5 0.38 1.3x10-4 0.15 KH550=3-Aminopropyltriethoxysilane,NH2(CH2)3Si(OC2H5)3; KH560=3-Glycidoxypropltrimethoxysilane, CH2OCHCH20(CH2)3Si(OCH3)3; DTDS=Diethoxydimethylsilane, (CH3)2Si(OC2H5)2

4.5.3 Optical limiting effect ofmetallo-phthalocyanines Significant optical multiple diffraction rings of the octa-PdPc-doped

148

Third-Order Optical Nonlinear Properties of Glasses

oraiosil samples under CW Ar+ laser illumination at wavelengths of 457.9, 488.0 and 514.5 nm were observed. Fig. 4.27 shows the diffraction rings at wavelength of 488 nm. Both the number of rings and the size of the outmost ring were intensity dependent. Under the same incident power of laser beams with different wavelengths there were different numbers of diffraction rings and different sizes of outmost rings. The largest number and size of diffraction pattern for the samples were observed at the wavelength of 457.9 nm. The second was at the wavelength of 488.0 nm and the last at 514.5 nm. This phenomenon of diffraction was used in optical limiter. Fig. 4.28 shows the optical limiting behaviours of octa-PdPc doped ormosil at different light wavelengths. The optical limiting threshold is different for different wavelengths.t55,56]

Fig. 4.27 Diffraction rings at wavelength of 488nm by octa-PdPc ormosil.

100

200 Input pewer/tnW

3#S

Fig. 4.28 The optical limiting behaviours of octa-PdPc-doped ormosil. (a) A,=514.5nm; (b) ?i=488.0; (c) X=457.9nm.

4.5 Optical Limiting Effects

149

Optical limiting response of octa-PdPc-doped organically modified sol-gels was also measured by using a Nd:YAG laser with 10ns pulse at 532 nm. It is shown that octa-PdPc-doped organically modified sol-gels is a kind of promising optical limiting material in pulsed laser operation. The optical limiting effect is induced by nonlinear refraction. Many attempts to increase the MPc concentration in ormosil have been reported. By using possible reaction between an amine group in 3-aminopropyl-triethoxysilane (APTS) and Chloride in SiPcCl2 high concentration of SiPcCl2 up to ~10"3M/L has been obtained/571 The obvious change of transmission has been occurred at the incident wavelength of 532 nm, the fluence is from 0.2 to 1.7 j/cm 2 for the SiPccontaining ormosil. 4.5.4 Optical limiting effect of micro-particles in sol and gel Because of coated micro-particles with high third-order optical nonlinearity, the optical limiting behaviours of PbS, Bi2S3, ln 2 0 3 and Bi 2 0 3 have been observed in sol and gel. For an example, nanometresized ln 2 0 3 coated with a layer of stearic acid (St.) possess large effective nonlinearity «2 in range of 10"7cm2/W, and strong lens effect. Fig. 4.29 shows the limiting behaviours measured at 632.8, 514.5, 488.0 and 457.9 nm, respectively. After exceeding the threshold for incident power, the output power of the limiter decreases as the input power increases rather than keeping a constant value. This means that the limiter will shut down the input once it exceeds the threshold. This property may be useful inprotecting sensitive optical detectors. From the limiting characteristics, we can obtain that the limiting threshold power for 632.8, 514.5, 488.0 and 457.9 nm are of -780, -650, -450 and -250 W/cm2, respectively. This indicates that the limiting result of the sample for 457.9 nm is better than others.1581 The sample of PbS nano-particles coated with DBS in organic sol was prepared via colloid Chemistry. The radius of PbS particles was about 3±0.3 nm. The thermo-optic coefficient of surface-embraced PbS nanoparticles was large, which can be measured with laser beams of 632.8 nm, 514.5nm and 457.9 nm emitted by He-Ne and Ar+ lasers. The measured results are listed in Table 4.10. The optical limiting behaviors of coated

150

Third-Order Optical Nonlinear Properties of Glasses

PbS particles are similar to that of ln 2 0 3 nano-particles, as show in Fig. 4.29.[59] Table 4.10 Experimental parameters and measured dn/dT and n2 values of PbS nanoparticles X/nm £«b/um Pin/mW VW . cm"2 a/cm"' S n2/cm"2 • W"1 (dn/dT) K"1

475.9 -30 1.1 77.8 3.2 0.5 -8.07xl0"7 -1.47xl0"4

0

5

514.5 -32 1.5 93.3 2.0 0.5 -4.95xl0"7 -1.26xl0"4

10 15 20 26 30 Input Power(mW)

632.8 -50 2.5 64.0 1.6 0.5 -4.34xl0"7 -5.69xl0"5

35

40

Fig. 4.29 Optical limiting behaviours of coated in In203 nano-particles. (a) A,=632.8, (b): ?L=514.5, (c) ^=488.0, (d) ?.=457.9nm.

References 1. Fuxi Gan, Nonlinear Optical Properties of Laser Materials, in: Laser Material, World Scientific Publishing Co. Ltd, Singapore, 1995,111-127.

References

151

2. Zhijian Liang, Fuxi Gan, Zhengrong Sun et al, Opt. Mater. 14 (2003) 13-17. 3. Y. Pang, M.Samoc, P.N.Prasad, J.Chen.Phys. 94 (1991) 5282. 4. M.S.Bahae, A.A.Said, T.H.Wei et al. IEEE J.26 (1990) 760-769. 5. Baolong Yu, Congshan Zhu, Fuxi Gan, Opt. Mater. 8 (1997) 249-254. 6. Gouhong Ma, Lijun Guo, Ye Liu et al. Phys. Lett. A 287 (2001) 385. 7. Qi-Ming Liu, Jun Mi, Shixiong Qian, Fuxi Gan, Chin Phys. Lett. 19 (2002) 575. 8. Fuxi Gm,Nonlinear Optical Properties of Glass, in Optical and Spectroscopic Properties of Glass, Springer-Verlag, Berlin, 1992. 9. L.Kang, S.Smolorz, T.Krass, et al. Phys. Rev. B-54 (1996- II) R 12641. 10. Q.Lin, J.Mi, S.Qian, F. Gan, Chin. Phys. Lett. 19 (2002) 575. 11. H.Kobyashi, H.Kanbera, J.Koga, K-Kubodera, J. Appl. Phys. 74 (1993) 3683. 12. M.Asobe, T.Kanamori and K.Kubodera, IEEE J.Quant. Electr-QE -29 (1993) 2325. 13. J.M.Harbold, F.O.Ilday, F.W.Wise et al, Opt. Lett. 27 (2002) 119-121. 14. H.Nasu, Y.Ibera and Kubodera, J.Non-Cryst. Solids, 110 (1989) 229. 15. F.Smektala, C.Quemard, L.Leneidre et al, J.Non-Cryst. Solids, 239 (1998) 139. 16. K.A.Corqua-Richardson, J.M.Mckinley,et al, Opt. Mater. 10 (1998) 155. 17. J.M.Harbold, F.O.Ilday, F.W.Wise et al. Proc. SPIE, vol.5061 (2003) 143-148. 18.Qiming Liu, Xiujian Zhao, Fuxi Gan et al, J.Non-Cryst. Solids, 351 (2005) 3147315. 19. Qiming Liu, Xiujian Zhao, Fuxi Gan et al., Solid State Commun. 134 (2005) 513-54. 20. Qiming Liu, Xiujian Zhao, Fuxi Gan et al., Opt.Commun. (2005 in press). 21. M.Frumar, J.Jedelsky, B.Frumarova et al, J. Noncryst. Solids. 326/327 (2003) 399404. 22. Qiming Liu and Fuxi Gan, Mater. Lett. 53 (2002) 411-414. 23. Qiming Liu and Fuxi Gan, Chin. Phys. Lett. 19 (2002) 124-126.. 24. S.R.Elliott, J.Noncryst.Solids, 81 (1986)71. 25. Fuxi Gan, Progress in Nature Science, 9 [8](1999)561-569. 26.1.Zyss, C.Dhenaut, I.Samuel et al, Digest IQEC'94, QTuB-2 (1994) 30. 27. Z.H.Jiang, H.Ye, L. L. Hu, Optical nonlinearity sol-gel glasses and composite, in Proceedings on XVII International Congress on Glass, edited by Gan Fuxi, Beijing, Chinese Ceramic Society, 1995, 1:203. 28. S.V.Gaponenko, V.P.Gribkovskii, L.G.Zimin, etal, Opt. Mater, 2 (1993) 53-58. 29. T.Wada, Y.Matsuoka, et al, MRSInt'l Mtg. on Adv. Mats., 12 (1989) 75.. 30. M.Hosoda, T. Wada, et al, Mat. Res. Soc. Symp. Proc, 175 (1990) 89. 31. F. L. Tang, F. X. Gan, C. S. Zhu, Sol-Gel Optics III, 1994, Proc, SPIE, 2288 (1994) 350. 32. Y.Z.Gu, Z.J.Liang, Fuxi Gan, Opt. Mater, 17 (2001) 471-475. 33.M.Yamane and Y. Asahara, Glasses for photonics, Cambridge University Press, 2000, 182-242. 34. Fuxi Gan, J.Non-Cryst.Solids, 129 (1991) 299-310. 35. R.K. Jain and R.C Lind, J.Opt. Soc. Amer., 73 (1983) 647. 36. T.Takada, T.Yano, A.Yasumori, etal, J.Non-Cryst.Solids, 147&148 (1992) 631. 37. M.Nogansi, Yiqing Zhu, Y. Tohyama and K.Nagasaka, J. Amer. Cer. Soc, 74 (1991) 238. 38. R.Reisfeld, SPIE, Proc. 1758 (1992) 546.

152

Third-Order Optical Nonlinear Properties of Glasses

39. Qingchun Zhao, Wensheng Shi, Liangying Zhang and Xi Yao, "Sol-gel preparation and nonlinear optical properties of semiconductor doped silica glasses", Proc. XVII International Congress on Glass, Beijing, Vol.4 (1995) 471. 40. Shuneuke Ohtsuka, Keiji Tsunetomo, Tadashi Koyama and Shukei Tanaka, Opt. Mater. 2 (1993) 209-215. 41. Baolong Yu, Congshan Zhu, Haiping Xia, Hongbing Chen, Fuxi Gan, J. Mater. Set Lett. 16(1997)2001. 42. Zhijian Liang, Yuezhong Gu, Fuxi Gan, et al, Acta Optica Sinica, 20 (2000) 414418, (in Chinese). 43. Zhonghong Jiang, HuiYe and Lili Hu, "Optical nonlinearity in sol-gel glasses and composites", Proc. XVIIInternational Congress on Glass, Beijing, Vol. 1,203 (1995). 44. Baolong Yu, Congshan Zhu, Fuxi Gan, Physica E, 89 (2000) 360-364. 45. Baolong Yu, Congshan Zhu, Fuxi Gan et ah, Opt. Mater. 8 (1997) 249-254. 46. Baolong Yu, Congshan Zhu, Fuxi Gan, J. Appl. Phys.82 (1997) 4532. 47. Haipmg Xia, Congshan Zhu and Fuxi Gan,"Study on permeation of fullerenes into porous glass", Proc. XVII International Congress on Glass, Beijing, Vol.3 (1995) 668. 48. Haiping Xia, Baolong Yu, Congshan Zhu, Fuxi Gan, Acta Optica Sinica, 18 (1998). 49. D.Richard, Rh. Roussignol and Chr.Flytzanis, Opt. Lett. 10 (1995) 511. 50. K.Uchiada, S.Kaneko, S.Omi, C.Hata et al. J.Opt.Soc.Amer.B-ll (1994) 1236. 51. Jan Matsuoka, Hiroyuki Nasu and Kanichi Kamiya, "Sol-gel processing and optical nonlinearity of Au microcrystal-doped oxide thin films", Proc.XVII International Congress on Glass, Beijing, Vol.1 (1995) 252 . 52. Haiping Xia, Congshan zhu, and Fuxi Gan, Study on permeation of fullerenes into porous glass, in Proceedings on XVII International Congress on Glass, edited by Gan Fuxi, Beijing: Chinese Ceramic Society, 1995, 3:668. 53. H.P. Xia, C.S. Zhu, F.X. Gan, Chinese Journal of Laser, B4 (1995) 247. 54. Haiping Xia, Congshan Zhu, Fuxi Gan, et al. Fullerene Sci&Tech. 5 (1997) 16211626. 55. Z.J.Liang, F.X. Gan, Z.R.Sun et al, Opt. Mater. 14 (2000) 13. 56. Y.Z. Gu, Z.J Liang, F.X. Gan, Opt.Mater. 17 (2001) 471-475. 57. Haiping Xia, Masayuki Nogami, Tomokatsu Hayakawa, Dai Imaizumi, J.Mater. Sci.Lett. 18(1999)1837-1839. 58. Baolong Yu, Congshan Zhu, Fuxi Gan, Yabin Huang, Opt. Mater. 7 (1997) 103-107. 59. Baolong Yu, Yuzong Gu, Yanli Mao, Congshan Zhu, Fuxi Gan, J. Nonlinear Opt.Phys&Mater. 9 (2000) 117-125.

Chapter 5

Second-Order Optical Nonlinear Properties of Glasses

5.1 Introduction Silica-based glass is one of the important optoelectronic and photonic materials. It exhibits extremely low optical loss over broad transmission band from the ultraviolet to the near infrared; it has high optical damage threshold; and it is compatible with the current fiber technology. So silica-based glass has very important applications as optical waveguides and other optoelectronic devices.11"61 Generally speaking, there should be no second-order optical nonlinearity in glass because of its inversion symmetry. This has usually brought the glass material only to the passive usages like fibers in photonics, while the second-order optical nonlinearity is the property that absolutely required for active applications in photonics. In 1986, however, Osterberg and Margulis discovered an efficient second harmonic generation (SHG) in silica-based optical fiber by irradiation with a laser light. Its second harmonic conversion efficiency was as high as 5%.[7] Later, in 1991, Myers et al. found that a permanent and large second-order optical nonlinearity could be generated in silica glass by thermal poling method.'13 A second-order nonlinear susceptibility x(2) as large as 1 pm/V was induced in bulk silica glass. The large second-order optical nonlinearity induced in fibers and glasses is of great interest since it opens new frontiers in optical material research. Since then, a large number of investigations have been done to search for novel glass materials and optimum poling conditions for the highest 153

154

Second-Order Optical Nonlinear Properties of Glasses

nonlinearity, and to make practical applications feasible. The second-order optical nonlinearity in glass materials could have potential applications such as parametric frequency converters for light source of shorter wavelength, and linear electro-optic modulators or optical switches that could monolithically integrated into optic fibers or planar glass waveguides.[3'4] In addition, the elucidation of the physical mechanism leading to the breakage of center symmetry in glass would be of considerable fundamental importance. 5.1.1 Inducing second-order optical nonlinearity in glasses Basically, a polarization P is induced in a medium by external optical electric field E with frequency co following: P = £oC/vE +f)EE

+ tf3)EEE +...),

(5.1)

!)

where s0 is the vacuum permittivity, rf is the linear susceptibility which accounts for the linear optical index, %(2> and x(3) represent the secondorder nonlinear susceptibility and third-order nonlinear susceptibility, respectively. However, the second-order nonlinear susceptibility %(2) is zero for conventional homogeneous glass because of macroscopic inversion symmetry. The centrosymmetry of glass can be broken via various poling techniques. Different methods of poling have been successfully used to induce permanent second-order optical nonlinearity in glasses.181 The methods include thermal poling,[1] corona poling,[2'9] electron beam implantation,'3' proton implantation,[10] UV poling[11'12] and photoinduced poling.[13] Among these methods, photo-induced poling was the first one used to induce second-order optical nonlinearity in glasses,17'14] and thermal poling is the most extensively used method to induce the nonlinearity in glass, because of its potentiality to induce stable and efficient nonlinearity in different kinds of glasses. 5.1.1.1 Photo-induced poling (optical poling) Photo-induced poling, or so-called optical poling, was first reported to induce second-order optical nonlinearity in Ge0 2 -Si0 2 glass fiber by

5.1 Introduction

155

Sasaki and Ohmori in 1981.[14] In 1986, Osterberg and Margulis discovered that an efficient SHG could be generated in a Ge-doped silica fiber of about lm in length by irradiation with 1.06um Nd:YAG laser light for a long time. The second harmonic conversion efficiency was as high as 5% after 12 hours irradiation.[7] A year later, Stolen and Tom showed that, when infrared fundamental light was introduced into a fiber together with its second harmonic light, the time to generate SHG was shortened to 5 min.[13] They believed that the incident fundamental and its second harmonic light beams caused a periodic dc electric field to build up in the glass and to permit periodic phase matching of SHG. Since that time, second-order optical nonlinearities were demonstrated using various kinds of glass fibers. Later, efficient SHG was also demonstrated to generate in bulk Ge-doped silica fiber preforms by optical poling.[15] Several other bulk glass systems, including commercial silicate and semiconductor-doped glasses, have been shown to exhibit the same effect.1161 Though optical poling has been used to induce second-order optical nonlinearity both in fiber and in bulk glass, the experiments on optical fibers gave best results. Since the induced nonlinearity is relatively small (%(2) ~ 10"4 pm/V), long interaction length is needed for high efficiency quasi-phase-matching SHG resulted from periodic electric field. 5.1.1.2 Thermal poling Thermal poling offers a simple and reproducible way to induce permanent second-order optical nonlinearity in glass. It is also called parallel-plate poling sometimes in order to distinguish with corona poling. The experimental setup is shown in Fig. 5.1. Basically the process of thermal poling of a glass is, heating the glass to high temperature of typically ~300°C while, at the same time, a strong external electric field ~107 V/m is applied across the sample. After a sufficient duration (about 30 minutes), the sample is cooled down to room temperature and the applied field is subsequently removed.tl] Thermal poling was first demonstrated by R. A. Myers et al. in 1991 to induce large second-order optical nonlinearity in bulk fused silica.[1] The induced nonlinearity, ~1 pm/V, which was evaluated from SHG

156

Second-Order Optical Nonlinear Properties of Glasses

measurement, is of similar magnitude to that from quartz. Since the nonlinearity was considerably large, the method has been applied widely to other oxide and chalcogenide glasses[17] and it can also be used for optical fiber poling.[18]

Power supply Silica glass plate 20mm in diameter 1.2mm in thickness

Fig. 5.1 Experimental setup for thermal poling of bulk silica glass. After Ref. [19].

5.1.1.3 Corona poling Corona poling is actually another method of thermal poling. It differs from parallel plate poling in the way that the electric field is applied. Instead of using plate electrode in thermal poling, a tungsten needle is placed about 1 cm above the grounded planar electrode and a voltage of ~5 kV is applied to the needle. Poling is carried out at temperature -200°C. The experimental setup[9] is shown in Fig. 5.2. Corona poling has been a technique commonly employed to orient the organic dye molecules in polymer films, which are generally called poled polymer. •HV

, Tungsten needle , Glass fUm Planar Electrode

Substrate

Fig. 5.2 Electrode configuration for corona poling. After Ref. [9].

5.1 Introduction

The first corona poling was et al.[2] With this method, they susceptibility x(2) ~ 1 pm/V in Corning7059 films on Pyrex glass

157

applied to glass films by A. Okata demonstrated second-order nonlinear radio-frequency sputtering deposited substrates.

5.1.1.4 Electron beam implantation Electron beam implantation was used to induce second-order optical nonlinearity in glass with charge implantation by exposure to a lowenergy electron beam. The first successful use of electron implantation was reported by Kanzansky et al. in lead silicate glass. The result showed X(2) ~0.7 pm/V.[3] In the experiment, a scanning electron microscope was used for electron-beam irradiation of the sample for 1 min, the beam currents used were in the range 0.3-10 nA, and the beam voltage ranged between 5 and 40 kV. The electron-beam spot size was 0.5 urn. The SH efficiency was found to increase with both the electron energy and the electron-beam current.[3] The injection of electrons into dielectrics can lead to the formation of a space-charge electrostatic field directed perpendicular to the surface of the sample. An advantage of electron beam implantation method is high resolution, which may be promising for fabricating optical waveguide or periodic structure for quasi-phase-matched SHG. In addition, proton implantation into silica glass has also been performed with minimum dosage of 0.25-0.5 mC and induced a x(2) of the order of 1.0 pm/V.[10] 5.1.1.5 UVpoling The method of UV poling was first realized by Fujiwara et al. in 1995.[11] In their pioneer work, Fujiwara et al. discovered that a particularly enhanced electro-optic effect could be induced in a Ge-doped silica fiber if the fiber was irradiated by nanosecond UV laser pulses at 193 nm in the presence of an applied electric field of >800 kV/cm.[11] The poling electric-field strength is at least one-order of magnitude higher than early reported works. Later, Fujiwara et al. reported the induction of a secondorder nonlinear coefficient d33 as high as 3.4 pm/V (i.e. x(2) = 6.8 pm/V)

158

Second-Order Optical Nonlinear Properties of Glasses

in highly Ge-doped (15.7 mol%) fused silica glass. for UV-poling is shown in Fig. 5.3.

The configuration

Fig. 5.3 Configuration for UV-poling in a Ge-doped silica glass. After Ref. [20].

Besides UV-poling, other auxiliary methods have also been used to enlarge second-order optical nonlinearity in silica glass, y ray and X ray irradiation[19] have been demonstrated earlier, irradiation of silica glass with infrared femtosecond laser also gave good results.[21] As thermal poling technique was used most frequently to induce nonlinearity in glass, the rest of this chapter will mainly emphasis on this technique. 5.1.2 Measurements of second-order optical nonlinearity in glasses Second harmonic generation measurement or the so-called Maker-fringe measurement^1 is the most commonly used technique to deduce the second-order nonlinear susceptibilities ^ of the poled glasses. Besides SHG measurement, interferometric measurement of the linear electrooptic (EO) effect is also widely used for probing second-order optical nonlinearity of glasses or fibers.[22] 5.1.2.1 Maker-fringe measurements The Maker fringe method has been widely used to determine the intensity and effective thickness of second-order optical nonlinearity of bulk materials.'1' 23] In the SHG experiment, as shown in Fig. 5.4, fundamental laser beam is incident onto the glass sample at different

159

5.1 Introduction

angle of incidence; the generated SHG signal is measured, which is proportional to the product (x(2)L)2, in which L is the material thickness contributed to the nonlinearity, i.e., the nonlinear region. The fundamental light used is usually the 1.064 um laser beam from a Nd:YAG laser operating at 10 Hz at an intensity of about 10 MW/cm2. By comparing the second harmonic signal of the glass sample with the signal generated by the same fundamental laser beam on a known nonlinear material, e.g. quartz, the second-order nonlinear susceptibilities "(2) of the glasses can be deduced. Ml

gMm,

Nd:YAG

made locked

Glam prism

yt

LI

L2

Rotatable Mage

pj| t e r

S 3 [| Q -fr

PMT

Hi!

Diode Boxcar

2IX Computer

HV

Fig. 5.4 Schematic of SHG measurement setup. [24].

To generate the Maker-fringe pattern, sample is rotated around an axis perpendicular to the laser beam. The resulting Maker-fringe pattern is the result of a phase mismatch Ak between the fundamental and the second harmonic waves: Ak =

{A,nlX){nacos0J-n2a)cos02o)'),

(5.2)

where na and n2m are the refractive index at fundamental and second harmonic wavelength, and 0m' and Q2m' are the angles of refraction inside the glass for the fundamental and the second harmonic waves, respectively, X is the wavelength of the fundamental beam. To get a clear fringe in Maker-fringe measurement, the thickness of second-order nonlinear region must be larger than the coherent length lc=n/Ak. In this bulk case, from the interval of the fringe, the thickness of nonlinear region L could be deduced. If only a thin layer of second-order optical nonlinearity is induced in the glass, i.e., a surface case with

160

Second-Order Optical Nonlinear Properties of Glasses

thickness less than the coherent length of silica glass (~24um), angular dependence of SHG measurement gives an envelope. Both results have been observed in poled silica glasses as shown in Fig. 5.5.[1'23] In most case, the later one was observed.

-60 -40 -20 0 20 40 Angle of Incidence / degre«

0 20 40 Angle of Incidence (degrees)

60

Fig. 5.5 Two examples of Maker-fringe patterns for fused silica: a clear fringe (left)[23) and an envelope (right)1'1.

The induced second-order optical nonlinearity in glass, both in bulk and surface case, is usually not homogenous over the nonlinear region. The second harmonic signal is thus described theoretically by integrating the second-order nonlinear susceptibility profile of sample x(2) (z) along z: I la, = A0)\ txi2)(z)exp[jAk(0)z]dz2,

(5.3)

where x(2)(z) is the second-order nonlinear susceptibility profile, z is the distance into the sample normal to its surface, 9 is angle of incidence, Ak is phase mismatch, A(9) is a function related to Fresnel function. In-order to obtain nonlinear profile and thickness of nonlinear region, especially in the surface case, other methods must be used together with SHG measurement, which will be shown in section 5.1.3. 5.1.2.2 EO measurements The linear electro-optic effect is also proportional to the second-order nonlinear susceptibility and offers a further potential for applications in modulators and optical switches. EO measurement gives the electro-optic

5.1 Introduction

161

coefficient y of the poled glasses, which is related directly to secondorder nonlinear susceptibility/^ via y=2rf2>/n4.[22] 5.1.3 Characterizations of the depth profile of the poled region Determining the second-order optical nonlinearity profile of poled glass is important, as the knowledge of this profile is essential to understand the physical origin of the nonlinearity, and it might also assist in developing methods to improve the strength and the uniformity of the nonlinear region. In thermal poling process, a dc electric field is recorded in the sample, which arises from a space-charge distribution in the depletion layer near the anodic surface.[1] Since the second-order optical nonlinearity is created near this layer, several studies have also been carried out to characterize this electric field and charge distribution. Different methods have been used to characterize the nonlinear profile or electric field distribution in poled glass. Apart from Maker fringe analysis, chemical etching of the poled region followed by either optical interferometric measurement^51 or atomic force microscope imaging1261 was used, as well as laser induced pressure pulse method[27] and optical second harmonic microscopy.[4] 5.1.3.1 Chemical etching method Chemical etching with hydrofluoric acid (HF) was first used by Myers et al.[1] to obtain the thickness of poled region. Later it was found that the built-in electric field distribution can be obtained by measuring the HF etching rate, and the variance in the etching rate has been suggested to be an indication of the onset of the frozen-in electric field.[25] By etching samples transverse to the poling direction together with AFM imaging, T. G. Alley and S. R. J. Brueck gave spatially resolved visualization of the space charge region directly, as showed in Fig. 5.6.[26] Interferometric measurement in real time could be used to deduce the HF removed thickness of silica with submicrometer precision.[25] It was used to determine the depth of poling region as a function of poling voltage, an average electric field Edc of 5,3x108 V/m was thus obtained.[28]

162

Second-Order Optical Nonlinear Properties of Glasses

Unlike SHG giving direct evidence of second-order optical nonlinearity profile of poled glass, HF etching alone can only give information of electric field or charge distribution in glass. An improved method to fully reconstruct the nonlinear profile with an accuracy of ~50 nm was subsequently proposed. It consists in HF etching the anodic surface of a poled sample, while the SHG intensity is recorded. The total nonlinear distribution is obtained after a step-by-step reconstruction.1291

Fig. 5.6 AFM image of a poled silica glass etched transversely to the poling direction, which reveals two ridges for regions of slower etch rates near anode. After Ref. [26],

The drawback of HF etching method is that it cannot distinguish changes in the etching rate from changes in the electric field and charge densities, which are both present within the nonlinear region. It also destroys the nonlinear layer during etching, therefore the measured nonlinear profile may not represent the original profile well. 5.1.3.2 Laser induced pressure pulse method Laser induced pressure pulse (LIPP) method was used to determine the space-charge distribution in poled glass. During LIPP measurements, a pressure pulse is produced by the impact of a short duration high power laser pulse on an absorbing layer deposited on the sample surface. This pressure pulse propagates inside the sample and gives rise to an electric

5.1 Introduction

163

current. The measurement of the current allows one to infer the sign and amplitude of the charges and their location.[27] Using LIPP method, Kazansky et al. have been able to accurately localize the electric charges responsible for the nonlinearity with a spatial resolution of ~3 um.[27] They have found that in a thermally poled silica glass, a 10 um thick space charged region consists of a positive layer at anodic surface and a negative one deeper in the poled sample. Later, they used LIPP method to study the difference of charge distribution in silica glasses poled in air and under vacuum.130' 5.1.3.3 Improved Maker fringe method When only a thin layer (L< 4) of second-order optical nonlinearity is induced in glasses, it is usually impossible to get second-order nonlinear susceptibility x(2) and the depth of the nonlinear region L separately from a simple Maker-fringe measurement, nor the nonlinear profile could be retrieved uniquely. Several improved SHG methods are issued, such as using prism,[31] glass hemispheres,132' or cylinder[33] to overcome the problem of total internal reflection at the rear glass-air interface, making L longer than lc at large angles. Another variant of Maker-fringe technique employed two fundamental beams at an angle to produce a non-collinear SH beam, which has much smaller coherent length (~2um in silica).[34] In both cases, clear fringe instead of envelope might be obtained, which was used for deducing the depth of nonlinear region. Later, Ozcan et al. showed that the Fourier Transform phase o$x(2)(z) could be retrieved by measuring the Maker-fringe generated by the interference of two samples.[35] This phase was used to get the unique recovery of^(z), which agreed well with results of Ref. [27]. 5.1.3.4 Second harmonic microscope Second harmonic imaging of poled silica glass was first performed by Kazansky et al.[4] The extent and distribution of nonlinear region have been investigated by SHG measurement when the incident laser beam was scanned transversely to the sample's poling direction. The result showed that SH spot was localized 12 um below the anodic surface, and

164

Second-Order Optical Nonlinear Properties of Glasses

the spread of the depth was ~ 7 urn. Later, SH microscopy has been used for investigating the nonlinear profile in thermally poled silica glass,[36] planar waveguides[37] and optical fibers[38] with sub-micron spatial resolution. Compared with other methods, the SH microscopy can provide direct visualization of the induced nonlinear profile no matter what mechanism is responsible. 5.1.4 Mechanism Many investigations have been devoted to the determination of the microscopic mechanism underlying the formation of second-order optical nonlinearity in silica glass.[8] Cations migration is usually assumed, and the occurrence of poling-induced charge (mainly Na+) migration has been reported.126'39_42] In most case, mobile alkali ions, such as Na+, Li+, or K+, migrate towards the cathode, leaving a depletion region of several micrometers thick at the anode. The negative charge centers are motionless in the silica matrix as compared to the positive charges. Due to the charge separation, a high permanent built-in electric field Edc as high as ~4xl0 8 V/m is frozen along the electric field direction (z axis) within the depleted region.[43] The rectification of third-order non-linear susceptibility x(3) by this built-in electric field induces an effective y2^ susceptibility as[1'13]: Xijk

=

^Xijkz^dc •

(5-4)

On the other hand, alignment of hyperpolarizable entities (such as bond, dipole, defect, or nano-crystallite) leads to the creation of a macroscopic %(2) susceptibility through relation'1'441:

where /? is the hyperpolarizability, p is the permanent dipole moment, N is the concentration of entities, k is Boltzmann's constant, and T is the temperature. Though dipole orientation model could not be excluded completely, a lot of experimental results supported the frozen-in electric field model.

5.1 Introduction

165

5.1.4.1 Charge migration model Charge migration models that predicts the electric field distribution have been given in order to understand the mechanism for x(2) creation.'44"461 The contribution of the built-in electric field to the x(2> susceptibility can be deduced from an understanding of how the charge concentration evolves with time and space during the poling process.[8] The evolution of the local charge concentration with time can be described by a set of partial derivative equations that involve diffusion, conduction and charge trapping and de-trapping. An analytical expression for the thickness w of the poled region' 40 ' 41 ' 45] has been obtained for the case of one mobile positive charge carrier: w=l—(V0-VT), Veco

(5.6)

where V0 is the applied potential, c0 is the initial charge concentration and e the elementary charge of electron. VT is a threshold voltage[45] that depends on the chemical glass composition, the poling temperature, the attraction energy between the mobile cation and the fixed anion (The VT was measured to be 900 V in Infrasil sample[45]). So the thickness of the cation depletion region depends mainly on the initial concentration of cations and the applied voltage. Neglect charge injection, values of w deduced from Eq. (5.6) are in good agreement with experimental ,„ [1,45-46]

ones. The time dependence of thermal poling and the values of the nonlinear thickness obtained in experiment[43] indicate that the poling process takes place in two stages: the rapid formation, within tens of seconds, of negatively charged region depleted with cations (Na+) followed by a slower process which is responsible for charge separation within the depletion region.141'47] To explain the experimental results, two charge carrier model was proposed and solved.[46] It has proposed that during thermal poling in air, If1" or H 3 0 + are driven by the high electric field at the anode into the region depleted by Na+, thus neutralizing the negative charges left behind by Na+. At the same time, the region depleted with Na+ increases and this ion exchange process results in the forward movement of the negative depleted region.[41] When using a

166

Second-Order Optical Nonlinear Properties of Glasses

non-blocking electrode, for increasing time, charge injection can no longer be neglected. In this case, the nonlinear thickness increases, thus leading to a decrease of the electric field, hence of /(2)- The poling stops when the electric field near the anode drops to values corresponding to which H+ /H30+ ions cannot be driven into the sample and when the total voltage drops across the depletion region, so that the movement of Na+ in the bulk stops. Fig. 5.7 shows one of the charge distributions in poled silica glass by LIPP measurements.[30]

Fig. 5.7 Charge distribution in the poled silica glass revealed by LIPP measurement. After Ref. [30].

5.1.4.2 Third-order nonlinear susceptibility modification In frozen-in electric field model, the second-order optical nonlinearity is shown to proportional to /3). %(i) varies from glass to glass, and is typically in the range (l-3)xl0" 22 m2/V2 in silica glass. Its value is generally assumed to be constant and unmodified after poling. In face, in some experiment, %(3) was shown to be the same for different samples poled with different time.[43] However, there are reports of third-order nonlinear susceptibility modification inside the nonlinear region after poling. Kashyap has reported that the x(3) susceptibility can be modified by a factor of 1.9 after poling in a Ge-doped silica waveguide.'481 The experiment in Ge-doped silica glass also showed an approximately 15 times enhancement of x(V due to the formation of crystallites after UVpoling.[12]

5.2 Second-Order Optical Nonlinearity in Silica Glasses

167

5.1.4.3 Structure characterizations Some studies have been performed in order to reveal the microscopic structure change in glass after poling. Here a few examples are given. Cabrillo et al. published neutron diffraction and inelastic neutron scattering results on poled silica glass.[49] They concluded that large microscopic alterations were found in silica glass after thermal poling. These resulted in the breakdown of isotropy which involved at least next nearest Si neighbors. The effectively evidenced an elongation of the shortest Si-Si distance from 3.06 to 3.23 A in this direction was attributed to the opening of Si-O-Si bond angles in the same direction. In Ref. [50], using X-ray absorption near-edge structure (XANES) and FTIR spectroscopy, Nazabal et al. have shown that Si-O-Si bridging bonds have been broken in the non-linear region during poling. At the same time, a decrease of hydroxyl Si-OH bonds associated with an increase of nonbridging oxygen (Si-CT) defect center was interpreted by a proton conduction during poling.[41] 5.2 Second-Order Optical Nonlinearity in Silica Glasses Second-order optical nonlinearity induced in silica is widely studied since the first report by Myers in 1991. Charge migration models are issued trying to elucidate the mechanism of second-order optical nonlinearity in silica glasses. Here we will give some results on silica glasses, which mainly come from our own work. Other results on silicate glasses are also briefly reviewed. 5.2.1 The influence of poling condition on the optical nonlinearity Poling condition must be optimized to obtain large second-order optical nonlinearity in poled glasses. We have studied the influence of poling condition on the second-order optical nonlinearity induced in silica glass.[51] The sample used in the experiments was commercial JGS] fused silica. During poling process, the sample was heated to the temperature of 150-450 °C, and a high voltage varied from 1.0 to 8.0 KV was applied across the sample plates for 0.5 to 6 hours.

168

Second-Order Optical Nonlinear Properties of Glasses

The SHG signal was obtained in poled silica glass by the Makerfringe method. In our experiment, a clear fringe pattern instead of an envelope was observed for the poled silica glasses under all poling conditions. Fig. 5.8(a) shows the dependence of SHG intensity on the poling temperature after the samples were poled with 7.5 KV for 1 hour. From 150°C onwards, the SH signal could be observed and the signal reached a maximum at 250°C and then quickly reduced to zero. The SHG intensity dependence on the poling voltage is shown in Fig. 5.8(b). When the voltage increases, the SHG intensity increases monotonically. The SHG intensity dependence on the poling time is given in Fig. 5.8(c). The samples were poled at 250°C and 7.5kV. This figure shows the SHG intensity increase as the poling time increases and approaches saturation after 3 hours. These experimental results revealed that there existed a optimum poling condition for generating a large second-order optical nonlinearity in silica glass.

Polna Temperature <°C)

POHCIQ Voltage (KV)

Poling Time (h)

Fig. 5.8 SHG intensity vs. poling temperature (a), poling voltage (b) and poling time (c) for silica glasses.

5.2 Second-Order Optical Nonlinearity in Silica Glasses

169

5.2.2 Inducing second-order optical nonlinearity in silica glasses by different poling method Different results of Maker-fringe, including both clear fringe[23] and envelope/11 have been reported by different groups in silica glasses. Henry et al.[52] also reported that different types of Maker fringe patterns could be observed in a same type of bulk silica glass after different processing. In some of their pre-annealed samples, multiple fringes emerged, while in as-received samples only a single envelope appeared. They claimed that the pre-annealing process removed the near-surface layer, but no detailed analysis was given. In our work,[24] we found that different types of Maker fringe patterns could be observed in the same type of un-preprocessed (i.e. as-received) silica glasses poled using different methods. The samples we used were JGS1 glasses with thickness of 1 mm. They were poled with two commonly used thermal poling methods: parallel plate poling (thermal poling) and corona poling. After poling, the Maker fringe was recorded by measuring the SHG signal at different angle of incidence. Fig. 5.9(a) shows a typical Maker fringe of a sample after parallel plate poling. The poling conditions were r=300°C and F=4.8 kV. Fig. 5.9(b) shows a typical Maker fringe of a corona poled sample (7=300 °C, V=12 kV). Note the peak intensity of curve (a) is 18 times larger than that of curve (b). 40

20

60

80

• ^ ?

A

0

20

",

fa

/ V v y 1/ 40

60

80

Angle of incidence (degree) Fig. 5.9. Typical Maker fringes of fused silica glasses: (a) plate-poled sample; and (b) corona-poled sample. The peak intensity of curve (a) is 18 times larger than that of curve (b), as written in the plot.

170

Second-Order Optical Nonlinear Properties of Glasses

According to the description given by Myers et al.[1], the formation of a depletion layer is the main cause of the second-order optical nonlinearity, which resulted in a nonlinear layer near the anode surface of the sample. From the model, two conditions must be satisfied in-order to build up a frozen-in field. One is that there must exist movable cations such as Na+ in the glass, the other is that no other movable ions enter the depletion layer after the layer is formed. It is clear that during corona poling, the second condition will no longer be satisfied. The continuous injection of positive ions from the air near the anode surface may destroy the formation of the depletion layer, and thus no high frozen-in electric field can be built up near the anode. Therefore, corona poling results in a bulk poling effect. 1.6 1.2 -

50

100

150

200

250

300

1.2

iMf, jitott poted sample b.fe': ooroia poled sample

W^%^**-?°° o
0.8

*\*Vb' ^ P (9

0.4 ,-v 3

0.8

0.0

"""a 0.4

J2

"g

- -0.4

0.0

'Tnt -0.4

50

100

150

200

250

300

-0.8

Temperature ( C) Fig. 5.10. OCTSD spectra of poled silica glasses: (a) plate-poled sample; and (b) coronapoled sample. The peak current at about 160°C of curve a is 2.5 pA and that of curve b is 4.6 pA. a', b' are the SHG decay of samples a and b.

It can easily be noticed that the second-order optical nonlinearity originated by a frozen-in field is much larger than that originated from bulk poling. The result in Fig. 5.9 shows that the SHG intensity of the plate-poled sample is at least one order larger than that of the coronapoled one. This is easy to understand since the bulk poling field is much weaker than that of thin layer poling.

5.2 Second-Order Optical Nonlinearity in Silica Glasses

171

An open circuit thermal stimulated discharge (OCTSD) has been used to show the bound charge distribution difference between samples poled by two different methods. The result in Fig. 5.10 clearly showed there exist large differences between the bound charge distribution of the sample poled by different method.[24] 5.2.3 Ge-doped silica glass Using UV-poling, Fujiwara et al. reported the induction of a secondorder nonlinear coefficient d33 as high as 3.4 pm/V in highly Ge-doped (15.7 mol %) fused silica glass.[20] In the experiment, they found that both d33 and d31 increase with increasing poling field up to about 150 kV/cm. The value of d33 is exactly three times as large as that of d3j, which should support the frozen-in electric field model.[20] However, the induced SHG in the sample decays after UV poling. The decay follows single-exponential function with an activation energy of 0.41 eV, exactly the same as that of the GeE' center. By extrapolating, the decay timeconstant at 15°C is approximately 280 days. This decay time could be prolonged to 18 years by the introduction of electron scavengers, such as hydrogen. In addition, crystallites of about 15-20 urn in diameter were also found in Ge-doped silicate glass, which leads to an increase in thirdorder optical nonlinearities approximately 15 times larger. These results suggested that the major origin of the second-order optical nonlinearity induced in Ge-doped silicate glass is a combined effect of a large third-order nonlinearity associated with the crystallites and an internal space-charge field, where the charges to build up the field are produced during the formation of Ge£" centers.1121 UV-poling was also applied to a sputtered GeC>2-Si02 film. Preheating the film in vacuum prior to UV-poling was found to further increase the second-order optical nonlinearity of the film. The value of second-order nonlinear susceptibility was obtained to be as large as 12.5 pm/V, which is the largest ever reported for poled glasses.[53] 5.2.4 Soft glasses The method for poling soft glass is a little different from that of silica

172

Second-Order Optical Nonlinear Properties of Glasses

glass. A step-like increase in the poling voltage must be used to limit the current flow during poling. The second-order optical nonlinearity has been induced in soft glass,[54] but it decays spontaneously after poling. It is also found possible to fabricate waveguides in soft glass by thermal poling, exploiting the change in refractive index caused by ion depletion.155 ] 5.3 Second-Order Optical Nonlinearity in High Refractive Index Glasses The frozen-field model given in section 5.1.4 predicts that x(2) should increase proportionally to /3\ for a given value of Edc. Enhancement of X(2) in proportion to x(3) is therefore expected in thermally poled glasses with higher /3) than silica. High refractive index glasses are potentially interesting for this purpose since they exhibit intrinsically high x(3)- Here we will give some of experimental results on high refractive index glasses. 5.3.1 Lead borate glass 5.3.1.1 Non-uniform bulk second-order optical nonlinearity distribution In most of the reported experiments on second-order optical nonlinearity in fused silica, the nonlinearity was regarded as coming from a thin layer near the anode surface.11'39] However, according to Le Calvez et al., the nonlinearity was also induced in the cathode area.[56] Furthermore, some groups reported clear Maker fringe observations on poled fused silica and other glasses.15'52] Although it was concluded that the second-order optical nonlinearity of these poled glasses may come from a bulk effect, the interference of the second harmonic signals generated from the two surfaces of a sample might result in a fringe pattern as well. In our paper, [57] a detailed Maker fringe investigation was reported on thermally poled PbO/B203 bulk glass. Our experimental results showed that the Maker fringes come from a bulk second-order optical nonlinearity. In particular, we showed that this bulk effect is strongly influenced by the negative charge layer frozen in the anode area. A theoretical model based on a

5.3 Second-Order Optical Nonlinearity in High Refractive Index Glasses

173

non-uniform second-order nonlinear coefficient distribution was proposed and the experimental results are explained. We used 0.43PbO:B2O3 glasses in the experiment. The samples were made through a standard melt-quenching method. The sample was thermally poled with a voltage of 5.5 KV at temperature of 400°C for 30 min. •

783 urn 634 fim —

7 M t*m calculated

100 90 80 70-

~. «o: 4 80^ -«' 40-

f\ki% A A R * tl

o

J *>: 20-

.^SL^Z'

to0-

,

^\~_

vvvv^^^

^J$Jv n.r-*^^^^ 20

fa)

i \ A A «,

40

WO

80

60

Incident angle (degree) » largest mmmna {cathode grind) B maximum (calhodr frind) O If rgesi Minimum (ftMdtgriad) a rruiimum (anode grind) -•—tartest minimum
grinding direction

SO

I

•A

40

a 20

200

(b)

400

600

600

1000

Thickness (jim)

Fig. 5.11 (a) Maker fringe patterns of a thermally poled PbO/B 2 0 3 glass sample at three different thickness in the process of grinding. The solid line shows the theoretical calculation according to the model proposed in this letter, (b) The maximum and largest minimum of the Maker fringes of the sample vs its thickness.

174

Second-Order Optical Nonlinear Properties of Glasses

After initial SHG measurement, the cathode side of the sample was ground to remove a thin surface layer, polished, and then the Maker fringes were measured again. This process was repeated and after about 200 um of the thickness was ground away, the anode side of the sample was similarly processed. The SHG measurement results during the grinding process are shown in Fig. 5.11, where Fig. 5.11(a) gives the typical Maker fringe patterns of the same sample at three different thickness, and Fig. 5.11(b) plots the maximum and the largest minimum of the Maker fringes of a sample versus its thickness. From the experimental results we found that with a reduction of the sample thickness, the interval between fringes increased, and even after more than one third of the thickness of the sample, i.e., about 300 um, was ground away, Maker fringes could still be observed. It indicates that the nonlinearity of the sample has a bulk distribution. When the cathode area of the sample was ground away, the SHG intensity kept roughly at a constant value, as shown in Fig. 5.11(b), however, when the anode area of the sample was ground away, the SHG intensity decreased rapidly. It is clear that the anode area strongly influences the bulk nonlinear distribution in the sample. To explain the experiment results, an anode-charge-layer dependent bulk electric field distribution model is proposed as follows. In the process of thermal poling, a carrier-deficient negative charge layer was formed near the anode area in the sample, as shown in Fig. 5.12(a), and positive charge would also accumulate in the sample near the anode surface according to the multiple-carrier model'411 or the highfield-induced discharge model.'39' The appearance of charge on both of the electrodes was due to the charge balance. The electric field distribution in the sample is shown as the solid line in Fig. 5.12(b). After thermal poling, when the poling voltage was removed, a screening process took place and the electric field distribution in the sample was like that shown by the dashed line in Fig. 5.12(b). Since the electric field in the sample was mainly determined by the negative charge layer frozen in the area near the anode, the electric field intensity will not decrease if the cathode area is ground away. However, when the anode area of the sample is ground away, the negative charge layer frozen in the sample became thinner, and the electric field intensity decreased due to a new

5.3 Second-Order Optical Nonlinearity in High Refractive Index Glasses

175

but weaker charge balance. According to either the dipole orientation model or the frozen-in electric field model, the second-order optical nonlinearity of a local area of the sample is proportional to the local electric field. Therefore, if the distribution of the electric field in the sample is known, the distribution of the nonlinearity in the sample can be obtained. anode

senile

cathode

Fig. 5.12 Theoretical model of thermal poling, (a) Charge distribution in the electrodes and in the PbO/B 2 0 3 glass while poling, (b) Electric field distribution during poling (solid line) and after poling (dashed line), (c) Electric field distribution in fused silica glass during poling, (d) Simplified electric field distribution used for theoretical calculation.

To simplify our calculation, we assumed that the electric field distribution in a poled sample is like that shown in Fig. 5.12(d). The relative intensity .Eanode and kathode are decided by the frozen-in negative charge layer. .Eanode /^cathode was assumed to have a fixed value, therefore, the effective coefficient defrfz) had a similar distribution, with

176

Second-Order Optical Nonlinear Properties of Glasses

detf(z=Oydefrfz=L) being a constant. The solid line curves in Fig. 5.11(a) are a theoretical Maker fringe pattern derived according to our model. The triangle and solid line in Fig. 5.11(b) show the calculated results of the sample at different thicknesses with the cathode side and/or the anode side ground away, as was done in the experiment. It can be found that the theoretically predicted SHG intensity with the sample ground to different thickness is in consistent with the experimental results. 5.3.1.2 Different poling condition Lead borate glasses have also been poled with different voltage and at different temperature, trying to find the optimum poling condition.[58] Fig. 5.13 shows the variation of SHG intensity with the poling temperature for a 0.43 PbO/B 2 0 3 glass. 140-

*

120•a? I O C 'S

/

.g BO'S.

Ica *>-. £ iocs i 55 20o200

250

M0

350

400

450

TfC) Fig. 5.13 Variation of SHG intensity with poling temperature.

From the figure, we can find that the SHG intensity reaches a maximal value at the poling temperature of 400 °C. This temperature is called optimal poling temperature, it was found to be 50 °C lower than the glass transition temperature. The relation between SHG intensity and poling voltage has also been found to obey a power relation ISHG = V , with t = 2.4. The SHG measurement results for the lead borate glasses poled at different poling voltage are given in Fig. 5.14.

5.3 Second-Order Optical Nonlinearity in High Refractive Index Glasses

111

We also found that different thermal poling scheme leaded to different nonlinearity distribution in PbO/B 2 0 3 glasses.[59] By simulating the movement of carriers with a multiple carrier model,[41'46] the distribution of the frozen electric field was obtained, which is supposed to be the same as the distribution of second-order optical nonlinearity. The glass samples used were 0. 43 PbO:B 2 0 3 with a thickness of 650 urn. During the process of thermal poling, the glass sample was placed between two electrodes that were in physical contact with the sample. The sample was poled with 4 kV voltage at 400 °C for 30 min. 5.5KV,1=230L=760, 240,0.21 4.0KV,1=230 L=760, 330,0.21 2.5KV,1=220 L=780, 560,0.21 - caculated

i-r

Incident Angle(degree)

Fig. 5.14 SHG intensity of lead borate glasses poled at different voltage.

Poling time(m in) Fig. 5.15 Current during the poling, (a) The open dots show the current during poling with uncovered electrodes. The solid line is the simulation result, (b) The filled dots show the current during poling with foil-covered electrodes. The dotted line is the result of simulation.

178

Second-Order Optical Nonlinear Properties of Glasses

Two kinds of electrodes were used in poling: uncovered stainless steel plate electrodes and stainless steel plates covered by soft aluminium foils. The soft foils were used to provide better contact and prevent ion injection from the atmosphere.[41J The foil on the anode side was found to be sticking to the glass after poling and could be removed. The Maker fringe method was used to measure the second-order optical nonlinearity of the poled glasses. Typical currents during poling are shown in Fig. 5.15. We can see that the initial current from poling with uncovered electrodes was smaller than that from poling with foil-covered electrodes. This may be caused by different contact area from the different electrodes. When the aluminium foil was used, the contact between electrodes and sample was better, and the contact area was larger, and the initial current was consequently higher. The current decayed during both types of poling, however, the current decayed more rapidly during poling with the foil-covered electrodes. Maker fringe patterns of a typical 650um-thick PbO/B 2 0 3 glass sample are shown in Fig. 5.16.

c .a

is.

"to c v>
100

I

CO

Angle of incidence(degree) Fig. 5.16 Maker fringes of a 650um thick lead borate sample (solid lines) and the results of theoretical fitting (dotted lines), (a) The sample was poled with uncovered stainless steel plate electrodes; (b) the sample was poled with foil-covered electrodes; (c) the sample was poled with foil-covered electrodes, and then a 15um-thick region on the anodic surface was ground off.

5.3 Second-Order Optical Nonlinearity in High Refractive Index Glasses

179

Fig. 5.16(a) shows the Maker fringe of the sample poled with uncovered stainless steel plate electrodes. The multiple peaks indicate a bulk nonlinearity. Fig. 5.16(b) shows the Maker fringe pattern of the sample poled with foil-covered electrodes. The single peak implies a near surface nonlinearity.[1,24'27] It is shown in Fig. 5.16(c) that, after grinding away a 15um-thick region on the anodic surface of the sample, we observed multiple peaks (bulk nonlinearity) again. The resultant SHG intensity was much weaker (400 times smaller); therefore we know that bulk nonlinearity could be generated in lead borate glass regardless of the types of electrodes used during poling. The near surface nonlinearity was induced only when foil-covered electrodes were used in poling. Note that the SHG intensity of Fig. 5.16(b) is about 120 times larger than in Fig. 5.16(a), which means that the near surface nonlinearity is much larger than the bulk nonlinearity. The relationship between SHG intensity and voltage has been studied when poling lead borate glasses with uncovered electrodes; only bulk nonlinearity was found.[58'60] So the different second-order nonlinear effects shown in Fig. 5.16 could not be explained by the slight difference of effective voltage during poling with different electrodes. Here we propose another model. There are two kinds of possible carriers in lead borate glasses: lead ions (Pb2+) and hydrogen (H+ or H 3 0 + , denoted as H+ below). H+ exists in PbO/B203 glasses because dissociation of water occurs during sample preparation. The hydrogen concentration in our samples is estimated to be -2000 ppm according to the OH* absorption around 3500 cm"'. The total concentration of Pb2+ in our samples is about 30%. Lead ions cannot move altogether at the same time, otherwise the Pb2+ depletion layer would be very thin and no bulk nonlinearity can be generated. We should also consider the transition between mobile lead ions and motionless lead ions. During poling, motionless lead ions can dissociate from negative charge sites (non-bridging oxygens on borate triangle units or borate tetrahedra) and become mobile. At the same time, mobile lead ions can recombine with negative charge sites and become motionless.'451 We simulated the poling processes with a multiple carrier model including Pb2+ions, H+ ions and motionless negative charge sites. During poling, the changes of ion concentrations and electric field can be

180

Second-Order Optical Nonlinear Properties of Glasses

described by [45] : ??fL-n

^ft-u

et ~

dx

Dpb

dn

H

n

2

Mn

^E'nn)

dx

d2nH

Ia a

+q

d(E-nH)

-^^"-^r-^—^x—' % " =%-«), ot q = K(C0 - nN 12), a = a-nN-nPb,

a

'

(5 7)

K

}

(5 8)

'

(5-9) (5.10) (5.11)

dE e — = -(2nPb+nH-nN), (5.12) ox e where t is time, x is the depth from the anodic surface. nPb, nH and nN are the concentrations of mobile Pb2+, H+ ions and motionless negative charge sites respectively. DPb and DH are the diffusion coefficients of Pb2+ and H4". juPb and fiH are the mobility of Pb2+ and it, respectively. E is the electric field strength. Q and a are the number of dissociated and recombined Pb2+ ions per unit of volume per unit of time, respectively. K and a are the dissociation and recombination coefficients respectively. Co is the total concentration of Pb2+. e is the electronic charge, e is the permittivity of glass. For simplicity, we also make the following assumptions in our simulation: The electric field in glass is limited by high field effect. We assume that the electric field cannot exceed 108 V/m, which is of the order of the breakdown strength. When poling with uncovered electrodes, we assume the ion injection just keeps the anodic surface neutral, which means no excess H+ enters the glass. When poling with foil-covered electrodes, we assume ion injection from atmosphere is completely blocked by the foils. The parameters used in the simulation are listed in Table 5.1. The results of the simulation are shown in Figs. 5.17 and 5.18. Fig. 5.17 shows the simulated carrier distribution when the glass is poled with uncovered electrodes. We can see that a Pb2+ deficient layer grows during poling. Its thickness reaches -1/3 of the sample thickness

5.3 Second-Order Optical Nonlinearity in High Refractive Index Glasses

181

after 30 minutes of poling. Some H+ ions accumulate in this layer to balance the negative charges generated by charge dissociation. After the high voltage is removed, charges will accumulate on the sample surfaces due to charge balance,[57] and a negative bulk electric field will be generated to balance the electric field in the Pb2+ deficient layer. 3-

!

'

a

=----. b

•

o

c

-

0

, ~ 0.6.

a

b ...

J- 0.4. © C 0.2.

^

J

J3

a.

c

0.0-

" 0 50

J 48-

~-—

—

^

^

a 42100

150

depth(pm)

Fig. 5.17 Simulation results: evolution of the electric field, mobile Pb concentration and H+ concentration during poling with uncovered electrodes. Different lines in the same plot stand for different poling time: (a) 1 min; (b) 10 min; (c) 30 min; (d) the glass is poled for 30 min and then the high voltage is removed.

The total electric field in the sample (line d in Fig. 5.17) can be written approximately as the following: E *

E

-(E

a ^ Ec,lPb

a

Ec)-x/lpb,0< <x
x< lPb

(5.13)

where Ea and Ec are the electric field on the anodic surface and cathodic surface, respectively, x is the depth, lPb is the thickness of Pb2+ deficient layer. L is the thickness of sample. Fig. 5.18 shows the simulated results of the poling with foil-covered electrodes. In this case, both a thick Pb2+ deficient layer and a thin H+ depletion layer can be found. But the electric field in the H+ depletion layer is very large, so the electric field in the Pb2+ deficient layer and the electric field caused by surface charges can be neglected.

182

Second-Order Optical Nonlinear Properties of Glasses 12-

fio. >

Eno'v/mi

a bc

i i i

8

° 62-

0.3. 0.0-

d

^ ^ _ « i • i > i > i - i • i 100 MO 300 400 5 M 600

depth
_,

<<~0.6»

\ i 0

af —

-

C

b;---

0.4-

^0.2. 0. c 0.0n~* 60.

1 so. a " ' <0~ 40*~0 30^ 20-

[

«FlO0-

50

0

100

150

200

250

depth(nm)

Fig. 5.18 Simulation results: evolution of the electric field, mobile Pb2+ concentration and H+ concentration during poling with foil-covered electrodes. Different lines in the same plot stand for different time: (a) 1 min; (b) lOmin; (c) 30min; (d) the glass is poled for 30min, the high voltage is removed, and then the H+ depletion layer on anodic surface is ground away. Table 5.1 Parameters used in the simulation. nPb Upb Dp b

a Co

5.5xl017/cm3 IO-WVV1 2.9xlO"11cmV1 6.1xl0_23cm3/s 6.6xl0 2l /cm 3

4.4x1019/cm3 7xlO"12cm2V"V 4.1xlO"13cmV 2.3xl0_7/s 10e0

nH RH DH K 8

Then the electric field distribution can be written simply as the following: E

* lE°>° [0,lH

~

X

~ l" ,

(5.14)

< x < L

where Ea is the electric field on the anodic surface, and lH is the thickness of H+ depletion layer. After the H+ depletion layer is ground away, the electric field distribution (Fig. 5.18, line d) is similar to that in the sample poled with uncovered electrodes (Fig. 5.17 line d), it can then be represented by Eq. (5.13).

5.3 Second-Order Optical Nonlinearity in High Refractive Index Glasses

183

The second-order optical nonlinearity in glass is proportional to the frozen electric field.[39' "• 5 6 ] According to Eqs. (5.13) and (5.14), the distribution of second-order nonlinear susceptibility can be written as: r(2)

A bulk

„

[x^-ixf'-zf'Yxllp^xZln (2)

[X.

^

(515)

lPb<x
y (2) A, surface

,0 < x < l„

0,lH

(5.16)

< x < L

where x(2>buik is the nonlinear susceptibility of the bulk nonlinearity induced by the electric field in Pb2+ deficient layer and the bulk electric field caused by surface charges, x(2)surface is the second-order nonlinear susceptibility of the near surface nonlinearity caused by the high electric field in FT depletion layer, x(2)a and /2)c are the nonlinear susceptibilities on the anodic surface and cathodic surface, respectively. /«, is the thickness of Pb2+ deficient layer, lH is the thickness of YC depletion layer. The experimental Maker fringes in Fig. 5.16 can be well-fitted with Eqs. (5.15) and (5.16). The results are shown in Fig. 5.16 as dot lines. The best fitting parameters are shown in Table 5.2. Because of the simplicity of our model, the thickness of the H+ depletion layer used in the fitting with Eqs. (5.15) and (5.16) is a little smaller than that obtained by the simulation with Eqs. (5.7)-(5.12). Table 5.2 Results of the fitting of Maker fringes in Fig. 5.16.

X™ (pm/V) %f (pm/V) a b c

lPb(\im)

//f(nm)

0.07±0.01

-0.012±0.002

200±10

-

0.8±0.2

-

-

13±2

0.03±0.01

-0.01±0.005

250tl00

-

We also simulated the currents during poling. After normalizing to the initial current in the experiments, the simulation results are shown in Fig. 5.15. The current simulation is in excellent agreement with the experiment. When poling with the foil-covered electrodes, the current decayed more because the voltage was partly applied on the H+ depletion layer.

184

Second-Order Optical Nonlinear Properties of Glasses

5.3.1.3 Raman spectroscopy measurements The polarized Raman spectra[60] of the glass samples were measured using a Labram-1 microscopic Raman spectrometer. The poled glass plate was broken along the middle of its main face and the exposed surface was polished for Raman measurements. Fig. 5.19(a) depicts the W polarized Raman spectra before ( W B ) and after (VVAA) poling, where AA means the spectra were taken near the anode side after poling. Fig. 5.19(b) depicts the HH polarized Raman spectra before and after poling. Fig. 5.19(c) shows the comparison of the W polarized Raman spectra and the HH polarized Raman spectra after poling. At the same time, we also did a comparison of W polarized and HH polarized Raman spectra before poling and found that the VVB spectra were nearly the same as HHB spectra.

(a)

(b)

2.5 -

, S

-

i

0

—

•

—

i

WAA HHM

—

2G0

•

—

i

—

400

'

—

i

—

600

•

—

i

—

GOO

•

—

i

—

•

—

1000

i

—

'

1200

—

i

—

•

1400

—

i

—

•

1000

—

i

1000

Wavenumber(cnV') (c)

Fig. 5.19 ( a ) W polarized Raman spectra, (b)HH polarized Raman spectra, and (c) comparison between VV polarized and HH polarized Raman spectra after poling.

5.3 Second-Order Optical Nonlinearity in High Refractive Index Glasses

185

5.3.2 Tellurate glass The optical properties of tellurate have attracted attentions because of its high refractive indexes and large third-order nonlinear optical susceptibility as well as its wide wavelength range of large gain as an optical fiber amplifier. Tanaka et al. found that thermal poled tellurate (Te02-based) glasses generated SHG.t6'61"651 A second-order nonlinear susceptibility as large as 2.1 pm/V was obtained in thermally poled W0 3 -Te0 2 glasses, which was presumably brought about by large x(3) by means of frozen-in electric field model. [62] They have also studied SHG in poled tellurate glasses to obtain information about the atomic origin of the second-order optical nonlinearity and argued the mechanism to induce SH signal. [6365] In some doped ZnOTe0 2 glasses, the second-order optical nonlinearity extends into the inside of bulk glass.[63] It was speculated that the large frozen electric field caused orientation of Te0 4 and Te0 3 tellurate structural units with nonbridging oxygen atoms that possess permanent electric dipoles. It was also found that there was a linear relationship between the optimum poling temperature and the glass transition temperature.[63] This revealed that the orientation of tellurate structural units occurs as a result of the structural relaxation of tellurate network around the glass transition temperature. In a glass with large flexible glass network structure, the orientation can take place more readily. Also, the relaxation time for the decay of induced x(2) w a s estimated to be as long as 9 years at room temperature for poled 30ZnO70TeO2 glass.[64] Other reports showed that SHG only existed in anode side.[62'64] It was revealed by XPS that penetration of Na+ into the anode-side surface of ZnOTe0 2 glasses took place during thermal poling and contributed to SHG, since the poling was carried out by rendering the tellurate glass sandwiched between two commercial borosilicate glasses containing Na+. Results suggested that the poling significantly affects both the glass network and compositional.[65]

186

Second-Order Optical Nonlinear Properties of Glasses

5.3.3 Chalcogenide glass The x(3) °f chalcogenide glasses is about two orders of magnitude larger than that of silica, thus large SHG can be expected.[66] Chalcogenide glasses have some advantage such as large infrared spectral range of transparency, low phonon energy, photosensitivity, high linear and nonlinear refractive indices and the possibility to perform optical waveguide. Therefore these materials present great potential for the conception of electro-optic modulators in the infrared spectral region.[67] In Ref. [17], first SHG was reported on AS3S3 glass by thermal poling and optical poling. Later on, second-order nonlinear susceptibility in GeAsS chalcogenide glass irradiated by an electron beam was also reported. It was as large as 0.8 pm/V and stable.[66] SHG has also been obtained in GaGeSbS glass submitted to a thermal poling treatment. x(2) susceptibility of 4.4 pm/V was obtained.[67] 5.4 Applications 5.4.1 Nonlinearity stability The possible use of poled glasses in telecommunication networks raises an important question about the long-term stability of the induced second-order nonlinear susceptibility. It was shown that the second-order optical nonlinearity of glasses induced by thermal poling is very stable. Samples kept at room temperature without special precautions for several months showed no noticeable nonlinearity degradation.1'1 The relaxation time of the secondorder optical nonlinearity has been investigated at high temperature, highlighting an Arrhenius dependence with an activation energy of 1.3 (2) e y [44] -j^e stability of the induced % susceptibility was found to be strongly depending on the chemical composition of the glass. In particular, it is related to the mobility of ions that can migrate during the poling process. If only Na+ ions are assumed to bring into play in the poling stability in silica glasses, theoretical models predict that the second-order optical nonlinearity should be stable for thousands of years at room temperature.[40]

References

187

5.4.2 Quasi-phase-matched second harmonic generation devices Phase matching condition must be fulfilled in order to have high conversion efficiencies in SHG device. The so-called quasi-phasematching could be realized much easier in glass by inducing a periodic nonlinearity along the propagation direction. Two methods were used to realize this quasi-phase-matching SHG in poled glasses: periodically patterning the nonlinearity by thermal poling using patterned electrode,[68] or selective erasure of poled nonlinearity with e-beam,[4] UV light[69] or with infrared laser beam.[70] The quasi-phase-matched SHG was demonstrated by periodic UV erasure of the second-order optical nonlinearity in Ge-ion-implanted channel waveguides on fused silica substrate, with SH conversion efficiency of 6.1x 10"4 %/W-cm2.[69] 5.4.3 EO devices Electro-optic devices have been reported. Liu et al. reported a phase shift of 32 mrad in a 4.8mm electro-optic phase modulation in silica glass waveguide. [71] Abe et al. demonstrated a Mach-Zehnder type electrooptic switch in Ge02-doped silica-based channel waveguides with a 36 cm-long phase shifter. The switching voltage was 1700 V and switching response time was 100 ns.[72]

References 1. 2. 3. 4. 5. 6. 7. 8.

R. A. Myers, N. Mukherjee, S. R. J. Brueck, Opt. Lett., 16 (1991) 1732. A. Okada, K. Ishii, K. Mito, K. Sasaki, Appl. Phys. Lett., 60 (1992) 2853. P. G. Kazansky, A. Kamal, P. St. J. Russell, Opt. Lett., 18 (1993) 693. P. G. Kazansky, A. Kamal, P. St. J. Russell, Opt. Lett., 18 (1993) 1141. H. Nasu, H. Okamoto, A. Mito, et al., Jpn. J. Appl. Phys., 32 (1993) L406. K. Tanaka, K. Kashima, K. Hirao, et al., Jpn. J. Appl. Phys., 32 (1993) L843. U. Osterberg, W. Margulis, Opt. Lett., 11 (1986) 516. Y. Quiquempois, P.Niay, M.Douay, et al., Curr.Opin.Solid State Mat.Sci., 7 (2003) 89.

188 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28.

Second-Order Optical Nonlinear Properties of Glasses

A. Okada, K. Ishii, K. Mito, K. Sasaki, J. Appl. Phys., 74 (1993) 531. L.J.Henry, B.V.McGrath, T.G. Alley, J. J. Kester, J. Opt. Soc. Am. B, 13 (1996) 827. T. Fujiwara, D. Wong, Y. Zhao, et al, Electon. Lett., 31 (1995) 573. A. J. Ikushima, T. Fujiwara, K. Saito, J. Appl. Phys., 88 (2000) 1201. R. H. Stolen, H. W. K. Tom, Opt. Lett., 12 (1987) 585. Y. Sasaki, Y. Ohmori, Appl. Phys. Lett., 39 (1981) 466. N. M. Lawandy, M. D. Selker, Opt. Commun., 77 (1990) 339. N. M. Lawandy, R. L. MacDonald, J. Opt. Soc. Am. B, 8 (1991) 1307. Y. Quiquempois, A. Villeneuve, D. Dam, et al., Electron. Lett., 36 (2000) 733. X. -C. Long, R. A. Myers, S. R. J. Brueck, Electron. Lett., 30 (1994) 2162. A. Kameyama, E. Muroi, A. Yokotani, et al, J. Opt. Soc. Am. B, 14 (1997) 1088. T. Fujiwara, M. Takahashi, A. J. Ikushima, Appl. Phys. Lett., 71 (1997) 1032. C. Corbari, J. D. Mills, O. Deparis, et al., Appl. Phys. Lett. 81 (2002) 1585. X. -C. Long, R. A. Myers, S. R. J. Brueck, Opt. Lett., 19 (1994) 1819. H. Nasu, H. Okamoto, K. Kurachi, et. al., J. Opt. Soc. Am. B, 12 (1995) 644. Zhiling Xu, Liying Liu, Zhanjia Hou, et al., Opt. Commun., 174 (2000) 475. W. Margulis, F. Laurell, Opt. Lett., 21 (1996) 1786. T. G. Alley, S. R. J. Brueck, Opt. Lett., 23 (1998) 1170. P.G. Kazansky, A. R. Smith, P. St. J. Russell, et al., Appl. Phys. Lett., 68 (1996) 269. A. L. C. Triques, C. M. B. Cordeiro, V. Balestrieri, et al., Appl. Phys. Lett., 76 (2000) 2496. 29. A. Kudlinsky, Y. Quiquempois, M. Lelek, et al, Appl. Phys. Lett., 83 (2003) 3623. 30. V. Pruned, F. Samoggia, G. Bonfrate, et al., Appl. Phys. Lett., 74 (1999) 2423. 31. D. Pureur, A. C. Liu, M. J. F. Digonnet, G. S. Kino, Opt. Lett., 23 (1998) 588. 32. Y. Quiquempois, G. Martinelli, P. Dutherage, et al., Opt. Commun., 176 (2000) 479. 33. A. Ozcan, M. J. F. Digonnet, G. S. Kino, Electron. Lett., 39 (2003) 1834. 34. D. Faccio, V. Pruneri, P. G. Kazansky, Opt. Lett., 25 (2000) 1376. 35. A. Ozcan, M. J. F. Digonnet, G. S. Kino, J. Appl. Phys., 97 (2005) 013502. 36. H. An, S. Fleming, G. Cox, Appl. Phys. Lett., 85 (2004) 5819. 37. J. Arentoft, K. Pedersen, S. I. Bozhevolnyi, et al, Appl. Phys. Lett., 76 (2000) 25. 38. H. An, S. Fleming, Opt. Lett., 30 (2005) 866. 39. P. G. Kazansky, P. St. J. Russell, Opt. Commun., 110 (1994) 611. 40. A. Le Calvez, E. Freysz, A. Ducasse, Eur. Phys. J. D, 1 (1998) 223. 41. T. G. Alley, S. R. J. Brueck, R. A. Myers, J. Non-Cryst. Solids, 242 (1998) 165. 42. T. G. Alley, S. R. J. Brueck, J. Appl. Phys., 86 (1999) 6634. 43. A.L.C.Triques, I.C.S.Carvalho, M.F.Moreira, et al., Appl.Phys.Lett., 82 (2003) 2948. 44. N. Mukherjee, R. A. Myers, S. R. J. Brueck, J. Opt. Soc. Am. B, 11 (1994) 665. 45. Y. Quiquempois, N. Godbout, S. Lacroix, Phys. Rev. A, 65 (2002) 043816. 46. N. Godbout, S. Lacroix, J. Non-Cryst. Solid, 316 (2003) 338. 47. H. Takebe, P. G. Kazansky, P. St. J. Russell, K. Morinaga, Opt. Lett., 21(1996) 468. 48. F. C. Garcia, L. Vogelaar, R. Kashyap, Opt. Express 11 (2003) 3041. 49. C. Cabrillo, G. J. Cuello, P. Garcia-Fernandez, et al., Phys.Rev.Lett., 81 (1998) 4361. 50. V. Nazabal, E. Fargin, G. Le Flem, et al., J. Appl. Phys., 88 (2000) 6245. 51. Jianhua Xu, Xingze Lu, Hongbin Chen, et al., Opt. Mater., 8 (1997) 243. 52. L. J. Henry, A. D. DeVilbiss, T. E. Tsai, J. Opt. Soc. Am. B, 12 (1995) 2037. 53. J. Khaled, T. Fujiwara, M. Ohama, A. J. Ikushima, J. Appl. Phys., 87 (2000) 2137. 54. F. C. Garcia, I. C. S. Carvalho, E. Hering, et al., Appl. Phys. Lett., 72 (1998) 3252. 55. W. Margulis, F. Laurell, Appl. Phys. Lett., 71 (1997) 2418.

References 56. 57. 58. 59.

60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72.

189

A. Le Calvez, E. Freysz, A. Ducasse, Opt. Lett., 22 (1997) 1547. Zhiling Xu, Liying Liu, Yue Fei, et al, Appl. Phys. Lett., 77 (2000) 70. Yangang Xi, Zhiling Xu, Zhanjia Hou, et al, Opt. Commun., 210 (2002) 367. Xiaowei Ye, Liying Liu, Lei Xu, et al., 2005 International Symposium on Glass in connection with the Annual Meeting of the International Commission on Glass (ISG/ICG2005), Shanghai, China, April 10-14, 2005. Yangang Xi, Lejun Qi, Liying Liu, et al., Phys. Chem. Glasses, 44 (2003) 103. K. Tanaka, K. Kashima, K. Hirao, et al., J. Non-Cryst. Solids, 185 (1995) 123. K. Tanaka, A. Narazaki, K. Hirao, Opt. Lett., 25 (2000) 251. K. Tanaka, A. Narazaki, K. Hirao, N. Soga, J. Appl. Phys., 79 (1996) 3798. A. Narazaki, K. Tanaka, K. Hirao, N. Soga, J. Appl. Phys., 85 (1999) 2046. A. Narazaki, K. Tanaka, K. Hirao, et al, J. Am. Ceram. Soc, 84 (2001) 214. Qiming Liu, Fuxi Gan, Xiujian Zhao, et al., Opt. Lett., 26 (2001) 1347. M. Guignard, V. Nazabal, J. Troles, et al, Opt. Express, 13 (2005) 789. R.Kashyap, G.J.Veldhuis, D.C.Rogers, P.F. Mckee, Appl.Phys.Lett., 64 (1994) 1332. S. Chao, H-Y. Chen, Y-H. Yang, et al, Opt. Express, 13 (2005) 7091. S. Montant, A. Le Calvez, E. Freysz, et al., Appl. Phys. Lett., 74 (1999) 2623. A. C. Liu, M. J. F. Digonnet, G. S. Kino, Opt. Lett., 19 (1994) 466. M. Abe, T. Kitagawa, K. Hattori, et al., Electron. Lett., 32 (1996) 893.

Chapter 6

Glass Fibers for Optical Amplification

6.1 Brief Introduction of Optical Fiber Amplifier The Erbium doped fiber amplifier (EDFA) enabled the development of worldwide fiber optic communication networks. Prior to the advent of optical amplifiers, the attenuation of optical signals is compensated by electronic regenerators along the optical fiber cable. This slow and expensive electronic regenerating process not only limits the full usage of large transmission capacity and bandwidth which are inherently associated with optical fiber communication networks, but also restricts the wide deployment because of its cost. The usage of optical fiber amplifier removes the bottleneck of optical fiber communication networks, resulting in repaid increasing in the capacity of systems. In addition to compensate the signal attenuation in optical fiber as an in-line amplifier, optical fiber amplifier can also be used to increase the received signal power to a level that is suitable for detection (preamplifier), and boost the signal immediately from transmitter and modulator to a level suitable for transmission (power amplifier). Recently, optical amplifier is also used for compensating the intrinsic losses of passive devices, such as splitter and switch array. The first optical fiber amplifier was demonstrated by E. Snitzer at the American Optical Company using neodymium doped silicate glass.[1] This amplifier was pumped with flashlamps and operated in pulsed mode. In 1985 Southampton University demonstrated single mode rareearth doped silica glass fiber using modified chemical vapor deposition (MCVD) process.[2] Neodymium doped single mode fiber laser was 191

192

Glass Fibers for Optical Amplification

demonstrated pumped by a GaAlAs laser diode in 1986. The first erbium-doped single mode fiber amplifier for traveling wave amplification of 1.5 micron signal was reported in 1987, which was pumped with a dye laser at 655 nm. This EDFA exhibited a gain of 20 dB for 20 mW of absorbed pump power. However, dye lasers and solidstate lasers are not suitable pumping source for practical EDFA. In 1989 Nakazawa et al. demonstrated efficient EDFA using high power 1.48 micron laser diode as the pumping source, removing the last major hurdle for wide application of EDFA.[4] Since then, the development of EDFA has gained considerable attention in the world, and commercial well-packaged EDFA were available shortly, bringing a revolution in optical fiber communication, especially in high capacity undersea and terrestrial fiber optical networks. The praseodymium (Pr3+)-doped fiber amplifier (PDFA) for 1.3 micron band and thulium doped fiber amplifier (TDFA) for 1.4 micron S-band are other two important rare-earth-doped fiber amplifiers for fiber optical communication applications besides EDFA. Many optical telecommunications systems in the world were operating in the 1.3 micron band, especially in Japan, because a large amount of 1.3 micron single mode fiber was installed. PDFA was proposed after the failure of overcoming the signal excited state absorption (ESA) at the 4F3/2 state of Nd3+ ions, which shifts the gain to a longer wavelength and limits the maximum gain.[5] Ohshi et al. demonstrated a PDFA with a gain of 30.1 dB and a saturation output power of 13 dBm using Pr3+-doped fluoride glass fiber.[6] Here fluoride glass host is selected because of its low phonon energy. Due to the large difference in thermal properties between fluoride glass and silica glass, fusion splicing process can not be used for fiber connection. So one of the critical developments for a practical PDFA is to mechanically splice two fibers with large index difference and mismatched mode size. TDFA became a hot research and development topic around year 2001 when the absorption peak in silica fiber near S-band was removed and many engineers worried C and Lband may be not sufficient to handle skyrocketing demand in communication capacity. The activities in TDFA development were significantly reduced in the last few years because of the slowdown of the fiber optical communication industry.

6.2 Er-Doped

Phosphate Glass Fiber Amplifiers

193

In additional to rare-earth-doped fiber amplifier, fiber pigtailed semiconductor optical amplifier (SOA), Raman amplifier and rare-earthdoped planar waveguide amplifier have also been investigated for applications in fiber optical communications. SOA exhibits many attractive features including directly electrical pump, small size, large gain over a large bandwidth, available for a variety of wavelengths, and easy integration with other semiconductor devices. At the same time, SOA has some drawbacks. The short lifetime of the excited state leads to distortion and crosstalk between channels when the amplifier is operated in saturation regime. The noise figure is relatively large due mainly to the high coupling loss between the SOA chip and the single mode optical fiber. The other problems are the polarization dependent gain and large gain ripple. Raman amplifiers are based upon the third-order nonlinearity of fiber, where the energy is transferred from the pump to signal by stimulated Raman scattering with frequency difference of approximately 15 THz. The key features of Raman amplifier is the low noise and tunable gain by changing the pump wavelengths. However, Raman amplifier requires a long fiber of several kilometers and exhibits large crosstalk between different wavelength signals in the saturation region. Raman amplifier is proposed to be used as distributed amplifier, where transmission fiber is used as the gain medium. The first rare-earth-doped waveguide was produced in 1972 by sputtering a Nd-doped glass on an undoped glass substrate.[7] Recently, the need for a cost effective lossless splitter for fiber optical networks has led to the integration of waveguide amplifiers with passive splitters and combiners.181 Various technologies are used for the fabrication of planar waveguide amplifier including ion-exchange,[9] rare-earth ion implantation,[10] flame hydrolysis deposition,111] plasma enhanced chemical vapor deposition (PECVD),[12] sputter deposition [13] and solgel.[I4] 6.2 Er3+-Doped Phosphate Glass Fiber Amplifiers The most popular glass host for EDFA is silica fiber. Numerous papers

194

Glass Fibers for Optical Amplification

describing spectroscopic properties of Er in silica glass and erbium doped silica fiber amplifier performance were published in the last twenty years. Readers who are interested in typical silica glass fibers and amplifiers are encouraged to read other well organized books.[15'16] Here the author summarizes the study of Erbium-doped phosphate glass fiber amplifiers, which were mainly carried out at Optical Sciences Center, University of Arizona and NP Photonics by Drs. Tao Luo, B. Huang, Y. Hu, K. Seneschal, et al.[17"241 Recently, much development effort in EDFA is geared towards meeting requirements of metro, access and fiber-to-the-home (FTTH) networks, namely, compact, integration, and low cost, especially after the recent dramatic changes in the fiber communication marketplace. The requirements for optical amplifiers in these emerging networks are very different from long-haul optical networks where specific performance is of supreme importance. Typical amplifiers consisting of erbium-doped silica fibers more than ten meters long achieve greater than 20 dB gain near the 1.54 \xra range. Obviously, the fiber management associated with such lengths is not ideal for assembly into next generation amplifiers for metro, access and FTTH networks. This challenging has motivated research and development into other technologies such as semiconductor optical amplifiers (SOA) and erbium doped waveguide amplifiers (EDWA). Although compact in size, SOAs have inherently large noise figures, typically greater than 8 dB, which makes them unattractive in usual network applications where a series of amplifiers are required. EDWA has better noise figure performance, but typically suffer from polarization dependence, internal loss due to coiling tens of centimeters of an erbium-doped waveguide on the planar chip, as well as high manufacturing cost. One solution is erbium micro fiber amplifier (EMFA) technology, which reduces the length of the active fiber to a few centimeters, combining the compactness of SOA and EDWA and the high performance of typical EDFA. 6.2.1 Cooperative upconversion and energy transfer of high Er3*- and Yb^-Er3*-doped phosphate glasses To achieve high gain within a few centimeters of length, high Er3+-

6.2 Er3*-Doped Phosphate Glass Fiber Amplifiers

195

doping concentrations (1020 ions/cm3) without serious ion clustering are required. Because of the high solubility of rare-earth ions in phosphate glass high Er3+ concentrations can be doped into phosphate glass without serious ion clustering.1251 The main loss channel that limits gain performance in the high Er3+-doped phosphate glasses is the so-called cooperative-upconversion process. Cooperative upconversion is an energy-transfer process between two excited Er3+ ions in close proximity that interact in the \ m manifold, as shown in Fig. 6.1.[26] -N E4

19/2

A NR AE43

NE3

Ml/2

NR ^E32

•N E 2

1-13/2

RiE13

'I 15/2

WE12

WE2:

AE21

N I El

Fig. 6.1 Energy-level diagram of Er ions with cooperative upconversion. Cooperativeupconversion energy-transfer processes are indicated by dashed arrows.

One excited Er3+ ion (donor) transfers energy to the other excited ion (acceptor), causing the acceptor to be promoted to the 4l9/2 manifold and the donor to be de-excited to the ground state (i.e., the 4Ii5/2 manifold) nonradiatively. The excited Er3+ ions in the %a manifold will nonradiatively decay to the 4I13/2 manifold. This reduces the Er3+ population in the 4Iu/2 manifold. The cooperative upconversion rate depends strongly on the host materials. In silica waveguide amplifiers, Er3+ ion concentrations higher than 4.4x1019 ions/cm3 cause a significant increase in the loss of the 4Ii3/2 level owing to the cooperativeupconversion processes. This plays an important role for the gain reduction for these devices.[27] The energy levels of the Er3+ ions for optical amplification at 1.5 (im form a three-level system, which requires a high pump rate to achieve

196

Glass Fibers for Optical Amplification

population inversion. In addition, the low-absorption cross section of Er3+ ions limits the pump efficiency. Ytterbium ions exhibit not only a large absorption cross section but also a broad absorption band between 800 and 1100 nm. Furthermore, the large spectral overlap between Yb3+ emission ( F7/2 - F5/2) and Er absorption (Ii 5 / 2 - I13/2) results in an efficient resonant-energy transfer from Yb to Er in the Yb -Er_3+ codoped system.[28] For these reasons, Er3+-codoped with Yb3+ is often used. The energy transfer mechanism in an Yb3+-Er3+-codoped system is similar to that of cooperative-upconversion processes in an Er3+-doped system. However, interactions are between Yb3+ (donor) and Er3+ (acceptor) ions instead of between two excited Er3+ ions. The schematic energy level diagram of an Yb3+-Er3+ co-doped system in phosphate glasses is shown in Fig. 6.2. Compared with that in silicate glasses, the larger phonon energy in the phosphate host increases the transition probability for 4ln/2 - 4IB/2 relaxation, which prevents the back energy transfer from Er3+ to Yb3+,[29] making phosphate glass an ideal host for an Yb3+-Er3+-codoped system. F9/2 K

c

j

NR

;AE54]

i-JF5/2 •

Kr

KF

AE32

r

T

NR ^E43 NR

ic 1 1 1

!

RY12

RY21

Ri

A-Y21

KF RE13

7/2

WE21

11/2

I

4T

13/2

R \A^ ^E21' C

±-% 5/2

JLl 3+

3+

Fig. 6.2 Energy-level diagram of the Yb -Er -codoped system. Cooperativeupconversion and Yb3+-Er3+ energy-transfer processes are shown by dashed arrows.

Many studies have been performed to design and optimize the Errand Yb3+-Er3+-doped amplifiers and lasers. [30'31] However, the parameters used in the modeling, such as cooperative-upconversion and energy transfer coefficients, were not experimentally determined. Studies related

6.2 Er3*-Doped Phosphate Glass Fiber Amplifiers

197

to energy-transfer efficiency to optimize the Er and Yb concentrations were performed by a Nd:YAG laser or a flash-lamp pump [32'33] The waveguide configuration and the development of 980-nm laser diodes make high-Yb3+ concentration unnecessary for device optimization.[34] Systematic studies on both cooperative upconversion and energy transfer in phosphate glasses are limited. Therefore the measurements on these parameters are essential to optimize device performance. Four Er3+-doped samples (Erl, Er2, Er3, and Er4) and three Yb 3+ 3+ Er -codoped samples (YEl, YE2, and YE3) were prepared for characterization of cooperative upconversion and energy transfer in phosphate glasses. The Er3+ and Yb3+ concentrations of the samples are listed in Table 6.1. Table 6.1 Er3+ and Yb3+ Concentrations of the Samples. Samples

Er3+Concentration (ions/cm3)

Yb3^ Concentration (ions/cm3)

Erl

2.0 xlO 20

Er2

3.0 xlO

20

0

Er3

3.5 x 1020

0

Er4

4.0 xlO 2 0

0

YEl

2.0 xlO

20

2.0 xlO 2 0

YE2

2.0 xlO

20

4.0 xlO 2 0

YE3

2.0 xlO 20

6.0 xlO 20

0

Pump intensity-dependent spontaneous-emission decays of the Er 4Ii3/2 level and of the Yb3+ 2F5/2 level were measured for the cooperativeupconversion and the Yb3+-Er3+ energy-transfer processes, respectively. The pump source is an InGaAs laser diode stabilized with a pigtailed distributed-Bragg-reflector fiber at a wavelength of 980 nm. The pump laser-diode driver was modulated by a function generator with square pulse duration of 40 ms and a repetition rate of 10 Hz. A long pulse duration of 40 ms enabled the populations of the Er3+ and Yb3+-Er3+ systems to reach steady state before the pump switched off (t = 0). Efforts were made to locate the focus area near the sample edge to minimize reabsorption and to enhance the detected signal. The

198

Glass Fibers for Optical Amplification

luminescence signals at 0.976, 1.030, and 1.535 urn were side collected, and then decay dynamics were measured with an InGaAs detector. There was no fluorescence detected for sample Er4 (highest Er3+ concentration) at 980 nm, and the lifetimes of co-doped samples measured both at 976 (Yb3+peak emission) and 1030 nm were very close within the measurement error range. These indicate that the detected signal at 976 nm was indeed from Yb3+ ions, and the effects from the pump beam were negligible. Rate equations were used to characterize the cooperative upconversion. Here is rate equation analysis for Er3+ only system and Yb3+-Er3+ co-doped system. 6.2.1.1 Er3+ only system Fig. 6.1 shows the Er3+ energy diagram. Considering the 980nm pump wavelength and the cooperative upconversion, the rate equations can be written as: ^

=

dNE2

dt dN

E3

dt

_*£_3 NEI _ w^

NEI + WEix NEI + A^

-WEnNEi ~WEiNEi -AREiNEi , ANR D A, ANRT,T ' _=R ENEx -A™NE3+A™NE4

+

,

NEI

+

CN\2 ,

A™NE} -2CN\

, (6.1)

2 ^ ± = -AENE3RN4E +CN El E , dt * NEi+NE2+NE}+NE4=NE,

where NE is the Er3+ concentration in the samples. N E) , NE2, NE3, NE4 are Er3+ populations in the energy levels 4Ii5/2 (^I), \yi (E2), \m (E3), and 4 I9/2 (E4). REU is the pump rate of Er3+ ions for the 980-nm pump wavelength. WEYI and ^£21 are the absorption and stimulated-emission rates at the signal wavelength. AE* and AEr are the nonradiative decay and spontaneous radiative rates between levels Et and Ej. C is the homogeneous-upconversion coefficient. Because of the small energy gaps between levels 4 In /2 and 4Ii3/2 and between 4I9/2 and 4 I n / 2 , and because of the large phonon energy for the

6.2 Er3'1-Doped Phosphate Glass Fiber Amplifiers

199

phosphate glass host, decay from levels %/2 and 4In/2 is dominated by fast nonradiative transitions, which are 3.6 x 105 and 3.5 x 108 s"1, respectively.1351 However, decay from level 4Ii3/2 is predominantly radiative. Since the decay for levels 4In/2 and 4I9/2 is fast compared with the pump rate REB, the populations at both levels are negligible. This is confirmed by calculation of the steady-state populations with the measured parameters in this study. The maximum populations in the levels 4 I„ /2 and 4I9/2 are less than 0.2% and 10"4 % of the Er3+ concentration for the highest Er3+ (Er4) concentration sample with 100mW pump power. The pump excited-state absorption from level 4In/2 is negligible because pump wavelength (980 nm) does not match any of the transitions to higher levels.[36] Furthermore, the pump excited-state absorption for level 4In/2 is neglected because of the negligible population in this level. In the spontaneous-emission measurements without signal reabsorption and stimulated emission, WEi2 and WE 2 I are neglected, and the rate equations can be simplified as ™^

=

_dNIL=

_AR

N

2

NEi+NE2=NE. When the lifetime is measured under weak excitation (i.e., NE is much smaller than AR/E2i), the quadratic term corresponding to the cooperative upconversion can be neglected. The population in the excited level becomes a single exponential decay with a lifetime of = 1//A TE(TE T~) a ft er m e pump switches off. When the pump power is increased, the cooperative-upconversion effects should be considered, and the population in the excited level can be solved from Eq. (6.2) as NEU):

1 E

!

+c exp-U-Q

,

(6-3)

4NE2(0)

where NEi (0) is the excited Er3+ ions at t= 0 (i.e., steady state in this case) and depends on the pump power. By solution of Eq. (6.2) in the steady-state condition, NE (0) can be solved as

200

Glass Fibers for Optical Amplification

_o

^(°) = W i + ^4

(6.4)

The only unknown parameter is C, which can be found by fitting of the luminescence decay curves with Eqs. (6.3) and (6.4). 6.2.1.2 Yb3+- E/+-codoped System An energy diagram including energy-transfer and cooperativeupconversion processes of an Yb3+-Er3+-codoped system is shown in Fig. 6.2. Because of the fast nonradiative decay in the levels 4In/2, \n, and 4 F9/2 in phosphate glasses, both the populations in these levels and backenergy transfer [Er3+(4in/2 - 4I15/2) -> Yb3+(2F7/2 - 2F5/2)] are negligible. The rate equations can be simplified (without a signal) as

^ " ^ ^

V * -Wr, ~42NYI -AlNE2

= - ^ L = REUNEI

~KFNENY2

+KFNENy2

-KcNEiNh,

-CNl2,

(65)

NEi+NEi=NE, Nyi+Ny2=Ny,

where NY is the Yb3+ concentration in the sample. NY andNY are Yb3+ populations in the energy level 2F7/2 (7i) and 2F5/2 (Y2). Rr and RY are the pump and stimulated-emission rates between level Y\ and Y2 of Yb3+ ions. AY is the spontaneous radiative rate of level 2F5/2. KF is the (forward) energy-transfer coefficient from Yb3+ to Er3+ ions, and Kc is the cumulative transfer coefficient from Yb3+ to excited Er3+ ions [Yb3+(2F5/2-2F7/2) -> Er3+(4Ii3/2-4F9/2)]. After the pump is switched off, the rate equations can be rewritten as d

-¥ = -^~N-{t^FNENY2

^

=dt

^ =d t

^ r»

+

-KcNE2NYi = ^ , ( 6 , 6 )

KFNENY-CNE, F

£,

Y2

E2,

6.2 Er3^-Doped Phosphate Glass Fiber Amplifiers

201

where r° =1/ A§ and zY is the measured lifetime. Values of 5.2 x 1021 and 6.8 x 1021 cm2/ion were used for the absorption cross section and the stimulated-emission cross section at 980 nm, respectively; a value of 2.2 ms was used for the Yb3+ lifetime (rr) in the samples without Er3+ doping. To retrieve KF, the following procedure was adopted. For low pump intensities the population in the level 4IB/2 is small, and both the cooperative upconversion and the cumulative energy-transfer effects for the Er3+ ions are negligible. Moreover, during the 250 \xs after the pump was switched off, the population of level 4Ii3/2 could be considered constant because of the long lifetime of the Er3+ ions in the metastable level. Therefore the spontaneous-emission decays of Yb3+ ions were measured and fitted between 20 and 250us. Assuming NE constant, the energy-transfer coefficients were calculated from the forward energytransfer rate (WF = KFNE ) and the Er3+ ground-state population at low pump intensity. The cooperative upconversion of Er3+ ions of an Yb 3+ Er3+-codoped system can be found by fitting luminescence decay of Er3+ ions at 1.5 um with Eq. (6.3) at higher pump intensity. However, NE2(0) needs to be solved numerically, and the fitting process is somewhat iterative. The cooperative-upconversion coefficient and lifetime were obtained as a function of Er3+ concentration, shown in Fig. 6.3.

2

2.5

3

3.5

4

4.5

Erbium Doping Concentration (10 20 ions/cm3)

Fig. 6.3 Cooperative-upconversion coefficient and lifetime as a function of Er3+ concentrations.

202

Glass Fibers for Optical Amplification

Experimental decay curves of the weak excitation were fitted very well with single exponential curves. The lifetimes are very close for different Er3+ concentrations, indicating that the concentration quenching is negligible. Cooperative upconversion coefficients were fitted with Eq. (6.3) at the highest pump intensity. The cooperative upconversion coefficients increase with higher Er3+ concentrations, as expected. This shows increasing interactions between the excited ions owing to reduced ion separation. The cooperative upconversion coefficient for an Er3+ concentration of 4 x 1020 ions/cm3 is -1.1 x 10"18 cm3/s, which is smaller than those in other hosts with lower Er3+ concentrations.137"401 Experimental data for lifetime and cooperative upconversion have shown that these Er3+-doped glasses are excellent candidates for gain media. In the Yb3+-Er3+ codoped system the energy-transfer process from Yb3+ to the metastable level of Er3+ ions is not instantaneous owing to the finite lifetime of level 4In/2. However, the residual pumping effect from the energy transfer of Yb3+ ions is negligible after 1 ms. Therefore, based on the steady-state population for t= 0, the luminescence decays were fitted from 1 to 30 ms for the cooperative upconversion coefficients to minimize the error from the residual pumping effect of the energy transfer process. The energy transfer and cooperative upconversion coefficients are shown in Fig. 6.4.

1

2

3

4

5

Yb3* Doping Concentration (1020 ins/cm3)

Fig. 6.4 Energy-transfer coefficient and cooperative upconversion coefficient as a function of Yb3+ concentrations.

6.2 Er3*-Doped Phosphate Glass Fiber Amplifiers

203

The cooperative-upconversion coefficient of sample Erl is also indicated for comparison. Owing to the stronger Yb3+-Er3+ interactions, the energy-transfer coefficient was found to increase with Yb3+ concentration, as expected. The energy-transfer coefficient for the Yb + concentration of 6 x 1020 ions/cm3 (YE3) was found to be 1.1 x 10"16 cm3/s, which is a reasonable value when compared with other results.'41' The cooperative upconversion coefficient that results from Er3+-Er3+ interaction was almost constant, as expected. This also indicates that the residual pumping effect was indeed negligible in our fitting procedure. Table 6.2 summarizes the cooperative upconversion and energy transfer coefficients. Phosphate glass has one of the lowest cooperative upconversion coefficients. A commonly used silica glass co-doped with Ge, Al, and P shows a much higher cooperative upconversion coefficient of 2 x 10"16cm3/s. It has been suggested that the relatively large amount of nonbridging oxygens in phosphate glasses contribute to a more homogeneous Er3+ distribution, which in turn leads to a smaller cooperative upconversion effect compared with that in other host materials. Because of the low Yb3+ concentration, the energy transfer coefficients obtained are smaller than those from other work. Table 6.2 Summary of the Cooperative-Upconversion and Energy-Transfer Coefficients. Er3+ Yb3+ Concentration Concentration (lO-'W/s) (10-"W/s) (1020 ions/cm3) (1020 ions/cm3) 2-4

2-6

0.8-1.1

-1

0

3.8

-1.2

0

4

0.4-5

0

0.7-4

0

0.4

Ref.

Phosphate

19

Soda-lime silicate

37

Er-implanted A1203

38

0.3-1

Soda-lime silicate

39

0.5-2

Alumino-silicate

39

Ge/Al/P-doped fused silica

40

Phosphate

41

-0.01

0

200

2

10

1.2

1.1

GlassHost

5

Measured lifetimes of Yb3+ ions and effective-transfer rates (1/x-r- l/xY°) from Yb3+ to Er3+ as a function of pump power. The energy transfer efficiency is calculated for low pump power by

204

Glass Fibers for Optical Amplification

ri = l-f.

(6.7)

The energy-transfer efficiency mainly depends on the ratio of the backward-transfer rate and the multiphonon relaxation rate of the Er3+ I11/2 level. Compared with silicate and germanate glasses, phosphate glasses have high efficiencies because of the small values of this ratio.[42] In tellurate glasses, low observed transfer rates indicate that a low transfer efficiency and/or a presence of back transfer from Er3+ to Yb3+ because the slow decay of the 4ln/2 level cannot provide an efficient sink for the excitation transferred from Yb3+.[43] The Ce3+ incorporated with the Yb3+-Er3+ system was studied to minimize the back-energy transfer by reduction of the lifetime of level 4In/2.[44'45] An energy-transfer efficiency of 45% was reported under weak excitation at 977 nm in borosilicate glasses with an Yb3+-Er3+ concentration ratio of 5 (Er3+, 8 x 1019 ions/cm3; Yb3+, 4 x 1020 ions/cm3). The low transfer efficiency of 45% is because of a low Yb3+-Er3+ ratio in borosilicate glass.[46] In phosphate glasses, energy-transfer efficiencies higher than 95% were measured for low pump power. The high transfer efficiency is not only because the back energy transfer is negligible, but also because both Yb3+ and Er3+ concentrations are high to cause high excitation diffusion among Yb3+ and Er3+ ions. Moreover, Er3+ concentration in phosphate glass is 2 x 10 ions/cm , which is higher than other glasses (<10 ions/cm ). These make high transfer efficiency even with a low Yb3+-Er3+ concentration ratio. The ytterbium measured lifetime increases with the pump intensity owing to the reduction in the total effective-transfer rate to the erbium. The effective transfer rate is approximately equal to WF + Wc, where WC(=KCNE2) is the cumulative-transfer rate. In the low pumppower region the effective-transfer rate for phosphate glasses can be considered as the forward-transfer rate only. As the pump power increases, the population of Er3+ 4Ii3/2 becomes significant and the cumulative transfer process takes place. The effective-transfer rate includes both forward- and cumulative-transfer processes as pump power increases. However, since the cumulative transfer process is less efficient than the energy transfer process, the effective-transfer rate decreases with increasing erbium inversion. The reduction in the direct-transfer rate for

6.2 Er^+-Doped Phosphate Glass Fiber Amplifiers

205

low pump power is, in fact, only partially compensated by the increasing of the cumulative-transfer process. Because of the increasing Er3+ 4Ii3/2 population, the effective-transfer rate tends to saturate at high pump intensity. A sample with higher Yb3+ concentration has a higher effective-transfer rate because of stronger interaction between Er3+ and Yb3+ ions. The small cooperative upconversion coefficients and high Yb3+-Er3+ energy transfer efficiencies indicate that these Er3+- and Yb3+-Er3+-doped phosphate glasses are excellent materials for active-device applications, such as fiber amplifiers. 6.2.2 Structure of rare-earth-doped phosphate glasses In order to get a better understanding of the properties of phosphate glasses, the structure of these glasses was investigated by solid state NMR. The environment of P atoms and Al atoms inside the glasses were studied. 6.2.2.1 31P NMR spectra The 15 KHz MAS and static spectrum are compared. By increasing the spinning speed, one can simultaneously average the chemical shift and reduce the strong dipolar interaction essentially due to 31P neighboring atoms. By focusing one can observe the asymmetric shape of the actual central 31P line (Fig. 6.5). Using an extended version of Bruker Winfit software this broad line is deconvoluted with two contributions.[49] Each of them has a Gaussian shape due to the disorder neighboring of the 31P sites in such a glass. The main contribution (86%) at -27.6 ppm is attributed to the so-called Q2 type 31p.t50'51] These P are involved in P0 4 tetrahedra corner shared to two other ones. The minority contribution (14%) at -37.8ppm is due to Q3 type 31p.[50-51] They correspond to P0 4 tetrahedra corner shared to three others.

206

Glass Fibers for Optical Amplification

Experimental ^ " " ^ spectrum

U\\

^,

Reconstructed spectrum

Q2 phosphorus contribution

Q phosphorus contribution

20

10

0

-10

-20

-30

-40

-50

-60 -70(ppm)

Fig. 6.5 Reconstruction of the 31P MAS NMR spectrum, evidencing Q2 and Q3 type of P atoms.

These results are in agreement with the well-known structural description of the phosphate glass family. Their glass network is expected to be built upon from more or less corner-shared P0 4 ~ tetrahedra. Here, the NMR spectra show a network that consists mainly of long tetrahedra lines cross-linked by one tetrahedra over 6 (86%/14%). 6.2.2.2

27

Al NMR spectra

The 27A1 15 kHz MAS spectrum is measured. The external transition spinning side bands are labeled with stars. The central lines indicate three types of neighborhood for Al: four-, five- and six-fold coordination to oxygen.[48'50'51] Note that the second order quadrupolar interaction lines are located at the right side of the isotropic chemical shift on a high spinning speed spectrum. The particular asymmetric line shape (trailing high field edge) of each contribution is attributed to the quadrupolar parameter distribution. The relative intensities of the three resolved sites were then evaluated from computer simulation of the experimental spectra (Fig. 6.6).[49] The calculation model assumes a Gaussian distribution of the quadrupolar constants VQ. In any case r|Q (asymmetric parameter) was

6.2 Er3*-Doped Phosphate Glass Fiber Amplifiers

207

fixed at 0.5 since its value does not play any role in the line shapes/481 The chemical shifts were also fixed for each line because they mainly depend on the number of O around each Al. It appears that the Al VI is in majority (70%) and that Al IV and Al V are equality represented (15%). This last coordination is rare in aluminate crystals but is regularly invoked in aluminate glass structures.[48'50'51] Experimental MAS spectrum Reconstruct ed spectrum

Al contribution

50

0

-50

-100 (ppm)

Fig. 6.6 Reconstruction of the 27A1 central lines thanks to a model that assumes a Gaussian distribution of the quadrupolar constants vQ.

6.2.2.3 Structural Model From the initial composition of the glass, 62% P 2 0 5 , 12% A1203 and 26% of network modifiers (Li 2 0, Na 2 0, BaO, CaO), it appears that the tetrahedron network is built up from 124 P, 372 O and some of the 24 Al. Since there are 86% P0 3 (Q2 type P) and 14% P0 2 , 5 (Q3 type P), 363 of the 372 O are involved in tetrahedra centered in P atoms (if there were 100% of Q2 type P, all of the 372 O would be used up in such tetrahedra). So, here, it remains 9 "unused" O over 372. Furthermore one got 0.15 x 24 = 3.6 Al IV. Then the Al IV can be considered as builder of the network in A1 36 0 9 = A102.5 tetrahedra. This role is consistent with their weak coordination number and coherent with the relative intensities obtained from NMR experiments. Finally each type of octahedron can be counted up as followed: 83.6%o of P0 3 octahedra bi-connected 13.6% of P0 2 . 5 octahedra tri- connected 2.8%o of A10 25 octahedra tri- connected

208

Glass Fibers for Optical Amplification

Al V and VI stand in interstitial positions between the chains. They rigidify the network via iono-covalent bonds O-Al-O. The rare earth ions Er3+ are also expected in interstitial positions with higher O coordination numbers. The phosphate glass network is open and very flexible. The linkage of the tetrahedron is very disordered which leads to the great stability of the glass. The high flexibility of the network explains the relatively low glass temperature and the low softening temperature of the phosphate glass. Thus it is possible to fabricate performs with more simple techniques and with lower costs like the rod in tube as CVD methods are used for silica glass for example. The structure of silica glass is based on Si0 4/2 tetrahedra, each corner being shared by Si-0 bonds which are covalent. From this rigid structure, the excellent chemical mechanical behavior of silica glasses can be explained, in particular their high resistance towards thermal shocks.[53] But the glass temperature and softening temperature are high. 6.2.3 Fundamentals of fiber amplifier A large portion of readers of this book are glass scientists and engineers who may not be familiar with fundamentals of fiber amplifier. Following several paragraphs describe gain and noise figures, which are two most important parameters for fiber amplifier without using many mathematic equations. Parameters like gain efficiency, polarization dependent gain, polarization mode dispersion, gain flattening, and transient gain dynamics are not discussed here, although all of them are extremely important for a practical fiber amplifier. Readers who are interested in these parameters are encouraged to read other books.'15'161 An atom can absorb or emit a photon by transitions between its energy levels. According to the energy conservation in the process, the interaction of light and atoms can be described by equation: d(j> = [N2B2lg2l (v)p(v) - N,Bngn

(v)p(v) + N2A2lg21 (v)]dz , (6.8)

where <) is the photon flux in the mode, Nt is population in the level i, B2i, B12 and A2i are the Einstein's A and B coefficients for stimulated emission, absorption, and spontaneous emission, respectively, gtj (v) is

6.2 Er +-Doped Phosphate Glass Fiber Amplifiers

209

the normalized line shape for the transition from level i to j , and p(v) is the radiation density (energy per unit volume). Eq. (6.8) can be re-written in terms of photon number density in the mode Np, cross section cry, and number of modes per unit volume per unit frequency interval m, dNp = [(N2cr2l -N,an)Np

+ mN2cr2l]dz .

(6.9)

Assume that the distance, L, is very short and the signal N p is also small, Nl and N2 can be considered as a constant in the interval of L, i.e., no population depletion caused by signal. We can solve Eq. (6.9) to find the relation between input and output photon number density with above assumption. Np(L) = GN{p)(0) + mnsp(G-l).

(6.10)

G = exp(N2cr2l-N1crn)L,

(6.11)

With the relation of N2

n

sP

N°nN Z

,

(6-12)

cr 2 l

n

where G is the overall gain in the interval L and sp is the inversion parameter or spontaneous factor. Eq. (6.10) includes the important parameters for an amplifier, gain and noise mnsp(G-l). In Eq. (6.10), the first term on the right side is the amplified signal and the second term is the amplified spontaneous emission output. The noise output power (PASE) in the bandwidth Av , around the frequency where the gain of the amplifier is G: ?ASE=mnsp(G-l)hvAv. (6.13) As mentioned, m presents the number of modes. There are two polarization modes in signal mode fiber amplifier, so m equals 2 in single mode fiber. However, m could be larger number in multi-mode fibers. Eq. (6.11) indicates that the gain is a function of the shape of inversion along the length of fiber. The pump intensity changes along the fiber, so the inversion changes along the fiber. Gain can be presented by the average population along the fiber by a simple relation as: G = cxp[(N2ase -N,asa)rsL]

,

(6.14)

210

Glass Fibers for Optical Amplification

where ase and asa are used for the emission and absorption cross sections for the signal wavelength, respectively. Ts is the overlap factor for the signal wavelength because the signal mode filed diameter is different from the diameter of core of the fiber, for example. JV, and N2 are the average lower-state and upper-state populations in the whole fiber, respectively. The average populations and the overlap factor are defined as following: N^j-^N^dz,

(6.15)

N2=j;^N2(z)dz,

(6.16)

00

r,=2*J/»^W,

(6.17)

o where F(r) is the normalized transverse mode intensity profile for signal, and N(r)/N is the normalized erbium ion distribution. So the average inversion rate in the fiber should be high and the overlap factor should be large in order to achieve a high gain. The total doping concentration has to be large in order to achieve a large gain in a short piece of fiber. Eq. (6.10) indicates that when a signal travels through the gain medium, output includes the amplified signal and amplified spontaneous emission. The amplified spontaneous emission acts as noise together with signal because it reduces the signal to noise ratio. A figure of merit to present the degradation for signal to noise ratio is noise figure (NF), which is defined as the ratio of the signal to noise ratio at the input to that at the output end of the fiber. NF = •§§*-,

(6.18)

The major noise sources are shot noise, signal-spontaneous beat noise, spontaneous-spontaneous beat noise, and reflection noise. Spontaneous-spontaneous beat noise can be significantly reduced by placing an optical filter before the photo-detector. Reflection noise can be reduce by fusion splicing process or angle contact. Therefore shot noise and signal to spontaneous beat noise are major noises. According to Eq. (6.18), the noise figure for single mode fiber can be expressed as: [16 '

6.2 Er3^-Doped Phosphate Glass Fiber Amplifiers

211

NF = 2nsp^+j;.

(6.19)

Typically, noise figure is described in decibels according to NF = 10 log(iVF). In a fully inverted fiber with a large gain, NF &2nsp from Eq. (6.19) and nspis 1 according to Eq. (6.12), which leads to a noise figure of 3dB. So in EDFA the lowest noise figure is 3dB. So an optical amplifier cannot improve the optical signal to noise ratio. 6.2.4 Fiber amplifier performance for highly Er3*-doped phosphate glass fiber Short optical amplifiers with high gain per unit length are ideal for integration with passive devices. In order to compensate losses due to passive devices, such as splitting loss, a gain greater than 10 dB within a few centimeters of length is required and the Er3+ concentrations must exceed 1 wt% (~1020 ion/cm3). At high Er3+ concentration, ion-ion interactions result in a reduction of 4Ii3/2 level measured lifetime with increasing pump power and tend to degrade the amplifier performance unless proper host material is selected. Undesirable high concentration effects can be minimized by using phosphate glasses due to high solubility of rare-earth ions and weak ion-ion interactions in phosphate glasses as discussed previously. Efforts have been made to design and fabricate short Er3+ doped amplifiers based on both planar waveguide and fiber technologies. Fiber amplifiers, however, have several advantages compared with planar waveguide amplifiers such as polarization insensitivity, good mode matching, and good Er3+ confinement. In addition, it is much more difficult to fabricate a good quality glass planar waveguide than a similar fiber waveguide with a comparable glass composition. Here, the performances of high concentration Er3+-doped and Er3+/Yb3+-co-doped phosphate fiber amplifiers are presented. Measurements of gain, noise figure, and signal saturation power were performed for a series of fiber lengths, pump powers, and signal wavelengths. The Er3+-doped phosphate fibers were fabricated by a two-step, rod in tube technique. The core glass rod was drilled from a glass specimen and

212

Glass Fibers for Optical Amplification

the barrel of the rod was polished. Both inside and outside surface of the glass tube made from the cladding glass were polished. A 3mm core rod and a 12 mm cladding tube were drawn into a rod with a 3 mm outside diameter. Then, this new rod was used as the core glass rod for next drawing step. The fiber drawing temperature was around 750°C. No plastic buffer was coated outside of the phosphate fiber. Fiber drawing was performed in an Argon gas atmosphere to reduce absorption of water from the air, which causes fluorescence quenching of \y2 level of Er3+ ions. The relationship between perform and resulting fiber diameter is based upon the equal volume in both cases. Neglecting the diffusion during the fiber drawing process, we can consider the doping is distributed uniformly inside core region. The refractive index was measured with bulk samples by a prism coupler at wavelength of 1550 nm, 1300 nm, 830 nm and 632.8 nm. The refractive index at 980 nm, interpolating from the measured index other wavelengths. The numerical aperture and mode field diameter are calculated. The near-filed intensity patterns of fabricated fibers at 1550 nm and 980 nm were also measured using infrared camera and microscope objective. The measured mode field diameters agree well with the calculated ones. The propagation loss was measured using the cutback method at 1.3 urn. The coupling loss between Er3+-doped phosphate fiber and input silica fiber was also estimated for gain calculation. The experimental set-up for fiber amplifier characterization is shown in Fig. 6.7. The signal source is a tunable external cavity laser diode with wavelength range from 1530 to 1570 nm. To maximize the input signal to noise ratio, the signal laser diode was operated at high driving currents, followed by a variable attenuator to adjust the input signal levels of the fiber amplifier. The 10% output port of the 90/10 coupler following the attenuator was used as signal monitoring and calibration. A multiplexed output from a 975 nm and a 980 nm laser diodes with an output power ratio of 1.3 were used as the pump source in the experimental.

6.2 Er -Doped Phosphate Glass Fiber Amplifiers

Tunable Laser 1530-1570 nm

Pump Laser 980 nm

213

Pump Monitor

a-i

"— Attenuator 90/10 Coupler

Signal Monitor

Optical Spectrum Analyzer

Fig. 6.7 Experimental set-up for fiber amplifier characterization.

A maximum pump power of 244mW was available at the multiplexed output. The signal and pump sources were combined by a multiplexer at the input of the doped fiber. There was another multiplexer at the output of erbium-doped fiber. The 980nm port of output multiplexer worked as a pump power dump and a pump power monitor. Without the 980nm power dump, the output isolator could be damaged by residual pump. Fusion splicing process between the phosphate glass fiber and silica fiber was developed.[54] But most results here were obtained with free space butt-coupling. Index matching oil was used between the two pieces of butt-coupled fibers in order to improve and stabilize the coupling efficiency, reduce the pump reflection to pump laser diode, and reduce the signal and ASE feedback to form a laser cavity. Input and output isolators at signal wavelengths were used during the experiment to prevent any unwanted reflections which might result in the laser action in the Er3+-doped fiber and instability for the signal laser diode. The output signal was sent to optical spectrum analyzer for gain and noise measurements. The optical amplifier gain was calculated as: G(dB) = 10Log10[(Pout - P ASE )/P in ],

(6-20)

where P;n and Pout are the amplifier input and output power at signal wavelength, and PASE is the noise power generated by the amplifier which lies within the optical bandwidth of the measurement. Noise figure was obtained by an interpolation method using the following Eq. (6.21). Eq. (6.21) is obtained by combining Eqs. (6.13) and (6.19).

214

Glass Fibers for Optical Amplification

NF = PASE/hvAvG + l / G .

(6.21)

6.2.4.1 Er3+ only fiber amplifiers Er3+ ions were doped uniformly in the core region with a concentration of 3.5 wt%. The fiber with 5 um core diameter is single mode at 1.5 |im with a numerical aperture of 0.22, while it is multi-mode at 0.98 um. The measured mode field radius at 1.5 and 0.98 um are 3.0 and 2.2 urn, respectively. The average background loss was measured to be 0.28 dB/cm at 1.3 urn. The coupling loss is ~1 dB between Er3+-doped phosphate fiber and input fiber. The small signal gain and noise figure were measured as a function of pump power for a series of fiber lengths shown in Fig. 6.8. An input signal power of -30 dBm at 1535 nm was used in the experiment. Internal gain is the amplification experienced by signal between the end and the beginning of Er3+-doped fibers. Net gain, including the coupling loss, was ~2 dB less than the internal gain. The noise figure was obtained by a spectral interpolation method based on the internal gain, but when referenced to net gain the noise figure will increase by ~ 1 dB. The highest internal gain obtained is 23 dB with a noise figure of 4.3 dB for a 71 mm long fiber at a pump power of 244 mW. A net gain of 10 dB using a 110 mW pump power can be also obtained. Gain can further be increased by increasing the pump power for long fibers, while gain saturates for short ones. The threshold pump power ranges from 20 to 55 mW depending on the fiber length. In order to investigate the amplifier behavior at different wavelengths, gain spectra were measured by tuning a signal wavelength from 1530 to 1570 nm at different pump powers and fiber lengths. Fig. 6.9 (a) shows gain spectra at different pump powers with a 31 mm long fiber. The gain thresholds for longer wavelengths are lower than those for shorter wavelengths due to the larger emission to absorption cross section ratio at the longer wavelengths. This can be explained as combination of three and four-level systems. The short wavelength region (-1535 nm) is considered as a near three-level system, which has ground level absorption. The long wavelength region (-1550 nm) is a

6.2 Er3*-Doped Phosphate Glass Fiber Amplifiers

215

near four-level system, which has a comparatively small ground level absorption. When the inversion level is low, the three-level medium exhibits high loss while four-level medium is relatively transparent. The gain peak moves to 1535 nm and becomes sharper as pump power increases. Phosphate glasses emit and absorb at longer wavelengths (peak at -1535 nm) compared to more ionic ones, such as fluoride glasses.[55]

- T O - i — i — . — i — . — i — i — i — . — i — . — i — i 0 50 100 150 200 250 Pump Power (mW)

Fig. 6.8 Small signal gain and noise figure vs. pump power for different Er-doped fiber lengths at 1535 nm. Signal input power is -30 dBm.

Fig. 6.9(b) shows small signal gain spectra for different fiber lengths at pump power of 224 mW. The gain difference between the peak and the shoulder is smaller for short fiber than that for long fiber. However, a short fiber cannot provide enough gain for practical applications. The noise figure of a 71 mm long fiber slightly decreases with wavelength is also reported in this figure. The saturation behavior has been characterized at different wavelengths, fiber lengths, and pump powers. Gain saturation at different pump powers for a 71 mm long fiber with a signal wavelength of 1535 nm is shown in Fig. 6.10(a). The input and output signal powers are linearly related at low pump power while output signal power saturates as pump power increases. The 3-dB saturation input power decreases as pump power increases due to the strong amplification experienced by signal. This also indicates that the photon conversion efficiency reduces

Glass Fibers for Optical Amplification

216

as pump power increases. 3-dB saturation output powers are 2.9, 5.3, 6.9 and 7.7 dBm for pump powers of 84, 147, 191 and 244 mW, respectively.

1520

1550

1560

1570

1590

Wavelength (nm)

24

Gain 71 mm 51 mm 32 mm 13 mm

21 18 •§, 15 c S 12 §

96 3 0

1530

1540

1550

1560

Wavelength (nm)

Fig. 6.9 Gain vs. wavelength for a 32 mm long Er-doped fiber with signal input power of -30 dBm at different pump powers (a). Gain and noise figure vs. wavelength for different fiber lengths with signal input power of-30 dBm and pump power of 244 mW (b).

6.2 Er3^-Doped Phosphate Glass Fiber Amplifiers

217

Fig. 6.10(b) shows gain saturation for 71 mm fiber at wavelengths of 1535 and 1550 nm for pump powers of 84 and 244 mW. The 3-dB saturation output powers at a 244 mW pump power are 7.7 dBm at 1535 nm and 12 dBm at 1550 nm. Higher saturation output power at 1550 nm compared with at 1535 nm for an 84 mW pump power is also observed. The difference in saturation power for the two wavelengths arises from the smaller cross section at 1550 than at 1535 nm. The saturation output power increases as pump power increases as a result of the three-level amplifier system of Er3+ ions.[16] 25 Pump Power (mW) • 244 » 191 T 147 • 84

20

t •

I

•

"

*

*

T

"

'

15 10

*•

*

5

f

0

O

-5

|

-10

' *

•J

-15 -20 -25

• * | *

-20

-15

-10

-5

0

5

10

Input/Output power (dBm) 242220-

• »

1535 nm 1550 nm

'

18-

•

16-

244 mW

1412-

''\j

10-

*A

"

84 mW

86-

••I

42-10

-5

*"'*.". _ , ! , ! , , 0

5

10

Output Power (dBm)

Fig. 6.10 Output vs. input signal power and internal gain vs. output signal power of a 71 mm long Er-doped fiber for different pump power at 1535 nm (a) and gain vs. output power of a 71 mm long Er-doped fiber for 84 and 244 mW pump powers at 1535 and 1550 nm(b).

218

6.2.4.2 Er

Glass Fibers for Optical Amplification

and Yb co-dopedfiber amplifier

For Yb3+ sensitization, phosphate glasses have decisive advantages over other glass types such as silicate because of low back energy transfer rates and low accumulative energy transfer rates in phosphate glasses.1281 Furthermore, the energy transfer efficiency from Yb3+ to Er3+ can be as high as 95% because of the large spectral overlap between Yb3+ emission spectrum (2F5/2-^ 2F7/2) and Er3+ absorption spectrum (4Iis/2~^ 4In/2)-[19] An obvious advantage of using Yb3+ sensitization is that it makes selection of pumping wavelength less critical because ytterbium ions exhibit not only a large absorption cross section but also a broad absorption band between 800 and 1100 nm.[19'56] It has been shown theoretically that Yb3+ to Er3+ energy transfer by cross relaxation provides an attractive indirect pumping mechanism that can strongly reduce the negative effects due to Er3+ concentration quenching, especially for high Er3+ concentration amplifiers.[57] The basic parameters for the Er3+-Yb3+-codoped phosphate fiber are very similar to that of Er3+ only doped fiber as described above. The core region of the fiber was uniformly doped with Er3+ concentration of 3 wt% and Yb3+ concentration of 2 wt%. The Er3+ lifetime at low pump power was measured to be 6.9 ms in bulk sample of the core glass. The measured small signal internal gain and noise figure at 1535 nm as a function of 980 nm pump power for fibers of lengths 3.6, 4.4, and 5.55 cm are shown in Fig. 6.11. The pump power indicated is the multiplexed output power from a 980 nm and a 975 nm laser diodes with a power ratio of 1.3. The multiplexing of the two pump laser diodes is meant to get higher pumping power. An input signal power of-30 dBm was used in the gain measurement. The maximum pump power available was 224 mW. At the maximum pump power, an internal gain of 18 dB was obtained from the 3.6-cm-long fiber, equivalent to an internal gain per unit length of 5 dB/cm. The net gain per unit length of the fiber is 4.4 dB/cm. The gain per unit length is significantly higher than that of Er3+ only doped fiber, which was ~3 dB/cm.[58] The noise figure shown in the figure is based on the internal gain. The corresponding noise figure at 224 mW pumping power for the 3.6-cm-long fiber is 3.8 dB, while it is about 1 dB higher

6.2 Er "* -Doped Phosphate Glass Fiber Amplifiers

219

when net gain is used. The gain curve shows that the gain can be further increased with increased pump power. Higher gain at the maximum pump power can be realized by increasing the fiber length, e.g. to 4.4 cm or 5.55 cm as also shown in Fig. 6.11.

16

20-

W - * >^g£-*~-

1510-

14

^ 12 ^

Intern alGa

— ^ - 5.55cm 5-

—•— 3.60cm

-

0-

10 iz1

1

-5-

/XA v _/ > - - ^ ^- t £ £ 2 - -

-10-15-

8

i

Z

6 4

-20-

0

50

100

150

200

250

Pump Power (mW)

Fig. 6.11 Small signal internal gain and noise figure vs. pump power for different fiber lengths at 1535 nm for signal input power of-30 dBm.

However, noise figure will be increased slightly because the same inversion level cannot be maintained through longer fibers. The optimized fiber length for maximum gain depends on the pumping power. The small signal gain spectra are shown in Fig. 6.12 for different fiber lengths at pump power of 224 mW. Similar gain spectra were reported in [58]. However, shorter fibers were used here, leading to possibly more compact integrated devices based on these Er3+-Yb3+doped phosphate fiber amplifiers. Fig. 6.13 shows the gain saturation characteristics of the Er3+-Yb3+doped phosphate fiber amplifiers for different lengths of fiber.

Glass Fibers for Optical Amplification

220

• A

_

—*— 3.60 cm —*— 4.40 cm —*— 5.50 cm

7

0\

1540

\"N-^ \ ! ^ \ ' \ ^ " \ .

1550

. -

1560

Signal Wavelength (nm)

Fig. 6.12 Internal gain spectra for different fiber lengths with signal input power of-30 dBm and pump power of 224 mW.

-25 24-

22 20

-15

-20 -3.60 -4.40 -5.55 -3.60 -4.40 -5.55

cm cm cm cm cm cm

-10

1535nm 1535nm 1535nm 1550 nm 1550nm 1550nm

18

—T-T-T-

*

•

*

.

T

,

'-V

"

T

x\

NL

16

\

1412 10

0-c3- o ^ n ^ L h - O - D - ^ n - D -

'-D-CL, Tk

A-^-A-A-fl~A-">-A-A-A-A.-A_a_

-20

Q,

f

-15

Output signal power (dBm)

Fig. 6.13 Internal gain vs. output signal power for different fiber lengths at both 1535 and 1550 nm at pump power of 224 mW.

Saturation behaviors for both 1535 and 1550 nm signals are shown. The 980 nm pump power used was 224 mW. The output 1535 nm signal powers at 3 dB compression are measured to be 5.6, 8.1, and 10.9 dBm for 5.55-, 4.4-, and 3.6-cm-long fiber, respectively. Although all three

6.2 Er +-Doped Phosphate Glass Fiber Amplifiers

221

1550 nm saturation curves have not reached 3 dB compression in this measurement, it can be seen that the saturation output powers are higher for the signal at 1550 nm because of its lower cross section.[16] 6.2.4.3 Multi-mode pumped fiber amplifiers Fig. 6.14 shows the layout of an amplifier in a reflective geometry by using a circulator. Pump light from a multimode 50-um broad-area laser diode is coupled into the first cladding of a passive double-clad fiber (DCF) through a proprietary coupler.[59] The DCF is then fusion-spliced to an 8-cm erbium-doped phosphate fiber (EDPF). Signal is launched from port 1 of the circulator and directed to EDPF through port 2. A dielectric mirror is coated on the tip of the DCF fusion-spliced to the other side of the EDPF and is used to highly reflect both the pump and the signal. Most of the residual pump is immediately dumped when it reaches the SMF and further reduced to -60dBm level by the built-in isolator in the circulator en route to the output of the amplifier. The inset in Fig. 6.14 shows a cross-sectional view of the EDPF. As shown in the figure, the core, the 1st and the 2nd cladding are all axially symmetric. The numerical apertures of the core and the 1st cladding were calculated to be 0.145 at 1550 nm and 0.24 at 980 nm, respectively. bUFJh^ Cross sectional view _Core — 1st cladding

Output Fig. 6.14 Schematics of the multimode pumped double pass amplifier. SMF: Corning single-mode fiber SMF-28; LD: laser diode; C: proprietary pump coupler; DCF: doubleclad fiber; EDPF: erbium-doped phosphate fiber; DM: dielectric mirror.

222

Glass Fibers for Optical Amplification

Amplifier performance was evaluated experimentally using fibers with a variety of lengths. Excluding the circulator and the dielectric mirror, the single-pass insertion loss with this EDPF was measured at 1310 nm to be 3 dB. Almost 1/3 of this insertion loss was from the propagation loss of the EDPF. Fig. 6.15 illustrates the measured and simulated amplifier gain spectra from the 8-cm-long EDPF at signal input powers of-30 dBm, -10 dBm and 0 dBm with a 1W pump power from the broad-area laser diode. A peak gain of 41 dB was achieved at 1535 nm with -30 dBm input signal. At the -30 dBm input signal level, the noise figures at 1530 nm, 1535 nm, 1550 nm and 1565 nm were measured to be 6.3 dB, 6.1 dB, 5.3 dB, and 4.8 dB, respectively. Fig. 6.15 shows that greater than 15dBm output power could be delivered over the whole C-band when the input power was 0 dBm. The output power increased to 17.5 dBm when the pump power increased to 1.5 W. The saturation output power (3 dB compression) at 1535 nm, 1550nm and 1565 nm are 11 dBm, 12.5 dBm and 14 dBm, respectively for the 8-cm-long fiber excited with 1W pump power. The output saturation power is 2 dB higher when the fiber is excited with 1.5 W pump power. •

1525

Pin=-30 dBm, exp. Pin=-30dBm, simulation

1535

1545 1555 1565 Wavelength (nm) Fig. 6.15 Measured and simulated gain spectra with various input signal powers.

The simulated gain performance agrees closely with the experimental results as indicated in Fig. 6.15. According to the modeling results, the core in the 8-cm EDPF absorbed approximately 15% of the multi-mode pump. The model also showed that such high absorption in such a short fiber was caused by the high pump absorption coefficient of the core.

6.2 Er * -Doped Phosphate Glass Fiber Amplifiers

223

Furthermore, the model indicated that an absorption efficiency of 30% is achievable with optimized EDPF design in terms of doping concentrations and fiber geometry. Fig. 6.16 illustrates the net gain and noise figure versus the pump power at 1550nm for different fiber lengths. The input signal power was -30 dBm. Fig. 6.16(a) indicates that the optimum fiber length is around 8 cm. Fig. 6.16(b) shows the gain saturation at 1530 nm, 1535 nm, 1550 nm, and 1565 nm for the 8-cm-long fiber excited with 1 W pump power. 45 23 40 —«G—L=7 cm — 0 » L = 8 cm *L=9 cm X L=7cm NF L=8cm NF A L=9cm NF

•

QHHQ H • 400

700

1000

1300

'1 "I

•

35

O1530nm • 1535im. A1550nm • 1565 nm

130 •a 25 k> ^ 0A O A O*
"i

0

*0A

• ^

L=8 cm

7 1600

Pump power (mW)

(a)

-10 0 10 Output power (dBm)

20

(b)

Fig. 6.16 (a) Net gain and noise figure versus the pump power at 1550nm for different fiber lengths, (b) Gain saturation for the 8-cm-long fiber excited with 1W pump power.

The saturation output power (3dB compression) at 1535 nm, 1550 nm and 1565 nm are 11 dBm, 12.5 dBm and 14 dBm, respectively. The output saturation power is 2 dB higher when the fiber is excited with 1.5 W pump power.

References 1. C.J. Koester, E. Snitzer, Appl. Opt., 3 (1964) 1182. 2. S.B. Poole, D.N. Payne, M.E. Fermann, Electron. Lett., 21 (1985) 737. 3. R.J. Mears, L. Reekie, S.B. Poole, and D.N. Payne, Elect. Lett., LT-4 (1986) 870. 4. M. Nkazawa, Y. Kimura, K. Suzuki, Appl. Phys. Lett., 54 (1989) 295.

224

Glass Fibers for Optical Amplification

5. M.L. Dakss, W.J. Miniscalo, IEEE Photon. Technol. Lett., 2 (1990) 650. 6.Y. Ohishi, T. Kanamori, T. Nishi, S. Takahashi, IEEE Photon.Technol. Lett., 3 (1991) 715. 7. H. Yajima, S. Kawase, and Y. Sekimoto, Appl. Phys. Lett., 21 (1972) 407. 8. Y. Jaouen, L. du Mouza, D. Barbier, J.M. Delavaux, and P. Bruno, IEEE Photon. Tech. Lett., 11(1999)1105. 9. S. Jiang, T. Luo, B.C. Hwang, G. Nunzi-Conti, M. Myers, D. Rhonehouse, S. Honkanen, and N. Peyghambarian, Opt. Eng., 37 (1998) 3282. 10. G.N. van den Hoven, R.J.I.M. Koper, A. Polman, et al., Appl. Phys. Lett., 68 (1996) 1886. 11. T. Kitagawa, K. Hattori, K. Shuto, et al., Electron. Lett., 128 (1992) 1818. 12. K. Shuto, K. Hattori, T. Kitagawa, et al., Elect. Lett., 29 (1993) 139. 13. J. Shmulovich, A. J. Bruce, G. lenz, et al., OFC 99, Postdeadline paper, PD42-1, 1999. 14. X. Orignac, D. Barbier, X.M. Du, et al., Optical Materials, 12 (1999) 1. 15. E. Desurvire, Erbium-doped fiber amplifiers: principles and applications, WileyInterscience, New York, 1994. 16. P.C. Becker, N.A. Olsson, J.R. Simpson, Erbium-Doped Fiber Amplifiers: Fundamentals and Technology, Academic Press, 1999. 17. Shibin Jiang, Tao Luo, Bor-Chyuan Hwang, et al., Journal of Non-crystalline Solids, 263&264 (2000) 364. 18. Shibin Jiang, Bor-Chyuan Hwang, Tao Luo, et al., Postdeadline papers-5, Optical Commumication Conference, March 7-10, 2000, Baltimore, Maryland. 19. B. Hwang, Shibin Jiang, T. Luo,, J. Opt. Soc. Am. B, 17 (2000) 833. 20. Y. Hu, S. Jiang, T. Luo, et al., IEEE Photon. Tech. Lett. 13 (2001) 657. 21. Shibin Jiang, S. B. Mendes, G. Nunzi-Conti, et al., Opt. Eng. 42 (2003) 2817. 22. K. Seneschal, F. Smektala, Shibin Jiang, et al., J. Non-Cryst. Solids 324 (2003) 179. 23. K. Seneschal, F. Smektala, B. Bureau, et al., Mat. Res. Bull. 40 (2005) 1433. 24. Y. Hu, Shibin Jiang, G. Sorbello, et al., J. of the Opt. Soc. of Am. B, 18 (2001) 1928. 25. S. Jiang, M. J. Myers, and N. Peyghambarian, J. Non-Cryst. Solids 239 (1998) 143. 26. J. C. Wright, in Radiationless Processes in Molecules and Condensed Phases, F. K. Fong, ed. (Springer-Verlag, Berlin, 1976), Chap. 4. 27. T. Kitagawa, K. Hattori, K. Shuto, et al., in Optical Amplifiers and Their Applications,'''' Vol. 17 of 1992 OSA Technical Digest Series (Optical Society of America, Washington, D.C., 1992); postdeadline paper PD1. 28. V. P. Gapontsev, S. M. Matitsin, A. A. Isineev, and V. B. Kravchenko, Opt. Laser Technol. 14(1989)189. 29. E. F. Artemev, A. G. Murzin, Y. K. Federov, and V. A. Fromzel, Sov. J. Quantum Electron. 11 (1981) 1266. 30. C. Lester, A. Bjarklev, T. Rasmussen, and P. G. Dinesen, J. Lightwave Techol. 13 (1995)740. 31. A. Shooshtari, T. Touam, S. I. Najafi, et al., Opt. Quantum Electron. 30 (1998) 249. 32. E. Townsend, W. L. Barnes, and K. P. Jedrezejewski, Electron. Lett. 27 (1991) 1958. 33. J. G. Edwards and J. N. Sandoe, J. Phys. D 7 (1974) 1078. 34. J. E. Roman, P. Camy, M. Hempstead, et al., Electron. Lett., 31 (1995) 1345.

References

225

35. T. Ohtsuki, S. Honkanen, S. I. Najafi, and N. Peyghambarian, J. Opt. Soc. Am. B 14 (1997) 1838. 36. W. J. Miniscalco, in Rare Earth Doped Fiber Lasers and Amplifiers, M. J. F. Digonnet, ed. (Marcel Dekker, New York, 1993), Chap. 2. 37. E. Snoeks, G. N. V. den Hoven, A. Polman, et al, J. Opt. Soc. Am. B 12 (1995) 1468. 38. G. N. V. den Hoven, E. Snoeks, A. Polman, et al., J. Appl. Phys. 79 (1996) 1258. 39. M. P. Hehlen, N. J. Cockroft, T. R. Gosnell, et al., Opt. Lett. 22 (1997) 772. 40. P. Blixt, J. Nilsson, T. Carinas, and B. Jaskorzynska, IEEE Photonics Technol. Lett. 3 (1991)996. 41. S. Taccheo, G. Sorbello, S. Longhi, and P. Laporta, Opt. Quan. Electron. 31 (1999) 249. 42. V. P. Gapontsev, S. M. Matitsin, A. A. Isineev, and V. B. Kravchenko, Opt. Laser Technol. 14(1989)189. 43. W. Ryba-Romanowski, S. Golab, L. Cichosz, and BJezowska-Trzebiatowska, J. Non-Cryst. Solids 105 (1988) 295. 44. B. Simondi-Teisseire, B. Viana, D. Vivien, and A. M. Lejus, Opt. Mater. 6 (1996) 267. 45. V. A. Lebedev, V. F. Pisarenko, Y. M. Chuev, et al., J. Lumin., 72-74 (1997) 942. 46. J. E. Roman, P. Camy, M. Hempstead, et al., Electron. Lett., 31 (1995) 1345. 47. D. Massiot, B. Cote, F. Taulelle and J.P. Couture, Application of NMR Spectroscopy to cement Science, edited by Colombet and Grimmer (Gordon and Breach Publishers, 1994)153. 48. C. Magnenet, D. Massiot, I. Klur and J.P. coutures, J. of Mat. Sci., 35 (2000) 115. 49. D. Massiot, H. Thiele, A. Germanus, Bruker Rep., 140 (1994) 43. 50. R.J. Kirkpatrick, R.K. Brow, Solid State NMR„ 5 (1995) 9. 51. R.K. Brow, J. Non-Cryst. Solids,, 263&264 (2000) 1. 52. R.K. Brow, R.J. Kirkpatrick, G.L. Turner, J. Am. Ceram. Soc, 76 (1993) 919. 53. H. Sholze, "Glass : Nature, Structure and Properties", Springer-Verlag, New York (1990). 54. Shibin Jiang, Jiafu Wang, US Patent 6,705,771. 55. W. J. Miniscalco, in Rare earth doped fiber lasers and amplifiers, ed. M. J. F. Digonnet, Marcel Dekker, 1993. 56. X. Zhou and H. Toratani, Phys. Rev. B, 52 (1995) 15889. 57. F. Di Pasquale, and M. Federighi, IEEE J. Quant. Elect, 30 (1994) 2127. 58. B. C. Hwang, S. Jiang, T. Luo, et al., IEEE Photon. Tech. Lett., 13 (2001) 197. 59. Y. Kaneda, S. Mendes, and Shibin Jiang, U.S Patent 6,529,318.

Chapter 7

Glass Fibers for High Power Lasers

7.1 Introduction of Optical Fibers The phenomenon of total internal reflection, responding for guiding of light in optical fibers, has been well known for two hundreds years.[I] Although glass fibers were made in the 1920s,[2] their use became practical only in the 1950s, when the use of a cladding layer led to considerable improvement in their guiding characteristics. Before 1970, optical fibers were used mainly for medical imaging over short distance. Their use for communication purposes was considered impractical at that time because of high loss (-1000 dB/km). However, the situation changed drastically in 1970's, following an earlier suggestion/31 the loss of optical fibers was reduced to about 20 dB/km. Nowadays the loss of optical fibers is only 0.1 dB/km at the 1.55 |am optical communication window. The availability of low-loss fibers led to a revolution in the field of optical fiber communications. Fiber laser has a history almost as long as that of the laser itself. Since its invention in 1963 by Elias Snitzer, almost two decades of development on fiber lasers were spent before the first commercial device appeared on the market in the late 1980's. These lasers used single-mode diode pumping, emitted a few tens of milliwatts, and attracted users because of their large gains and the feasibility of singlemode continuous-wave (CW) lasing for many transitions of rare-earth ions not achievable in the more-usual crystal-laser version. The most well-known application of fiber-laser technology is 1550-nm erbiumdoped fiber amplifiers/41 227

228

Glass Fibers for High Power Lasers

For many laser applications, however, high optical power in the order of Watts rather than milliwatts is required. The jump to Watt-level fiberlaser output occurred in 1990, when a 4-W erbium-doped fiber laser was reported. This development laid the groundwork for ten-watt and higher single-mode fiber lasers suitable for micromachining and other applications. Output powers of more than 1 kW have been achieved for diode pumped systems in CW-operation with diffraction limited beam quality based on solid core rare-earth doped, low numerical aperture, large mode-area-fibers (LMA-fibers). The capability of fiber systems for extracting pulse energies as high as several milli-Joules at ns-pulse durations has been demonstrated most recently in Yb-doped fibers. Additional sufficient pulse stretching in the time domain using the chirped pulse amplification technique enables reduction of nonlinear pulse distortions in the fiber and extracting energies in the order of the saturation fluence limit of rare-earth doped fibers even in ultrashort pulse operation. A further power scaling of fiber lasers and amplifiers is possible by using novel highly doped air-clad photonic crystal fibers with increased mode field diameters of the active core. This novel class of optical fibers is based on a wavelength-scale morphological microstructure along the fiber, resulting in attractive guiding properties compared to conventional solid core fibers. For such an air-micro structured single-mode fiber with a mode-field diameter of 35 urn, a diffraction-limited output power in air-cooled operation of more than 3 kW can be achieved, which is limited by optical damage, thermal load and nonlinearity. Water-cooled air-micro-structured fibers will allow output powers to go up to 10 kW from a single fiber. Even in ultrashort pulse operation, pulse energies of more than 200 uJ at repetition rates up to 1 MHz are accessible. These parameters will allow for novel applications in science and technology. 7.1.1 Reflection of light in optical fibers Generally, an optical fiber consists of a core surrounded by a lower refractive index cladding, at the core-cladding interface of a step-index fiber, the refracted angle 6r of the ray for an incident angle 0i is given

7.1 Introduction of Optical Fibers

229

by: n0 sin 0. = nx sin 0r,

(7.1)

where nQ and nx are the refractive indices of the fiber core and cladding, respectively. For angles larger than a critical angle <j)c, defined by:

sin £ = « , / « „ .

(7.2)

Total internal reflection occurs at the core-cladding interface. Since such reflections occur throughout the fiber length, all rays with <j> > <j)c remain confined in the fiber core. This is the basic mechanism behind light confinement in optical fibers. From Eq. (7.2), the maximum angle that an incident ray can propagate in the fiber core satisfying total internal reflection condition is « 0 sin(^) = (n 1 2 -n 2 2 ) 1/2 .

(7.3)

Numerical aperture (N.A.) of the fiber, which represents the lightgathering capacity of an optical fiber. For n{ ~ n2, NA is defined as 'S.A. = nl[2(nl-n2)/nl]U2.

(7.4)

When the optical fiber is curved, the reflective angle ^ of an optical ray from the core boundary will be changed to (f> = ±\l g, where g is the radius of curvature. The guidance condition 9 <(PC does not satisfy any more. The rays with the refractive angle larger than the critical angle will be leaked outside the fiber and become loss to optical field. Because this loss comes from a bend of optical fiber, it is named as bending loss. Other operations, such as strain, temperature gradient and magnetic field, can also introduce losses into the optical fibers. These loss mechanisms provide opportunities to manage the propagation characteristics of optical fibers. The dispersion effects can be explained on the basis of behavior of group velocities of the guided modes in the optical fiber. Group velocity is the velocity at which the energy in a particular mode travels along the fiber. The group velocity is given by _ da _ dX da r-j ^\ Vs ~~dp~~dp~dl' Thus, the group velocity is different from the phase velocity in an optical fiber.

230

Glass Fibers for High Power Lasers

The signal losses are defined as the ratio of optical input power (pj) to the optical output power (p0). Optical input power is the power injected into the fiber from an optical source. Optical output power is the power received at the fiber end with an optical detector. The following equation defines signal attenuation per unit of length: f

Loss = | — | log

A

\

(7.6)

Po

Signal attenuation or loss is a log relationship. Length (L) is expressed in kilometers. Therefore, the unit of attenuation is decibels/kilometer (dB/km). 7.1.2 Numerical aperture of optical fibers Fig. 7.1 depicts a section of a clad cylindrical fiber showing the core with refractive index of ni and the cladding with index of n2. Also shown is a light ray entering the end of the fiber at angle 0, reflecting from the fiber core/cladding interface. However, if angle 6 becomes too large, the light will not reflect at the interface as we described in the previous section. This angle 6 is the critical angle. TYPICAL 50 U FIBER CORE, »,

INTERFACE CLAD, » 3 Fig. 7.1 Numerical aperture of an optical fiber.

7.1 Introduction of Optical Fibers

231

The N.A. is a universal description of light propagation angle. For example, the N.A. of a light source is given by N.A. = sin a, (7.7) where a is the project angle of the light source. The brightness of light source is inversely proportional to the N.A. of the light source. The light from a source with a N.A. can be collected totally by an optical fiber with the N.A. larger than that of the light source. Note that the capacity of an optical fiber collecting light is determined by the product of the N.A. and the fiber diameter, not the N.A. itself only. The lens or a focal system can also have the N.A. The F number equivalent of the N.A. of a lens is calculated as follows: F „ =——. "umber

(7.8)

2N.A.

7.1.3 Propagation in optical fibers Electromagnetic field propagation in optical fibers is governed by Maxwell's equation. For a non-conducting medium without free charges, these equations provide a general wave equation for wave propagation in optical fibers, ^

_

_

VxVxE =

\ d2E

d2P

-slF-M°W>

(7 9)

-

where the speed of light in vacuum is defined as usual by c = (jU0e0)~ The solution of the wave equation that satisfies the appropriate boundary conditions and has the property that its spatial distribution does not change with propagation is called fiber modes. The fiber modes can be classified as guided modes, leaky modes, and radiation modes. The total number of guided modes in an optical fiber is governed by the normalized frequency or simply the V parameter, V = k0a(n2-n22)U2
(7.10)

where um is the mth root of the first order Bessel functions Jm (u) . The first root is 2.405. When an optical fiber is designed with 0
(7.11)

Glass Fibers for High Power Lasers

232

the optical fiber supports only the fundamental HE 11 mode. This is the single-mode condition. Otherwise, the optical fibers will support a number of guided modes and are called multimode fibers. The general spatial distribution of an optical field in the fibers is: E

z = Jm (k*P) exp(i/H0 exp(z/? z ) .

(7.12)

The propagation constant /? and kx satisfy the relationship

kl=n\kl-p\

(7.13)

Obviously, the field distribution in optical fibers is different from that in free space. However, the fundamental mode in optical fiber is often approximated by a Gaussian distribution with an effective field size or usually referred to as the spot size. Ez = A exp(-/? 2 / w2) exp(//? z ) .

(7.14)

Note that the approximation to a Gaussian distribution is only valid for the fundamental mode. High order modes cannot be described by a high order Gaussian beam in most cases. 7.1.4 Double clad fibers Fibers widely used in the telecommunication are single-mode fibers, which consist of one cladding and one core. The core diameter of singlemode fibers is limited by the single-mode condition (see Eq. (7.10)), usually 5—10 urn. For the applications in optical communications, the fiber core can transport up to giga-byte signal over a distance of hundreds of kilometers. However, for the applications of high power fiber lasers, the core diameter is too small to support high pump power. To increase the pump power capacity of a single-mode fiber, double clad fiber is constructed. As its name shown, a double clad fiber consists of two claddings. The inner cladding is a multi-mode fiber, used to retain much more pump power while the core still satisfies single-mode condition to supply single-mode laser output. The refractive index profile of a double clad fiber was shown in Fig. 1.10 in Chapter 1 of this book. The refractive index of the fiber core is larger than that of inner claddings, while the inner cladding refractive index is larger than that of outer

7. / Introduction of Optical Fibers

233

claddings. In contrast to the fiber core of 5—10 urn with the N.A. of ~ 0.1, the inner cladding usually has a size of several hundred (j,m with a N.A. of- 0.4. The pump power capacity of double clad fibers is thus increased by hundreds times Inside inner cladding, optical rays are reflected from cladding boundary with cylindrical symmetry. Most of them are trapped in the neighbor of cladding boundary and do not pass through the doped core. In such a way, the pump beam absorption is very small. Destroying this symmetry will increase the absorption efficiency. Several types of inner claddings, e.g., D-type, rectangle, triangle, hexagonal and other irregular shapes, are proposed. In practice, D-type and rectangular inner claddings are widely used. A D-type inner cladding with the same diameter as the circular inner cladding can double the pump absorption. The pump beam absorption in a double clad fiber can be approximately given by P(z) = i»exp(-i7a^-z),

(7.15)

where S; and S0 are the cross-sectional area of the inner cladding and fiber core, respectively, a is the absorption coefficient in the bulk material which is used to fabricated the fiber core and TJ is a coefficient determined by the shape of the inner cladding. The absorption coefficient is proportional to the ratio of cross-sectional area between the inner cladding and fiber core. The thicker the inner cladding, the lower the absorption coefficient is. The double clad fiber design has become standard because the large inner cladding can efficiently collect pump power, while the smaller core concentrates laser power in a smaller volume for a higher-quality output beam. The geometry of the inner cladding is chosen for efficient pumping of the core. Similar structures are used in pulsed and continuous-wave lasers. Practically the fiber is coiled in one or more loops. The design details depend on the applications and power requirements.

234

Glass Fibers for High Power Lasers

7.2 Fabrication and Materials 7.2.1 Fabrication of optical fibers How to fabricate low loss fiber is a long-history and still interesting topic. Conversional optical fibers are made by drawing a preform which has a structure similar to that of the required fibers. There is a variety of methods for preparing preforms. They are MCVD (modified chemical vapor deposition), OVD (outside vapor deposition), VAD (vapor axial deposition), etc. The most popular method is MCVD. In MCVD fabrication, a silica film is deposited on the inside surface of a fused silica tube. This tube is put on a glass working lathe and is rotated. Raw halide material vapor carried by oxygen gas is introduced into the tube at a constant speed. A flame heats the tube from the outside to about 1600°C. The halide vapor is oxidized to fine glass particle and deposited on the inner surface of the tube. The heated zone can be kept at a constant temperature by controlling the flow rates of oxygen and hydrogen. The raw materials used for high silica glass are silicon, germanium and boron chlorides and phosphorus chlorate. To reduce the refractive index to be lower than that of silica, dopant such as BCI3 is used; for higher refractive indices, GeCl4 or POCl3 are used. The flow rate is controlled by a gas supply system and the mixing ratio of halide vapor is controlled to a programmed value. By changing GeCLt concentration, the refractive index of each glass layer can be controlled. The metal halides (expect BC14) remain in the liquid state at less than 50 °C giving porous material. Dehydration and consolidation are the next steps. The reacted gas is extracted from the silica tube. After a sufficient thick film is deposited, the tube is collapsed so that the deposited layers form a high index core and the tube provides the low index cladding. Silicon dioxide, or pure silica, is usually obtained in the form of small particles (about 0.1 um) called soot. This soot is deposited on the target rod or tube. The deposition of the silica soot, layer upon layer, forms a homogeneous transparent cladding material. To change the value of a cladding's refractive index, some dopants are used. For example, fluorine (F) is used to decrease the cladding's refractive index in a depressedcladding configuration.

7.2 Fabrication and Materials

235

The soot material for the fiber core is made by mixing three gases— SiCU, GeCl4, and 0 2 — which results in a mixture of SiC>2 and GeC>2. The degree of doping is controlled by simply changing the amount of GeCl4 gas added to the mixture. The same principle is used for doping other materials. The fabrication procedure of a double clad fiber is similar to that of the single clad fiber. The difference is only in preparing the prefrom. The double clad preform is fabricated by introducing impurities into thin layers of glass tube so that the refractive index in the preform resembles a set of concentric rings, which shows a programming value as the double clad fibers. Preparing a preform with pure silica free from OH- hydroxyl ion is important to obtain low loss optical fibers. After using additional processes to remove residual OH- hydroxyl ion from prepared perform, fibers with attenuation lower than 0.1 dB/km can be drawn. 7.2.2 Materials Commercial optical fibers are manufactured mainly by glass or plastics. The silica glass is the most important material in fabricating telecommunication optical fibers. More than 95% telecommunication optical fibers are made of silica glass. The silica system is a mixture of silicon dioxide, Si0 2 and other metal oxides. To generate a difference in refractive index between the core and cladding, various dopants have been used to increase the refractive index of silica including titanium, aluminium, germanium, and phosphorus oxides. Silica glass has excellent thermal properties with an extremely low coefficient of expansion 0.55x106 cm/°C (0-300°C). Another related property is its high resistance to thermal shock. Thin sections can be heated and cooled rapidly without cracking. In silica glass, the wavelengths of operation range is from 700 to 1600 nm. The wavelength of operation is between two intrinsic absorption regions. The first region is the ultraviolet region (below 400-nm wavelength). The second region is the infrared region (above 2000-nm wavelength). The main cause of intrinsic absorption in the infrared

236

Glass Fibers for High Power Lasers

region is the characteristic vibration frequency of atomic bonds. Meanwhile, intrinsic absorption in the ultraviolet region is caused by electronic absorption bands. Chalcogenide glasses have a longer cut-off wavelength than oxide and fluride glasses. They are solid solutions of metal sulphides, selemides and telluride of arsenic, germanium and antimony. Chalcogenide glasses are melted directly in quartz ampoules using purified chemicals through distillation/sublimation. Typical melting temperature ranges from 600 to 900 °C. They have a stable vitreous state and a wide range of transmitting wavelength. Fluoride glass is made from beryllium, zirconium, mercury, aluminum and barium fluorides. They are promising materials for use in the infrared wavelength range up to 2 um because of very low loss in this region. Phosphate glass is made from P 2 0 5 , Si0 2 and Ge0 2 , B 2 0 3 , Al 2 0 3 , and other alkali metal oxide. Phosphate glass is almost the same to the silica glass, in which some of Si0 2 has been replaced with P 2 0 5 . It has a low melting point and is transparent over a wide range of wavelengths. It has a high index of refraction so that the fiber has small bending radius. Fluorozirconate glass comprises pre-selected molar proportions of ZrF4, ThF4 and BaF2 that are essentially free of hydroxide and oxide impurities. Fluorozirconate glass has adsorption bands at 2.9 |a,m wavelength. Casting from a melt in the presence of a mixture of He and CCU to thereby compensate for fluorine deficiencies, which cause a formation of the color-center of Fluorozirconate glass, by the introduction of a chlorine dopant. Fluorozirconate glass is continuously transparent to light from the near UV all the way to the near infrared. ZBLAN (zirconium, barium, lanthanium, sodium) glass is a standard fluozirconate glass system composition (ZrF4-BaF2-LaF3-AlF3-NaF). ZBLAN glass is one of many fluoride glass composition used to make fluoride glass and fiber. The ZBLAN glass is the most promising of these materials since its fiber-drawing region lies on the edge, or possibly just outside its crystallization region. ZBLAN Fluoride fiber has a more specific composition that uses zirconium. By varying the composition, optical or mechanical properties of the fiber can be optimized depending.

7.2 Fabrication and Materials

237

7.2.3 Rare-earth doped fibers Optical fibers play an important role in carrying optical signals, they also act as host medium in fiber lasers and amplifiers when rare-earth elements are doped into the fiber core. Neodymium (Nd 3+ ) and Erbium (Er 3+ ) are the most common rare earth elements used as dopants. Other ions like Ce 3+ , Pr3+, Tb 3+ , Dy 3+ , H 0 3 + , Tm 3+ and Yb 3+ can be used in host materials like heavy metal fluoride glass fibers (ZBLAN, ZBLANPb, ZBLAB P). All these fibers can be fabricated by rare earth vapor phase deposition at high pressure, or by impregnation and diffusion of rare earth doped solutions followed by drying. Generally, the concentration of rare earth materials is less than those of the classical dopants (300-1000 ppm) for the highest gain. The most common rare earth dopant in silica glass fiber is the Yb 3+ . The relevant energy level diagram of Yb 3+ is very simple and consists of the 2 F 7 / 2 ground state and 2 F 5/2 excited state manifolds separated by about 10000 cm"1. The laser wavelength is around 1.05-1.1 um in the silica glass. The emission spectrum for 1000 ppm Yb 3+ doping concentration silica glass sample is shown in Fig. 7.2. cm" g 11630 f 11000 e 10260

H

5Q

1

J

915nm

c

975nm

D •

B •

•

1r

1490 1060 600

F7/2

'' Fig. 7.2 Energy-level structure of Yb3+ in fiber.

The laser transition is 2 F 5/2 to 2 F 7/2 with the terminal level 623 cm"1 above the ground state. The thermal energy at room temperature is 200 cm"1, therefore the terminal state is thermally populated which makes

238

Glass Fibers for High Power Lasers

Yb a quasi three level system. In comparison, the terminal laser level in Nd3+ is about 2000 cm"1 above the ground state. Being a quasi three-level system, the fiber should be pumped intensely to overcome the reabsorption problem. At room temperature the thermal population of the lower laser is about 5%. The absorbed pump power per volume needed to achieve and maintain transparency at the laser wavelength is / = fantVp/Tf, where fa is the fraction of the total ion density n, occupying the lower laser level, h vp is the energy per pump photon, ty is the lifetime of the upper level. With fa= 0.055, n,= 1.38X1020 cm"3 at 1000 ppm doping, hvp= 2.11 xlO'9 J and tj=0.95 ms the absorbed pump power needed to reach inversion is 1.7 KW/cm3. Of course, a higher power density is required to overcome optical loss and reach laser threshold, and for an efficient operation the laser has to be pumped about 5-6 times above threshold. Typically, in this laser, small volumes of material are pumped at the order of 10 KW/cm3. Optical fibers with small core and long length are suitable for Yb3+ laser emission. Yb3+ laser performance is strongly dependent on the temperature and excellent cooling characteristics of optical fiber lasers can reduce the thermal degradation and increase the conversion efficiency. Pumping of Yb-doped glass fiber around 915, 945 or 975 nm produces the smallest amount of heating compared to any other major laser system. Actually the pump in this material generates only about one-third of the heat compared to Nd-doped glass laser. The fractional thermal loading is around 11% for Yb-doped glass laser pumped at 945 nm and 32% for Nd-doped glass laser pumped at 808 nm. This substantially reduced thermal dissipation is a result of a very small energy difference between the photons of the pump and laser radiation. This quantum defect or Stokes shift, is 9% in Yb-doped glass vs. 24% in Nd-doped glass. The thermal load generated in a laser medium is of primacy concern for high-power applications. The reduced thermal heat load can potentially lead to higher-average-power system with better beam quality. Fig. 7.3 shows the fluorescence spectrum of Yb-doped fiber DCYDF-400, the corresponding characteristics of the china made double cladding fiber is shown in Table 7.1.

7.3 High Power Lasers Based on Rare-Earth Ions Doped Fibers

980

1000

1020

1040

1060

1080

1100

1120

1140

239

116

Wavelength (nm) Fig. 7.3 The fluorescence spectrum of Yb-doped fiber DCYDF-400. Table 7.1 Characteristics of two types china made double cladding fiber. Characteristics/Types Yb-concentration Core diameter Core NA Geometry Inner cladding diameter Inner cladding thickness Cladding numerical aperture Slope efficiency

DCYDF-400 6000 ppm 8~30um 0.06—0.20 D 400um 350um 0.38, 0.46 >75%

DCYDF-650 6000 ppm 25~50um 0.07-0.15 D 650um 600um 0.46 >75%

7.3 High Power Lasers Based on Rare-Earth Ions Doped Fibers 7.3.1 Principle Rare-earth doped silica fiber has been proven to be an excellent solution for all-optical network and has been extensively and reliably employed in the telecommunication area for many years. Silica fibers are also attractive for high power amplifiers and lasers. Silica based glasses have high damage thresholds. The surface-to-volume ratio of an optical fiber is high so that heat dissipation is straightforward, and the gain medium is incorporated in a waveguide so it is possible to maintain single mode

240

Glass Fibers for High Power Lasers

wave propagation. Fiber manufacturing technology allows producing long fibers, and fiber laser cavities can be tailored to provide the beam quality for almost all applications. Recently there have been significant developments in the area of cladding pumped double-clad doped-fiber for very high power fiber lasers and amplifiers. In this approach the fiber core is heavily doped with active ions (e.g. Yb, Er, Nd, Tm) and the silica-clad fiber is coated with low refractive index material; pump light is launched into the cladding of the optical fiber (typically in the range 125 [im to 650 urn diameter). The silica inner cladding and the low index coating give rise to a high numerical-aperture multimode waveguide for the pump light, allowing high-power multi-mode semiconductor diode lasers be used. This offers advantages both in terms of cost-per-Watt, and of the availability of much high-power high-brightness pump lasers. In operations, the pump light propagates in the undoped cladding of the optical fibre and is absorbed by the active ions in the core. The pump power into the inner clad can be expressed as P = 4BD2NA\

(7.16)

in which D2 and NA are the area and numerical aperture of the inner cladding respectively, B is the brightness of the pump source. To increase the maximum pump power coupled into the fiber, the inner cladding is designed with a high NA and a large area (typically in the range of 350 to 650 urn). From Eq. (7.16), in order to realize high-power output for a certain double-clad fiber, the most import thing is to have a high brightness pump source. The pump absorption efficiency changes with the inner cladding shape. The inner cladding shape is normally non-circular to let more pump power enter the fiber core. If the inner cladding is designed to be circular, then only few pump modes cross the doped core. In this case, the pump efficiency is low. However, the core can be offset from the center, or the inner cladding can be designed to have no-circular shape to improve the absorption efficiency. To optimize the pump absorption, various shapes of the inner cladding have been proposed. The most common shapes are described in Fig. 7.4. In which the inner cladding are circular (a), offset (b), square(c), rectangular (d), hexagonal (e), flower shape (f), D shape (g), "unstable cavity" shape (h). The absorption

7.3 High Power Lasers Based on Rare-Earth Ions Doped Fibers

241

characteristics are simulated for double-clad fibers with (a), (b), (g) and (h) inner cladding shapes by the 2-D ray tracing method.[5]

(c)

(d)

(e)

(f)

Fig. 7.4 Typical double clad fibers.

The pump light absorption efficiency is 10%, 50%, 80% and 98%, correspondingly. The results are shown in Fig. 7.5.

40

60

Times of reflection N Fig. 7.5 Variation of absorption efficiency versus different inner shape fibers.

A fundamental limit to power scaling of double-clad fiber lasers is the damage of fiber core end facet in building high-power fiber lasers. In the multimode regime (large core), it is relatively easy to get a high output power, but single-mode output is a great challenge due to the optical damage. Also, nonlinear optical processes in the fiber degrade the laser performance. Several important nonlinear processes that limit the output power and energy are stimulated Brillouin scattering (SBS), stimulated Raman scattering (SRS), and self-phase modulation. Their influences are determined by the pulse duration, spectral linewidth, and fiber length.

Glass Fibers for High Power Lasers

242

The simplest solution to overcome these problems is to increase the size of the core diameter. As shown in Fig. 7.6, a large dc0re correspond to larger V. when V>2.4, the beam quality is poor due to the propagation of higher order modes. Recently, many scientists pay attention to find a way to make use of the multimode power capabilities but remaining the single mode beam quality. 60

..y=8.o 20

... V=4.0 V=2.405 — i —

0.14

0.06

0.16

COTeNAj

Fig. 7.6 Doped core diameter Vs. Core NA (The wavelength is near l.Oum region). The size of dcore for single mode fiber is less than 20um with an NA,.,,,.,, 0.06.

7.3.2 Laser diodes and beam shaping The choice of pump source is very important in the overall performance of a high-power fiber laser. It influences the reliability and cooling methods, as well as the efficiency. Historically bulk solid state lasers have been widely used for high power applications. They can be easily pumped with lamps. But for the high-power fiber lasers, laser diodes are the best choice for the pumping. Fortunately, the developments of highpower pump lasers and low-loss rare-earth doped fibers make high power fiber lasers possible. The difficulty in high-brightness delivery of high-power diode laser beams stems from the geometries and structures of these devices. Thus, a high-power LDA has a broad-area light-emitting aperture of about 1 cm

7.3 High Power Lasers Based on Rare-Earth Ions Doped Fibers

243

x 1 |im. With such a configuration, some commercially available high power LDAs can provide 40 to 50 W of power. Much higher power can be achieved by layering high-power LDA bars in stacks. The raw output beam from an LDA is highly divergent and suffers from two asymmetries—astigmatism and an elliptical beam profile. The divergence angles are different in two axes, the so-called "fast axis" and "slow axis." Typically the fast axis divergence is about 40° full-width half-maximum (FWHM) while that of the slow-axis is about 10° FWHM. Sometimes the divergence can be as high as 50° x 15°. To improve the beam quality and brightness, sophisticated beam rearrangement mechanisms are normally used. Two typical examples are the step mirror approach used by the Fraunhofer Institute for Laser Technology (Aachen, Germany) and the two-reflector approach invented by researchers in the University of Southampton (Southampton, England). Both methods have been used commercially to provide fibercoupled laser-diode devices. A new efficient approach is developed by using groups of thin prisms for beam shaping at Shanghai Institute of Optics and Fine Mechanics (China patent No. ZL 03115584.7 by Qihong LOU). 7.3.3 End-pumping fiber laser Coupling the pump beam into the inner cladding of double-clad fiber through its ends by so-called end-pumping scheme is the simplest and most efficient way to pump double-clad fibers with high-power pump sources. In end-pumping configuration, a large inner cladding is required in order to accommodate the large pump beam of the high-power laser diode source. The laser diode pump source is either coupled to the double-clad fiber with bulk optics or fiber optics.[6] The bulk optics approach, in which the pump light is launched into the end of the fiber, is often used in laboratory applications. A major drawback is that one or both of the fiber ends are obstructed by the bulk optics used to launch the pump light. In addition, this approach lacks scalability (one fiber has only two ends) and is difficult to implement in a compact and rugged manner. However, most double-clad fibers made to date use a low-index

Glass Fibers for High Power Lasers

244

polymer as the outer clad material to achieve the desired high numerical aperture (NA 0.3-0.45). These polymers have much poorer thermal stability than glass. Under high power pump, the polymer near the innerouter clad interface can easily burn or gradually degrade. Because of any faulty steps in beam shaping and assembling, the high-power collimated pumping beam is not so good for pumping the double-clad fiber directly. A spatial filter can be used to improve the beam quality of the highpower pump light. In order to inject the pump light into the inner cladding with high coupling efficiency, a special aspheric lens is designed and fabricated. Benefiting from the diffraction limited performance of the aspheric lens, as well as the optical spatial filter, the focus spot and the cone angle of the pump light match well with the corresponding parameters of the double clad fiber. And the pump light can be safely and efficiently coupled into the inner-cladding. The experimental setup is shown in Fig. 7.7.

a

Yb-doped fiber

Pump source

Dichroic mirror Spatial filter

Collimation lens

Fig. 7.7 Experimental setup of fiber laser with one-end-pumping using spatial filter.

The output power is a function of fiber length when the injected pump power is fixed. Optimum fiber length of 20-m was achieved theoretically as shown in Fig. 7.8. The experiment results of 6-, 21-, and 52-m fibers are presented in Fig. 7.9. The experimental results are in agreement with the calculated ones. For one end pumping, the maximum laser output power of 20-m DCF is more than 200W at 1.1 um with a slope efficiency more than 69%. With two-end scheme, the output power was more than 440 Watts.[7S1

7.3 High Power Lasers Based on Rare-Earth Ions Doped Fibers

245

20 26 Fiber length L (m)

Fig. 7.8 Output power as a function of fiber length. 50

FiberLengh Slope Efficiency - • - 22m 67.1% - A - 53m 53.7% -A- 72m 52.6% - a - 6m 45.9%

40

30

a. 3 20-

/ & *

talK&. 20

30

40

50

Launched pump power (W) Fig. 7.9 Output power of different fiber lengths.

The fiber optics approach, or so-called fused fiber bundle (pump combiner), in which several multi-mode(MM) fibers are bundled together, fused and drawn into a taper, fusion spliced to a double-clad fiber, and then recoated with a low index polymer; pump light is launched into the double-clad fiber from individual diode lasers that are coupled to the MM fibers. Optionally, the fiber bundle can include a single-mode (SM) fiber that is used to couple signal light into or out of

Glass Fibers for High Povxr Lasers

246

the core of the double-clad fiber. This method is stable and robust and can provide high coupling efficiency (ultimately limited by the efficiency of fiber-coupling of the pump diodes). The approach allows unidirectional pumping and is scalable. The shape and size of the fiber bundle and of the SM pigtail must be matched to the double fiber being pumped (see Fig. 7.10, from IPG catalog).

\^vA

%

&JJ)

or ; J>& &3*^ : A

;.'*&/*****" ««*' Jii* *LD

""* ^ ^

Fig. 7.10 Fiber laser with fused fiber bundles pumping schematic.

7.4 High Power Pulsed Fiber Lasers 7.4.1 Introduction Recent improvements in the modified chemical vapor deposition (MCVD) process and diode-pumped lasers offer the possibility of incorporating a large variety of rare-earth ions in low-loss silica glass fibers. Interests in Er3+ ions were stimulated by the fact that its peak laser transition near 1.5 um falls into the low-loss window of silica fibers. Erbium doped fiber amplifiers (EDFA) are developed rapidly and used widely in broad-band optical communication system. In recent years, high power pulsed fiber lasers have caused particular interest as a reliable, efficient and compact, low cost source for a variety of applications in industry. There are several methods to get pulsed operation of fiber laser, one is the Q-switch fiber laser with lower power output, another one is a master oscillator power amplifier (MOPA) system. MOPA system is used frequently comparing with CW fiber

7.4 High Power Pulsed Fiber Lasers

247

lasers, pulse fiber laser need higher peak power intensity at the fiber surface, also how to reduce amplified spontaneous emission(ASE) signal is important for MOPA system. In this section, we will discuss these problems in detail. In the past few years, Yb-doped fiber amplifiers have made great strides in the optical power amplification field. Performances have been improved firstly thanks to new pumping technologies: end pumping has been progressively replaced by side pumping, making it possible to leave the double clad fiber ends free for fiber splicing. Moreover, the use of large emitting area laser diodes makes it possible to launch very high pump power, typically from a few to several tens of watts. The inner clad shapes of the fibers lead to different absorption. High level saturated signal output power can thus be reached, but with reduced efficiency. This is mainly due to the fact that the shape of the cross section of the fiber and/or the transverse distribution of rare-earth ions are not optimized, leading to an inefficient absorption of the pump power. In 2000, A Q-switched, 5W average output power amplifier using 3+ Yb -doped double clad fiber was reported by J.Alvarez-Chavez.[9] Further research was carried out by J. Limpert et al. at Jena, Germany. Single-frequency (MOPA) emits up to 20W. Picosecond pulse duration fiber amplifier is capable of generating 51.2 W. Nanosecond fiber amplifier can even produce up to 100 W average power at 50 KHz repetition rate, corresponding to pulse energy of 2 mJ. [1012] We launched the research of high power and high energy Yb-doped double clad fiber and amplifier at Shanghai Institute of Optics and Fine Mechanics, CAS. We used the MOPA system and homemade large mode area (LMA) double clad fibers, realize 133.8 W average power of amplified radiation at the wavelength of 1064 nm and a repetition rate of 100 KHz, limited only by the available pump power. Peak power of 300 KW at 20 kHz with the pulse duration of 15 ns is obtained. Compared to the previous results with similar arrangement, it is the highest single-fiber output average power to our knowledge.[7] 7.4.2 Transient response of Yb-doped fiber Amplifier Quasi-three-level systems such as erbium and ytterbium-doped glass

Glass Fibers for High Power Lasers

248

fibers take part in commonly used technologies in many applications such as telecommunication devices and high-average-power systems. Because of the importance of these devices, much theoretical research has been undertaken to understand and optimize. Two main areas have been prospected in order to characterize the behavior of the fiber amplifiers. On one hand, fully numerical models allow us to predict accurately the gain and the ASE spectra, but these approaches usually take a lot of computing time and are not very suitable for a better understanding of the fiber device properties. On the other hand, several analytical or semi-analytical models have been developed. The two-level rate equations neglecting of ASE can be given as : 9N /

=

) dt

(RP°+W^Ni-(Rl*+W-

+

^

N

i

'

Nl+N2=p, ^

-

(7-17>

(7.18)

^

= ±?;-rp[N2*pe-N]apa],

(7 .19)

dP± 1 8P± ~^ + ^- = ±P;Ts[N2ase-NlaJ, oz vg at

(7.20)

The transition rates can be described as follows:

R

/T

T~"

= pa'e

p

p

T^

w =^L±P

A= y

Here, Ni and N2 are the population of upper and lower status of laser. p is total Yb3+ density. apa!e is the pump absorption and emission cross section respectively, &sa,e is the signal absorption and emission cross section. Aeff is the effective core area of the YDFA and Ts and Tp give a measure of the overlap of the optical modes with the Yb distribution. x2i is the fluorescence lifetime of the metastable level of the two-level system. Pp and Ps are pump and signal power respectively. vp and vs are pump frequency and signal frequency, vg is the group velocity, h is Planck constant.

7.4 High Power Pulsed Fiber Lasers

249

We suppose nt = Nt j p (i = 1, 2), the above equations can be written as: ^

= (Rpa + WJn, ~(Rpe + Wse + Ae)n2,

«i+«2=l.

(7.21) (7.22)

And we can obtain: ^

+

Zn2=V,

(7.23)

at t = Rp„+Wsa+Rpe+Wse + A2l , t} = Rpa+W„. The general solution is n2 = ce'1'" + TJ/% , c = n\- n/4 . So the characteristic time constant t0=\/Z = T/l + q + p, q = Pp/P?, P = PjPrThe initial conditions can be calculated under steady-state conditions, the pump is constant for pumping scheme, the Gaussian signal power is 0.3 W and the duration is 1 (is, the pulse repetition rate is 20 KHz, the length of fiber is 10 meters, P psat =2xl0" 3 W, P ssat =1.5xl0" 2 W. We use finite-difference method mentioned above to integrate numerically with the parameters below: Xp=9\5nm,

As =\064nm , opa =2.5x1 (T2W ,

crpe = 3 xl(T26 m2, z2l=0Mms, <jsa=3x\0-26m2, ase=5.5xl0-26m2,

p = 1.9xl025w3, ^#=5xl0""w2,

T p = 0.0012, T s =0.82, P = 1000, ^ = 100. The calculated result is shown in Fig. 7.11. The peak of pulse shift to the rising edge, which is due to the long population inversion recovery time of Yb ions. The experimental result is shown in Fig. 7.12, which is in agreement with the calculated results. The phenomenon demonstrates that the low repetition rate pulse has obvious transient gain. The gain saturation and recovery time depend on pump rate.

Glass Fibers for High Power Lasers

250

Fig. 7.11 The transient changes of Gauss pulse as functions of time in YDFA.

^ 3

0.020

i?

0.015

d

0

1

2

3

4

Time( urn) Fig. 7.12 Amplified pulsed profile from pulse fiber laser.

7.4.3 High power Yb-doped fiber amplifier We have done some research on double-clad fiber lasers and amplifiers. A 133 W fiber laser and amplifier was fabricated using double-clad fiber made by the Fiberhome Technology, Inc, Wuhan, China.[1319] Increasing the size of core appears is one of the main directions of the

7.4 High Power Pulsed Fiber Lasers

251

technological advancement towards high pulse energy and power, and also short length fibers are preferred as the Raman scattering can be reduced and stimulated Raman scattering threshold increased. The double-clad large-mode-area Yb-doped fibers used in the experiment were designed by independent technology and fabricated by the standard MCVD method. The fiber has a 43 urn diameter Yb-doped core with a NA of 0.08, centered in the preform and 650/600 um D-shaped inner cladding with a NA of 0.37. The doping Yb3+concentration is evaluated to be -6500 ppm. The setup of MOPA system is shown in Fig. 7.13. A Q-switched laser is applied as the seed source. The laser delivers average powers up to 1 W with repetition rate between 20 and 100 KHz. A Faraday isolator was used to protect the seed laser from back-reflections. The length of double-clad fiber is 4 m. The large cladding size ensures that more than 90% of the launched pump light is absorbed in the fiber. Mi*'

Seed source

LD

~l

isolator

\

Ett^a^ DCF

M^

Fig. 7.13 Experiment setup of MOPA system.

The fiber was coiled to a 10-cm diameter cylindrical mandrel in air, without any special cooling device. Both fiber ends were polished at angles of 5-10° to suppresses ASE arising from the Fresnel reflections. The fiber amplifier was pumped by a laser diode which was water-cooled and the operating temperature was from 18 to 22 °C. The central wavelength of the diode was about 975nm. Two lenses, which have short focal length, were used to couple the pump light into the inner cladding with a coupling efficiency of -90%. A dichroic mirror with AR for pump light and HR for amplified light was placed by an angle of 45° to separate the pump and amplified light. Two reflective mirrors Ml and M2 were used to decrease the length of the total system. An aspheric lens

252

Glass Fibers for High Power Lasers

was used to couple the seed light into the active core with high efficiency. The output characteristics are shown in Fig. 7.14.

50

100 150 Launched pump power, W

200

250

Fig. 7.14 Output power against launched pump power.

At a repetition rate of 100 KHz, we were able to produce an average output power up to 133.8 W at the maximum diode driven current. The slope efficiency with respect to the launched pump power was 56 %, and the output power increased linearly with the launched pump power. Compared to the previous results with similar arrangement, it is the highest output average power to our knowledge. We have not found any facet damage at the maximum power. So more output power can be realized if we increase the pump power. Due to transient gain of the MOP A system, the amplified pulse duration is reduced from 30 ns to 15 ns at the repetition rate of 20 KHz, corresponding to a peak power of 300 KW. The pulse duration decreases with the decrease of repetition rate. Fig. 7.15 shows the emission spectrum at the maximum output power, against the seed source spectrum and ASE on the logarithmic scale at a repetition rate of 100 kHz. When we pumped the fiber without injection the seed source, the output spectrum was ASE spectrum and centered at 1040nm with 3 dB bandwidth of 20 nm, but when the seed source was coupled into the core, the output peak spectrum was shifted to 1064 nm due to mode competition. The ASE spectrum was reduced effectively and no stimulated Raman scattering occurs. Coiling the fiber in a diameter less than 10 cm suppressed higher order transversal mode

7.4 High Power Pulsed Fiber Lasers

253

through bending losses, and only lower order modes are amplified. The M value is characterized to be 3.2. This value can be improved with a smaller diameter of the cylindrical mandrel.

M

1000

'

1

1020

1

1

1

h

1040 1060 Wavelength, nm

"n

1

1080

1

1

1100

Fig. 7.15 Emitted spectrum of seed source, amplifier output and ASE.

The modeling and experiments presented here are not the last word on any of these subjects. There is still a great deal of work to be done to scale fiber amplifier system to high power. Additional work must be done to improve beam quality. The variation in amplifier performance with temperature is unlikely to be a major difficulty. However, as amplifiers are scaled to KW levels, eliminating waste heat becomes a concern. Short fibers, while reducing SBS, make heat removal more difficult. A very short fiber will have some of the same thermal problems as a rod laser. In addition, elevated core temperatures may cause diffusion of the core dopants and changes in the guiding properties. A great challenge that has not received much attention is the need for high power, low-loss isolators. Low-loss fiber pig-tailed isolators can only operate at the hundred mW regime. High-power bulk isolators require free-space propagation. It also introduces a place subject to misalignment, negating the greatest advantage of fiber systems.

254

Glass Fibers for High Power Lasers

7.5 Recent Development and Applications of Fiber Lasers 7.5.1 Recent development Fiber lasers are technological revolution in the fields of solid state lasers. There has been a rapid, large increase in the power produced by fiber laser systems.[20"231 Recently, the level of CW power available in single double-clad fiber has been increased from hundreds of Watts to more than 1 KW.[24"261 This breakthrough has been driven by two factors, one is the progress in development of high brightness semiconductor diode pumping sources; the second is by using large mode area ( LMA) fibers. The remarkable rises in CW fiber laser output power that has been reported by several research groups over the past few years are illustrated in Fig. 1.9 of Chapter 1. This figure shows only the results achieved in a single double-clad fiber. An array of combined fiber laser with an output close to 10 KW has been reported. The achievement of higher laser powers is limited by three effects: first, optical damage of the fiber material by high power laser, second, limitation by the usable pump powers with higher brightness, and third, the effects of nonlinear interaction in the fiber core.[27] The development of the optical fiber has traditionally been driven by the need of the long distance optical communication networks. During 1970's, the trend was to move from large multimode to small singlemode core, the modal dispersion is much smaller in single mode core compared to multimode fiber core. However, the need to obtain high CW power or high average pulse fiber lasers is pushing the double clad fibers to the opposite direction: increasing the fiber core dimensions. For standard telecommunication fiber, the core diameter is usually less than 10 um, a typical high power LMA double cladding fiber core diameter is around 30 um. The ten times larger core area allows ten times more output power. The pumping power with high brightness is another key point in developing the high power fiber lasers. Recently, the high power diode laser module with total output of more than 750 Watts and centre wavelength of 975-980 nm is available. The output can be transmitted by a fiber with core diameter of 600 um. By using two end pumping configuration, 1 kW laser output can be obtained. On the other hand,

7.5 Recent Development and Applications of Fiber Lasers

255

several fiber pigtail diode pumps can be coupled into a double cladding fiber by using a fiber combiner. This combiner consists of several relatively large diameter pump fibers fused together with signal-carrying fiber, which contains a core matching that of a LMA fiber. The fused pump fibers are tapered at the end to match the dimensions of the double cladding fiber, being cleaved and then spliced to a rare-earth doped LMA fiber. With this diode -combining scheme, several low power diodes can be used instead of a single high power one. The maximum power of this kind of combining system is around 200 Watts. Since the magnitude of laser - signal distortion induced by nonlinear effects (such as stimulated Raman scattering, SRS) is inversely proportional to the fiber mode area, the LMA fiber has a mode area more than ten times larger than the standard one. For standard single-mode fiber at 1.064 um typically supports a fundamental mode with diameter ~ 6.2 um, and corresponding mode area of 30 um2, when LMA fiber is used with core diameter of 30 um, the mode field diameter is ~ 24 um, corresponding mode area of 452 um2. The SRS threshold for CW fiber laser is typically in the 100 W range and that of LMA is in KW range. For pulsed fiber laser, the SRS threshold for standard fiber is only ~ 20 KW peak power, and for LMA (30 um core), it can increased to ~ 300 KW. Although the LMA fibers have advantages described above to increase the total output power from a single double cladding fiber laser with higher SRS threshold, the beam quality of the LMA fiber laser is decreased. Because LMA fibers with core sizes larger than 10 um support more than one transverse mode, it is necessary to employ some specific techniques to achieve single-mode operation of LMA fiber laser. For example, in a coiled fiber, higher order mode can be filtered to offset the effects of intermodal scattering, Practically by using short length high quality LMA fiber, it has proved possible to get a nearly diffractionlimited beam output. In contrast to all other optically pumped solid state lasers, fiber lasers use guided-mode propagation to create a new kind of laser structure with very long resonator lengths. The laser cavity is extremely simple without need for complex alignment. What are the limiting factors for scaling the power of the fiber laser in future? It is likely that within the next few

256

Glass Fibers for High Power Lasers

years, research and development efforts will continue to extend output power with a single fiber into the several KW range, since current approach to produce LMA fibers are closed to their limits in terms of the mode area that can be achieved, to design and develop new, large mode area fiber would be highly advantageous. One new idea is hollow core rare-earth doped photonic crystal fiber (PCF), which can reduce the nonlinearity of the fibers, the other one is short centimeter-size fiber lasers with large core areas. Fiber geometry offers another promising avenue of power scaling, several fibers can easily be packaged together, optical combination of the outputs of these fibers could lead to higher power than what individual one can provide. A variety of techniques for incoherent and coherent combination of fiber lasers is now being studied and developed. Much higher power can be obtained in future. 7.5.2 Applications of fiber lasers Fiber lasers have become the "hotbed of solid state laser development" with advance in powerful output, better ways to generate short pulses at high average power and higher efficiency. According the annual market review and forecast, the LASER FOCUS WORLD starts a report on fiber laser sales only a few years ago, in 2004 total sales of fiber lasers reached $ 84 million, it is 14% over 2003, additional growth of 47% is expected in 2005. The reasons for this strong growth are new technology and ideas, including improvements in fiber design and fabrication, steady improvements in pumping diodes source. In various applications, fiber laser can be used in the fields of materials processing, medical applications, instrument and basic research, in which, the main field is the material processing. It includes welding, cutting, drilling, marking and semiconductor or microelectronics manufacturing. The requirements for the flexible tool are as follows[28]: 1. Good beam quality: compare with other lasers, fiber laser has very nice beam quality with M2 less than 1.5 with core diameter < 10 |j,m, for large core diameter, more than 600 W with M2 < 1.2 was demonstrated recently.

7.5 Recent Development and Applications of Fiber Lasers

2.

257

Flexibility in spot geometry: the fiber laser is delivery by the double cladding itself, no additional transmission fiber is needed. 3. Flexibility in time: CW and pulse output are available from fiber laser, usually, a master oscillator power amplifier scheme is used to generate pulse output. Laser processing technology is widely used in car manufacture, aerospace industry and many other fields. It is the tendency of material processing in the future and has most advanced advantages comparing with other processing method. Three dimensional laser processing makes it possible to realize the processing of three dimensional workpiece by high intensity laser beam along the complicated special curve for welding, cutting and other treatment. The fiber laser is held by robot arm, normally, five or six degree of freedom to reach any position of the workpiece. In this system, the focus characteristics of the fiber laser beam keep constant at different processed position during the processing. Laser micro-machining is another application field for fiber lasers, as microelectronic circuits continue to shrink, the use of laser in circuit production continues to grow, based on the fiber laser can be focused to very small point, it can be used for laser trimming of thick- and thin-film resistor on ceramic and silicon substrates and laser repair of redundant memory device, the fiber laser rapidly becomes a viable tool in a wide variety of micro-machining applications. Typical lamp pumped laser for trimming applications required high energy input power supplies ( for example 3 KW ), large cooling water system, periodic lamp and water filter replacement, it also need frequently output power monitoring and adjustment. In comparison, a LD pumped fiber laser requires only a few hundred watts of input power and provides constant, stable laser energy for thousands of hours. With improved quality of nonlinear crystals, such as BBO and LBO, the process in the laser micromachining can get smaller focal point by using second harmonic or third harmonic generation in the green or UV regions. Better results can be obtained with much cleaner cut and less thermal effected zone. Laser marking is one of the most important applications for laser material processing especially in China. Fiber laser can be used in laser marking system with several advantages: good beam quality with almost

258

Glass Fibers for High Power Lasers

diffraction limited beam spread angle, compact size with flexible fiber output, high plug efficiency (more than 20 %) with low operating cost and air-cooling system. Fig. 7.16 shows the fiber laser marking system developed by Shanghai Institute of Optics and Fine mechanics.

Fig. 7.16 Laser marking system developed by SIOM.

The system can make marks in the region of 100mm x 100mm with scanning speed of 30 m/sec. The fiber laser used in this system is a 10 Watts pulsed fiber with repetition rate between 20 KHz to 100 KHz, the laser beam spread angle is 2 mrad.

References 1. M.Born and E. Wolf, "Principle of optics," (Cambridge University Press, Cambridge; New York) 1999. 2. S. Chuang, "Physics of optoelectronic device," (John Wiley & Sons, Inc., New York) 1995. 3. Kao, K..C. and Hockham, G.A., "Dielectric-fibre Surface Waveguides for Optical Frequencies", Proc. I.E.E.E 113 (1966) 1151.

References

259

4. S.Millar and I.Kaminow, Optical fiber telecommunications-II, New York: Academic, (1988). 5. J.Zhou, Q.H.Lou, Z.J.Wang, J.X.Dong, Y.R.Wei, SPIE, 4914 (2002) 141-145. 6. V.Dominic, S.MacCormack, R.Waarts et al., Electronics Letters, 35 (1999) 11581160. 7. Qihong Lou, Jun Zhou, Jianqiang Zhu, CLEO-PR 2005, Tokyo, Japan, Post Deadline paper. No. 00989 (2005). 8. D.Xue, Q.Lou, J.Zhou, L.Kong, J.Li, S.Li, Chinese optics Letters, 3 (2005) 345-347. 9. J.A.Alvarez-Chavez , H.L.Offerhaus, Opt.Lett, 25 (2000) 37. 10. S.Hofer, A.Liem, Opt.Lett, 26(2001) 1326-1329. 11. J.Limpert, A.Liem, Opt.Lett, 26 (2001) 1849-1851. 12. J.Limpert, S.Hofer, Appl.Phys.B75 (2002) 477-479. 13. Jun Zhou, Qi-Hong Lou, Lingfeng Kong, Chin.Phys.Lett, 21 (2004) 1083-1085. 14. Lingfeng Kong, Qihong Lou, Jun Zhou, Zhonglin Wu, Chinese Optics Letters, 2 (2004) 98-99. 15. Dong xue, Qihong Lou, Jun Zhou, High power laser and particle beams, 17 (2005) 665-668. 16. Lingfeng Kong, Qihong Lou, Jun Zhou, Dong Xue, Optics & Laser Technology, 37 (2005) 597-600. 17. Qihong Lou, Jianqiang Zhu, Jun Zhou, Chinese J. of lasers, A32 (2005) 552-555. 18. Lingfeng Kong, Qihong Lou, Jun Zhou, Actaphotonica sinica, 33 (2004) 1286-1289. 19. Lingfeng Kong, Qihong Lou, Jun Zhou, Chinese J. of lasers, Ail (2004) 93-95. 20. Zhonglin Wu, Qihong Lou, Jingxing Dong, Yunrong Wei, SPIE, 5623 (2004) 137140. 21. Zhonglin Wu, Qihong Lou, Jingxing Dong, Yunrong Wei, Chinese J. of lasers, A 32 (2005) 953-955. 22. Qihong Lou, Jun Zhou, Jianqiang ZhuChinese J. of lasers, A 31 (2004) 1029 . 23. Qihong Lou, Jun Zhou, Jianqiang Zhu, Chinese J. of lasers, A 32 (2005) 20-21. 24. Lou Qihong, Zhou Jun, Zhu Jianqiang, Wang Zhijiang, CLEO-Pacific-Rim, Japan, (2005). 25. Lou Qihong, Sino-German Workshop on Advances in Diodes and diode Pumped Lasers, Beijing (2005). 26. Lou Qihong, Zhou Jun, Zhu Jianqiang, Wang Zhijiang, The 2ed China-Korea Workshop on Optical Technologies and 3rd Joint Steering Committee Meeting of OSTCC, Sanya, (2005). 27. A.Galvanauskas High Power Fiber lasers, Optics & Photonics News, July (2004) 4247. 28. Y. Sun and E. J. Swenson , Laser Micro-machining in the microelectronics industry. SPIE 4915 (2002) 17-22.

Chapter 8

Hybrid Organic-Inorganic Solid-State Dye Laser Glasses

Since the born of the first laser in 1960 with ruby as active medium, several kinds of laser materials have been developed. They include crystals, gas, semiconductors and dyes. Crystal media are boasted for their high strength, however, the "thermal lens" effect resulted from their poor optical homogeneity deteriorates the laser performance. The strength and lifetime of gas as laser materials are limited. The lifetime and strength of semiconductors are both high and the laser configuration can be very compact, but the tuning range is only several nanometers. Laser dyes can be used in solid or liquid, their concentration can be readily controlled and the laser wavelength can be tuned continuously over a very wide spectrum range, producing narrow linewidth laser pulses without substantial energy loss. Also, organic dyes are very cheap. At present, dye lasers are still the major tunable laser sources. Because of the above characteristics, dye lasers have found extremely wide applications in industry, military, environment monitoring, medical diagnoses and therapy, laboratory and optical communication. In scientific research, tunable dye laser is especially useful in spectroscopy, biology, holography, photochemistry, isotope separation, non-linear optics and integrated optics, etc. Dye lasers in early stage mainly use liquid solution as active media. Since mid 1980s, intensive efforts have been devoted to the research on all solid state dye laser media and devices. In this chapter, a brief review is given on the history, recent progress and developing trends in this field, especially for hybrid organic-inorganic solid state dye laser glasses.

261

262

Hybrid Organic-Inorganic Solid-State Dye Laser Glasses

8.1 Organic Dyes and Liquid Dye Lasers 8.1.1 Structure Characters and spectra ranges of Organic laser dyes Dyes are the substances containing conjugated double bonds. Generally, the absorption and fluorescence band of a dye molecule is often very wide, covering more than tens of nanometers.[1] The output wavelength covers a wide spectrum range and is tunable. The spectra ranges of some organic laser dyes are shown in Fig. 8.1.[2]

360

400

440

480

520

560 600 640 680 WAVELENGTH (nm)

720

760

840

920

Fig. 8.1 Wavelength ranges of different laser dyes.

8.1.2 Synthesis and selection of new laser dyes From approximately 500 organic dyes identified to exhibit laser action, only a few show efficient laser output,[3] such as coumarin, rhodamine, pyrromethene and perylene family dyes. The majority of current laser dyes are in the visible region. In recent years, the most prominent progress is the synthesis of two kinds of efficient dyes, perylene family dyes and pyrromethene family dyes.[4"U] The former has extremely high photostability,[12"1316] and the latter has high quantum yields.[1415]

8.1 Organic Dyes and Liquid Dye Lasers

263

Pyrromethene family dyes have extremely high fluorescence quantum yields (near 1) whilst very low triplet yields.[17] Up to now, the most efficient dye laser output was obtained from pyrromethene dyes.[1215'18] The transition mechanisms and spectra characteristics of these dyes in various hosts have also been investigated^17'19"201 M. D. Rahn et al. summarized that the recent progress in the performance of solid-state dye lasers was brought by the newly synthesized pyrromethene dyest21] The molecular structure of pyrromethene dyes is shown in Table 8.1. The photostability of pyrromethene family dyes was investigated and the order of photostability of several kinds of pyrromethene dyes can be listed as follows: P567
.CH,

Dye P546 P556 P500 P567 P580 P597 P650

R2, 6

H NaS0 3 CH3 C2H5 CH3(CH2 C(CH3)3 CH3

R8 CH3 CH3 CH3 CH3 CH3 CH3 CH3

^max

492nm

/ 515nm 516nm 518nm 524nm 588nm

By varying the substitute groups, the electron density on the benzene ring structure could be varied, which also changed the spectrum characteristics and photostability of pyrromethene dyes. Currently, the attempts to graft various substitutes so as to improve the photostability or spectra characteristics of pyrromethene dyes are still underway.110"11] 8.1.3 Typical liquid tunable dye lasers1261 In liquid dye lasers, transverse and longitudinal cavities are the most common configurations. With wavelength selective elements, narrow linewidth output from dye lasers can be achieved. Up to now, 4 classes of wavelength selective resonators have been employed: (1) resonators including devices for spatial wavelength separation; (2) resonators including devices for interferometric wavelength discrimination; (3)

264

Hybrid Organic-Inorganic Solid-State Dye Laser Glasses

resonators including devices with rotational dispersion; (4) resonators with wavelength-selective distributed feedback. 8.2 Hybrid Solid-State Dye Laser Glasses and Preparation Techniques 8.2.1 Solid-state dye laser and its advantages Liquid dye lasers have some disadvantages[14]: the whole system is bulky and rather complicated, the lifetime of dye solutions are short, and the volatile solvents are toxic. Another problem is the dimers of dye molecules in solutions, which leads to the decrease of quantum yields and other problems. Since mid 1980s, intensive efforts have been devoted to solid-state dye laser materials, which can make the system compact and miniaturized and the laser medium replacement simple. Compared with the liquid counterpart, the thermal lens effect is prohibited, and the populated probability of triplet and photodegradation yields decrease. The laser lifetime of dye molecules is prolonged after being solidified, which also reduces the possible toxicity. After solidification, because the dye molecules are trapped into the skeleton of the solid matrix, thus effectively separates the dye molecules from dimerization, and the quantum yield increases. So the research on solid-state dye lasers has not only theoretical meaning, but also widespread applications. At present, the fabrication techniques and the resulting hybrid solidstate dye laser materials can be catalogued as 3 kinds: (1) doping laser dyes into polymers and epoxies, (2) post-doping laser dyes into porous glass, and (3) doping laser dyes into organically modified silicate glasses (ORMOSILs) through sol-gel method. 8.2.2 Solid-state dye laser materials based on polymers Because of their high homogeneity, high transmittance, good mechanical properties, low cost, simplicity of fabrication technique, and capability of obtaining high optical quality surfaces, organic polymers were chosen as the first host media of solid-state dye lasers.t27~291 However, the poor photostability of early dye-doped polymer prevents it from practical use.

8.2 Hybrid Solid-State Dye Laser Glasses and Preparation Techniques

265

Compared with other solid hosts, polymeric matrix is superior for the better compatibility with organic dyes. Meanwhile, the mechanical properties of polymers enable them to obtain high quality surface, which can then be coated with various reflectance.[30] Currently, the majority of polymers used as solid-state dye hosts are polymethylmethacrylate (PMMA) derived from the monomer methyl methacrylate (MMA). However, lower laser damage threshold and thermal conductivity of PMMA result in fast photodegradation and poor photostability of dyes, preventing it from commercially practical. A. Costela et al. proposed that the main cause of dye degradation in polymeric matrix was the thermal-degradation resulted from the poor heat dissipation in polymeric hosts.[31] In polymeric matrix, in order to improve the laser damage threshold, two methods are adopted: to fabricate the polymer through the copolymerization of different monomers, or to modify the polymer with additives, i.e. the modified polymethylmethacrylate (MPMMA). The most commonly adopted monomer for co-polymerization is the polar 2-hydroxyethyl methacrylate (HEMA), which improves the compatibility between host and polar dyes, for instance, rhodamine B (RB).[32] W. J. Wadsworth et al. investigated the optical and thermal properties of P(HEMA:MMA) copolymerized by MMA and HEMA, and MPMMA modified by ethanol and illustrated that the micro-fracture and bubbles in polymers have been prohibited after modification.133' The laser damage threshold of MPMMA and P(HEMA:MMA) improved by twofolds. The effects of HEMA and ethanol were just like plasticizer. F. J. Duarte et al. have also investigated the thermal properties of polymeric gain media.[34] The measured 9n/ST, temperature coefficient of refractive index, for PMMA doped with R6G is -1.4±0.2xlO-4K_1, which is very close to P(HEMA:MMA) copolymers. A. Costela et al. have investigated the laser performance of dyes doped into various copolymers.'351 By co-polymerize ethylene glycol dimethacrylate (EGDMA) with HEMA, the laser efficiency of R6G increased by one-fold and the laser lifetime of R6G doped in P(MMA:HEMA) reached 7.5 xlO4 pulses. The laser lifetimes of dyes doped into other copolymers, which 2-hydroxyethyl acrylate (HEA), 2phenoxyethyl acrylate (PEA), l-vinyl-2-pyrrolidone (VP) and 2,2,2-

266

Hybrid Organic-Inorganic Solid-State Dye Laser Glasses

trifluoromethyl methacrylate (TFMA) co-polymerized respectively with MMA, were also improved because of the reduced mobility of dye molecules due to the greater degree of cross-linking in host matrix.[3639] The laser damage threshold and photostability of modified polymer matrices are very high. R. S. Anderson et al. [30'40] and S. C. Picarello et al. [41] have doped pyrromethene and rhodamine dyes into MPMMA, which have laser damage threshold as high as 3.0J/cm2. At the pump intensity of 1.3J/cm2 and repetition rate of 10Hz, the laser output energy of P597 decreased to 80% of its initial value after 1.5xlO5 pulses. At the pump intensity of 2J/cm2, the laser output of P597 and P650 kept steady for 5xl0 4 and 2><104 pulses at 1250mJ and 150mJ, respectively. With commercially available PMMA, K. C. Yee et al. developed a mould casting technique, i.e. fabricated R6G and rhodamine 640 (R640) doped PMMA monoliths without polymerization process.[421 8% of slope efficiency and 0.012% of reduction in output per pulse were obtained under the excitation of N2 laser at the pump intensity of 0.025J/cm2. Epoxy is another kind of polymer hosts. R. E. Hermes et al. have doped P567, P570 and P612 into modified acrylic plastic, and 77%, 85% and 52% of slope efficiency were obtained respectively.115' P567 exhibited laser output as high as 141mJ, and the output decreased to 66% of its initial value after 2xl0 4 pulses, at the pump intensity of 0.6J/cm2 and the repetition rate of 3.3Hz. D. Pacheco et al. adopted Epoxy 310 as host material, which has higher laser damage threshold than MPMMA.[43] At the pump intensity of 0.16J/cm2 and the repetition rate of 5Hz, the laser output of P597 decreased to 70% of their initial value after 3.3xl0 4 and 9.5xl0 4 pulses respectively, when doped in Epoxy 310 and MPMMA. There are large variations in performances of dyes doped in PMMA by different researchers. For instance, M. Ahmad et al. obtained 2.7xl0 5 pulses of laser lifetime with P567 doped in PMMA,[44] while the results by some researchers were only lxlO 4 pulses.136'45"461 Such a variation may be attributed to differences in fabrication and test conditions. Recently, 1.9xl06 pulses of laser lifetime and 350GJ/mol of normalized photostability have been obtained for P567 co-doped with coumarin 540A (C540A) into ethanol modified MPMMA.[47] lxlO 5 pulses of laser lifetime has been obtained by A. Costela with R6G doped

8.2 Hybrid Solid-State Dye Laser Glasses and Preparation Techniques

267

into organic-inorganic hybrid materials, which were copolymerized by HEMA with various content of TEOS and TMSMA.[48] 8.2.3 Solid-state dye laser materials based on porous glass derived from sol-gel method Sol-gel process is another method for laser dye solidification, fabricating films and monoliths through the hydrolysis and polycondensation of Si, Al, Zr and Ti alkoxide.[4950] The final products are defined as inorganic gel glass. The metallic alkoxides form polymers during the process of hydrolysis and polycondensation, and then metal oxide with large molecular weight are obtained after the dehydration of final products, leading to three dimensional "metal-oxygen" chains, i.e. -M-0-M-0-.[51~ 521 So, glass could be obtained after heat treatment at the temperature far below melting point and reached the homogeneity at molecular level. During hydrolysis and polycondensation, the water content, pH value and temperature are critical because the final structure of gel glass is thus determined.153' The acid or base, added to adjust the pH value of the solution, plays the role as catalysts, which have great influence on the process of gelling and the gel structure. J. Lin et al. investigated the influence of anion on hydrolysis and polycondensation reaction of metallic alkoxides.[54] Anions with smaller radius, for instance, F- could accelerate the hydrolysis reaction rate of precursors. D. Lo et al. investigated the laser properties of coumarin dyes doped in Si0 2 gel glass.[55] The results illustrated that by prolonging the gellation and aging time, the homogeneity of gel glass was improved. Since Reisfeld et al. made the first attempt to dope R6G into SiC>2 gel glasses/561 solid-state dye laser media based on inorganic gel glasses have generated much interests, for its advantages listed as follows: (1) High photostability and high transmittance. (2) Preventing the reaction between dye molecules and surroundings. (3) Improving the permissible dye concentration. (4) Reducing the mobility of dye molecules. Because of the absorbance in blue and UV region, the thickness of dye doped polymers in this region is less than lmm.[31] However, the UV cut-off of Si0 2 sol-gel glass can be less than 260nm.[57] The

268

Hybrid Organic-Inorganic Solid-State Dye Laser Glasses

transmittance of the sol-gel glass was even higher after 600°C of heat treatment.[55] So in this spectrum region, the main solid- hosts for laser dye are inorganic sol-gel glasses rather than polymers. As the hosts for active species, the inorganic gel glass is a porous material. Besides its lack of good mechanical property, other defects are: (1) The porous nature of gel glass resulting in thermal trapping. (2) Optical loss resulting from the porous structure. So, good laser performances are hard to obtain with such materials. Dipping is the most commonly adopted method for post-doping lasers dyes into porous glass. First, the porous glass was immersed into organic dye solutions. Then, dye molecules diffused into the pores of porous glass. After proper heat treatment, dye molecules solidify in the pores. K. S. Lam et al. doped UV laser dyes PTP, E351 and PBD into Si0 2 sol-gel glasses by immersing them in dye solutions.'581 Laser output at 342nm, 352nm and 364nm were obtained, respectively. This result is the shortest wavelength of dye laser output obtained in solid hosts up to now. Another method is to adopt dry control chemical additive (DCCA) in the hydrolysis and polycondensation process to eliminate the adverse effect of micro-pores in sol-gel glass on luminescence, thermal and mechanical properties. C. S. Zhu et al. decreased the pore volume of silica gel glass by adopting N, N'-dimethylformamide (DMF) as DCCA.[5960] In such a material, the laser output of UV laser dye PQP at 374nm was obtained with a slope efficiency of 8% and a laser lifetime of 5000 pluses. When doping laser dye E376 in this gel glass, laser output at 364nm with slope efficiency of 4% was also observed. When the inorganic gel glasses are dipped into polymer monomer solutions dissolved with laser dyes, the monomers may diffuse into the pores of gel glass and be in-situ polymerized, and thus the dye molecules might be trapped in the pores. Such porous glass/polymeric host materials are defined as composite glass or polycom, in which the adverse influence of pores on the optical properties of the medium might be eliminated by the in-situ polymerization. E. J. A. Pope et al. immersed porous glass monoliths into MMA solutions dissolved with laser dyes.[61] Under the catalysis of initiators and irradiation of UV lamp, MMA polymerized in the pores, and a transparent composite material with 30% content of PMMA was

8.2 Hybrid Solid-State Dye Laser Glasses and Preparation Techniques

269

obtained. By varying the fabrication parameters, the pore size, pore volume and pore distribution can be adjusted, thereby the relative content of inorganic phase to organic phase can be controlled. They also investigated the transparency, refractive index and mechanical properties of these materials, and found that the composite integrated the properties from both PMMA and Si0 2 gel glass. Through the above process, H. R. Aldag et al. fabricated a polymerfilled microporous glass (PFMPG) doped with pyrromethene and rhodamine dyes.[62] At the pump intensity of 0.625J/cm2 and repetition rate of 5Hz, the laser output of P597 and Rl IB decreased to 70% of their initial values after 6.0><104 and l.lxlO 5 pulses respectively. Results illustrated that the thermal lens effect in PFMPG was much lower than that in MPMMA due to its better thermal conduction properties. However, due to the slow diffusion rate, the size of such composites are limited by homogeneity, for instance, the polycom fabricated by T. A. King et al. is only 2mm thick.[63] 8.2.4 Solid-state dye laser materials based on ORMOSILs Organic materials have good mechanical properties, large non-linear coefficient and quick response, and can be molecular designed. However, they also have defects such as low melting point, poor thermal stability and low laser damage threshold.1641 Inorganic sol-gel glass has good optical and thermal properties, but the thermal lensing and optical loss due to high porosity limit the photostability, and the poor mechanical properties prevents cutting, grinding and polishing. The ORMOSILs may combine the advantages of two kinds of materials whilst the defects being overcame.1651 Organically modifying agents can be catalogued as two kinds. One can participates the hydrolysis and polycondensation reaction and chemically bond with inorganic network, for instance, methyltriethoxysilane (MTES), vinyltriethoxysilane (VTES) and yglycidyl trimethoxisilane (GPTMS).[66] Another kind of modifying agent is physically mixed with the inorganic component, and in-situ polymerizes during sol-gel process and then fills in the pores with weak bonds connected to inorganic network, for instance, MMA.[67"68]

270

Hybrid Organic-Inorganic Solid-State Dye Laser Glasses

The structure of ORMOSILs can be classified as 3 cataloguesLb/' W"/UJ: (1) Class I corresponds to hybrid system in which organic molecules or oligomers are embedded into the inorganic hosts and connected with inorganic hosts by weak bonds such as Van der Waals force, H bond, etc. (2) Class II corresponds to hybrid compounds where the organic and inorganic components are bonded through covalent or ion-covalent bonds, thus the organic components are grafted to inorganic network. (3) Class III represents a combination of the former two classes. The organic active molecules are isolated in the "cages" of a host due to its cross-linking structure, which improves the photostabihty and thermalstability of dyes.[71"74] M. Canva et al. adopted VTES and MTES as precursors to fabricate dye doped monoliths.[18'75"76] 2.5 xlO5 pulses of laser lifetime of P-red were obtained in deoxygenated MTES-derived ORMOSIL at the pump intensity of 0.2 J/cm2 and repetition rate of 5Hz. 5><105 pulses of laser lifetime were obtained for P597 doped into MTES-derived ORMOSIL at repetition rate of 20Hz and the pump intensity of 0.2 J/cm2. 86% of slope efficiency was also obtained for P580 doped MTES-derived ORMOSIL Q. Y. Zhang et al. fabricated the organically modified titanium gel glass with GPTMS and MMA as modifying agents.[7779] With MTES, GPTMS and MMA as modifying agents, the laser output of P597 decreased only 10% after 3><104 pulses. Highly efficient and stable output of various laser dyes doped into various precursors derived ORMOSILs have also been obtained in our lab. [12 ' 8081] Dye-doped samples were prepared by multi-step process, i.e. acid-catalyzed hydrolysis and basic-catalyzed condensation of the MTES and VTES precursors. Laser dyes included P567, P-red, P-orange, C440 and C500. The MTES- and VTES-derived ORMOSILs could be machine cut, grinded and polished. Fig. 8.2 shows the photographs of samples with the size more than 3 cm in diameter and 2cm in thickness. We systematically investigated the laser performances of pyrromethene and perylene dyes doped in MTES- and VTES-derived ORMOSILs.

8.2 Hybrid Solicl-Slale Dye Laser Glasses and Preparation Techniques

271

Fig. 8.2 Photos of MTES- and VTES-derived ORMOSILs doped with various laser dyes before (left) and after (right) grinding and polishing.

The laser slope efficiency as high as 78.8% were obtained for P567 doped in MTES-derived ORMOSIL whilst the output energy was kept at 85% of its initial value after 1.06xl05 pulses. 53.2% of laser slope efficiency and 7.8><104 pulses of laser lifetime were obtained for Porange doped in VTES-derived ORMOSIL. Normalized photostability of P-red doped into VTES-derived ORMOSIL was 466.7GJ/mol. The evolution of laser output of P567 and P-orange doped into MTES- and VTES-derived ORMOSILs respectively were shown in Fig. 8.3. These results were among the best values at the same time.

r" **&&**?&** I |

I £

-

0.4

' '

* P567

0.2 - I * 0 0

P-orange |

i i i 1 1 I i i 11 i I i 11 i i I i i i i i I i i i 11 I i i i i i

0

20000

40000

60000

80000 100000 120000

Number of pump pulse? Fig. 8.3 Normalized output of P567 and P-orange doped in MTES- and VTES-derived ORMOSIL (16mm and 15mm thick, respectively), as a function of pump pulses.

272

Hybrid Organic-Inorganic Solid-State Dye Laser Glasses

Recently, similar results were also reported by T. H. Nhung et al. on the laser performances of pyrromethene and perylene dyes doped into MTES-derived ORMOSILs, which had been aged for several years.[82] 8.2.5 Solid-state dye laser materials based on chemically doping technique T. Suratwala et al. illustrated that the more the inorganic content in host materials, the lower degradation rate of dyes molecules.[83] This phenomenon, i.e. the lower degradation rate of dyes in inorganic hosts, clearly reflects the "cage" effect of inorganic component on dye molecules. If the dye molecules are covalently bonded onto inorganic network, the chance for dye molecules being caged will increase while the mobility and dimerization of dyes decrease, and the stability of dyes doped in inorganic hosts will be further improved. This can be realized by grafting the hydrolysable alkoxyl groups onto the dye molecules; thereby the dye molecules can participate the hydrolysis and polycondesation reactions and be covalently bonded with hosts. Compared with the physically doping dyes into host media as mentioned above, we defined such a fabrication method as "chemically doping". By grafting organic molecules onto inorganic glass network through chemically doping technique, solid-state dye medium with higher photostability might be fabricated[84] The technique is to covalently bond dye molecules onto the alkoxy groups of metallic alkoxide, thereby the molecules can be grafted onto the inorganic skeletons during sol-gel process through hydrolysis and polycondensation reactions. The advantages of chemically doping technique are: (1) Improving the solubility of dyes in solid hosts. (2) Higher luminescence intensity resulting from the higher degree of dye molecules being separated by the matrix. (3) Improving the photostability of dye molecules due to the faster energy dissipation from dye molecules to surroundings. M. Canva et al. modified RB and coumarin dyes with hydrolysable groups, thereby covalently bonded the dye molecules onto inorganic backbones.[75] The results illustrated that the laser lifetime of RB increased from 6500 to 1.3 xlO4 pulses.

8.3 Photostabilities and Photodegradation Mechanisms

273

T. Suratwala et al. also synthesized coumarin dyes with organic groups which could participate in hydrolysis and polycondensation reactions.[84] The fluorescence decay process of dyes doped into polydimethylsiloxane (PDMS) modified ORMOSIL under the excitation of laser pulses illustrated that the phototstability of dyes improved greatly after it being covalently bonded onto the glass network, which proved the validity of chemically doping technique. Meanwhile, the research on coumarin dyes with various numbers of hydrolysable groups on various molecular positions illustrated that the photostability of these dyes did not increase with the increase in hydrolysable groups. Recently, a variety of P567 derivates were successfully synthesized,[8586] on which an acetoxy or a polymerizable methacryloyloxy group substituted the 8 position of the chromophore core. These monomeric pyrromethene dyes were then copolymerized with MMA, yielding linear copolymers of high optical quality where the covalently bonded chromophore was separated from the polymeric main chain by a spacer of variable length. Improvement in photostability of these newly synthesized dyes was observed compared with that of P567, due to additional modes for heat dissipation along the polymeric chain. 8.3 Photostabilities and Photodegradation Mechanisms of Hybrid Solid-State Dye Laser Glasses 8.3.1 Photodegradation of organic laser dyes in hybrid glasses After a period of irradiation, the fluorescence intensity and laser output of solid-state dye laser materials decrease with the increase in irradiation time or pump laser pulses.[23] This phenomenon is defined as "photodegaradation" or "photobleaching". The photobleaching of organic compounds in solutions is mainly attributed to the photochemical reactions of organic active compounds, or the quenching resulted from dimers, or the photodecomposition of organic compounds. In solid hosts, the organic species are trapped in the cages and isolated by the matrix. Thus, the probability for organic dyes being photobleached would be reduced. M. Canva et al. found that the

274

Hybrid Organic-Inorganic Solid-State Dye Laser Glasses

photostability of dyes in solid hosts was improved by 1 or 2 orders of magnitudes compared with that in liquid solutions.[87] Some scientists suggested that photostability was determined not only by the composition and structure of hosts and dyes but also by other factors, such as dye concentration, pump wavelength, repetition rate, pump intensity and the size of samples.'88"891 M. Canva et al. characterized the photostability of laser dyes with absorption cross-sections.[87] Up to now, several mechanisms about the photodegradation of laser dyes doped in solid hosts have been suggested. T. Suratwala et al. investigated the decay process of the fluorescence intensity of P567 under laser excitation.'84] The results illustrated that at the pump intensity of 0.1J/cm2, the dyes photobleached gradually without any indication that the hosts have been damaged, while at the pump intensity of 0.53J/cm2, crazing and darkening areas were observed on the sample surface, indicating the damage of the matrix. M. D. Rahn et al. suggested that the photodegradation was mainly caused by the photochemical reactions of excited state dye molecules.[17'44'90] The possible participants were oxygen, free radicals, active carbon hydrogen groups and impurities. Moreover, other pathways might also be possible, such as the photochemical reactions of triplet state dye molecules and multi-photon absorption, inducing more active excited state or bond broken. R. Reisfeld et al. also suggested that the photodegradation process was not a single photon process because of the non-exponential decay behavior of fluorescence intensity.'911 Their research demonstrated that most of the photochemical reactions of dyes doped into solid hosts were diffusion controlled dependent/18' The photostability of dyes could be improved by deoxygenation[I8,46] or by doping proper singlet oxygen quencher into solid hosts.'44'901 An excited state dye molecule has several pathways to be deactivated, as shown in Fig. 8.4.'241 Among them, the main pathways that lead to photodegradation of dyes are isomerization, dissociation, dimerization and photochemical reactions with other species. So, theoretically, controlling all the above pathways could reduce the photodegradation rate. However, due to the variation in photodegradation process, the effect of controlling one pathway on the enhancement of the photostability of different dyes is varied. The research of photodegradation

8.3 Photostabilities

and Photodegradation

Mechanisms

275

mechanisms and the controlling routes should base on specific dyes and matrix environments. Excited Molecule

Exchange short-range trqslet-energy transfer Fluorescence Phosphorescence / Substitution reactions

Oxidation

Addition Reaction \ Hydrogen abstraction

Dipote-dipole long-range singlet-energy tansfcr

\ .

Exciplex

Dimenzation

Fig. 8.4 Decay pathways for an excited dye molecule.

8.3.2 Photodegradation mechanisms of pyrromethene family dyes in polymeric hosts M. D. Rahn et al. investigated the photodegradation of P567 dye under the excitation of laser irradiation.[17] They suggested that the main photochemical reaction is between the ground state P567 and singlet state oxygen, which was sensitized by triplet state P567 through energy transfer. So, the rate of the above photochemical reaction and its content were responsible for the decay rate of dye molecules. The photochemical reaction process of P567 molecules is shown in Fig. 8.5.[24] The final product is 3-ethyl-2-methylmalemide or similar compounds. At low energy intensity, the main mechanism responsible for the photochemical reaction of P567 can be attributed to self-sensitized and radical-sensitized photo-degradation. Both mechanisms are related with the micro-environment of dyes.

276

Hybrid Organic-Inorganic Solid-State Dye Laser Glasses

Self-sensitized photo-degradation mechanism refers as the photochemical reactions of the ground state P567 molecules with singlet oxygen, which is sensitized by triplet state P567 molecules. The photochemical reaction processes are listed as below.[92]

D + hvWD* '£>* -^D

+ hv

l

D*^D*

(8.1) (8.2) (8.3)

3

D*+302 -*D+X02 [

02+D^>

D02

02 + £)-> j 0 2 + D

(8.4) (8.5)

l

(8.6)

l

o2^o2

(8.7) where D represents the ground-state dye molecules, 'D* and 3D* represent the excited singlet-state and triplet-state dye molecules respectively, and ' 0 2 and 3 0 2 represent the singlet-state and triplet-state oxygen molecules respectively.

Fig. 8.5 Degradation routes for (a) P567 by oxidation to (d) 3-ethyl-2-methylmaleimide.

8.3 Photostabilities and Photodegradation Mechanisms

111

From above equations, it is found that the photodegradation of P567 is related with the number of triplet-state dye molecules, oxygen concentration and the lifetime of singlet-state oxygen. The prohibition of photodegradation can be realized by controlling these factors. Choosing solvents in which the lifetime of singlet oxygen is short or the oxygen concentration is low, the photostability of P567 may be improved/461 In polymeric hosts, the photochemical reaction rate of dyes is accelerated by the free radicals resulted from laser pumping. In the ORMOSILs with MMA as modifying agent, similar organic fragments resulted from laser pumping can also accelerate the photodegradation process. G. Jones II et al. investigated the phototransients, intermediates and the role of energy and electron transfer mechanisms in the photodegradation process of P567 dye.[23' When co-doped with pyromellitic dianhydryde (PMDA), P567 lost one electron due to electron transfer. It is suggested that free radicals were an important factor responsible for the photobleaching of P567, which was supported by the researches of adding free radical scavenger into laser media.[24'43] To withhold the laser dyes from photo-oxidations, some researchers deoxygenated the samples in the fabrication process. M. D. Rahn et al. t46] and M. Canva et al.[181 have attempted to deoxygenate the dye-doped polymer, composite and ORMOSILs by bubbling N2 or vacuumized. However, such methods are costly and the exposure in air is inevitably. So, the photostable agents seem to be more feasible. Under the precondition that the laser performances of the dyes should not be deteriorated, main photostable agents could be catalogued as follows. (1) Interceptor, which slows oxidation by preferentially reacting with 0 2 or by quenching singlet state 0 2 . (2) Free radical scavenger, which reacts with free radicals. (3) Deactivator, which quenches the dye in its excited state. Recently, the influences of photostable agents on the photostability of dyes doped into polymeric hosts have been investigated. M. Ahmad et al. have doped 1,4-diazobicyclo [2, 2, 2] octane (DABCO) into PMMA with P567.The laser lifetime increased by 100% compared with that of the sample which has no DABCO.[44'90] At optimized DABCO concentration, the laser lifetime increased by 3-folds.[931

278

Hybrid Organic-Inorganic Solid-State Dye Laser Glasses

8.3.3 Photostabilities and photodegradation mechanisms of laser dye P567 in hybrid gel glasses From above discussion, it is found that the host media adopted to study the photostabilities of laser dyes are mainly polymers. For investigating the influence of photostable agents on the photostability of laser dyes in hybrid gel glasses, we doped two promising additives, DABCO and 2,2,6,6 tetramethylpiperdin (TMP), into MTES-derived ORMOSILs with P567 dye respectively. The influence of the additives and the concentration effects on the fluorescence, laser efficiency and photostability of P567 in MTES-derived ORMOSILs were studied.[94] For fear that the introduction of basic agents may deteriorate the performances, the fluorescence and laser performances of P567 in the absence and presence of photo-stable additives were characterized. Varying the additive concentration, only slight differences in fluorescence spectra of P567 could be observed. The laser efficiencies of P567 doped in MTES-derived ORMOSIL with the presence of various contents of TMP agents also indicated that the introduction of a certain amount of TMP, smaller than 4><10"4 mol/1, lead to a slight increase in the slope efficiency of P567. These results clearly indicated that the addition of photo-oxidants had little negative effects on the performance of P567. A continuous UV lamp was adopted as the excitation source to investigate the photostability of P567. Recently, a similar method was also adopted to investigate the photostability of dyes in polymers.[86] To evaluate the photo-stable effect of TMP on P567 dye, the exposure time dependence of fluorescence intensity of P567 doped in MTES-derived ORMOSIL with various content of TMP additive, was obtained, as was shown in Fig. 8.6. The introduction of TMP additive and its content in MTES-derived ORMOSIL did have great influence on the photo-decay process of P567 laser dye. The fluorescence intensity of P567 dye with no or a small TMP content (P567 and TMP1) decreased at the beginning of exposure, however, those with more TMP content were kept at their original level for 200 hours, and then began to decrease with different decay rate. The best result obtained was the sample labeled as TMP4, corresponding to an optimized initial concentration of 4><10"4 mol/1 of TMP in ORMOSIL, which decreased to 80% of its initial

8.3 Photostabilities and Photodegradation Mechanisms

279

fluorescence intensity after 370 hours of irradiation whereas the fluorescence intensity of P567 without additive decreased to 50% of its initial value after 200 hours. It could be suggested that the photostability of P567 had been improved by 100%.

" 0

50 100 150 200 250 300 350 400 T i m e (hours)

Fig. 8.6 Exposure time dependence of fluorescence intensity of P567 doped in MTESderived ORMOSIL with various content of TMP additive. TMP1, TMP2, TMP3, TMP4 and TMP5 are corresponding to the molar ratios between TMP and P567 of 1:1, 2:1, 3:1, 4:1 and 5:1 respectively.

The exposure time dependence of fluorescence intensity of P567 doped in MTES-derived ORMOSIL with various content of DABCO was also obtained, as shown in Fig. 8.7. Similar photo-stable effect on P567 with the presence of DABCO was observed and the optimized initial concentration of DABCO was 1.5 xlO"3 mol/1, corresponding to the additive/dye ratio of 15:1. At the optimized additive concentration, the fluorescence intensity of P567 dye decreased to 60% of its initial value after 370 hours of UV irradiation, i.e. the photostability of P567 laser dye was believed to be improved by 100%. Recently, improved photostability of P567 incorporated into the MPMMA polymer with C540A laser dye was achieved.[47] In our lab, enhanced laser efficiency and photostability have also been obtained with the pyrromethene and perylene dyes co-doped with coumarin dyes respectively in ORMOSIL. The exposure time dependence of

280

Hybrid Organic-Inorganic Solid-State Dye Laser Glasses

fluorescence intensity of P567 doped in MTES-derived ORMOSIL with the presence of C440 dye, as shown in Fig. 8.8, indicated an enhancement of at least 4 times in the photostability of P567. The photostable mechanism of C440 is believed to be the quenching of triplet-state P567 by ground-state C440 dye molecules.

1.0 _ 0.8 £' 0.6

I 0.4 0.2 0

50 100 150 200 250 300 350 400 Time (hours)

Fig. 8.7 Exposure time dependence of fluorescence intensity of P567 doped in MTESderived ORMOSILs with various content of DABCO additive. DABCOl, DABC05, DABCOIO, DABCO 15 and DABCO20 are corresponding to the molar ratios between DABCO andP567 were 1:1, 5:1, 10:1, 15:1 and 20:1 respectively.

1.2

it

1.0

£ 1

0.8

I

04

J

0.6

0.2 OJ 0

200

400 600 800 Time (hours)

1000

1200

Fig. 8.8 Exposure time dependence of fluorescence intensity of P567 doped in MTESderived ORMOSIL in absence and presence of C440.

8.4 Hybrid Solid Dye Laser Glass Based on Energy Transfer

281

8.4 Hybrid Solid Dye Laser Glass Based on Energy Transfer Between Laser Dyes With a specific laser dye, narrow linewidth laser output tunable in tens of nanometers can be obtained.181'95"971 The feasibility of a mixture dye laser was first demonstrated by O. G. Peterson et al. [98]. Energy transfer in laser dye mixture can improve the efficiency and broaden the tunable ranges.199"1001 Energy transfer dye lasers using numerous donor-acceptor pairs and the mechanisms have been extensively investigated.[10M07] Recently, energy transfer from C440 and C500 to P-red, P-orange and P567 respectively in hybrid gel glasses was investigated in our lab. The energy transfer mechanisms between laser dyes and the effect of donor concentration on energy transfer were studied. The laser performances such as laser efficiency, tunable range and the photostability of the energy transfer dye laser media were also investigated.'108"1091 The fluorescence spectra of P-red doped in VTES-derived ORMOSILs in the absence and presence of C440 were presented in Fig. 8.9, and the variation of the peak fluorescence intensity of the acceptor with donor concentration was shown. Energy transfer from C440 to P-red was observed. The maximum peak fluorescence intensity was observed at the donor concentration of 250

500

550

600 650 Wavelength { mn )

700

Fig. 8.9 PL spectra of P-red doped in VTES-derived ORMOSILs in the absence and presence of C440 or C500. (solid: with C440 of 5.0x10"4mol/l; dash: with C440 of 1.0xl0"3mol/l; dot: with C440 of 2.0*10"3mol/l; dash dot: without C440 or C500; dash dot dot: with C500 of S.OxlO^mol/l).

282

Hybrid Organic-Inorganic Solid-State Dye Laser Glasses

1.0xlO"3mol/l. At lower donor concentration, the peak fluorescence intensity of the acceptor laser dye increased with the increase of the donor concentration because the intermolecular distance between the donor and acceptor molecules was large and reduced continuously with the addition of donor dye. Above a certain concentration, the peak fluorescence intensity decreased, which might be attributed to the effect of a non-radiative fluorescence-quenching between the donor molecules.[110] In other words, increasing the donor concentration up to a certain value will not in any way enhances energy transfer to the acceptor. Greater enhancement in the fluorescence intensities of P-red and Porange doped in VTES-derived ORMOSILs using C500 as donor were observed, respectively compared with that of C440 as donor. Such greater enhancement was attributed to the greater overlap between the absorption bands of acceptor dyes and the fluorescence band of C500. Higher laser efficiencies were obtained with the perylene family dyes in the presence of coumarin dyes as shown in Table 8.2. Table 8.2 Laser performance of the dyes doped into ORMOSILs. Dye pair (molar ratio) C440: p-red (10:l) a C500: p-red(10:l) a p-red a C500:p-orange(5:l) a p-orangea C440: p-red (10:l) b C500: p-red (10:l) b p-red b C440:p567(5:l) b C500:p567(5:l) b P567 b a

Host matrix (thickness) VTES(7mm) VTES(8mm) VTES(8mm) VTES(6mm) VTES(6mm) VTES(7mm) VTES(8mm) VTES(8mm) MTES(8mm) MTES(8mm) MTES(8mm)

Slope efficiency (%) 3.4 3.6 1.4 9.9 6.3 15.5 15.1 6.2 40.9 37.3 34.6

Laser threshold

(vJ) 518.6 567.1 498.2 115.1 120.7 333.3 379.0 324.9 121.1 156.9 138.7

: pumped by 355nm Nd:YAG laser; : pumped by 532nm Nd:YAG laser

At least 60% improvements in the slope efficiency of the perylene family dyes have been observed in the presence of C440 or C500 dyes at the optimized dye concentration, especially for that of P-red, which increased at least 100%. Only marginal improvements in slope

8.4 Hybrid Solid Dye Laser Glass Based on Energy Transfer

283

efficiencies of P567 were observed in the presence of coumarin dyes. The coumarin dyes, as triplet-state quenchers, can increase the laser efficiency and reduce the possibility of photochemical reactions leading to photo-degradation of P567. It was found that the addition of C440 led to at least 3 times of enhancement in the photostability of P567, as shown in Fig. 8.8. The tunable properties of P-red doped into MTES-derived ORMOSILs in the absence and presence of C440 and/or P567 have also been studied, as shown in Fig. 8.10. When solely doped, the tunable range of P-red is narrow, from 586.12nm to 613.42nm, with conversion efficiency lower than 0.2%. In the presence of P567, the tunable ranges are broadened. Moreover, the energy conversion efficiency increases sharply when P567 co-doped. A factor of 2 improvements in conversion efficiency is observed when the p567 concentration increases from 5.0xl0"5mol/l to l.OxlO^mol/l. In the presence of C440, the tunable range of P567 and P-red mixture is even broader, from 561.80nm to 619.08nm, with peak conversion efficiency as high as 1.5%.

Wavelength ( urn ) Fig. 8.10 Laser tunable ranges of p-red doped in MTES-derived ORMOSILs in the absence and presence of C440 and/or p567. ( T : P567(5xl0"5mol/l)+p-red; •: P567(lxl0" 4 mol/l)+p-red; •: P567(2xl04mol/l)+p-red; A: C440+P567(lxl0"4mol/l)+p-red).

The influence of multi-dyes co-doping on the laser lifetime was also investigated. As shown in Fig. 8.11, the laser output of P-red in the absence and presence of P567 and C440 were compared at the pump

284

Hybrid Organic-Inorganic Solid-State Dye Laser Glasses

intensity of 0.5 J/cm2 and the repetition rate of 2 Hz. When solely doped, the output laser energy of P-red decreased steadily during the pump process. After about 30000 pulses of pumping, the output of P-red decreased to 50% of its initial value, corresponding to a normalized photostability of 253.8 GJ/mol. In the presence of P567 and C440, the initial output energy of P-red was improved and after 45000 pulses of pumping, the laser output of P-red was still above half of its initial value. And after 45000 pulses, the output of P-red in the presence of P567 and C440 was still higher than that of P-red solely doped, though the concentrations of P-red in all cases were the same. It is suggested that the decay of the output energy of P-red in the presence of P567 and C440 during the first period of pump may be partly due to the photodegradation of the latter two dyes. It is supposed that with more P567 and C440 dopants, the laser lifetime of P-red may be even longer. 0.50

0.15 15000

30000

45000

Pump Pulses

Fig. 8.11 Laser output of p-red doped in MTES-derived ORMOSILs in the absence and presence of C440 and p567 as a function of pump pulses, (triangle: p-red; square: C440(2.5xl0"4mol/l)+p567(lxl0-4mol/l)+p-red).

Laser dyes co-doped materials have potential application as energy transfer dye laser to broaden tunable range and improve the efficiency and photostability. In recent years, interaction between dyes co-doped in solid hosts has generated much interest. T. H. Nhung et al. observed energy transfer from RB to P-red in ORMOSILs.[111] A. K. Sheridan et al. observed efficient energy transfer from P-orange to P-red in organic

8.5 Solid-State Dye Lasers and Parameter Optimization

285

thin films. J Based on studies on the photostability of pyrromethene dyes in solutions and solid host,[113"115] W. N. Sisk et al. observed electron transfer from P597 and P567 to P546, which accelerated the photodegradation rate of the donor dyes, P597 and P567.[116] In short, designing and controlling the energy and charge transfer process between dyes co-doped in solid matrices may be an important issue at present to develop new highly efficient and stable solid-state dye laser materials. 8.5 Solid-State Dye Lasers and Parameter Optimization 8.5.1 Effect of pump conditions on solid-state dye laser performances Pump conditions including the kind and wavelength of pump source, repetition rate, pump intensity, pump pulse duration and its modes have significant influence on the laser performances of solid-state dye lasers, no matter how they are pumped longitudinally or transversely. Up to now, the highest energy conversion efficiency obtained by flashlamp pumping was only 1%.[117"120] D. P. Pacheco et al. have demonstrated that the laser efficiency and lifetime of solid-state dye materials under us pump pluses were much lower than that under ns pulses due to higher triplet-state yield.[43] Other investigations showed that the laser lifetime of solid-state dye laser was longer and the laser threshold was low under ps pump pulse than those under ns pump pulses.[121122] In short, a shorter pulse pump source is superior. 8.5.2 Cooling and circulating equipments of solid-state dye lasers Due to the movement of the media, different positions are excited alternatively. The time interval between two successive pump pulses onto a same position is long enough for the accomplishment of heat dissipation. So, the laser samples are cooled and circulated. The halt or change of the movement may leads to the cease of laser output. By rotating P597 doped MTES-derived ORMOSIL, M. Canva et al. obtained 107 pulses of laser output at the repetition rate of 30Hz.[18] By

286

Hybrid Organic-Inorganic Solid-State Dye Laser Glasses

rotating and translational moving the polymeric media, A. Costela et al. realized the laser output at a repetition rate as high as 10KHz.[123] 8.5.3 Cavity design and beam quality of solid-state dye lasers Key laser oscillator parameters include configuration, cavity length and coupler reflectance. An optimized cavity favors to obtain better laser performances including beam quality. The beam quality can be characterized as the stability of output energy, laser modes and the content of amplified simultaneous emission (ASE), etc. For good beam quality, some optical elements lead to sharp decrease in laser efficiency. S. K. Lam et al. obtained single longitudinal mode output of C460 dye by inserting an F-P etalon.[55] But the slope efficiency was only 3% because of the high reflectance of the etalon. By optimizing the cavity, the balance between beam quality and conversion efficiency could be reached. J. A. Russell et al. investigated the beam quality of solid-state dye lasers with multimode resonator, hard-edge unstable resonator and unstable resonator with a gradient reflective mirror (GRM).[124] The laser efficiency in multimode stable cavity was high, but M2 was about 140-150. In unstable cavity, the divergence of output laser decreased and M2 was about 11-12. With GRM as output coupler, the laser efficiencies of dye-doped polymeric media in unstable cavity increased, approaching that in stable resonator, while good beam quality with M2 about 5-6 was also obtained. When the laser media were deployed at Brewster angle or the sample surfaces were anti-reflection (AR) coated, the degree of parasitic oscillation due to the reflection of sample surfaces could be reduced, and thus the laser efficiency and beam quality could be improved.[125] 8.5.4 Narrow linewidth tunable output of solid-state dye lasers To realize narrow linewidth output, the wavelength selective resonator is the key technical issue. Up to now, several kinds of cavities have been employed to obtain tunable solid-state dye laser output. The cavity adopted by F. J. Duarte is typical, as shown in Fig. 8.12.[95'n6]

8.5 Solid-State Dye Lasers and Parameter Optimization

287

The cavity consists of an output coupler, beam expander composed by several prisms and a holographic grating deployed in Littrow configuration. Wavelength tuning is realized by rotating the grating to select a wavelength for laser oscillation. With a R6G doped P(MMA:HEMA) copolymer sample, 350 MHz of linewidth and near TEMQO laser output was achieved with 2% of conversion efficiency. Solid-state gain medium Grating

s: M

Fig. 8.12 Schematic of the solid-state multiple-prism Littrow grating dye-laser oscillator (The multiple-prism expander is deployed in a compensating configuration).

This cavity is very compact and the tuning method is simple. However, in order to obtain narrower linewidth output, precisely designed prisms expander is needed to cover the full length of the Littrow deployed grating. The resolution of output laser can be further improved by deploying the grating to grazing incidence. D. Lo et al. realized the tunable solid-state dye laser output with a double-grating oscillator, as shown in Fig. 8.13.[96]

Pumping laser | j «e====2S==» Cylindrical lens Grating

<*-• Satire

-«*

-*-To detection system Laser output

Output coupling mirror

Grating Fig. 8.13 Schematic of the solid-state double grating dye-laser oscillator.

288

Hybrid Organic-Inorganic Solid-State Dye Laser Glasses

The optical arrangement consists of a holographic grating set in grazing incidence for beam expanding and a Littrow grating for tuning. Narrow linewidth as short as 3.3 GHz and in 2 or 3 longitudinal modes were realized with the R6G, C460 and Exalite 377 doped in gel glass, respectively. These results were comparable to those obtained from liquid dye laser using the same double grating cavity. The linewidth was limited by grating cavity. Considering the low energy conversion efficiency due to the transversal pumping and low resolution of the Littrow gratings, we developed a kind of solid-state tunable dye laser instrument pumped longitudinally,1127"1281 as was shown in Fig. 8.14.

Fig. 8.14 Schematic of the longitudinally pumped solid-state dye laser oscillator (1: reflector; 2: aperture; 3: beam splitter; 4: focus lens; 5: high reflectivity coupler; 6: experimental sample; 7: holographic grating; 8: output coupler).

The optical arrangement of the cavity consists of a dichromic mirror, an holographic grating set in grazing incidence for beam expanding and an output coupler for tuning. Tuning is achieved by rotating the output coupler, which feedback a specific wavelength for laser oscillation. With the P567 doped MTES-derived ORMOSILs, the tunable range from 547.2nm to 598.2nm and energy conversion efficiency of 11.4% were obtained.[81] Efficient tunable output of RB, R6G, P-orange and P-red doped into various ORMOSILs have also been obtained. The typical linewidth of the tunable laser was 0.014 nm, which is expected to be narrower by improving cavity. However, the direction of output laser was changing in the tuning process because of the rotation of output coupler, which also brought inconvenience to practical use. Another kind of narrow linewidth operation solid-state dye laser equipment is the distributed feedback (DFB) laser established by W. J.

8.5 Solid-State Dye Lasers and Parameter Optimization

289

Wadsworth et al.,[97] as shown in Fig. 8.15, utilizing the spatial distribution of the gain of laser dyes, which was first established by C. V. Shank et al.[129] Two identical pump beams converge onto the sample surface with the same incidence angle 6, which is controlled by the two mirrors. A prism is deployed in front of the sample to eliminate the difference in refractive index between medium and air. Thus, the output wavelength of the DFB laser can be given:

Aoutput =

l<„/sin0'

(8.8)

Telescope x 5

a> i %

t-y Fig. 8.15 DFB laser configuration: A, aperture; C, cylindrical lens; G, diffraction grating; Mi and M2, high reflectance steering mirrors; P, right angle prism; D, dye doped slab.

Tuning of the DFB laser is achieved by altering the convergence angle of the two pump beams. For tuning over large ranges, mirrors Ml and M2 are rotated symmetrically and the dye slab is re-positioned so that the dye slab face is remained at the intersection of the two pump laser beams. However, such a tuning operation is a little complicated. In such a configuration, tunable range from 604 nm to 649 nm, 4x 10" rad of divergence angle, and 20% of conversion efficiency were obtained for P-red doped in MPMMA.[97] Among the above cavities, DFB lasers are advantageous, because it eliminates the need for lossy dispersive elements. The wavelength selective feedback of DFB laser for laser action is provided by Bragg reflection from a grating structure within the gain medium, which may be dynamically written, or may be permanently written into the gain medium, as be common in DFB semiconductor diode lasers.

290

Hybrid Organic-Inorganic Solid-State Dye Laser Glasses

Recently, D. Lo et al. made some interesting modifications on the DFB laser.'130"1461 These research efforts are promising for further miniaturization of the solid-state dye lasers, integrated optics. 8.6 DFB Laser Based on Sol-Gel Derived Organic-Inorganic Hybrid Thin Film Waveguides Wave Division Multiplex (WDM) technique is widely used in modern optical communication network and integrated optics. DFB technique and laser devices are very suited for application. In place of a conventional resonator cavity, optical feedback was provided via backward Bragg scattering from periodic perturbations of the refractive index and the gain of the laser medium.'1471 This mechanism results in a compact laser structure and realizes narrow lasing emission easily. Recently, passive and active planar optical waveguides based on solgel materials have generated much interest because of the versatility of sol-gel technique and the potential applications in integrated optics. Since 2000, D. Lo started systematic investigation on DFB laser actions based on sol-gel glass materials. They realized dye-doped silica DFB lasers tunable in the visible and UV region '130' m i and demonstrated temperature tuning of solid-state DFB dye lasers.1131'138'144] Later, they demonstrated DFB lasers based on sol-gel glass planar waveguides: wavelength-tunable DFB titania-silica waveguide lasers,'1341 single- and multi-wavelength DFB zirconia and zirconia-ORMOSIL waveguide lasers,'135"1361 DFB sol-gel glass symmetric waveguide lasers,[137] DFB sol-gel channel waveguide lasers [143'145] and two-photon-pumped DFB zirconia wavesuide laser.[146] A brief review is presented on the main DFB sol-gel dye laser research works by D. Lo in recent years. 8.6.1 Multi-wavelength DFB laser based on sol-gel glass thin film Lasers with multi-wavelength output have found applications in sensors and lidars for environment monitoring. Multi-wavelength DFB laser output has been observed by D. Lo et al. with rhodamine dye doped titania-silica glass thin films and zirconia-titania-ORMOSIL thin

8.6 DFB Lasers Based on Sol-Gel Derived Hybrid Thin Film Waveguides

291

films. J Lasing with multiple guiding modes was achieved at a film thickness of 1.1 urn with a R6G doped titania-silica waveguide spincoating on glass substrate. In R6G doped zirconia waveguides, dualwavelength output was achieved for a film of 1.1 um thickness, whereas quadruple-wavelength output was observed for a 2.2um thick film.[136] 8.6.2 DFB laser based on zirconium-ORMOSIL thin film channel waveguides Of the several types of waveguides in use, the planar waveguide lacks confinement and the circular waveguide is not compatible with planar processing technology. In integrated optics, the rectangular dielectric waveguide is the most commonly used structure on which many of the active or passive devices are in fact based. The rectangular waveguides are usually rectangular dielectric strips embedded in other dielectrics of lower refractive index. Active centers or junction structure must be built in the dielectric strips to render the waveguides optically active (e.g., waveguide lasers). Casalboni et al. reported light amplification in dyedoped sol-gel channel waveguides.[147] Recently, D. Lo et al. fabricated dye-doped zirconia channel waveguides using wet or dry etching of quartz substrates followed by sol-gel deposition of R6G-doped zirconia or LDS-doped zirconiaORMOSIL in the channels. By crossing two identical nanosecond Nd:YAG laser beams at 532nm, near infrared DFB laser output was obtained in the LDS 925-doped zirconia-ORMOSIL channel waveguide. Wavelength tuning was achieved from 787nm to 933nm. The output laser mode was identified as the fundamental mode. 8.6.3 Two-photon pumped zirconium waveguide DFB laser Compared with lasers obtained from one-photon pumping, the solid-state dye lasers achieved from Two-photon pumping are advantageous: (1) the degradation probability of dye molecules resulting from the photodissociation decreases; (2) the problem of heat dissipation resolved because the absorption and scattering of host matrix are low in the longer wavelength spectrum region whereas the transmission is high. Recently,

292

Hybrid Organic-Inorganic Solid-State Dye Laser Glasses

intensive efforts have been devoted to the synthesis and characterization of new laser dyes with large two-photon absorption cross sections.1148"1501 Among dozens of newly synthesized laser dyes, trans-4-[p-(Nhydroxyethyl-n-methylamino)styryl]-N-methylpyridinium p-toluene sulfonate (HMASPS) demonstrates strong two-photon upconverted emission when pumped at 1.06um. D. Lo et al. fabricated HMASPS doped zirconia thin films and realized upconverted lasing. By crossing two identical nanosecond 1.06um Nd:YAG laser beams, wide band tuning of the DFB lasing was achieved, from 618nm to 640nm. The twophoton DFB laser linewidth was also found to be 45GHz (or 60pm). 8.6.4 Solid-state dye lasers and DFB laser in the NIR and IR region Recently, IR26 dye-doped zirconia-ORMOSIL thin film planar waveguides were fabricated in our lab. Depending on the viscosity of the sol, spin speed and the thermal treatment, films of thickness varying from 2.4 to 2.8um were obtained. The absorption peak of the IR26 doped in zirconia-ORMOSIL thin film was at about 1.08um. Excited by a Nd: YAG laser at 1064nm, the fluorescence of IR26 doped in zirconia-ORMOSIL waveguide was observed, with a peak at about 1.1 lum. Increasing the pumping density above 8.6mJ/cm2, ASE centered at about 1.14um with FWHM of ~15nm was achieved. The photostability of IR26 doped in zirconia-ORMOSIL film was investigated at a repetition rate of 10Hz and pump intensity of higher than 50mJ/cm2. The ASE output intensity decreased to 50% of its initial value after 1200 pump pulses. The IR26 doped zirconia-ORMOSIL waveguides were promising for future DFB operations. 8.7 Summary and Future Prospects After decades of development, solid-state dye lasers have obtained great progress in both materials and devices and some have reached the commercial market place. These were made possible mainly by three factors: synthesis of new laser dyes and host matrix, development of fabrication techniques and new findings in photophysical and

8.7 Summary and Future Prospects

293

photochemical process. Before the late 1990s, the major obstacle in this field is the photostability of laser dye itself. For instance, there is rarely a laser dye more efficient and stable than the traditional rhodamine family dyes in the yellow to red region, until the synthesis of pyrromethene and perylene dyes. Two order of magnitude or even greater improvement has been observed in the photostability of dyes doped in solid matrix in this spectrum region. In the future, high performance laser dyes emitted at other regions are highly needed, for instance, stable infrared laser dyes will be synthesized. New kind of dyes with various non-linear optical properties will also generate interests. As mentioned before, each laser dye has its specific preferred host matrix.[25] Photophysical and photochemical processes and the mechanism responsible for the dye degradation have been extensively investigated during the last decade. These efforts provided the theoretical and experimental basis for the designing and controlling on the microenvironment of dye molecules. New solid host matrix and the fabrication techniques will be developed. Organic-inorganic hybrid materials via sol-gel process are superior due to versatility and low cost. Concerning the actual use in integrated optics, planar and channel waveguides and the DFB configuration will generate more interests. Permanent DFB structures will be optically written on the hybrid waveguiding materials, concerning the need for device simplicity, as have been studied by some researchers.[15M53] Y. Oki et al. fabricated grating structure in polymer thin films through the method of UV irradiation lithography. In recent years, periodic structures in solid materials fabricated by femtosecond laser through multi-photon process have also been investigated.11541 Permanent grating structures were also written into VTES-derived ORMOSILs by femtosecond laser with diffraction efficiency as high as 35% in our lab, which provides another possibility for the establishment of DFB lasers.[155] It is not a surprise that hybrid organic-inorganic solid-state dye glass may find more actual applications in the near future. Acknowledgements: The authors gratefully acknowledge the financial support for this work from the National Natural Science Foundation of China (Nos. 59572020, 69890230, 59902005, 90101007 and 50532030),

294

Hybrid Organic-Inorganic Solid-State Dye Laser Glasses

the Foundation for the Author of National Excellent Doctoral Dissertation of P. R. China (No. 200134), Trans-Century Training Program Foundation for the Talents by the Ministry of Education of P. R. China and Education Foundation of FOK Ying Tung.

References 1. M. Bass, T. F. Deutsch and M. J. Weber, "Dye Lasers" in Lasers, Vol. 3, ed. A. K. Levine and A. DeMaria, Marcel Dekker, Inc., New York, 1971, p275 2. Laser Photonics, Inc., 12351 Research Parkway, Orlando, FL 32826, formerly, Molectron Corporation and Cooper LaserSonics, Inc. 3. T. G. Pavlopoulos, Prog. Quan. Electr.. 26 (2002) 193. 4. T. G. Pavlopoulos, J. H. Boyer, M. Shah, et al., Appl. Opt. 29 (1990) 3885. 5. J. H. Boyer, A. Haag, M. L. Soong, et al., Appl. Opt. 30 (1991) 3788. 6. T. G. Pavlopoulos and J. H. Boyer, SPIE Vol. 2115 (1994) 231. 7. G. Seybold and G. Wagenblast, Dye and Pigment 11 (1989) 303. 8. R. Gvishi, R. Reisfeld, Z. Bruhstein and E. Miron, SPIE Vol. 1972 (1990) 390. 9. R. Reisfeld, R. Gvishi and Z. Bruhstein, J. Sol-Gel Sci. Technol. 4 (1995) 49. 10. G. Jones II, Z. Huang, S. Kumar and D. Pacheco, SPIE Vol. 4630 (2002) 65. 11. A. Bergmann, W. Holzer, R. Stark, H. Gratz, A. Penzkofer, F. A. Guerri, A. Costela, I. G. Moreno and R. Sastre, Chem. Phys. 271 (2001) 201. 12. Y. Yang, M. Q. Wang, G. D. Qian,et al, Opt. Mater. 24 (2004) 621. 13. R. Reisfeld, Opt. Mater. 4 (1994) 1. 14. M. P. O'Neil, Opt. Lett. 18 (1993) 37. 15. R. E. Hermes, T. H. Allik, S. Chandra, et al, Appl. Phys. Lett. 63 (1993) 877. 16. A. A. Manpnkov and V. S. Nechitailo, Sov. J. Quan. Electr., 10 (1980) 347. 17. A. A. Gorman, I. Hamblett, T. A. King and M. D. Rahn, J. Photochem. & Photobio. A: Chem. 130 (2000) 127. 18. M. Faloss, M. Canva, P. Georges, A. Brun, et al., Appl. Opt. 36 (1997) 6760. 19. F. L. Arbeloa, T. L. Arbeloa, I. L. Areloa, et al., Chem. Phys. 236 (1998) 331. 20. F. L. Arbeloa, T. L. Arbeloa, I. L. Areloa, et al, Chem. Phys. Lett. 299 (1999) 315 21. M. D. Rahn and T. A. King, SPIE Vol. 3613 (1999) 94. 22. T. G. Pavlopoulos, SPIE Vol. 3613 (1999) 112. 23. G. Jones II, O. Klueva, S. Kumar and D. Pacheco, SPIE Vol. 4267 (2001) 24. 24. T. Suratwala, K. Davidson, Z. Gardlund and D. R. Uhlmann, SPIE 2986 (1997) 141. 25. G. D. Qian, Y. Yang, Z. Y. Wang, et al., Chem. Phys. Lett. 368 (2003) 555.

References 26. 27. 28. 29. 30.

295

F. P. Scafer, Dye lasers, Springer-Verlag, 1973. B. H. Soffer and B. B. McFarland, Appl. Phys. Lett. 10 (1967) 266. O. G. Perterson and B. B. Snavely, Appl. Phys. Lett. 12 (1968) 238. V. I. Bezrodnyl, Sov. J. Quantum. Electron. 12 (1982) 1602. V. S. Nechitailo, R. S. Anderson, S. C. Picarello, G. A. Matyushin and J. H. Bohn, SPIE Vol. 3613 (1999) 106. 31. A. Costela, I. G. -Moreno, J. Barrpsp and R. Sastre, J. Appl. Phys. 83 (1998) 650. 32. S. M. Giffin, I. T. Mcninie, W. J. Wadsworth, et al., Opt. Commun. 161 (1999) 163. 33. W. J. Wadsworth, S.M. Giffin, I. T. Mcninie, et al., Appl. Opt. 38 (1999) 2504. 34. F. J. Duarte, A. Costela, I. G. -Moreno and R. Sastre, Appl. Opt. 39 (2000) 6522. 35. R. Sastre and A. Costela, Adv. Mater. 7 (1995)198. 36. A. Costela, I. G. -Moreno, J. Barroso and R. Sastre, Appl. Phys. B 70 (2000) 367. 37. A. Costela, I. G. -Moreno, C. Gomez, et al, Appl. Phys. 90 (2001) 3159 38. A. Costela, I. G. -Moreno, M. L. Carrascoso and R. Sastre, Opt. Commun. 201 (2002) 437 39. A. Costela, I. G. -Moreno, C. Gomez, et al., Chem. Phys. Lett. 369 (2003) 656. 40. R. S. Anderson, R. E. Hermes, G. A. Matyushin, et al., SPIE 3265 (1998) 13. 41. S. C. Picarello, R. S. Anderson, V. S. Nechitailo, SPIE 3590 (1999) 43. 42. K. C. Yee, T. Y. Tou and S. W. Ng, Appl. Opt. 37 (1998) 6381. 43. D. P. Pacheco and H. R. Aldag, SPIE 3265 (1998) 2. 44. M. Ahmad, M. D. Rahn and T. A. King, Appl. Opt. 38 (1999) 6337. 45. G. Jones II, O. Klueva, S. L. -Samanniego et al., SPIE 3265 (1998) 51. 46. M. D. Rahn, T. A. King, A. A. Gorman and I. Hamblett, Appl. Opt. 36 (1997) 5862. 47. M. Ahmad, T. A. King, D. -K. Ko, et al, Opt. Laser Tech. 34 (2002) 445. 48. A. Costela, I. G. Moreno, C. Gomez, et al., Chem. Phys. Lett. 387 (2004) 496. 49. R. Reisfeld, Opt. Mater. 16 (2001) 1. 50. E. Yariv, S. Schultheiss, T. Saraidarov and R. Reisfeld, Opt. Mater. 16 (2001) 29. 51. H. Dislish, Argew. Chem. Int. Ed. 10 (1971) 363. 52. B. E. Yoldas, J. Non-Cryst. Solids. 51 (1982) 105. 53. H. Nasu, Opt. Eng. 26 (1987) 102. 54. J. Lin, J. Inorg. Mater. 12 (1997) 363 (in Chinese). 55. S. K. Lam, X. L. Zhu and D. Lo, Appl. Phys. B 68 (1999) 1151. 56. D. Avnir, D. Levy, R. Reisfeld, J. Phys. Chem. 88 (1984) 5956. 57. K. S. Lam, D. Lo and K. H. Wong, Appl. Opt. 34 (1995) 3380. 58. K. S. Lam and D. Lo, Appl. Phys. B 66 (1998) 427. 59. S. X. Wu and C. S. Zhu, J. Mater. Lett. 18 (1999) 281. 60. S. X. Wu and C. S. Zhu, Opt. Mater. 12 (1999) 99. 61. E. J. A. Pope and J. D. Mackenzie, J. Mater. Res. 4 (1989) 1018. 62. H. R. Aldag, S. M. Dolotov, M. F. Koldunov.et al., SPIE 3929 (2000) 133. 63. M. D. Rahn and T. A. King, J. Mod. Opt. 45 (1998) 1259. 64. F. J. Durate and R. O. James, Opt. Lett. 28 (2003) 2088. 65. J. C. Altman, R. E. Stone, F. Nishida and B. Dunn, SPIE 1758 (1992) 507.

296 66. 67. 68. 69.

Hybrid Organic-Inorganic Solid-State Dye Laser Glasses

Y. Zhang and M. Q. Wang, J. Non-Cryst. Solids, 271 (2000) 88. X. C. Li and T. A. King, SPIE 2288 (1994) 216. M. D. Rahn and T. A. King, SPIE 2288 (1994) 382. C. Sanchez and F. Ribot, In: Proceedings of 1st European Workshop on Hybrid Organic-Inorganic Materials (Synthesis, Properties and Applications), Chateau de Bierville, France, Nov. 8-10, 1993, p9 70. H. Schmidt and B. Seiferling, In: Better Ceramics through Chemistry 0, C. J. Brinker, D. E. Clark and D. R. Ulrich(Eds), Mater. Res. Soc, Pittsburgh, 73 (1986) 739. 71. B. Dunn, F. Nishida, J. C. Altman and R. E. Stone, Chemical Processing of Advanced Materials, L. L. Hench and J. K. West(Eds), John Wiley and Sons, NY, 1992, p941. 72. D. Avnir, V. R. Kaufman and R. Reisfeld, J. Non-Cryst. Solids, 74 (1985) 395. 73. J. C. Altman, R. E. Stone, F. Nishida and B. Dunn, Proc. SPIE Vol. 1758 (1992) 507. 74. C. A. Capozzi, Proc. SPIE Vol. 970 (1988) 135. 75. M. Canva, A. Dubois, P. Georges, A. Bran, F. Chaput, A. Ranger and J. P. Boilot, SPIE Vol. 2288(1994)298. 76. A. Dubois, M. Canva, A. Bran, F. Chaput and J. P. Boilot, Appl. Opt. 34 (1995) 428. 77. Q. Y. Zhang and Z. H. Jiang, J. Inorg. Mater. 13 (1998)145 (in Chinese). 78. Q. Y. Zhang and Z. H. Jiang, Mater. Chem. Phys. 39 (2001) 95. 79. Q. Y. Zhang, W. X. Que, S. Buddhudu and K. Pita, J. Phys. Chem. Solids, 63 (2002) 1723. 80. G. D. Qian, M. Q. Wang, Y. Yang,, Chinese Patent, ZL01117640.7. 81. Y. Yang, G. D. Qian, Z. Y. Wang and M. Q. Wang, Opt. Commun. 204 (2002) 277. 82. T. H. Nhung, M. Canva, T. T. A. Dao, et al., Appl. Opt. 42 (2003) 2213. 83. T. Suratwala, Z. Gardlund, K. Davidson and D. R. Uhlmann, J. Sol-Gel Sci. Tech. 8 (1997)953. 84. T. Suratwala, Z. Gardlund, K. Davidson and D. R. Uhlmann, J. Sol-Gel Sci. Tech. 8 (1997)973. 85. F. L. Arbeloa, J. B. Prieto, I. L. Areloa, et al., Photochem. Photobio. 78 (2003) 30. 86. F. A. Guerri, M. Liras, M. L. Carrascoso and R. Sastre, Photochem. Photobio. 77 (2003) 577. 87. A. Dubois, M. Canva, A. Bran,et al., Appl. Opt. 35 (1996) 3193. 88. B. Dunn, F. Nishida and R. Toda, Mat. Res. Soc. Symp. Proc. 329 (1994) 267. 89. M. Canva, P. George and J. Perelgritz, Appl. Opt. 34 (1995) 428. 90. T. A. King, M. Ahmad, A. A. Gorman, et al., SPIE Vol. 3929 (2000) 145. 91. E. Yariv and R. Reisfeld, Opt. Mater. 13 (1999) 49. 92. M. S. Mackey and W. N. Sisk, Dye and Pigment 51 (2001) 79. 93. M. Ahmad, T. A. King, D. K. Ko, et al., Opt. Commun. 203 (2002) 327. 94. Y. Yang, G. D. Qian, D. L. Su, et al., Opt. Commun., 239(2004) 415-420. 95. F. J. Duarte, T. S. Taylor, A. Costela, et al., Appl. Opt. 37 (1998) 3987.

References

297

96. D. Lo, S. K. Lam, C. Ye and K. S. Lam, Opt. Commun. 156 (1998) 316. 97. W J. Wadsworth, I. T. Mckinnie, A. D. Woolhouse and T. G. Haskell, Appl. Phys. B 69 (1999) 163. 98. O. G. Peterson and B. B. Snavely, Bull. Am. Phys. Soc. 13 (1968) 397. 99. H. K. Law, T. Y. Tou and S. W. Ng, Appl. Opt. 37 (1998) 5694. 100. M. Fukuda and K. Mito, Jpn. J. Appl. Phys. 39 (2000) 3470. 101. P. J. Sebastian and K. Sathyanandan, Opt. Commun. 35 (1980) 113. 102. T. Urisu and K. Kajiyama, J. Appl. Phys. 47 (1976) 3563. 103. D. Mohan, S. Sanghi, R. D. Singh, et al., J. Photochem. Photobiol. A., 68 (1992) 77. 104. G. A. Kumar and N. V. Unnikrishnan, J. Photochem. Photobiol A., 144 (2001) 107. 105. L. Tharejja, K.Sharma and R. D. Singh, Opt. Commun. 26 (1978) 81. 106. M. Ali, SA. Ahmed and K. Mitwally, Appl. Opt. 28 (1989) 3709. 107. R. Ghazy, S. A. Zim, M. Shaheen and F. El-Mekawey, Optics & Laser Tech. 34 (2002) 99. 108. Y. Yang, G. D. Qian, D. L. Su, et al., Chem. Phys. Lett. 402 (2005) 389. 109. D. L. Su, Y. Yang, G. D. Qian, et al, Chem. Phys. Lett. 397(2004) 397-401. 110. D. C. Lai, B. Dunn and J. I. Zink, Inorg. Chem. 35 (1996) 2152. 111. T. H. Nhung, M. Canva, F. Chaput, H. Goudket, G. Roger, A. Brun, D. D. Manh, N. D. Hung and J. P. Boilot, Opt. Commun. 232 (2004) 343. 112. A. K. Sheridan, A. R. Buckley, A. M. Fox, A. Bacher, D. D. C. Bradley and I. D. W. Samuel, J. Appl. Phys. 92 (2002) 6367. 113. K. -S. Kang, W. N. Sisk, M. Y. A. Raja and F. Farahi, J. Photochem. & Photobio. A., 121(1999)133. 114. W. N. Sisk, N. Ono, T. Yano and M. Wada, Dyes & Pigment 55 (2002) 143. 115. W. N. Sisk and W. Sanders, J. Photochem. & Photobio. A., 167 (2004) 185. 116. N. Tanaka and W. N. Sisk, J. Photochem. & Photobio. A., 172 (2005) 109. 117. A. J. Finlayson, N. Peters, P. V. Kolinsky and M. R. W. Venner, Appl. Phys. Lett. 75(1999)457. 118. A. Mandl, A. Zavriyev and D. E. Klimek, SPIE Vol. 3265 (1998) 38. 119. A. Mandl and D. E. Klimek, SPIE Vol. 3613 (1999) 119. 120. A. J. Finlayson and M. R. W. Venner, SPIE Vol. 3929 (2000) 154. 121. J. C. Altman, R. E. Stone, B. Dunn,et al., IEEE Photo. Tech. Lett. 3 (1991) 189. 122. Y. Oki, S. Miyamoto, M. Tanaka, et al., Opt. Commun. 214 (2002) 277. 123. K. M. Abedin, M. Alvarez, A. Costela, et al., Opt. Commun. 218 (2003) 359. 124. J. A. Russell, D. P. Pacheco, H. R. Aldag, et al., SPIE Vol. 4267 (2001) 36. 125. F. J. Duarte, Opt. Laser Tech. 29 (1997) 513. 126. F. J. Duarte, Appl. Opt. 38 (1999) 6347. 127. Z. Hong, Z. Y. Wang, Y. Yang, et al., Chin. Phys. Lett. 18 (2001) 225. 128. Y. Yang, M. Q. Wang, G. D. Qian, et al., Mater. Lett. 57 (2002) 660. 129. C. V. Shank, J. E. Bjorkholm and J. Kogelnik, Appl. Phys. Lett. 18 (1971) 395. 130. X. L. Zhu, S. K. Lam and D. Lo, Appl. Opt. 39 (2000) 3104. 131. X. L. Zhu, S. K. Lam and D. Lo, SPIE Vol. 4267 (2001) 102.

298 132. 133. 134. 135. 136. 137. 138. 139. 140. 141. 142. 143. 144. 145. 146. 147. 148. 149. 150. 151. 152. 153. 154. 155.

Hybrid Organic-Inorganic Solid-State Dye Laser Glasses X. L. Zhu and D. Lo, Appl. Phys. Lett. 77 (2000) 2647. J. Wang, G. -X. Zhang, L. Shi, D. Lo and X. L. Zhu, Opt. Lett. 28 (2003) 90. X. L. Zhu and D. Lo, Appl. Phys. Lett. 80 (2002) 917. D. Lo, L. Shi, J. Wang, et al, Appl. Phys. Lett. 81 (2002) 2707. J. Wang, G. -X. Zhang, L. Shi, et al, Opt. Lett. 28 (2003) 90. L. Shi, G. -X. Zhang, J. Wang and D. Lo, J. Opt. A: Pure Appl. Opt. 5 (2003) LI X. L. Zhu, L. Shi, J. Wang, C. Ye and D. Lo, SPIE Vol. 5177 (2003) 146. X. L. Zhu and D. Lo, Chin. Opt. Lett. 1 (2003) 99. D. Lo, C. Ye and J. Wang, Appl. Phys. B 76 (2003) 649. C. Ye, L. Shi, J. Wang, D. Lo and X. L. Zhu, Appl. Phys. Lett. 83 (2003) 4101. C. Ye, J. Wang, L. Shi and D. Lo, Appl. Phys. B 78 (2004) 189. F. Chen, J. Wang, C. Ye, D. Lo and X. L. Zhu, Appl. Phys. Lett. 85 (2004) 4284. X. L. Zhu, L. Shi, J. Chan, et al., Opt. Commun. 251 (2005) 322. F. Chen, J. Wang, C. Ye, et al., Opt. Exp. 13 (2005) 1643. J. Wang, C. Ye, F. Chen, L. Shi and D. Lo, J. Opt. A: Pure Appl. Opt. 7 (2005) 261. M. Casalboni, F. De Matteis, V. Merlo,et al., Appl. Phys. Lett. 83 (2003) 416. G. S. He, L. X. Yuan, Y. P. Cui, et al., J. Appl. Phys. 81 (1997) 2529. G. S. He, T. -C. Lin, V. K. S. Hsiao, et al., Appl. Phys. Lett. 83 (2003) 2733. G. Y. Zhou, D. Wang, Y. Ren, et al., Appl. Phys. B, 79 (2004) 341. T. Kobayashi, J. -B. Savatier, G. Jordan, et al., Appl. Phys. Lett. 85 (2004) 185. B. Darracq, F. Chaput, K. Lahlil, et al, Adv. Mater. 10 (1998) 1133. Y. Oki, T. Yoshiura, Y. Chisaki and M. Maeda, Appl. Opt. 41 (2002) 5030. J. F. Zhai, Y. Q. Shen, J. H. Si, et al, J. Phys. D: Appl. Phys. 34 (2001) 3466. G. D. Qian, J. Y. Guo, M. Q. Wang,et al., Appl. Phys. Lett. 83 (2003) 2327.

Chapter 9

Optical Glass Waveguides

Optical waveguides are micron size channels that can be used to guide light. Although optical fiber is actually one special type of waveguides, usually waveguides refer to optical integrated circuits on substrates. The concept of optical waveguide came out in early 1960s. However, the first widely successfully used waveguide is semiconductor diode laser, where the device was etched to channels and sandwiched between p-type and ntype semiconductor materials with a little lower refractive index. Therefore, light is well confined in a thin gain layer and as the result, power threshold to achieve laser action was drastically reduced. Glass waveguides are important devices. In the past 20 years, with the rapid development of optical communication systems, requirement for compact, high density optical devices increased tremendously. Especially after the wavelength division multiplexing (WDM) technique was widely accepted as a new generation of optical communication standard platform, many new device concepts were proposed, devices with excellent performance were put into market with vast quantities. An optical waveguide is actually a three or more layer thin film assembly, consists basically a substrate, a lower buffer layer, a guiding layer and a top cladding layer. Glass materials for waveguides fabrication can be categorized to silica glass, multi-component glass and polymeric glass materials. Various methods were developed in the past 40 years to fabricate waveguide films, these methods include chemical vapor deposition (CVD), ion-exchange, flame-hydrolysis deposition (FHD), Sol-gel, etc. Conventional glass waveguides are used simply to guide and split light, just like electric current wires in IC chips. However, more 299

300

Optical Glass

Waveguides

functions were explored, which widely expand the utilities of glass waveguides. They can be used as optical switches, tunable frequency filters. Even more fancy works have been done to investigate the possibility of using glass material in on-chip optical signal amplifiers and electro-optical modulators. 9.1 Principles of Optical Waveguides A standard planar waveguide has three layers, buffer layer, core layer (thickness d) and top cladding layer as shown in Fig. 9.1.

>Y n3 Fig. 9.1 Diagram of a three layer waveguide structure with refractive indices of rii, n2, n3 respectively. Light is guided in the n2 medium.

The refractive index of the three layers is nlt ri2, n3 respectively, they satisfy the following conditions: n2 > nx,n2 > n3.

(9.1)

Propagation of light in waveguides can be understood when considering light as electro-magnetic wave which follows Maxwell's wave equation V2E(r,t) = n2(r)/c2d2E(r,t)/dt2,

(9.2)

where E is the electric field vector, r is the radius vector, n(r) is the refractive index distribution, and c is the speed of light in vacuum.

9.1 Principles of Optical Waveguides

301

Considering monochromatic wave approximation, and the three layer structure shown in Fig. 9.1, Eq. (9.2) can be simplified to be: x>d, d2E(x,y)/d2x2+(k2n2-j32)E(x,y) 0<x
+(k2n22 +(k2n32

= 0, - j32)E(x, y) = 0,

- J32)E(x,y)

(9.3)

= 0,

where j3 is the propagation constant in z direction, E(x,y) is the electric field in one polarization. Considering further the boundary conditions at the interface between buffer layer/core layer and cladding layer/core layer, light wave with limited discrete number of P can propagate through the waveguide. These propagating waves are called waveguide modes. Propagation constant pn of the nth waveguide mode satisfies the following mode equation: And IX^n2

- N2 =

fa+fc+lmx,

(9.4)

where the phase shifts at n]/n2 and n3/n2 interface are:

0, = 2tarT

,1=1,3

(9.5)

where Xo is the vacuum wavelength, N is the effective refractive index which corresponding to propagation constant /? with the relation fi^ioN (k0= vacuum wave vector), F=\ if the E-field is polarized perpendicular to the plane of incidence (TE modes), and F = n2ln2,

(9.6)

for E-field polarization parallel to the plane of incidence (TM modes). From Eq. (9.4), it can be seen that for a waveguide structure of rti=n3 (so-called symmetric waveguide), there is in principle no limitation of core layer thickness d to support the fundamental mode propagation. On the other hand, for an asymmetric waveguide (nj ^ n3), a minimum thickness dc exists to sustain at least one mode. dc is named the cutoff thickness. We will show later that the concept of cutoff thickness can have very important utilities.

302

Optical Glass Waveguides

9.2 Glass Waveguides Fabrication and Optical Properties 9.2.1 Chemical Vapor Deposition (CVD) CVD is a frequently used technique to prepare glass waveguides, especially silica or silicon nitride films. Take the fabrication of silica waveguides as an example, precursors such as SiCU, GeCl4 are mixed with 0 2 and react at very high temperature (1100-1300 °C) to form Si0 2 and Ge0 2 soot. After annealing at 1300°C for some time, the soot turns to dense optical films. Plasma technique was introduced later to fabricate films with better uniformity and lower the fabrication temperature. The technique is thus named plasma enhanced CVD or PECVD.[2] Films with thickness 10-20 urn can be obtained. 9.2.2 Sputtering The advantage of sputtering is the possibility of depositing multicomponent glass materials. Deposition happens in vacuum chamber where both a target and a substrate are put in. The target is bombarded by inert ions (like argon gas) driven by discharge field, target material is sputtered out and deposited on to the substrate. In principle, the deposited film keeps the same composition as the target. To deposit oxide films, oxygen gas is used instead of inert ions to compensate the lost oxygen component in the material during sputtering. Magnetron radiao frequency (rf) sputtering allows discharge and sputtering to occur at very low pressure (typically 1 mTorr of discharge gas pressure).[3] 9.2.3 Flame hydrolysis deposition (FHD) Flame hydrolysis deposition (FHD) is a technique to prepare silica films on silicon wafer.[4] The process of FHD is very similar to the method of preparing fiber performs. Precursors like SiCl4, GeCU are burned in H 2 /0 2 flame and hydrolyzed by hot water vapors generated by the torch, soot of Si0 2 and Ge0 2 are formed on silicon substrate. After deposition of the lower cladding and core layer, the substrate with these porous glass layers is heated to about 1300°C for consolidation. Channel

9.2 Glass Waveguides Fabrication and Optical Properties

303

waveguides are then formed by conventional photolithography and reactive ion etching (RIE). Finally, FHD is used again to cover the ridges with an overcladding. The whole process is schematically shown in Fig. 9.2.[5] FHD

IH

J

WIS

SiC>2-Ge02 glass particle Si0 2 glass particle

Hi Consolidation

-Core - Underdaddng

ithograph y&RIE W Core

FHD

Consolidation

*

•

• a o 2 glass particle

Overcladding

Fig. 9.2 FHD Frication process of silica-based waveguides. After Reference [5].

FHD is one of the most successful techniques to produce silica optical waveguides with high optical quality. Propagation loss as low as 0.01 dB/cm was achieved. Nippon Telephone and Telecom Co. (NTT) has developed excellent performance dense channel arrayed waveguide grating using silica on silicon systems fabricated by FHD technique. Active elements like rare-earth-element doped waveguide amplifiers and lasers were also explored. In these cases, ErCl3, YbCl3 were used as rare-earth precursors together with other core channel precursors (SiCl4 and GeCl4).

Optical Glass Waveguides

304

9.2.4 Ion-exchange Since the first successful ion exchange waveguide report in 1972,t6] this technique has become a relatively mature waveguide fabrication technique.^1 The principle is to replace the ions in a multi-component glass substrate with ions of larger polarizability, therefore increase the surface refractive index of the glass substrate. Most frequently used ion pair for exchange is Ag+ and Na+. A normal ion exchange process is, heat a salt mixture (like AgNC>3 and NaNOs) over melting temperature (300-350 °C, depending on the composition), put cleaned glass samples into the melted salt for several minutes. Na+ in the glass drifts to the solution, while Ag+ in the melted salt moves into the glass. Ion exchange occurs in a 10 um surface region, generating a gradient refractive index distribution with its maximum on the surface. The refractive index change can be well described by equations :[8] n(x) = n0+An- erfc(x/d)

,

An = n(0)-n0,

(9.7)

where n0 is the refractive index before ion exchange, d is the diffusion depth, De is the diffusion coefficient, erfc(x) is the error function. Table 9.1 Ion exchange parameters for Schott glass IOG-1. Surface An

8wt%AgN03+92wt%KN03 335 °C, 7 minutes

0.034

Diffusion depth Diffusion coefficient d(um) D(u,m2/min) 6.32 1.33

Table 9.2 Ion exchange parameters for BK7 glass. Surface An

5wt%AgN03+95wt%KN03, 360 °C 180 minutes

0.101

Diffusion depth Diffusion d(u.m) coefficient D(um2/min) 8.3

0.0958

9.2 Glass Waveguides Fabrication and Optical Properties

305

Table 9.3 Ion-exchange parameters for soda lime glass. Surface An

2xl0_3mol%AgNO3+ 9.9mol%NaN03 330 °C, 60 minutes

0.045

Diffusion depth Diffusion d(um) coefficient D(um2/min) 5.94

0.15

The asymmetric refractive index distribution can be modified by a second step ion-exchange, either decrease the surface refractive index through reverse Ag+-Na+ exchange, or push the maximum Ag+ distribution into glass by applying external electric field. Tables 9.1-9.3 give the ion exchange parameters and the final refractive index change, as well as the refractive index profile parameters for 3 kinds of frequently used glasses. Schott glass IOG-1 is a phosphate glass doped with Er 2 0 3 (2.31 wt%) and Yb 2 0 3 (3.57 wt%). BK7 and soda lime glasses are popular silicate glasses. All the three kind of glasses are commercially available. 9.2.5 Sol-gel Sol-gel technique was first realized as a chemical method to generate Si0 2 structure through hydrolysis and condensation of precursors (like tetraethoxysilane, TEOS) in acidic environment. After exploration for over 100 years, sol-gel technique has been developed to be a key wet chemical method to produce inorganic materials or organic/inorganic hybrid nanocomposites.[9] Sol refers to nano size colloids suspended in liquids, when these nano-size colloids grow up and connect to each other, the suspension can no longer flow freely, the product is called Gel. Since 1980, researches started to realize that sol-gel technique could be a possible and convenient way to produce thin films with excellent optical quality. A new division of sol gel was developed very quickly and soon formed an important community called Sol-gel Optics.[10] Conventional sol-gel reactions are used to fabricate multi-component oxide films with precisely controlled refractive index. Active elements, like rare-earth elements can be doped in liquid phase with pretty high concentration, thus active waveguides are formed. The most difficulty in

306

Optical Glass Waveguides

this approach is, when depositing thin films on substrate (usually silicon wafer is used as substrate), deposited film thickness should be below 100 nm to avoid cracking, crackingfrequentlyhappens in films prepared by sol-gel technique. To overcome cracking, multi-cycle deposition process is necessary, For each cycle, film thickness is controlled to be less than 100 nm. After coating, the film must be annealed at high temperature (1100-1300 °C), then the next coating cycle starts. A normal waveguide structure consists three layers, the total structure thickness is around 4050 um. Therefore, 400-1000 cycles is needed. Syms et al. developed a spin-coating and rapid thermal annealing (SC-RTA) technique to repeatedly coat thin films on silicon wafer.[u] This technique requires that after each coating, the sample should be treated at 1300°C for a few minutes in a oven that use high power Xe lamps to rapidly heat the samples. We modified the SC-RTA process by a fully computercontrolled multi-layer deposition-annealing apparatus, the apparatus has been successfully used to prepare thick sol-gelfilms.Fig. 9.3 is the AFM image of a 10-um thick sol gel film that we fabricated following the SC-RTA procedure. The film composition was 10%P2O5, 5%A1205, 0.5%Er2O5, 0.5%Yb2O5. The surface roughness is less than 2 nm.

Fig. 9.3 AFM image of a 10 pxa thick sol gel film.

SC-RTA has turned out to be one of the best way to produce erbium doped multi-component sol gel glass films and waveguides. Refractive index of the material can be easily controlled by slightly adjust the ratios of the components. Table 9.4 shows the refractive index of several compositions.

9.2 Glass Waveguides Fabrication and Optical Properties

307

Table 9.4 Refractive indices of erbium doped multi-component sol gel glass. Composition

Refractive index @ 632.8 nm

10%P2O5, 5%A1205, 0.5%Er2O5, 0.5%Yb2O5 10%P2O5 5%P205, 5%B205

1.4668 1.4676

5%P205, 10%B2O3

1.4647 1.4614

9.2.6 Fabrication techniques for non-oxide glass waveguides 9.2.6.1 Fluoride glass waveguides Fluoride glasses were received attention because of their optical transmission in the infrared, theoretical low losses and potential as hosts for active rare-earth ions.[12] In 1990, these glasses emerged as an outstanding class of materials for fiber laser sources and optical fiber amplifiers. Lucas and his group described an ion exchange procedure to generate planar and channel waveguides in fluoride glasses.[13] The process is based on the idea that the F" anion in the glass matrix can be exchanged with its equivalent in charge and size, namely OH" and OD". The basic reaction occurring at the surface of the glass is MFn+H20^

MFn_x (OH)x + xHF

(9 g)

The anionic exchange process takes place at temperatures below, or very close to glass transition temperature Tg. The replacement of F" ions by OH+ or OD" ions produces an increase of the surface refractive index in an order of 0.01 by controlling the exchange time and temperature. R.Almeida el al reported vapor-phase film deposition of rare-earth doped fluorozirconate glasses.tl4] Zirconium fluoride based glassy films include ZrF4-PbF2, ZrF4-PbF2-NdF3 systems. As the vapor pressure of ZrF4 and PbF2 are much higher than those of NdF3, it is necessary to find adequate evaporation boats, as far as the material type and its electric resistance. The deposition was operated at a pressure near 2xl0"5 mbar. Single crystal silicon wafers and microscope glass slides were used as substrates. 3 minutes of deposition yielded a film thickness of about

308

Optical Glass Waveguides

3 micron. The propagation loss of the planar waveguides thus fabricated is around 4 dB/cm at He-Ne laser wavelength (632.8 nm). B.Boulard and his group explored the possibility of fabricating ternary fluoride glass films PZG (PbF2-ZnF2-GaF2) by physical vapor deposition/151 From a thermodynamic point of view, a multi-component glass can be seen as a eutectic mixture. The non-congruent vaporization of a binary compound causes a composition displacement, and crystallization phenomena occur in the melt on solidification. To prevent such crystallization of the melt, which stops the vaporization, they looked for a starting glass composition which would allow the melt to remain in a glassy domain during the entire evaporation process. This can be accomplished by diluting the PZG glass into another flux glass which is fully compatible with PZG and has very low vapor pressure and good stability towards crystallization. One kind of flux material (BaF2InF3-MnF2) has a melting temperature of 591 °C, which is obviously higher than PZG glass (547 °C).[16] 9.2.6.2 Chalcogenide glasses Interest in chalcogenide materials, particularly in the form of glass, is continuing to increase as their usual and varied properties are ever more understood. Chalcogenide materials are generally classified as those in which sulfur, selenium, or tellurium form one of the basic constituents, usually in the form of crystal, ceramic, or glass. The transmission window for chalcogenide glasses extends from 1 um to longer infrared wavelengths compared with oxide and fluoride glasses. The highrefractive index, enhanced mid-IR transmission in chalcogenide glasses with reduced maximum phonon energy and ability to exhibit localized reversible or irreversible material changes when irradiated with light suggest these glasses are indeed interesting candidates for thin film development.117"181 A.K.Mairaj and co-workers described inverted spin coating technique to deposite Ga:La:S films (see Fig. 9.4).[19]

9.2 Glass Waveguides Fabrication and Optical Properties stage 1 depositon boron nitride vacuum chuck substrate (dadding glass)

stage 2 withdrawal withdrawal of substrate

309

stage 2 withdrawal spinning of molten glass

: * I msniseus

thin film is formed

molten class

Fig. 9.4 Three-step process for inverted deposition spin coating of thin film glass waveguides. After Ref. [19].

The fabrication of spin-coated amorphous thin films of Ga:La:S glass can be separated into three stages: the deposition of fluid onto the substrate, the withdrawal of the coated substrate from the melt, and spinning of the coated substrate. The surface uniformity obtained has a variation of 0.5 jim across 25 mm. The propagation loss at 1.064 um is 0.3 dB/cm over a 15 mm path length, making this technique attractive for a number of optical applications. This technique allows the deposition of thin Ga:La:S films ranging in thickness from 1 rnm to 10 ^m, by varying temperature and spinning speed. B. Luther-Davies and his group used high-repetition rate short-pulse laser to deposit chalcogenide thin films for waveguide application.[20"21] Pulse laser deposition (PLD) is a process that evaporates a target material by high intensity laser irradiation and deposition of the evaporated materials onto a substrate. A unique advantage of PLD is its stoichiometric transferring of target composition to the deposited film. However, conventional PLD encounters the problem of depositing small clusters or particles onto the substrate. The Australian group successfully overcame the problem by using ultra-fast lasers for PLD. In ultra-fast PLD, the laser pulse energy is lowered so that it is no longer possible to create large particles. On-target intensity is chosen so that sufficient energy is available to ablate the surface to a depth equal to the penetration of the thermal wave. This ensures that the ablation process reaches its maximum efficiency and occurs without significant heating of the target. They used second harmonic light of a mode-locked Nd:YAG laser (532 nm, 50 ps pulse duration) at 76 MHz repetition rate (pulse

310

Optical Glass Waveguides

energy 80 nJ) to deposit As2S3 films. After focusing the light onto the target, the peak intensity reached was about 5xl0 8 W/cm2.

Fig, 9.5 SEM micrographs showing the profile of AS2S3 waveguides etched by ICP using photoresist mask (a) AS2S3 waveguide with photoresist mask (b) coating with polysiloxane cladding, its width is well controlled as required. After Ref. [21].

The deposition rate was 2 nm/second. As-deposited film had propagation loss of 0.1 dB/cm at 1550 nm wavelength. Film was also dry-etched by using either helicon plasma etching or ion-coupled plasma (ICP) etching. Fig. 9.5 shows the SEM images of the etched waveguide. The side wall smoothness was better than 50 nm, the channel waveguide had propagation loss of 0.25 dB/cm at 1550 nm for 4 and 5 um wide waveguides. 9.3 Organic/inorganic Hybrid Glass Waveguide Materials Organic/inorganic hybrid materials refer to a class of materials that combine organic and inorganic moieties at the molecular level. These materials are also known as polycerams, ormosils or ormocers.[22] Hybrid materials are synthesized at low temperatures by wet-chemical techniques, and the resulting products can be produced with exceptional purity and homogeneity. The chemistry of hybrid materials typically involves chemical reactions (hydrolysis/condensation) between metal alkoxides and functionalized polymers. Their structures can be varied, depending on the nature of the constituents and the processing parameters. Methacryl-oxypropyl trimethoxysilane (MAPTMS),[23"24] Hyroxyl-terinated polydimethylsiloxane (PDMS)[25] are among the first few of hybrid materials for optical waveguides fabrication. Hybrid materials are interesting waveguide materials because they keep the good

9.3 Organic/inorganic Hybrid Glass Waveguide Materials

thermo-stability and chemical stability of inorganic materials, whiles the organic entities enable thick film forming and incorporation with functional organic dopants. In the following, we will take MAPTMS as an example. MAPTMS consists a silane end group and an organic end group with C=C double bond that can be polymerized under heat or light exposure. MAPTMS is hydrolyzed together with zirconium propoxide (ZPO) in acidic environment. ZPO is introduced to adjust the refractive index of the final product, ZPO also helps in producing uniform films with good optical quality. After hydrolysis and condensation, structures such as -Si-O-Siand -Si-O-Zr- are formed. Fig. 9.7 is the flow chart of preparing the organic/inorganic hybrid sol.

Fig. 9.6 Flow chart for preparation of hybrid waveguide materials.

After hydrolysis and condensation, films as thick as 10 um can be spin-coated on silicon wafer. UV light exposure then polymerize the hybrid andfilmsbecome glassy hard. Spin-coated film thickness depends on the viscosity of the solutions. A convenient way to control film thickness is let the solvent (here it is propanol) to evaporate in vacuum so as to adjust the viscosity of the final solution. Fig. 9.7 shows for two MAPTMS/ZPO ratio (one used as waveguide cladding material, the other for waveguide core), the relation between thickness and residual solution weight after evaporation.

311

Optical Glass Waveguides

312

;

!

i

icladdingj/

!

1

|

1/

L cor©

:^i2fer: 100

95

90

S5

80

75

Solution weight ratio [%] Fig. 9.7 Spin-coated film thickness versus residual solution weight after evaporation.

The spin speed was kept at 600 tums/min. We found that this method is more controllable than the generally accepted method of changing spin speed.

o<*>

Fig. 9.8 Film morphology of hybrid film from AFM.

Fig. 9.8 is the film surface morphology obtained from AFM measurement. The fluctuation in height is around 3 nm. Fig. 9.9 shows the absorption spectrum of the hybrid thin film in the near infrared region. The absorbing structure at 1400 nm is due to the residual water in the film. A sharp peak at 1610 nm is the residual C=C vibrational mode absorption, and the structures in the 1700 nm band are C-H absorptions.

9.3 Organic/inorganic Hybrid Glass Waveguide Materials

313

1100 1200 1300 1400 1500 1600 1700 1800

Wavelength [ran] Fig. 9.9 Absorption spectrum of a MAPTMS/ZPO hybrid film.

The thermal optical property of the hybrid materials was also measured. The dependence of a material's refractive index on temperature is called the thermo-optic effect. This effect is very useful in the fabrication of optical waveguide switches, which are key devices in intelligent optical communication network systems. According to Prod'home's theory, the thermo-optic coefficient relates to the density and electronic polarizability changes of a material with temperature :[26] dn (n2-\)(n2+2) 6n ~dT

(«>-/?),

(9.9)

where

314

Optical Glass Waveguides

stability and high transmission losses prohibit their commercial development. Recently, sol-gel-derived organic/inorganic hybrid materials composed of silica and polymers began to receive attention for potential applications in integrated optics. Excellent thermo-optic properties as well as much better thermal stability are expected. The thermo-optic coefficients of the hybrids were measured by a prism coupling method. A thin film of hybrid material was deposited on a fused silica plate for measurement. A home-made computer-controlled prism-coupler setup (See Fig. 9.10), which could measure refractive indices of materials with an accuracy of 0.0001 at a wavelength of 632.8 nm was used. A ring-type heater was butted on the back of the substrate to change the film temperature. Fig. 9.11 shows the dependence of refractive index and dn/dT of the hybrid on MAPTMS/ZPO ratio. In Fig. 9.11(a), it can be clearly seen that the refractive index of the hybrid increases linearly with the concentration of ZPO, as Zr02 has much higher refractive index than SiOz. photodiode

I

i

Fig. 9.10 schematic diagram of prism coupling technique for refractive index and dn/dT measurement.

9.4 Functional Glass Waveguide Devices

-10

0

10

20

30

40

50

60

ZPO molar concentration [%mol]

0

10

315

20

30

40

SO

ZPO molar concentration [%mol]

Fig. 9.11 Dependence of refractive index (Left) and dn/dT (Right) on ZPO concentration.

However, the dependence of dn/dT on ZPO concentration is highly nonlinear. Moreover, very interestingly, dn/dT of the hybrid is much larger than that of PMMA. It is known that the dn/dT of Si0 2 , Zr0 2 and PMMA are Ixl0"5/°C, 4.8xlO~7°C and -Ixl0' 4 /°C respectively. Since the hybrid materials comprise of the above three components, the measured dn/dT of the hybrid materials are unexpected. One possible reason is that the interactions between the organic and inorganic parts changed either the electronic polarizabilities or free volumes inside the hybrid materials, high values of the thermo-optic coefficient was also obtained for other organic/inorganic materials.[30] 9.4 Functional Glass Waveguide Devices 9.4.1 Silica glass waveguides and arrayed waveguide grating Arrayed waveguide grating (AWG) is an integrated wavelength filter/router that are key devices in dense wavelength division multiplexing (DWDM) system. In a DWDM system, light with different wavelengths is launched into one single mode fiber. One wavelength corresponds to one communication channel. These wavelengths have equal interval and are precisely fixed according to International Telecommunication Union (ITU). AWG is used to combine different channels (wavelengths) coming from different sources to a single fiber (so called multiplexing), or de-couple the channels to individual slot at

Optical Glass Waveguides

316

the end of transmission (so called de-multiplexing). 16 to 64 channel AWGs have already been marketed and are widely used in DWDM systems in North America and Japan. Fig. 9.12 shows the configuration of a AWG. An AWG comprises of input waveguides, slab waveguides and arrayed waveguides. The arrayed waveguides has a constant path length difference AL between neighboring waveguides. The input light is launched into the first slab waveguide and then excites the arrayed waveguides. After traveling through the arrayed waveguides, the light beam interferes constructively at one focal point in the second slab. The location of the focal phase delay in each arrayed waveguide is given by AL/X. The slab and array waveguides act as a lens and a grating respectively. Since the focusing position differs with wavelength, lightwaves with different wavelengths in the input waveguide are output in specific output waveguides determined by their wavelengths. Arrayed waveguides

4

Input waveguides

Fig. 9.12 configuration of N x N AWG multiplexer After Ref. [5].

The wavelength separation from two neighboring channels at the output is determined by[5] . „ ndDXn AX = —c°-,

NJAL

(9.10)

9.4 Functional Glass Waveguide Devices

317

where nc is the effective index of the arrayed waveguides, Nc=ncMnJdX,f\s> the focal length of the output slab waveguide, D and d are the in- and out- waveguide spacing of the output slab. If AX is fixed (100 GHz, 50 GHz, 25 GHz, etc.), AL can be deduced from Eq. (9.1). The spatial free spectral Range (FSR) of the device is XFSR=^~, nsd

(9.11)

where ns is the effective refractive index of the slab waveguide. Therefore, the number of available wavelength channels is given by XFSR/D.

The device size is determined by bending radius of waveguides. Increasing the refractive index contrast A of core/cladding, the bending radius can be reduced. If A can be raised to 1.5%, the bending radius decreases to 2 mm. The circuit size of a 100-GHz-spacing 16-channel AWG can be reduced from 580 mm2 to 200 mm2. Fig. 9.13 shows the transmission spectrum of a 25 GHz-spacing 400-channel AWG fabricated by NTT Photonics Laboratory. Table 9.5 summarizes the design parameters of the device. The device comprises over 1000 array waveguides, the circuit size is 12x6 cm2. 6 inch Si wafer was used as the substrate. The device has favorable demultiplexing properties covering the whole C band and L band (1530 -1610 nm). The chip loss is between 3.8-6.4 dB, crosstalk -20 dB between adjacent channels and about -30 dB in the background. Table 9.5. Design parameters of a 25-GHz-spacing 400-channel AWG. After Ref. [5]. Center wavelength Diffraction order Path length difference Slab focal length Number of arrayed waveguides Circuit size

1570 nm 18 19.4 urn 47 mm 1175 124 mm x 64 mm

Optical Glass Waveguides

318 < ——•—•——

c-band

™—

—»<

- •"

L-band

»

Wavelength (tun)

Fig. 9.13 Transmission spectra of a 25-GHz-spacing 400-channel AWG. After Ref. [5].

9.4.2 Optical switch from hybrid glass waveguides We have shown in the previous section that organic/inorganic hybrid glass materials maintain the excellent thermo-optic properties as that of polymers, while their thermal stability will be presumably much better. Moreover, as the materials are polymerized under UV light exposure, the material itself is actually a kind of photoresist, photo-patterning turns out to be very simple. Therefore, hybrid glass materials are very suitable to fabricate functional waveguide devices using their peculiar thermo-optic properties. Our first approach is a 2x2 Mach-Zehnder interferometric thermooptic switch.[31] The organic/inorganic hybrid waveguide material used was synthesized by hydrolysis and poly-condensation of MAPTMS, methacrylic acid (MAA), and ZPO, as have been described in the previous section. Two types of sol were prepared, with molar ratios of MAPTMS: ZPO: MAA to be 10:4:4 (HYBRID I) and 4:1:1 (HYBRID II). They were used as waveguide core and cladding/buffer materials respectively. Thin films were formed by dip coating the sols on silicon substrate and heated. After pre-heating, they were exposed to UV light for polymerization of the organic part of the hybrids. Direct UV patterning was also done by placing a mask on top of the film and illuminating the film with UV light. After UV exposure, another heating process was conducted for densification.

9.4 Functional Glass Waveguide Devices

319

Fig. 9.14 shows the dispersion curves of HYBRID I and HYBRID II. The refractive indices of the two materials at 1550 nm are 1.5179 and 1.5111 respectively. Refractive index difference is Zl«=6.7xl0"3.

ID CH 1515 1.510 400

600

800

1000

1200

1400

1600

Wavelength [nm]

Fig. 9.14 Dispersion curve of two hybrid materials, they are used as core and cladding material respectively.

Cross port

Electrode Top cladding

buffer

Fig. 9.15 Schematic diagram of the (Right).

Z.AZ. UjJUUiU

switch (Left) and the waveguide structure

320

Optical Glass

Waveguides

The 2x2 MZ interferometric switch was fabricated on a silicon substrate. Fig. 9.15 is the schematic diagram of the waveguide structure. The switch composes of two 3 dB directional coupler and a MZ interferometer. Two arms of the interferometer can be heated to change and balance the optical path. In the device design, the waveguide spacing was selected to be 5.5 |j,m. The 3 dB coupling length is 1692 |am. Fig. 9.16 shows the 3 dB coupler part of a fabricated 2x2 optical switch. The picture shows that the waveguide fabricated has pretty good cross-section, that guarantees good performance of the device. + 4.DO0DD -

+ 2.7SDDQ

-

E" I 3.150DQ0 4—I

en
x

-1.00DDD

: -

: I—i—i—i—i—I—i—i—i—i—|—i—i—i—i—I—i—i—i—i—I—i—i—i—i—I—

Q.QOQ

Q.QQ5

D.Q1 Q

0.Q15

D.Q2D

Q.025

Distance (mm) Fig. 9.16 3-D structure of a fabricated 3 dB coupler.

Fig. 9.17 shows light output from the device when a laser diode at 1550 nm wavelength was used as the light source. When applying different electric power, bar state, cross state and balance state could be obtained. The 3D power distribution clearly proves that the device is working at single mode. Due to the optical-path differences caused by fabrication errors, the bar-port did not extinguish when no power was applied. An extinguished bar-port state could be reached by applying a small offset power to heater II. Fig. 9.18 is the result with power applied to heater I and no offset power applied to heater II. From the figure, the extinction ratio can be deduced to be 18.1dB and 19.4 dB for the bar-port and cross-port respectively. When the applied power is 3.7 mW, the bar-port is on and

9.4 Functional Glass Waveguide Devices

321

Fig. 9.17 Output from a 2x2 optical switch. Top left: Bar state, top right: cross state, top middle: balance state. Lower figure: 3D power distribution of the balance state shows single mode performance of the device.

Switching Power (mW) Fig. 9.18 Switching performance of the 2x2 optical switch. Switching power of less than 6 mW can be read from the plot.

the cross-port is off. When the applied power reaches 9.3 mW, the barport is off and the cross-port is on. So the switching power, the electiic power required for the switch to pass through both the on and off states, is 9.3 mW. Note that for a silica optical switch with the same structure, the switching power is about 500 mW. So the power consumption of our device is two orders of magnitude smaller, mainly due to the peculiar thermo-optic property of the hybrid materials. The temporal response of the switch was also tested by applying a

Optical Glass Waveguides

322

square wave voltage to heater I to switch the device between the "on" and "off states. Fig. 9.19 shows the switch response. The lower and middle traces are the output light detected from the bar-port and crossport, respectively, and the upper trace is the square wave voltage applied. The rise and fall times are 3.5 ms and 4.2 ms respectively. These results are very good compared to the response times of polymer switches. Compared to the results of the switch fabricated with a hybrid material, the response time of our switch is better, and our switching power is only about 20 percent, since an interferometric structure instead of a Y branch was adopted. 3.0 2.5

I

1

I"

v r

i: i

/ \

0.5

0 0

10

12

14

16

18

20

22

24

Time (ms) Fig. 9.19 Switching time of the 2x2 optical switch. Upper curve: applied electric power. Lower curve: response of the device.

9.4.3 8-channel variable optical attenuator Variable optical attenuators (VOAs) are basic components of optical transmission system. It is useful in flattening the gain of optical amplifiers, equalizing channel power levels in wavelength division multiplexed (WDM) system, reducing power fluctuation, and attenuating the power for the receivers. Among the developed VOAs, the integrated optical waveguide VOAs have attracted most attentions for its flexibility of structure design, compact volume and the ability to hybrid integrate with other photonics devices. In most cases, either silica[32] or polymer[33] were used as the thermo-optic waveguide materials.

9.4 Functional Glass Waveguide Devices

323

We fabricated a Mach-Zehnder interferometer thermo-optic VOA using sol-gel derive organic/inorganic hybrid materials. The structure of the proposed VOA is shown in Fig. 9.20.

Fig. 9.20 upper part: A VOA structure based on a MZ interferometer, lower part: design of a 8-channel VOA, dark lines are electric wires, horizontal shadow lines are wavguides.

The single mode waveguide channel is connected by S-band waveguides to two-phase arms spacing 250 um, to form a Mach-Zehnder Interferometer (MZI). The length of the arm is 3000 (xm and the total length of the devices is 10 mm. The electrode heaters are located directly on the top of both arms. When voltage is applied to the electrode, the refractive index under the heater would change due to the thermo-optic effect which will cause the phase difference. Fig. 9.21 shows the fabricated device on a 3 inch silicon wafer, together with a scanning-electron microscope (SEM) photo of the waveguide. Fig. 9.22 shows the VOA attenuation curves versus the electrical power applied to one electrode. As the power increased, the attenuation increased for both TE and TM polarization. After the attenuation reached a peak value, it gradually decreased and returned to the original value. The TM polarization

324

Optical Glass Waveguides

reaches the maximum attenuation of -25 dB for 12.5 mW. The TE polarization reaches the maximum attenuation of-31 dB for 13.2 mW, It can be noted that these power consumptions are one order smaller than the VOAs based on silica or silicon materials130' and similar to polymerbased device.[31]

Fig. 9.21 Left: device on a 3 inch silicon wafer. Bright lines are electric wires, waveguides are in the vertical direction. 8 channel VOA is in the white box. Right: SEM image of a waveguide cross section.

Power [mW] Fig. 9.22 VOA attenuation curve.

The best attenuations achieved is over -24 dB for the wavelength range 1480-1630 nm (see Fig. 9.23).

9.4 Functional Glass Waveguide Devices

325

However, fluctuation of TM attenuation is obviously larger than TE polarization, mainly due to the vertical heat flow from top electrode to the substrate. -15

ST a-20 c

o ro -25

;

^

D

^

^

s= CD ^

o — O — TE

-30

\

—•—TM

£ -35

o

v/V

TO ^

-40

1470

1500

1530

1560

1590

1620

1650

Wavelength [nm] Fig. 9.23 Wavelength dependence of the maximum attenuation.

14.0

13.5

o -o-TE — TM

-

13.0 O

O"

7v

cr Q- 12.5

12.0

n-n-n

: 1470

,o s

0

' D

X

-

^ 1500

1530

1560

1590

1620

1650

Wavelength [nm] Fig. 9.24 Wavelength dependent power consumption to achieve 7i phase change.

The power consumptions of the wavelengths for TE and TM polarization are plotted in Fig. 9.24. It is seen that the longer wavelength, the more power needed to reach the maximum attenuation. The highest power consumptions are about 14 mW for TE polarization and 13 mW

326

Optical Glass Waveguides

for TM polarization at 1630 nm. The wavelength sensitivity of the devices can be improved by using a feedback system. The temporal response of device was also tested by the same method shown in the previous section. The rising time is 3.7 ms and the falling time is 4.7 ms. The data are comparable to the devices made of polymer. 9.4.4 Tunable waveguide dye laser1341 Lasing from organic dye doped planar waveguide was investigated widely these years,[35"37] as dye-doped solid-state waveguide lasers provide low cost, high efficiency tunable coherent light sources. Conventionally, wavelength selection was realized by using distributed feedback Bragg reflector (DFB) or distributed Bragg reflector (DBR) as coupling mirrors in resonant cavities. However, wavelength tuning is quite limited in DFB or DBR lasers. Wide range wavelength tuning was accomplished in external cavity waveguide laser that uses grating as the dispersion element. We developed a novel wavelength tuning method that can be used to tune the lasing wavelength for glass waveguide laser. The technique is based on the fact of the existence of cutoff thickness for an asymmetric waveguide. Eq. (9.4) describes the minimum waveguide thickness dc that is necessary for light of wavelength X to be guided by an asymmetric waveguide. For given waveguide thickness d, we can also determine from Eq. (9.4) the light of longest wavelength Xc that is allowed to propagate in the waveguide. Xc is named the cutoff wavelength for a given waveguide thickness. In principle dc(X) decreases with the decreasing of X. This is also the case in our experiment. Fig. 9.25 plotted the relation of dc by using parameters in our experimental condition (see below for detailed experiment description). Considering an optically pumped dye-doped waveguide, for a given waveguide thickness d, if the cut-off wavelength is at the blue side of gain maximum, lasing wavelength will be exactly at the cut-off wavelength where the gain coefficient is the largest for allowed guiding wavelength. Therefore, wavelength tuning can be realized by simply changing the thickness of waveguide.

9.4 Functional Glass Waveguide Devices

327

1.466 y.

.3 u .;> o 1.465 re

,t-<

"8

1.464 560

570

580

590

600

610

Wavelength (nm) Fig. 9.25 Cutoff thickness as a function of wavelength. Also shown are dispersion curve for film and substrate.

Experimentally, the wavelength tuning in an optically pumped dyedoped waveguide was investigated by observing lasing wavelength change. For light propagating in the guiding layer, a length of several hundreds of micrometers will be sufficient to obtain the necessary gain and hence ASE indicated by spectral narrowing at comparable low threshold can be observed easily. PMMA (poly(methyl methacrylate))/ silica-gel hybrid composite was used as the host matrix, RhB was the gain media. The reason to use this hybrid material as the waveguide material is, its refractive index can be precisely tuned by controlling the ratio of organic/inorganic composition. In order to push the cutoff thickness of an asymmetric waveguide to several microns, the refractive index of the core should be very close to buffer layer (in this case is the glass substrate), therefore fine-tuning of the refractive index is important. Reasons for using hybrid composite matrix are the flexibility of tuning refractive index and better thermal stability of the hybrid material. The dye-doped waveguide was transversely pumped. The pump light was focused to a strip of 2-3 cm in length. Amplified spontaneous emission (ASE) emitted from end face was collected. Fig. 9.26 shows the absorption spectrum(D), fluorescence spectrum(C), and ASE emission spectra (A,B) of DCM-doped organic/inorganic hybrid planar waveguide. The absorption peak is around 460 nm and the absorption edge extended

Optical Glass Waveguides

328

pump be am

0.8 0.7 0.6

s 0>

400

450

500

Ik 550

600

650

0.5 0.4 0.3 0.2

•a

<

0.1 0.0 700

Wavelength (nm) Fig. 9.26 Absorption and fluorescence spectra of DCM doped film. A,B are ASE spectra obtained at different film thickness. Inset: schematic pump probe diagram.

to as far as 605 nm. The fluorescence emission peak is around 582 nm and its bandwidth (FWHM) is about 90 nm. Fig. 9.26 shows that the fluorescence spectrum overlaps with the absorption spectrum. If the thickness of waveguide was so large that the cutoff wavelength of the waveguide was far beyond the region of present interest, then the position of ASE peak was determined by absorption and fluorescence spectra. Consequently, ASE peak appeared at 608 nm, rather than at the position of the most intense fluorescence. Apparently, the net gain coefficient at this wavelength was the largest. In addition, ASE bandwidth was around 13 nm, much narrower than its fluorescence bandwidth. On the other hand, if the thickness of the waveguide was carefully controlled so that the cut-off wavelength lied on the blue side of the ASE peak wavelength mentioned above, it is understandable that ASE peak wavelength would move to the cut-off wavelength. This is the position where the waveguide can sustain one mode and the absorption is the least. Changing the pumping position on the wedge waveguide, i.e., changing the thickness of waveguide, we observed a continuous wavelength shift of ASE from 608 nm to 578 nm (tuning range of 30 nm). Fig. 9.27 plotted changes of ASE peak intensity and linewidth with wavelength. Peak wavelength blue shift also resulted in decrease of ASE

9.4 Functional Glass Waveguide Devices

329

linewidth. As loss (mainly absorption) was more severe at blue side, ASE profile was compressed at the cut-off wavelength. Fig. 9.27 shows that linewidth of ASE reduced from 13 nm to around 4 nm.

•d 1000 -

Wavelength (ran) Fig. 9.27 ASE intensity and linewidth changes as a function of wavelength.

9.4.5 Rare-earth doped waveguide lasers and amplifier Rare-earth doped waveguide lasers and amplifiers are compact active light sources that can have broad applications. The use of single-mode guided wave structures also enable low pump power and excellent coupling with fiber systems.[38] The first Nd doped glass waveguide was reported in 1974 by using ion-exchange method to form buried single mode waveguide in Nd-doped glasses.[39] With the invention of high power fiber-coupled solid laser diodes working at 808 nm and 980 nm, various kinds of Nd-doped and Er/Yb doped glass waveguide lasers and amplifiers have been explored/40"431 Other rare-earth elements doped waveguide lasers were reported as well, one example was the Tm3+ ion doped glass waveguide laser working at 2 \xm range which are very important wavelength for medical science/441 In the following, we will concentrate on the overview of erbium doped glass lasers and amplifiers. Erbium doped glass waveguide lasers were generally formed by ion-exchange in silicate or phosphate glasses/39'421 Other techniques like FHD[45! and ion implantation/441 laser ablation[47] were investigated as well.

330

Optical Glass Waveguides

Veasey and his group reported an array of distributed-feed- back (DFB) waveguide laser.1481 The waveguide laser was formed by K+-Na+ ion exchange on commercially available phosphate alkali glass that was co-doped with 0.99xl020 /cm3 Er3+ and 3.97xl020 /cm3 Yb3+ ions. Gratings were etched on the waveguides by using reactive ion etching with Ar-ion plasma, are frequently used laser cavities(see Fig. 9.28).

1536-1536.3 rail

Fig. 9.28 Distributed-Bragg-reflector waveguide laser array realized using a single pitch grating and diffused waveguides with varying effective index. After Ref. [48].

The slope efficiency was 26% and the threshold power was 50 mW. An output power of 80 mW was achieved with 350 mW of coupled pump power. Each laser exhibits stable operation on a single longitudinal mode and all had linewidths less than 500 KHz. Erbium doped waveguide amplifiers (EDWA) are important devices in all optical communication systems to compensate the signal loss after long distance propagation and boost signals after splitting in access networks. On the contrary to well-known erbium doped fiber amplifier (EDFA) which usually has a fiber length of several tens of meters, EDWA requires amplification in a very short length, typically one centimeter long. Therefore, much high concentration of erbium ions, together with Yb ions, used to enhance the absorption and efficiently transfer the energy to nearby Er ions. Different approaches are explored. In Chapter 6 of this book, short length fiber amplifier and laser are described to function as a waveguide amplifier. Here traditional EDWA techniques will be presented.

9.4 Functional Glass Waveguide Devices

331

EDWA films can be prepared by various methods. Up to now, magnetron rf sputtering/491 ion-exchange[501 and PECVD[51] are the most frequently used techniques. On the other hand, Sol-gel technique in fabricating Er doped glass waveguide films have achieved significant progress in the past 5 years.[52] YC Yan and A Polman reported a net gain of 4.2 dB/cm on their EDWA made by magetron sputtering.[49] The glass matrix was phosphate glass, their erbium concentration was 5.3xl020 cm"3 (0.75%). By pumping the waveguide at 980 nm with a power of 21 mW, a net optical gain of 4.1 dB/cm at 1535 nm was achieved. F.D Patel et al. reported EDWA fabricated by ion-exchange.[53] They used commercial Shott phosphate glass with very high concentration of Er (as high as 8 wt %) and Yb (up to 12 wt %). They used two-step ionexchange (first step Ag+-Na+ exchange, second step electric field assisted ion exchange) to produce buried waveguides. A net gain of 4.1 dB in a 3mm length waveguide at 1534 nm was obtained. Fig. 9.29 shows the obtained net gain versus pump power. Although potentially high concentration of rare earth elements is allowed to be incorporated, sol gel technique was not considered to be a suitable way to fabricate EDWA, because various channels to quench the infrared fluorescence of Er ions exist. However, in the past few years, quenching was successfully reduced by carefully select the glass matrix.

50 75 100 125 150 Incident Pump Power [mW]

175

200

Fig. 9.29 Performance comparison of low- and high-doped glass amplifiers showing measured gain at 1534 nm versus pump power for waveguide amplifier fabricated in 4wt.% Er /3-wt.% Yb and 8-wt.% Er /12-wt.% Yb phosphate glasses. After Ref. [53].

Optical Glass Waveguides

332

A multi-component glass matrix consisting P2O5, B2O3, AI2O3, Ga 2 0 3 was explored, in which P2O5 has lower phonon energy, therefore nonradiative decay rate can be reduced. AI2O3 is used to control the Er3+ environment (avoid clustering) and Yb3+ is co-doped to improve the pump efficiency. In order to avoid cracking which happens frequently in sol-gel films, a repetitive SC-RTA technique was used. P2O5 also served to lower down the annealing temperature of the SC-RTA process. Fig. 9.30 shows the transmission measurements for a 5-cm-long EDWA pumped at 980 nm pump. Pump power (mW) 174.2 75.2 32.0 18.3 7.3 0.0

1450

1500

1550

1600

1650

Wavelength (nm) Fig. 9.30 Transmission measurements for a 5-cm-long EDWA with a 4-um core width, for different values of 980-nm pump power. After Ref. [54].

The core composition contains 10 mol% P 2 0 5 , 2.5 mol% A1203, 0.25 mol% Er 2 0 3 , 0.25 mol% Yb 2 0 3 , giving an asymmetric guide with an index step of 1.5xl0"2. It can be clearly seen that at the pump power of 174 mW, a maximum net gain of about 6 dB was obtained.[54] 9.4.6 Optical sensors based on waveguide technology In a classical picture, light propagates through a waveguide by total reflection on the core-cladding interfaces, strong evanescent waves existing on interfaces provides excellent channel for very sensitive optical monitoring of environment changes. Minor changes of optical properties such as absorption, fluorescence, refractive index can be measured as the evanescent wave is generally intense and the working

9.4 Functional Glass Waveguide Devices

333

length can be long in principally ( several centimeters). The first optical waveguide sensor was introduced in 1979 using polymer waveguide to sense blood ammonia.[55] After that, optical sensors based on different waveguide techniques have been developed. The special wet chemistry process of sol gel technique enables chemical sensing through optical property changes.[56] Sol-gel-derived materials can be both optically transparent (into the ultraviolet spectrum) and highly porous; organic indicator groups can be incorporated (via physical entrapment and/or covalent bonding) into the material prior to gelation yet remain sterically accessible to analytes that diffuse into the glass; the high surface area of the material provides a means to concentrate an analyte by adsorption from a dilute sample; the chemical and physical properties of sol-gel glasses can be systematically adjusted by varying the precursor composition and processing parameters.1571 S.S.Saavedra and his group developed optical sensors based on Solgel glass waveguides.[58"59] Fig. 9.31 shows the schematic diagram of a planar waveguide sensor for gaseous iodine sensing. The sensing principle is based on detection of a charge transfer complex formed between iodine and phenyl groups that have been covalently incorporated into a sol-gel indicator film. The film was coated over a single-mode planar waveguide, and light was coupled into and out of the laminate structure using a pair of integral diffraction gratings. The sensor exhibited a linear response to t i n the range of 100 ppb to 15 ppm with response and recovery times less than 15 s. The response was selective to 4 ppm iodine in the presence of 10 ppm chlorine and was stable for at least 3 months. Most recently, M.Zourob described a metal cladding leaky mode waveguide (MCLW) sensor for the detection of particles.[59] In their waveguide structure, 8.5 nm thick titanium layer was deposited on BK7 glass, then a 300 nm thick silica was deposited. The advantage of the insertion of the thin layer of metal between the substrate and the spacer serves to increase the sensitivity by pushing the waveguide mode further into the sensing region. As the result, MCLW shows a greater extension of evanescent field than the surface plasmon resonance (SPR), large particles such as bacteria will be more illuminated within the evanescent field of the MCLW than other configurations.

Optical Glass Waveguides

334

f

lotfne

charge transfer complex

moaen! lighi

outcoupted ight

Fig. 9.31 schematic diagram of a waveguide sensor. After Ref. [57].

High sensitivity of waveguide sensors relies on the intensity of evanescent wave in the sensing region. However, as high intensity evanescent wave also generates high propagation loss, Itoh and his group carried out another approach by separating sensing waveguide with the propagating waveguide.[60] In their demonstration, propagating waveguide was conventional single mode K+-exchanged glass waveguide. A second step local Ag+-exchange was followed to lift the refractive index in the sensing region (see Fig. 9.32). Tapered thin film dielectric layer with high refractive index like Ti02 could also be deposited on the ion-exchanged waveguide (see Fig. 9.33). The taper served as a vertical mode coupler, as a result, guided light is pushed to the high refractive index region and increase the evanescent wave intensity in the sensing region. In an application to ammonia gas sensing, ammonia at concentrations of 1 ppt was easily detected. As a biochemical application, the detection of immonuglobulin G at concentrations of 70 pg/cm3 was demonstrated.

References

335

Ag°4oisH«cltan|p4 layer

Fig. 9.32 Structure of the ion-exchanged composite OWG. The arrow show the guided light is transferred from one part of the OWG to another part via adiabatic transition. After Ref. [60].

thia film layer «r ewaeseent wsvt

Fig. 9.33 Structure of the composite OWG and the principle of operation. m,nm, and nt are respectively, refractive index of the substrate (1.515), the K+-ion-exchanged layer (1.5195), and of the thin film. After Ref. [60].

References 1. Buckmao, "Guided-Wave Photonics", (Saunders College Publishing, Fort worth, 1992) 2. J.Hubner, S.Guldberg-Kjer, M.Dyngaard, et al, Appl.Phys.B, 73 (2001) 435. 3. G.C.Righini, S.Pelli, M.Ferrari, et al, Opt. Quantum Electronics, 34 (2002) 1151. 4. M.kawachi, OptQuantum electron., 22 (1990) 391. 5. Y.Hibino, IEEE J.Select. Toptics Quantum Electronics, 8 (2002) 1090. 6. T.Izawa and H.Nakagome, Appl.Phys.Lett., 21 (1972) 584. 7. R.V.Ramaswamy and R.Srivastava, J.Lightwave Tech., 6 (1988) 984. 8. G.Sorbello, S.Taccheo, M.Marano, M.Marangoni, et al, OptMat, 17 (2001) 425. 9. L.L.Hench and J.K.West, Chem. Rev., 90 (1990) 33.

336

Optical Glass Waveguides

10. The first Sol-Gel Optics Conference was organized in 1990. Proceedings of the conference was published in J.Non-crystalline Solids., 1990. 11. R.R.A.Syms, A.S.Holmes, W..Huang, et al, J.Sol-Gel Sci.Tech., 13 (1998) 509. 12. G.Rault, J.L.Adam, R.Smektala, J.Lucas, J.Fluorine Chemistry, 110 (2001) 165. 13. E.Josse, J.E.Broquin, G.Fonteneau,et al., J.Non-Cryst. Solids, 213-214 (1997) 152. 14. R.Almeida, P.J.Morais, M.C.Goncalves, J.Non-Cryst. Solids, 213-214 (1997) 251. 15. B.Boulard, Y.Gao, C.R.Chimie, 5 (2002) 675. 16. C.Jacoboni, O.Perror, B.Boulard, J.Non-Crystalline Solids, 184 (1997) 184. 17. A.Zakery, S.R.Elliott, J.Non-Crystalline Solids, 330 (2003) 1. 18. RJ.Curry, A.K.Mairai, C.C.Huang, et al, J.Am.Ceram.Soc, 88 (2005) 2451. 19. A.K.Mairaj, R.J.Curry and D.W.Hewak, Electronics Lett., 40 (2004) 412. 20. A.V.Rode, A.Zakery, M.Samoc, et al., Appl.Surf.Sci., 197-198 (2002) 481. 21. Y.Ruan, W.Li, R.Jarvis, et al, Opt.Express, 12 (2004) 5140. 22. H.Schmidt and M.Popall, SPIE, 1328 (1990) 249. 23. H.Krug, f.Tiefensee, P.W.Oliveira, and H.Schmidt, SPIE, 1758 (1992) 448. 24. SI Najafi, T.Touam, R.Sara, et al., J. Lightwave Tech., 16 (1998) 1640. 25. S.Motakef, J.M.Boulton, and D.R.Uhhnann, Opt.Lett., 19 (1994) 1125. 26. L.Prod'homme, Phys.Chem.Glasses, 4 (1960) 119. 27. M.Okuno, N.Takato, T.Kitoh and A.Sugita, NTT Review, 7 (1995) 57. 28. R.S.Moshrefzadeh, M.D.Radcliffe, T.C.Lee and S.K.Mohapatra, J.Lightwave Tech., 10(1992)420. 29. M.B.J.Diemeer, Opt.Mater., 9 (1998) 192. 30. E.S.Kang, T.H.Lee and B.S.Bae, Appl.Phys.Lett., 81 (2002) 1438. 31. X. J. Wang, L. Xu, D. X. Li, et al., J. of Appl. Phys., 94 (2003) 4228. 32. M. Svalgaard, K. Farch, and L.-U. Andersen, J.of Lightwave Tech., 21 (2003) 2097; T. Hurvitz, S. Ruschin, D. Brooks, et al., J. of Lightwave Tech., 23 (2005) 1978. 33. S. M. Garner, and S. Caracci. IEEE Photonics Tech. Lett., 14 (2002)1560, and references therein. 34. X.Peng, L.Liu, J.Wu, et al., Opt. Lett., 25 (2000) 314. 35. Y.Sorek, R.Reisfeld, I.Finkelstein and S.Ruschin, Appl.Phys.Lett., 66 (1995) 1169. 36. H.Hillmer, H.L.Zh, A.Grabmaier, et al, Appl.Phys.Lett., 65 (1994) 2130. 37. J.E.Roman and K.A.Winick, Appl.Phys.Lett., 61 (1992) 2744. 38. E. Lallier, Appl.Opt, 31 (1992) 5275. 39. M.Saruwatari, T.Izawa, Appl.Phys.Lett., 24 (1974) 603. 40. T. Feuchter, EK Mwarania, J. Wang, IEEE Photon.Tech.Lett., 4 (1992) 542. 41. JE Roman, P.Camy, M.Hempstead, et al., Electron. Lett., 31 (1995) 1345. 42. DL Veasey, DS Funk, PM Peters, et al., J.Non-Crystalline Solids, 263 (2000) 369. 43. CTA Brown, CL Bonner, TJ Warburton, et al., Appl.Phys.Lett., 71 (1997) 1139. 44. DP Shepherd, DJB Brinck, J.Wang, et al, Opt.Lett., 19 (1994) 954. 45. T.Kitagawa, K.Hattori, M.Shimizu, et al., Electon.Lett., 27 (1991) 334. 46. G.N.van den Hoven, RJLM Koper, a.Polman, et al., Appl.Phys.Lett., 68 (1996) 1886. 47. R.Serna, J.M.Ballesteros, M.Jimenez de Castro, J.Appl.Phys., 84 (1998) 2352. 48. DL Veasey, DS Funk, NA Sanford, JS Hayden, Appl.Phys.Lett., 74 (1999) 789. 49. YC Yan, AJ Faber, H. DeWaal, et al, Appl.Phys.Lett., 71 (1997) 2922. 50. A.Benn, Lightwave Europe, March 2000, 38.

References 51. 52. 53. 54. 55. 56.

337

K.Shuto, K.Hattori, T Kitagawa, et al., Electronics Lett., 29 (1993) 139. W.Huang and R.R.A.Syms, J.Lightwave Tech., 21 (2003) 1339. F.D.Patel, S.DiCarolis, P.Lum, et al, IEEE Photon.Tech.Lett., 16 (2004) 2607. W.Huang, R.R.A.Syms, IEEE Photon.Tech.Lett., 14 (2002) 959. P.L.Smock, T.A.Orofino, G.W.Wooten, W.S.Spencer, Anal.Chem., 51 (1979) 505. BD MacCraith, C, McDonagh, AK Mcevoy, et al., J.Sol-Gel Sci. Tech., 8 (1997) 1053. 57. L.Yang, S.S.Saavedra, N.R.Armstrong, Anal.Chem., 68 (1996) 1834. 58. L.Yang, S.S.Saavedra, N.R.Armstrong, J.Hayes, Anal.Chem., 66 (1994) 1254. 59. M.Zourob, S.Mohr, B.J.T.Brown, et al, Sensors and Actuators B, 90 (2003) 296. 60. A.Yimit, A.G.Rossberg, T.Amemiya, K.Itoh, Talanta 65 (2005) 1102, and references therein.

Chapter 10

Glass Photosensitivity and Fiber Gratings

Over the past three decades, optical fibers achieved great success in the field of telecommunications. Its ideal properties such as low transmission loss, high optical damage threshold and low optical nonlinearity enabled long distance communication to be a reality. In recent years, the ability to alter the refractive index of the core of a single mode optical fiber by irradiating the fiber with UV light, which is originated from the photosensitivity of glass materials, reshaped the fiber-optic technology. The photosensitivity of optical fibers has been used to generate permanent structural changes in the core of a fiber. A good example is, a periodic refractive index modulation along a fiber forms a fiber Bragg grating, it can act like a wavelength selective mirror that satisfies the Bragg reflection condition.[1'2] The grating period and length, together with the refractive index modulation contrast, determine whether the grating has a high or low reflectivity over a wide or narrow range of wavelength.'3"51 Different fiber gratings were used for various purposes, such as wavelength division multiplexers in telecommunication systems, narrow-band high-reflective mirrors for laser or sensor applications, or wavelength-selective filters in fiber amplifiers.'61 The field of fiber grating study is almost thirty years old. Ken Hill and coworkers reported the photosensitivity in germanosilica fibers in 1978.'11 They found that launching a 488 nm Ar+ laser into a fiber could introduce Bragg grating in the fiber core. However, the discovery remained dormant for about ten years until Gerry Meltz et al reported holographic writing of gratings in fibers using single-photon absorption at 244 nm.'21 The discovery renewed worldwide interest in this subject, after people realized that the Bragg grating period can be shifted to 339

340

Glass Photosensitivity and Fiber Gratings

response for a specific wavelength of interests, for example: at 1550 nm for optical communications. The photosensitivity and fiber gratings technology opened a new era in the field of fiber-optic based devices. In this chapter, the photosensitivity of optical fibers, the principles of fiber Bragg gratings, the fabrication of fiber gratings and their specific applications will be reviewed, special focus will be on our works about photosensitivity of Sn-doped and Pb-doped glasses, applications of fiber grating will also be discussed. 10.1 Glass Photosensitivity Material photosensitivity invariably refers to a permanent change of refractive index or opacity induced in the material by exposure to light radiation.[7] Photosensitivity is the basis to fabricate fiber gratings, and fiber grating is the most important device using the effect of photosensitivity. For nearly ten years after the Hill's discovery, the phenomenon of photosensitivity was considered to present only in the special Bell Northern research fiber until Stone observed photosensitivity in many other fibers, all of them have a relatively high concentration of germanium.[8] Today, Ge-doped silica glass is still the fundamental photosensitive material in this field. Techniques like hydrogen loading, flame brushing and boron codoping are widely used to enhance the photosensitivity of Ge-doped silica fibers by as much as two orders of magnitudes.[9"11] The mechanism of glass photosensitivity has been widely studied. Novel photonic materials with higher photosensitivity are being searched. Recently, Sn-doped silica glass received more and more attentions as a hopeful material,'121 Pb-doped silicate glass is another good candidate.'131 In addition to fibers, technologies to prepare photosensitive planar films attract more and more interests in recent years to expand the application field from fiber optics to integrated optics.[14'15] The fabrication methods of photosensitive glasses are similar to other photonic glass materials. For bulk materials, the main techniques include

10.1 Glass Photosensitivity

341

the casting method and various CVD (chemical vapor deposition) techniques such as MCVD (modified chemical vapor deposition), OVD (outside vapor deposition), VAD (vapor-phase axial deposition), PCVD (plasma chemical vapor deposition) etc. As for glass films fabrication, the usual methods include radio frequency-sputtering, PLD (pulse laser deposition), Sol-gel technique, PECVD (plasma enhanced chemical vapor deposition), FHD (flame hydrolysis deposition), HARE (heliconactivated reactive evaporation) etc. 10.1.1 Ge-doped glass For high concentration Ge-doped silica glasses, the reported photoinduced refractive index changes (An) were in a range of -6.OX 10"3 to 2.1 X 10"3.[16'171 If a germanosilica fiber is hydrogenated, its photosensitivity can increase by two orders of magnitude. For example, R. M. Atkins et al reported results for a silica fiber contained 3 mol% Ge0 2 and 2.4 mol% H2. After irradiated by 248 nm KrF laser light, an An of 5.9 X 10"3 was obtained.191 While an irradiated 3 mol% Ge0 2 doped standard telecom fiber only have a refractive index change less than 3 X 10-5[.8]

The mechanism of photosensitivity in Ge-doped glasses was study widely for a long time. The origin of the photosensitivity is very complicate and not fully understood yet. Generally, the photosensitivity is higher in Ge-doped glasses with higher Ge0 2 concentration. But An actually also depends on sample preparing processes and UV irradiating conditions (wavelength, pulse energy, repetition rate, irradiating time etc) closely. Even the sign of the An also relies on fabrication technologies. The situation is the same to the Sn-doped and Pb-doped glasses. Several models were proposed. It is generally accepted that the defects in the glass matrix are responsible for the photo-induced refractive index changes. These defects form easily during the fabricating and photosensitizing processes. But it was also believed that more than one mechanism involves. Most popular models include color center model and compaction model. Color center model: Color center model was proposed by Hand and Russell in 1990.[19] They considered that An came from the change of

342

Glass Photosensitivity and Fiber Gratings

absorption. They are related by the Kramers-Kronig (K-K) relationship: An(X')

1 1

"In

&1

L Aa(A)

'*. 1- {x/x'J

-dX..

(10.1)

where An(X') is the refractive index change at the wavelength X', Aa is the photo induced change of the absorption, X\ and X2 are boundaries of the whole absorption spectrum. The main idea of the color center model is that UV light converts the color centers in the material and alters the absorption spectrum, resulting in the refractive index change via K-K relationship. This model bases on some experiment phenomena and is accepted widely. R. M. Atkins et al calculated the refractive index change in a germanosilica fiber according to the color center model.[20] The result was in good agreement with the value estimated from the formed fiber grating. Fig. 10.1 shows typical UV absorption spectra of a Ge-doped silica glass before and after UV exposure. Both curves can be divided into three parts. 180 nm is the absorption edge corresponding to the material band gap. The glass is highly transparent for wavelength longer than 280 nm. Absorption peaks between 180-280 nm are from defects. These defects are often called color centers because of their strong absorptions. The color centers we will discuss are listed in Table 10.1.

180

200

220

240

260

280

Wavelength (nm)

Fig. 10.1 UV absorption spectra of 3 mol% Ge0 2 MCVD optical fiber preform core, (i) before and (ii) after 30 min of UV exposure. The induced UV spectral changes are shown in (iii). After reference [20].

10.1 Glass Photosensitivity

343

Table 10.1 Color centers in Ge-doped silica glass related with photosensitivity. Absorption Name and Absorption Energy abbreviation of the Peak (nm) (eV) defects 4.6

270.0

5.08

244.0

5.16

240.0

5.8

213.8

6.4

193.8

Germanium Electron Centers (GEC) Germanium Oxygen Deficient Centers(GODC) Germanium Oxygen Deficient Centers(GODC) Germanium Electron Centers (GEC)

Name and abbreviation Ge(l) Neutral Oxygen Monovacacy (NOMV) Neutral Oxygen Divacacy (NODV) or Germanium Lone Pair Center (GLPC) Ge(2) GeE'

According to Table 10.1, there are two kinds of Germanium Oxygen Deficient Centers (GODC): NOMV and NODV. Their overlapping absorption bands around 240 nm can be found in Fig. 10.1. After UV exposure, the 240 nm absorption band decreases while the absorption around 190 nm rises up. This phenomenon is usually accompanied with refractive index change. The 190 nm absorption band is due to GeE' center absorption. It was supposed that after UV light irradiation, GODC centers in material convert to GeE' centers. The conversion dynamics was studied carefully in theory and has been partially proved by some measurements.[21 "24] Compaction model: this model points that laser irradiation induces density alteration in photosensitive materials, which may also be a reason for the refractive index change. This theory was well testified on a thin film of a-Si0 2 in 1986.[25] M. G. Sceats and M. V. Bazylenko introduced it to describe the origin of photosensitivity in glasses.126'271 A. I. Gusarov assumed that color center model plays a key role under the low irradiating dose condition. If the UV dose is high, compaction model may have a high weight.[28J In addition to the above models, dipole model[29] and stress-relief model'301 can also explain part of experimental observations.

344

Glass Photosensitivity and Fiber Gratings

Some techniques are also developed to efficiently enhance the photosensitivity of germanosilica glasses. Hydrogen loading and flame brushing are the most frequently used techniques. Hydrogen loading: hydrogen loading sensitizes the Ge-doped fibers by diffusing H2 into fiber cores at high pressure and temperature.[9] But the increased photosensitivity easily reduces as hydrogen diffuses out. Hydrogen loading will also introduce losses at 1.39 and 1.41 urn from the increased OH concentration in the fiber.[9'18'31] Using deuterium instead of hydrogen can shift the absorption band out of 1.55 um.[31] But the cost is presumably much higher. Flame brushing: this technique is to brush the fibers or waveguides repeatedly by a hydrogen-rich flame.[10] The flame temperature is approximately 1700°C, making the hydrogen diffuse into the core of the fiber very quickly. The sensitizing process only need about 20 minutes. The increased photosensitivity by flame brushing is permanent. But fiber strength is weakened after high temperature treatment. 10.1.2 Sn-doped glass Sn0 2 has been used as a codopant to increase the photosensitivity of germano-silicate optical fibers.[32] Sn-doped silica optical fibers showed higher photosensitivity and excellent thermal stability than germanosilica fibers and the doped glass has low absorption on the telecom transmission widow.[12'33] In 1995, L.Dong reported An of 1.2 X 10"3 in a silica fiber doped with 0.6 mol% Sn0 2 and 9 mol% P205.[34] It is one of the earliest results about the photosensitivity of Sn-doped glass fiber (without germanium). The fiber was irradiated by a 50 mJ/cm2, 20 ns, 20 Hz, 248 nm KrF laser. G Brambilla et al proved in 2000 that the photosensitivity of a 0.15 mol% Sn0 2 doped silica fiber was almost the same as a 10 mol% Ge0 2 doped silica fiber, both of them were about 3 X 10"4.[35] K.Gaff et al prepared the Sn-doped silica glass with the Sn0 2 concentration between 5 mol% to 25 mol% by HARE technique.1151 An of -2.7 X 10"3 was observed under the exposure of a 32 mJ/cm2, 248 nm KrF laser.

10.1 Glass Photosensitivity

345

10.1.2.1 Mechanism The discussion about the mechanism of the photosensitivity in Sn-doped glass approximately started in 2000.[33'35] From then on, more and more research works indicated that the origin of the photosensitivity in Sndoped silica glass is more complicate than germanosilica, photo induced defects conversion mechanism only plays an insignificant role. G. Brambilla suggested that the similar defects conversion such as TODC centers (tin oxygen deficient centers) to SnE' centers like in Gedoped glass occurred. However, calculation by K-K relationship indicated that the absorption change was too small and mostly leaded to negative An, while An in the fiber was found positive. Brambilla pointed out that the photosensitivity in Sn02:Si02 fibers may be related to the fiber itself (fabrication, geometry, stress, etc.) and is not only a consequence of material properties.[35] A. Anedda measured VUV (Vacuum Ultraviolet) absorption spectra of Sn-doped silica fiber preforms.[36] He found that the photosensitivity cannot be solely explained by photo-conversion of defects and VUV absorption changes. Structural modifications accompanying the defect photo-conversion process were suggested to be responsible for the positive refractive index changes. N. Chiodini et al proved the existence of SnE' centers in x-ray irradiated Sn-doped silica samples in 1998 by EPR (electron paramagnetic resonance) technique.[37] The possible substitutional role of Sn in Si02 is confirmed. They carefully checked the Sn-doped silica fiber preforms exposed to the 248 nm KrF laser. Methods like EPR spectroscopy, micro-Raman mapping, PL (photoluminescence) measurements, and stress polarimetry were used. They found the evidences about the photo induced SnE' centers, nonbridging-oxygen sites and stress relief. Their results supported that the photosensitivity in Sn-doped silica preforms mainly comes from structural rearrangement.138' In 2001, they reported that the UV bleaching of the 5 eV absorption band due to TODC centers was unessential to the photosensitivity. Medium range structures of (Si0 4 ) n rings have been proposed to be the structural units involved in photosensitive process. Conversion of Sn-containing rings with out-of-average dimension toward reduced size rings was

346

Glass Photosensitivity and Fiber Gratings

proposed to account for the photosensitivity through a process of structural compaction.'395 We prepared 0.1 mol% SnOa doped silica films by MCVD method.[40] Fig. 10.2 shows the film structures.

Fig. 10.2 Cross-section of a Sn-doped silica film made by MCVD and solution-doping method. 0.1 mol% Sn0 2 was doped.

The photosensitivity of Sn-doped silica films was measured by using a set-up shown in Fig. 10.3. RefefBtte«a#sl

Fig. 10.3 Set-up to measure the photosensitivity of Sn-doped silica films.

The setup allows us to measure the in situ refractive index change during the UV laser irradiation. A KrF excimer laser at 248 nm was used as the irradiation source. The laser pulse width was 15 ns and the repetition rate was 50 Hz. The laser energy density we used was about

10.1 Glass Photosensitivity

347

150 mJ/cm2. A standard single-mode fiber pigtailed 1550 nm laser diode (LD) was used as light source to measure the reflectivity of the film. The light from the LD was split into a probe beam and a reference beam by a directional coupler (DC). The probe light was launched into one arm of a two-fiber collimator through an optical isolator (ISO). The collimator carrying probe light was placed normally to the film surface. The light reflected from the film was collected by the same collimator and propagated through another arm to the detector. The reference light and reflected light were detected by a two-channel optical power meter simultaneously and the data were processed by a computer. After the reflectivity of the irradiated spot was measured, the refractive index «2 of the Sn-doped silica film was estimated by: ("LZZty

+ ( Vi)» +

2(^) ( ^)cos ( 2^ 2 /0

2 — ( ~y H - ) + 2(— -)(— -) cos(2 — n2h) +1 w, + n2 n2 + «3 nx + n2 n2+ n3 A0

where R(n2) and h are the reflectivity and the thickness of the Sn-doped silica film, «/ and n3 are refractive index of air and pure silica substrate respectively, Xo is the wavelength of probe light. The advantages of our novel setup are simple and robust. Fluctuation of the probe light source has been taken care and corrected. In addition, as the measurement was carried out in real time during the irradiation, the same spot measurement can be guaranteed, therefore the measurement error was effectively reduced. The changes of refractive index estimated above are more reliable. After irradiating the Sn-doped glass film by the 150 mJ/cm2, 248 nm laser, positive refractive index change about 2><10"4 at 1550 nm was observed (see Fig. 10.4). The absorption spectra of the Sn-doped silica film were shown in Fig. 10.5. The peak at 252 nm is related to the TODC absorption. The absorption spectrum after the sample was annealed in 0 2 at 700°C for 120 hours was shown in dash. The disappearance of the 252 nm absorption band supported the above conclusion. We found that the photosensitive behavior was distinct under different UV exposure strength. The absorption spectra exposed to 266

Glass Photosensitivity and Fiber Gratings

348

e c

nm laser with high energy density (about 50 mJ/cm2) are shown in Fig. 10.6. It can be seen that the absorbance below 295 nm decreased obviously as irradiating time increased. At the same time, the absorption edge shifted toward shorter wavelength. In the inset, the largest negative absorbance change appeared at 256 nm after irradiation. No positive change in the whole absorption bands was observed.

ange

)r •4-pil

.S 1.5x10 *

"

-v/ 9

V

•v . « 5.0x10"° u <M
°*

.

r

J 1.0x10"*

I

I

1 1 "

J

0.0- \

,

,

,

4000

-T

8000

'

1

12000

16000

Laser pulses Fig. 10.4 Evolution of the photo-induced refractive index change in the Sn-doped silica film as a function of irradiating laser pulses. 15 ns, 150 mJ/cm2, 248 nm KrF laser was used. 800 before annealing after annealing in 0 2 for 120 hat700°C

200

250

300

350

Wavelength (nm) Fig. 10.5 Absorption spectrum of the Sn-doped silica film. The absorption spectrum of the sample after annealing in 0 2 at 700°C for 120 hours was shown in dash. The peak at 252nm disappeared after annealing.

10.1 Glass Photosensitivity

349

1000 <-N

- before exposure - after 15 mins exposure to 266 nm, 50 mj/ (cm ) laser

800- \

- after 30 mins exposure to 266 nm, 50 mJ/(cm ! ) laser

& u - ^ 600- A c o A •-C a 400Vl o <» A -< 200-

1U-.

1

•

•/n<:cr^z; -10-

\

-20.

-30 J -40-50-

-so*

/

/'

N

1

'

' after 15 mins exposure after 30 mins exposure

-

-7(1.

250

300

Wavelength (nm)

200

250

300

350

Wavelength (nm) Fig. 10.6 Absorption spectra of the Sn-doped silica film exposure to 266 nm pulse laser with high energy density (about 50 mJ/cm2). The inset shows the spectra difference with the beginning.

1000 before exposure — after 5 mins exposure to 266 nm, 10 mJ/ (cm ) laser • - - after 20 mins exposure to 266 nm, 10 mJ/ (cm ) laser after 80 mins exposure to 266 nm, 10 mJ/ (cm ) laser

250

300

Wavelength (nm)

250

300

350

Wavelength (nm) Fig. 10.7 Absorption spectra of the Sn-doped silica film exposure to 266 nm pulse laser with low energy density (about 10 mJ/cm2). The inset shows the spectra difference with the beginning.

Glass Photosensitivity and Fiber Gratings

350

The absorption spectra with low energy density (about 10 mJ/cm ) are shown in Fig. 10.7. The inset gives the absorption changes after irradiation. In contrast to above result, the absorbance below 247 nm and over 270 nm increased with the UV light dose. Three peaks around 195 nm, 215nm and 280 nm and a valley at 256 nm emerged. The absorption edge shifted toward longer wavelength. On the other hand, Raman spectra (excited by He-Ne laser at 632.8 nm) of a Sn-doped silica film and a pure silica plate was compared in Fig. 10.8. Before irradiation, both of the samples have the typical spectra as that of the Si0 2 . After irradiation, no detectable structural change appears in the Raman spectra except that in the spectrum of the Sn-doped silica film, luminescence background generated a strong broad spectral structure. These results indicate that the breaking of network bonds exist in the film due to Sn-doping. Together with Raman spectra, we believe that under high energy density irradiation, photosensitivity in the sample is mainly from the microscopic structural modifications originating from bond-breaking of oxygen deficient defects. On the contrary, under low energy density irradiation, photo-conversion of optically active defects may play an important role in the generation of photosensitivity. 1

1

1

1

A„

1

„

i*

6000.** ,*

O5000-

s d

.

1

1

1

1

Sn-Doped, before irradiation 1 — — Sn-Doped, after irradiation " . • - * Sn-Free, before irradiation . Sn-Free, after irradiation V

/

^4000-

1 3000-

1

* * -•*% >

1

fc

20001000-

1 \fszzZ>~

'\^^

\ 1

**

V 00

^•^^—^'•^N.

""*•......••-« i

200

'

i

400

'

i

600

i

800

^~ '

i

1000

•

1200

Wavenumber (cm") Fig. 10.8. Raman spectra of the Sn-doped silica film and a pure silica plate before and after UV laser pulses irradiated.

JO.] Glass Photosensitivity

351

10.1.2.2 Thermal stability It has been confirmed that the photo-induced refractive index changes in Ge-doped glasses will vanish entirely after annealing at 900 °C or higher temperature.'16'21'261 The situation in the hydrogenated germanosilica fibers is much worse. R. M. Atkins et al reported that for a 10 mol% Ge0 2 , 2.8 mol% H2 doped fiber, An can be erased totally after annealed at only 100°C.[41] Regeneration of refractive index change is possible in thermally-erased Ge-doped fibers, but the photosensitivity in most cases decreased. G. Bramibilla et al found that the An of a 5 mol% Sn0 2 , 20 mol% Na 2 0 doped fiber kept stable up to 600 °C. Their experimental results showed that the thermal stability of photosensitive glasses with different dopants could be sorted as Sn0 2 >Ge0 2 >B 2 03. The temperature dependent photosensitivity of these three fibers are shown in Fig. 10.9.[33] X. C. Long studied the thermo stabilities of the fiber gratings written in the Pb-doped fiber. The An modulation still existed under 600 °C annealing.[13] 1.0 0.8 O

4°- 6 o

0.2

0.0 0

200

400

600

800

1000

Temperature (°C) Fig. 10.9 Comparison of temperature stability of gratings written in three core glass compositions: Si0 2 :Sn0 2 :Na 2 0 (SSN), Si0 2 :Ge0 2 (SG), and Si0 2 :Ge0 2 :B 2 0 3 (SGB). The refractive-index modulation normalized to the initial value at room temperature. (Anmod/An0) was measured during step heating: The samples were heated in steps of ~45°C (starting from 205CC) in 2 min per step and kept at that temperature for 28 min before the temperature was increased in another step. After reference [33].

Glass Photosensitivity and Fiber Gratings

352

10.1.2.3 Sn, Yb codoped silica fiber preform The Bragg gratings written in rare-earth ions doped fibers have the dual functions: select wavelength and boost signal level. They can be useful as narrow-band filters, fiber lasers and sensors.142' N. Chiodoni reported the properties of an E^Ch-SnCVSiC^ silica glass prepared by sol-gel method.[43] The results showed that the Er, Sn codoped material had the similar photosensitivity as solely Sn doped silica glass. Moreover, photoluminescence (PL) at 1.55 um was also observed. We prepared a silica fiber preform codoped with Yb3+ and Sn in the core by the MCVD technique followed by a liquid doping process. The PL spectrum and photoinduced refractive index changes are shown in Figs. 10.10andl0.ll. It was found that Sn did not change the PL properties of Yb3+ around 1.1 urn. And the Sn doped silica glass co-doped with Yb still had good photosensitivity. A An of 2*10"4 was observed. In addition, after UV irradiation, the absorption and Raman spectra of the co-doped sample changed in the same way as Sn-doped sample. The results confirmed that, Yb and Sn have no negative effects to each other.

-before exposure, Peak X=1070 nm • - after 10 mins exposure, Peak A.=1068 nm — after 30 mins exposure, Peak X=1066nm

1100

1150 1200 1250

130C

Wavelength (nm)

1000

1050

1100

1150

1200

1250

1300

Wavelength (nm) Fig. 10.10 PL spectra of the Yb, Sn coped fiber preform before and after 266 nm UV irradiation. The inset is the spectrum of silica glass doped Yb only.

353

10.1 Glass Photosensitivity 0.0020

— after 30 mins exposure - before exposure

o o

a U ,U 0.0015

•£

0.0010-

,g .> §

0.0005 -2.500000

0.0000

0.000000

2.500000

5.000000

Radius (mm)

Fig. 10.11 The refractive index changes after exposure to 266 nm laser.

10.1.3 Pb-doped glass X. C Long found that the maximum An could reach 0.21 in a lead silicate glass with 57 mol% PbO concentration. In his work, 25 mJ/cm2, 10 ns, 10 Hz, 266 nm YAG laser was used.[13] We observed close result. An of0.25 was measured in a 50 mol% PbO lead silicate glass after the glass was irradiated by light pulses from a 50 mJ/cm2, 10Hz, 266 nm Nd:YAG laser.[44] The knowledge about the mechanism of the photosensitivity in Pbdoped glass is very limited. But it was believed that bond breaking and structure changes play the key roles.[13] Although color centers were reported to be formed in lead-silica glass through two-photon absorption of 532 nm photons coming from a Nd:YAG laser, no photosensitivity was observed in these bulk glasses at laser intensities up to its damage threshold^451 We observed large negative refractive index change in lead silicate glasses with different lead concentrations (from 30 mol% to 50 mol% PbO) by irradiation with fourth harmonic output of a Q-switched Nd:YAG laser (266 nm, 10 Hz repetition rate) at the energy density of 50 mJ/cm2. The largest An was -0.25 ± 0.04 in lead silicate glasses with 50 mol% PbO.

354

Glass Photosensitivity and Fiber Gratings 1.90

1

10

100

Time (min)

Fig. 10.12 The refractive index change versus irradiating time in the lead silicate glass with 50 mol% PbO.

The relationship between photoinduced index changes of refraction and PbO concentration in samples was shown in Fig. 10.13. An increases exponentially to the PbO concentration. Similar results were also obtained by other groups.[13'46]

PbO mol%

Fig. 10.13 The relationship between refractive index changes and PbO contents.

Energy density (mJ/cm ,

Fig. 10.14 The relationship between absorption coefficients at 600 nm and 266 nm pulse energy density. The sample was a lead silicate glass with 43mol% PbO contented. The irradiation time was 1 minute.

10.2 Principles of Fiber Gratings

355

UV-visible absorption spectra of the lead silicate glass with 43 mol% PbO were measured after exposure to the 266nm laser beams with energy density from 50 mJ/cm2 to 350 mJ/cm2. We found that (see Fig. 10.14) the absorption coefficients in visible wavelength increased suddenly when the energy density of the laser beam was larger than a threshold value. This indicates that the photoinduced structural changes of lead silicate glasses are different with varying of energy densities. 10.2 Principles of Fiber Gratings 10.2.1 Basic properties of uniform Bragg grating The simplest form of a fiber Bragg grating is a periodic modulation of the refractive index in the core of a single-mode optical fiber (see Fig. 10.15). Uniform fiber gratings are fundamental building blocks for most Bragg grating structures. Light propagating along the optical fiber will be scattered by each grating plane. If the light wavelength satisfies the Bragg condition ~kt+lc = ~kr,

(10.3)

light will be reflected by the grating. In Eq. (10.3), kt is the incident wave vector. K=2n/A is the grating wave vector that has a direction normal to the grating planes. A is the grating spacing. kr is the reflected wave vector. Bragg grating

Fig. 10.15 Illustration of a uniform Bragg grating with constant refractive index modulation amplitude and period. After reference [6].

356

Glass Photosensitivity and Fiber Gratings

In the first-order Bragg condition for uniform fiber gratings, the center wavelength of the back-reflected peak is called Bragg wavelength, defined as XB=2neffA, (10.4) where neff is the effective refractive index of the fiber mode at wavelength XB. 10.2.1.1 Bragg grating reflectivity Consider a uniform Bragg grating with an average refractive index n0 in the core. The refractive index profile can be simplified as 2.7ZX

n(x) = n0+An cos(

), (10.5) A where An is the amplitude of the induced refractive index (typically 10~5~10~2) and x is the distance along the fiber longitudinal axis. The reflectivity of this grating can be given by coupled-mode theory as the following expression:[3'4] R(L,A)=

f ^f , 2 2 AA:2sinh2(5Z) + 5 2 c o s h 2 ( ^ )

(10.6)

where R(L, X) is the reflectivity, it is a function of the grating length L and wavelength A, Q=nAn/AB is the coupling coefficient, Ak = k~n/A is the detuning wave vector, k = 2rni0/A is the propagation constant and s2 = ff-Ak2. At the Bragg wavelength XB, there is no wave-vector detuning, so Ak = 0. Therefore, the reflectivity reaches the maximum: Rmm=R(L,AB) = tanh2(^).

(10.7)

According to Eq. (10.6), the fiber grating reflectivity increases as An or L increases. A calculated reflection spectrum as the function of wavelength detuning is shown in Fig. 10.16.

10.2 Principles of Fiber Gratings

1548.8

1550.0

357

1550.2

Wavelength (nm)

Fig. 10.16 Bragg grating reflection spectrum as the function of wavelength detuning. After reference [6].

The approximate full width at half-maximum bandwidth is given by[5]

AA =

ABaA>+(±)\

,

(10.8)

2n

o

where L/A is the number of the grating plane. The parameter a ~1 for strong modulated gratings (for grating with near 100% reflection) whereas a -0.5 for weak modulated gratings. 10.2.1.2 Strain and temperature sensitivity^6'47] The Bragg grating wavelength XB, which is the center wavelength of the light back reflected from a Bragg grating, depends on the refractive index of the core and the periodicity of the grating. Both parameters are sensitive to strain and temperature. According to the Eq. (10.4), the shift in AB due to strain and temperature change is given by dn dAs dn dA (10.9) AXB = 2 ( A — + n—)AL + 2 ( A — + n—)AT dT 8T 8L dL The first term in Eq. (10.9) represents the strain effect on an optical fiber, corresponding to a change in the grating spacing and the strain induced change in refractive index. The above strain effect term can be expressed as A/LB=/lB(l-peK,

(10.10)

358

Glass Photosensitivity and Fiber Gratings

where p e is an effective strain-optic constant defined as P.=n2[p12+Kpn+p12)]/2,

(10- 11 )

in which pn and pn are components of the strain-optic tensor, n is the refractive index of the core, and v is the Poisson's ratio. For a typical optical fiber pn=0.113, pi2=0.252, v=0.16 and n=1.482, using these parameters and the above equations, the expected sensitivity at 1550 nm is 1.2pm/|j.£. The second term in Eq. (10.9) represents the temperature effect on an optical fiber. AB is shifted when the grating spacing change due to thermal expansion and the refractive index change due to thermo-optic effect. The wavelength shift for a temperature change AT can be written as Al^ = ^(a

+ £)AT,

(10.12)

where a=(l/A)(dL/dT) is the thermal expansion coefficient for the fiber, which is 0.55 XI0~ 6 for silica. The quantity £=(l/n)(dn/dT) represents the thermo-optic coefficient and is about 8.6 X 10~6/°C for the Ge-doped silica fiber core. Clearly, the index change is the dominant effect. From Eq. (10.12), the expected sensitivity at -1550 nm is approximately 13.7pm/°C. 10.2.2 Types of fiber Bragg gratings There are several distinct types of fiber Bragg grating structures such as the common Bragg reflector, the blazed Bragg grating and the chirped Bragg grating. These fiber Bragg gratings are distinguished by their grating pitch (spacing between grating planes) or tilt angle (between grating planes and fiber axis). 10.2.2.1 Common Bragg reflector Uniform Bragg gratings are the simplest and most frequently used fiber gratings. Depending on the parameters such as grating length L and magnitude of refractive index change An, the Bragg reflectors can function as narrow-band transmission/reflection filters or broadband reflectors. In combination with other Bragg reflectors, these gratings can

10.2 Principles of Fiber Gratings

359

be arranged as band-pass filters. An important application of Bragg reflector is in fiber lasers or external cavity laser diodes, where fiber gratings are used at one or both ends of the laser cavity and tune the laser wavelength by varying the Bragg resonance feedback.'481 Bragg reflectors are considered as excellent strain and temperature sensing devices because the measurements are wavelength encoded'491 to avoid amplitude or intensity fluctuations. Fiber Bragg grating lasers can also be used as sensors where the Bragg reflector serves the dual purpose of wavelength tuning and sensing. A series of Bragg gratings can be written on the same fiber, each having a distinct Bragg wavelength. This configuration is used for wavelength division/multiplexing or multipoint sensing. 10.2.2.2 Blazed Bragg gratings Tilting the Bragg grating planes at angles to the fiber axis will couple the light satisfying Bragg condition into loosely bounded guiding modes or leaking modes. The tilt of the grating planes and the refractive index modulation depth determine the coupling efficiency and bandwidth of the tapped out light. For blazed gratings, not only different wavelengths emerge at different angles, but different modes at the same wavelength also emerge at slightly different angles due to their different propagation constants. Therefore the blazed fiber grating acts as a spectrometer and mode discriminator. Multiple blazed gratings were used to flatten the gain spectrum of erbium doped fiber amplifiers.'50' Another interesting application of blazed gratings is mode conversion.'511 Cladding

/

Topped Light

-r© Blazed index modufcrtlon

Core

Fig. 10.17 Schematic diagram of a blazed grating. Light is directed either upward or downward depending on the propagation direction of the bound mode. After reference [6].

Glass Photosensitivity and Fiber Gratings

360

10.2.2.3 Chirped Bragg grating A chirped Bragg grating is a grating that has a monotonically varying grating period. This can be realized by axially varying either the period of the grating or the refractive index of the core or both. In a linearly chirped grating, the ^ B is a linear function of the axial position along the grating so that different frequencies are reflected at different points acquiring different delay times. Chirped gratings, therefore, can be used as a dispersion-correction and compensation devices like compressing temporally broadened pulses for high-bit-rate transmission over long distance. Other applications include chirped pulse amplification, chirp compensation of gain-switched semiconductor lasers, sensing, higherorder fiber dispersion compensation, ASE suppression, amplifier gain flattening, and band-blocking and band-pass filters.[7] M

Core

. ^

re*

Oocfclng /

shottX» tongX* (b)

longX-"

•"• "

^ = 3 <

Fig. 10.18 (a)A chirped grating with an aperiodic pitch for forward propagating light, (b) A cascade of several gratings with increasing period, which simulates a long chirped gratings. After reference [6].

10.2.2.4 Apodization of Fiber Gratings Fiber gratings are not infinite in length. Therefore side-lobe structure

10.3 Fiber Grating Fabrications

361

associates in the reflection spectrum (see Fig. 10.16). Keeping a constant average refractive index throughout the length of the grating while gradually altering the refractive index modulation depth on both sides of the grating can diminish the side lobes substantially. The technique is called apodization. The advantages of apodization are not only controlling the smoothness of the reflection spectrum, but also adjusting the dispersion characteristics of the fiber grating.[6'7] 10.3 Fiber Grating Fabrications The fabrication methods to write fiber gratings broadly fall into two categories: holographic methods, mainly by bulk interferometers or phase mask interferometers; and non-interferometric methods such as point-by-point techniques. The former technique divides a single input UV beam into two, make them interfere at the fiber; the latter depends on periodic exposure of a fiber to pulsed sources or through a spatially periodic amplitude mask. 10.3.1 Bulk interferometer method Standard holography is a ready-to-use technique for fiber grating writing, with the UV beam divided into two by a beam splitter and then brought together at a mutual angle of 0, by reflections from two mirrors. This method allows the Bragg grating wavelength to be chosen independently on X-uv (UV wavelength). The Bragg wavelength is adjustable from ^uv to infinity (with 0=0 ).[2] 10.3.2 The phase mask method One of the most effective methods for inscribing Bragg gratings in photosensitive fiber is the phase-mask method. A phase mask is formed holographically or by electron-beam lithography. A phase-mask grating has a one-dimension surface-relief structure fabricated in a high-quality silica glass plate transparent to the UV writing beam. The profile of the periodic gratings is designed that when the UV beam is incident on the

Glass Photosensitivity and Fiber Gratings

362

phase mask, the zeroth-order diffracted beam is much suppressed (typically < 5%) and the diffracted plus and minus first orders are maximized (typical >35% for each). A near-field fringe pattern is produced by the interference of the two diffracted beams. The interference pattern photo-imprints the refractive index modulation in the core of the photosensitive optical fiber that placed in contact with or in close proximity immediately behind the phase mask (Fig. 10.19). A cylindrical lens can be used to focus the fringe pattern along the fiber core.1521 Incident Uttrvtolet Laser Beam

Grating ligations Corrugations

I Slca Glass Phase Mask

Fiber

•sS:":::::::1::::::;:::":::1'-"—-1 (1111 I I I ' ^ ' ' ' '

S

!

xCWacted Beams

Fringe pattern -1 order Zero order (<3% of throughput)

Fig. 10.19 Schematic of the phase-mask technique for photo-imprinting a refractive-index Bragg grating in a photosensitive optical fiber. After reference [6].

Phase mask technology has many advantages. It greatly reduces the complexity and instability of the fiber grating fabrication system. The phase mask forms a very stable interferometer since there are no adjustable parts. Mechanical vibrations problems are also minimized because the fiber is usually placed directly behind the phase mask. Phase mask technology is insensitive to the translation of the UV beam too. Several gratings can be written on a single phase-mask plate, each at the required wavelength. 10.3.3 The point-by-point method The point-by-point technique for fabricating Bragg gratings is to change the refractive index a step along the fiber core at one time. Each grating

10.4 Fiber Grating Applications

363

plane is produced separately by a single focused UV pulse passed through a slit. The refractive index of the irradiated core changes locally. The fiber is then translated through a distance to repeat the next step gain. Essential to the point-by-point fabrication technique is a very stable and precise submicron translational system.1531 Flexibility is the main advantage of the point-by-point technique. Since the grating structure is built up point by point, variations in grating length, grating pitch and spectral response can be easily incorporated. Because the UV pulse energy can be varied between points, the refractive-index profile of the grating can be provided for any desired spectral response. Compared with other techniques, point-by-point writing technique is especially useful in fabricating long period gratings (period >10 um). m 10.4 Fiber Grating Applications Fiber gratings are widely utilized as wavelength selective elements, sensors and chromatic dispersion compensation devices. Here we give some examples that we developed. 10.4.1 Hyper-narrow bandwidth fiber grating filter

Port 2

1

in

X-X-iout Port4

Fig. 10.20 Schematic of a fiber grating filter with the bandwidth of 0.02 nm.

The schematic diagram shown in Fig. 10.20 is a fiber grating filter with a hyper-narrow band of only 0.02 nm. This filter is composed of three fiber Bragg gratings and a 3 dB fiber coupler. The FWHM bandwidth of the three fiber gratings is 0.1 nm. FBG3 and FBG4 have the same

Glass Photosensitivity and Fiber Gratings

364 hfer1(A) 1051.93nm 150.30 pW

MhrU 1G50.67nm 75.33 pW

hfclR IQ54S3nm 7304 pW

BW 4.26nm 5012 E-3

Mktt(A) 1053.166 nm 214.62 nW

CWl 1052.30 n

?

dftflL 1053.107 nm 10757 nW

MkrlR 1053220 nm 105.94 nW

~ ^ ™ ~ ~ ™ ~ ™ -

T

BW O.Mlnm 5012 E-3

~ ™

F

CWL 1053.IG3 rr

™ -

—

—-I-—j-—f—i—[—j-—4—4—|—

/! \ . \ / i \

__-j-—|—4—Ju_L-ya-._4--4---t—1 ;

:

lr<

|

.

'

• . ,-1-7

—^~J~ ~ - ^ ~

•..i,'^"F ! ., 104321 fiSW. VBV.-1

02 nm 135 Hz

.*'. Sat*: ST:

105321 758 pW 3.36 »

M^.

U--

_ , -f

. -,<••<..-.

.<.

jSSfi^ / \"

i:.:i-v 106321 85W VSW:

Off

0.06 im 49 Hz

Sem: ST

V;

Y

K_

1053.159 8.63 pW 4.17 s

—

te^

1053.409

Off

Fig. 10.21 input spectrum in port 2 (left) and output spectrum from port 1 (right). YDF

Q

77nm LD Signal Input

+-E&

Signal Isolator Output

WDM

« - *

Isolator

(a) Signal Input 4

o

YDF

lG53.2nm FBO

WDM

Circulator

Signal! Output

977nm LD

(b)

Fig. 10.22 Illustration of the fiber amplifiers, (a) Traditional setup (b) Double-pass structure base on a fiber Bragg grating.

specifications. The Bragg wavelength of FBGl is tunable. Carefully choosing the distance between coupler and fiber gratings (L^c and Lc_3) and the Bragg wavelength of FBGl, 0.02 nm narrow bandwidth output could be obtained at Port 1 when a broad-band laser is launched to Port 2. This filter is equivalent to a Michelson interferometer cascaded with a F-P resonant cavity. The input and output spectra are shown in Fig.

10.4 Fiber Grating Applications

365

10.21. If only one fiber Bragg grating is used to achieve the same effect, the grating length should be 120 mm which is very difficult to realize. 10.4.2 Double-pass fiber amplifier based on a 1053 nm Bragg grating A fiber amplifier with double-pass structure uses the pump energy more efficiently and increases the signal gain. If a filter is inserted, the ASE noise can be suppressed. Fig. 10.22 shows a double-pass Yb-doped fiber amplifier working near 1053 nm. A fiber Bragg grating is used as the wavelength selective reflector. A traditional fiber amplifier setup is also given for comparison. 30 25-

-A— Single Pass -•— Double Pass


20-

^—' 15a •i-H

O a •SP i/S

10-

SS

5

£

0-5-10 35

—i—

40

45

50

55

—i—

60

65

70

Pump Power (mW) Fig. 10.23 The signal gains versus the pumping power.

As shown in Fig. 10.22 (b), the 1053 nm signal was coupled into a Yb-doped fiber through a circulator and was amplified. When the light reached the fiber Bragg grating, the signal was reflected and amplified again while the ASE passed the grating and was eliminated. The doublepass amplified signal was coupled out through the same circulator. The relationship between the gains and noises with the signal and pumping powers in both amplifier structures were shown in Fig. 10.23 and Fig. 10.24.

Glass Photosensitivity and Fiber Gratings

366

The results show that the signal gain from the double-pass amplifier is higher than the traditional one when the pump power is higher than 61 mW. More pump power is needed in the double-pass amplifier, otherwise the absorption in the Yb-doped fiber will add when the pump energy is totally absorbed. On the other hand, with high enough pump level (e.g. 68 mW in Fig. 10.24), both the signal gain and noise coefficient are better in double-pass amplifier. Signal gain decreased with input signal power was due to the saturation effects. 30 —•—Double Pass —*— Single Pass

2826

f

24

14

-

|

2CM

S

18

20 18 16 ^

^

\

a

o

-A—A—£

.2P 16

-14

14 12 10-45

3

12 "g - 10 M

2 -40

-35

-30

-25

-20

-15

-10

-5

Input Signal Power (dBm) Fig. 10.24 The signal gains and noises coefficients versus the input signal powers. The pumping power was constant at 68 mW.

10.4.3 Narrow band fiber laser with the longitude modes selected by a 1053 nm Bragg grating Fig. 10.25 gives an illustration of a fiber ring laser. The basic ring structure is composed of a 1053 nm, 3 dB fiber coupler (C), a 980 nm/1053 nm WDM, the Yb-doepd fiber used as the active media (YDF1) and a 1053 nm isolator (ISOl). The other two arms of the fiber coupler are connected with another Yb-doped fiber plus 1053 nm FBG and an isolator (IS02) respectively. FBG feedbacks the laser light to the ring cavity, the port with IS02 is the output port. YDF2 and 1053 nm FBG together fixed the wavelength and compress the output spectrum. The

10.4 Fiber Grating Applications

367

center wavelength is 1053.132 nm, FWHM is 0.006 nm. 0.006 nm is the resolution limit of our Optical Spectrum Analyzer.

FBG YDF2

u

IS02

Fig. 10.25 Illustration of the fiber ring laser. The longitude mode of this laser was selected by a fiber Bragg grating (FBG).

10.4.4 Simultaneous pressure and temperature measurement using high birefringence fiber Bragg gratings Fiber grating sensors have many significant advantages and applications. But in general, as a wavelength-encoding optical fiber sensor, the Bragg wavelength shift in a fiber grating comes from two different effects, i.e., temperature and strain. It is hard to distinguish these two effects when only the wavelength shift is measured. Many works to discriminate the strain and temperature effects have been carried out, such as dualwavelength fiber gratings,'541 hybrid FBG/long-period gratings,[55] and different polymer-coated fiber Bragg gratings.[56] When a Bragg grating is written in a birefringent fiber, there exists two Bragg reflection peaks, corresponding to fast axis and slow axis respectively (see Eq. (10.3)). The slightly different relation of the two Bragg reflection position to temperature and stress gives a way to determine simultaneously temperature and stress. Here we describe the result of using a novel Hi-Bi optical fiber to measure gas pressure and temperature.'57'58! Fiber Bragg grating was written in a new kind of Hi-Bi fiber that was fabricated by MCVD method. The novel Hi-Bi optical fiber is called "quasi-rectangle" (China Patent No. ZL95243466). Its birefringence

368

Glass Photosensitivity and Fiber Gratings

comes from two borosilicate stress-inducing elements on the sides of the core (see Fig. 10.26). The two elements have different thermal expansion coefficient from their surroundings, so that when they cool down to room temperature after manufacturing process, anisotropic stress is set up across the core of the fiber. This fiber has much higher birefringence (7.2X10"4) comparing with other kinds of fibers such as PANDA fiber (4.5X10"4) and bow-tie fiber (5.5X1Q"4) because its two stress-inducing elements are closer to the core.[59,60]

Fig. 10.26 The cross-section of the "quasi-rectangle" Hi-Bi fiber. Two dark spots are stress-inducing elements.

Fig. 10.27 shows the reflection spectrum. Two similar reflection peaks correspond to the fast-axis mode and slow-axis mode respectively. Their wavelengths are 1556.82 nm and 1557.59 nm at 20 °C. The difference is 0.77 nm. The two Bragg wavelengths of the Hi-Bi fiber grating have slightly different sensitivities to the hydrostatical pressure and temperature. The shifts of two Bragg wavelengths, AXFB and AA,SB are given by AAra=KFrxAT+KFpxAP,

(10.13)

AXsB=KsTxAT+KspxAP,

(10.14)

where K F T and KsT are the temperature sensitivities of the two Bragg wavelengths corresponding to the fast-axis mode and slow-axis mode, KFP and KSP are the hydrostatical pressure sensitivities. It can be seen from the above two equations that the applied hydrostatical pressure and

10.4 Fiber Grating Applications

369

temperature can be uniquely determined if the shifts of two Bragg wavelengths in Hi-Bi fiber are measured.

m -o o a.

1555

1556

1557

1558

1559

Bragg Wavelength (nm) Fig. 10.27 The transmission spectrum of a FBG written in the "quasi-rectangle" Hi-Bi fiber. The left trough, X,FB =1556.82nm, corresponds to the fast-axis mode, the right, >.SB=1557.585nm, corresponds to the slow-axis mode.

*

1

1

1

1

1

'

1

1

|

6 1558- i^ /;ciT=0.0088(nm/'C) SB so

a >

|

1

|

1

-__*-—•"-"*""""^ "

1557A

.

—**

0.79

1556-

• - , 0.78 \ .

!%FB/;nT=0.0093(iWC) 1 ™

SO 00 l l

1

Iq^aTNJ.OOOSfiWC)

'ra ° - 7 5

1555-

"^

m

; . -

0.74 0.73

-

>0

1554-60

- 1

-40

-20

20

-40

-20 1

40

0 1

20 1

60

40 .

60 1

80

80 1

100

Temperature (°C) Fig. 10.28 The temperature dependence of the two Bragg wavelengths under a normal atmosphere. (•) slow-axis mode, (A) fast-axis mode, lines are the linear fittings. Inset: Bragg wavelength difference A.SB-A,FB dependence on the surrounding temperature.

The temperature characteristics of the FBG were studied experimentally. When the temperature rose from -50 °C to 80 °C, both of the Bragg wavelengths red-shifted linearly, but the difference of the two

370

Glass Photosensitivity and Fiber Gratings

Bragg wavelengths decreased (see Fig. 10.28). The temperature sensitivities of these two Bragg wavelengths were measured to be KFT=0.0093 nm/°C and KsT=0.0088 nm/°C, respectively. The reflection spectra were recorded for the pressure change from 0 to 10 MPa (see Fig. 10.29 for the measurement results). The two curves showed good linearity and the linear fit error was less than 0.3%, which mainly came from the errors of the barometer. The pressure sensitivities of both Bragg wavelengths were almost the same, K P F =K S F =0.020 nm/MPa.

-,

0

1

1

2

1

1

4

1

,

1

6

1

8

,

f—

10

Pressure change |nP£"MPa£©

Fig. 10.29 The gas pressure dependence of the two Bragg wavelengths at room temperature. (•) slow-axis mode, (A) fast-axis mode, lines are the linear fittings. Inset: Bragg wavelength difference X,SB-A.FB dependence on the applied gas pressure.

From the above equations and data, the applied gas or fluid pressure (in MPa) and temperature (in °C) could be obtained from the following equations: AT=2000(AA. FB -AX S B),

(10.15)

AP=930A2iSB-880A?iFB.

(10.16)

We measured temperature and gas pressure change simultaneously using the Hi-Bi fiber Bragg grating. Fig. 10.30 shows the typical measurement results when the gas pressure applied on the fiber grating

References

371

was changed at the two different temperatures (-50 °C and 80 °C). The deviations are less than 1 °C and 0.5 MPa respectively.

0

2

4

6

8

10

0

2

4

6

8

10

Pressure change AP (MPa) Fig. 10.30 The gas pressure dependence of the two Bragg wavelengths at two different temperature of-50°C and 80°C. (•) slow-axis mode, (A) fast-axis mode.

References 1. K.O. Hill, Y. Fujii, D.C. Johnson, and B. Kawasaki, Appl. Phys. Lett. 32 (1978) 647. 2. G. Meltz, W. W. Morey, and W. H. Glenn, Opt. Lett. 14 (1989) 823. 3. D. K. W. Lam and B. K. Garside, Appl. Opt. 20 (1981) 440. 4. P. St J. Russell, J.-L. Archambault, and L. Reekie, Phys. World 6 (1993) 41. 5. Agrawal G P, Radic S, IEEE Phot. Tech. Lett, 6 (1994) 995. 6. A. Othonos, Rev. Sci. Instrum. 68 (1997) 4309. 7. Raman, Kashyap, "Fiber Bragg Gratings", (Publication: San Diego Elsevier, 1999). 8. J. Stone, J. Appl. Phys. 62 (1987) 4371. 9.P. J. Lemaire, R. M. Atkins, V. Mizrahi, and W. A. Reed, Elect. Lett. 29 (1993) 1191. 10. F. Bilodeau, B. Malo, J. Albert, et al, Opt. Lett. 18 (1993) 953. 11. D. L. Williams, B. J. Ainslie, R. Armitage, et al, Electron. Lett. 29 (1993) 45. 12. G. Brambilla and V Pruneri, IEEE J. On Sel. Topics in Quan. Elect., 7 (2001) 403. 13. X.C. Long and S.R.J. Brueck, Opt. Lett. 24 (1999) 1136. 14. P. Niay, M. Douay, P. Bernage, et al, Opt. Mat. 11 (1999) 115. 15. K. Gaff, A. Durandet, T. Weijers, et al, Electron. Lett. 36 (2000) 842. 16. M.V. Bazylendo, D. Moss, and J. Canning, Opt. Lett. 23 (1998) 697.

372

Glass Photosensitivity and Fiber Gratings

17. Jarvis RA, Love JD, Durandet A, et al, Electron. Lett. 32 (1996) 550. 18. R.M. Atkins, P.J. Lemaire, T. Erdogan, and V. Mizrahi, Electron. Lett. 29 (1993) 1234. 19. D.P. Hang and P.S.J. Russell, Opt. Lett. 15 (1990) 102. 20. R. M. Atkins, V. Mizrahi, and T. Erdogan, Electron. Lett. 29 (1993) 385. 21. B.G. Potter, Jr., and K. Simmons-Potter, Nucl. Instr. And Meth. B 166 (2000) 771. 22. M. Fujimaki, T. Watanabe, T. Katoh, et al, Phys., Rev. B 57 (1998) 3920. 23. M. Essid, J. Albert, J.L. Brebner, K. Awazu, J. Non-cryst. Solids 146 (1999) 39. 24. T. Uchino, M. Takahashi, and T. Yoko, Phys. Rev. B 84 (2000) 1475. 25. C. Fiori and R. A. B. Devine, Mater. Res. Soc. Symp. Proc. 61 (1986) 187. 26. M.V. Bazylenko and M. Gross, J. Appl. Phys. 81 (1997) 7497. 27. M.G. Sceats, G.E. Atkins, and S.B. Poole, Annu. Rev. Mater. Sci. 23 (1993) 381. 28. A.I. Gusarov and D.B. Doyle, Opt. Lett. 25 (2000) 872. 29. D. L. Williams, B. J. Ainslie, R. Kashyap, et al, Proc. SPIE 2044 (1993) 55. 30. M. G. Sceats and S. B. Poole, Aust. Conf. Opt. Fiber Technol., (1991) 302. 31. K. Noguchi, N. Shibata, N. Uesugi, and Y. Negishi, J. Lightwave Technol. LT-3 (1985)236. 32. L. Dong, J. L. Cruz, L. Reekie, et al, IEEE Photon. Technol. Lett., 7 (1995) 1048. 33. G. Brambilla, V. Pruned, L. Reedie, et al, Opt. Lett. 25 (2000) 1153. 34. L. Dong, J.L. Cruz, J.A. Tucknott.et al, Opt. Lett. 20 (1995) 1982. 35. G. Brambilla, V. Pruned, and L. Reekie, Appl. Phys. Lett. 76 (2000) 807 36. Anedda, CM. Carbonaro, A. Serpi, et al, J. Non-Cryst. Solids 280 (2001) 287. 37. N. Chiodini, F. Meinardi, F. Morazzoni, et al, Phys. Rev. B 58 (1998) 9615. 38. N. Chiodini, F. Meinardi, F. Morazzoni, et al, J. Non-Cryst. Solids 261 (2000) 1. 39. N. Chiodini, S. Ghidini, and A. Paleari, Phys. Rev. B 64 (2001) 073102. 40. Chen GH, Li YG, Liu LY, He YJ, Xu L, Wang WC, J. Appl. Phys. 96 (2004) 6153. 41. R.M. Atkins and R.P. Espindola, Appl. Phys. Lett. 70 (1997) 1068. 42. G. G. Karapetyan, A. V. Daryan, D. M. Meghavoryan, N. E. Gevorgyan, Opt. Commun. 205(2002)421. 43. N. Chiodini, A. Paleari, G. Spinolo, et al, J. Non-Cryst. Solids, 311 (2002) 217. 44. Jia HZ, Chen GG, Wang WC, J. Non-cytal. Solids 347 (2004) 220. 45. K. W. Delong, V. Mizrahi, G. I. Stegeman, et al, J. Opt. Soc. Am. B 7 (1990) 2210. 46. S. Mailis, A. A. Anderson, S. J. Barrington, et al, Opt. Lett. 23 (1998) 1751. 47. W. W. Morey, G. Meltz, and W. H. Glenn, Proc. SPIE 1582 (1992) 36. 48. G. A. Ball and W. W. Morey, Opt. Lett. 17 (1992) 420. 49. Byoungho Lee, Optical Fiber Technology, 9 (2003) 57. 50. R. Kashyap, R. Wyatt, and P. F. Mckee, Electron. Lett. 29 (1993) 1025. 51. K. O. Hill, F. Bilodeau, B. Malo, and D. C. Johnson, Electron. Lett. 27 (1991) 1548. 52. K. O. Hill, B. Malo, F. Bilodeau, et al, Appl. Phys. Lett 62 (1993) 1035. 53. B. Malo, K. O. Hill, F. Bilodeau, et al, Electron. Lett. 29 (1993) 1668. 54. M. G. Xu, J.-L. Archambault, L. Reekie and J. P. Dakin, Electron. Lett., 30 (1994) 1085. 55. H. J. Patrick, G. M. Williams, A. D. Kersey, et al, IEEE Photo. Tech. Lett., 8(1996) 1223. 56. Y. Liu, Z. Guo, Y. Zhang, K. S. Chiang and X. Dong, Electron. Lett., 16 (2000) 564.

References

373

57. Chen GH, Liu LY, Jia HZ, Yu JM, Xu L, Wang WC, Opt. Comm. 228 (2003) 99. 58. Chen GH, Liu LY, Jia HZ, et al, IEEE Photo. Technol. Lett. 16 (2004) 221. 59. M. Sudo, M. Nakai, K. Himeno, S. Suzaki, A. Wada, and R. Yamauchi, Proc. 12th Int. Conf. on Optical Fiber Sensors, IEEE/LEOS and OS A (1997) 170. 60. L. A. Ferreira, F. M. Araujo, J. L. Santos, F. Farahi, Opt. Eng. 39 (2000) 2226.

Chapter 11

Glass Fibers for Photonic Crystals

Photonics plays an important role in modern technological systems, such as telecommunications and sensing systems. Traditionally, the manipulation of optical photons has relied in general on the mechanism of total internal reflection (TIR).[1] Since the 1980s, the research in new purpose-built materials has opened up the possibilities of localizing and controlling light in cavities and waveguides by a new physical mechanism/2'3] namely the photonic band-gap (PBG) effect, there has been a growing interests in the realization of photonic crystals or photonic band-gap structures as optical components and circuits/4' Photonic band gap that excludes light transmission in all directions for a specific wavelength range, is analogous to the electronic band structure of semiconductor materials,'51 where the periodic arrangement of ions on a lattice gives rise to the energy band structure and the energy bands control the motion of charge carriers through the crystal. The periodic arrangement of refractive index variation in photonic crystal controls how photons are able to move through the crystal. A simple example is a simple-cubic lattice, whose lattice points are occupied by dielectric spheres. The crux of photonic crystals is that the lattice constant lies in the order of an optical wavelength/1' The most striking effect is that for a certain range a frequencies (colors) light can not propagate in the photonic crystal. Natural examples can be found, such as opal. Light of frequency within the photonic band-gap of the opals is reflected, which causes the stone to appear colorful, although the stone is made of transparent silicon dioxide. The photonic band-gap edges in an opal shift as a function of direction, thus its color changes when you look at in different ways. The beautiful iridescent wings of some butterflies and 375

376

Glass Fibers for Photonic Crystals

beetles are also caused by photonic band-gap incorporated into the wing surface. The main concept of photonic band-gap structure or photonic crystal was first proposed by Yablonovitch[2] as a possible way of inhibiting spontaneous emission and JohnP] for localization of light in strong scattering dielectric structure. Photonic crystals are a promising technology platform for realizing compact optical components and advanced/novel functions in integrated photonic circuits. They have many potential applications, such as quantum optical devices, optoelectronics, inhibition of the spontaneous emission, single-mode light emitting diodes, waveguides in the optical domain, filters, polarizers, planar antenna substrates. Photonic crystals have three structural forms: one-dimensional (ID), two-dimensional (2D), and three-dimensional (3D) periodic structures. A one-dimensional photonic crystal is repetitive along one direction, forming a stack of dielectric sheets. They are used as interference filters or in distributed Bragg reflectors. A three-dimensional photonic band-gap crystal has periodic structure in all directions and possesses one or more complete photonic band-gaps. They are described as stacked cheesecloth or stacked wood. Since the periodicity of the medium must be comparable to the wavelength of the electromagnetic waves to inhibit their propagation, photonic band-gap in optical or infrared range requires sub-micron size structures. However, the fabrication of photonic crystals with threedimensional (3D) periodicity is still a challenge at this tiny scale. Two-dimensional photonic band-gap crystals are structures that exhibit photonic band-gaps for waves traveling within a two-dimensional plan. Two-dimensional photonic crystals can be fabricated in two different forms: photonic crystal fibers (PCFs) and planar photonic crystals. Optical fibers are an excellent approximation of twodimensional (2D) structures because they are effectively infinite in the third dimension. Optical fibers are nominally invariant along their length, with all interfaces parallel to the fiber axis. Russell first proposed the concept of PCF[6'7] and Knight and co-worker first reported a fabricated example in 1996.[8] The new class of optical fibers is featured by an array of air holes running along the fiber length, so it also named holey fiber or microstructure fiber. By using this particular structure, various unique

11.1 Light Guidance in PCF

m

features, which are not realized in conventional fibers, ~ J such as the single-mode cut-off wavelength from the UV to IR, numerical aperture (NA) values ranging from very lov/ to about 0.9[!5] and air core guidance, extremely low bending loss, manipulated dispersion and nonlinear characteristics, can be provided. The design flexibility for PCFs is very large, and designers can use many different, fascinating, and odd air-hole patterns to achieve specific PCF parameters. Furthermore, doped glasses are now used in a variety of fiber lasers and amplifiers; these can be combined with the unique capabilities of PCFs to provide even more useful devices. Hence, PCFs are now receiving interests from areas as diverse as spectroscopy, metrology, imaging, telecommunication, and industrial machining. 11.1 Light Guidance in PCF PCFs contain a hollow core or solid core and an arrangement of air holes that ran along the fiber length, and example is shown in Fig. 11.1.[9'16] Light can be guided using either one of two quite different mechanisms. [11-13] ^ | j e r i igj-gg aij-.hoieg are arranged in a periodic lattice, guidance can be obtained through photonic band-gap effects and it is therefore called PBG-PCF or hollow core PCF.

Fig. 11.1 Cross-section of photonic crystal fibers (a) hollow core PCF, (b) index-guiding PCF. After References [16, 9].

378

Glass Fibers for Photonic Crystals

The second form of guidance arises from TIR, so the core must have a higher average refractive index (solid core) than the holey cladding; and it does not rely on any periodicity of the air holes. This fiber is thus called TIR-PCF or index-guiding PCF. 11.1.1 Total internal reflection Traditional optical fibers or step-index fibers have a silica glass core surrounded by a cladding layer. The cladding has a lower refractive index than the core (n cladding < ncore), so that light can be reflected at the boundary by total internal reflection. Index-guiding PCF guides light following the principle of TIR as well. [11~131 Holes act to lower the effective refractive index in the cladding region, and light is confined in the solid core, which has a relatively higher index. Unlike conventional fiber, index-guiding PCF can be made entirely from single material, typically un-doped silica. The basic operation of index-guiding fibers does not depend on having a periodic array of holes. However, practically the holes are arranged typically on a hexagonal lattice, so these fibers are sometimes referred to as photonic crystal fibers. The parameters that characterize a PCF profile are the hole-to-hole spacing A, the hole diameter d as shown in Fig. 11.2 and the number of rings of holes used to define the cladding region. The average refractive index of the cladding is lowed by the air holes.

Photonitic crystal fiber

Step index fiber

Fig. 11.2 Cross-section of a photonic crystal fiber and step-index fiber.

11.1 Light Guidance in PCF

379

A P-parameter has traditionally been applied to derive the higherorder mode cutoff as well as the mode field diameters for step-index fibers. In a standard step-index fiber with core radius a, and core and cladding indices ncore and nciadding at the operation wavelength X, the number of guided modes is determined by the Vsi value: n.

2a7r / AJnc

-n

(11.1)

cladding

Vs\ must be less than 2.405 for the fiber to be single mode. Thus single-mode fibers are in fact multimode for light of sufficiently short wavelength. Core diameter for step-index single mode fibers of silica is about 4-11.0 urn. In analogy with conventional step-index fibers, a normalized frequency parameter (F-parameter) for a PCF can be defined . [9,17-18]

as:

VPCF = Ink I ^ncore2{X)-ndadding\X).

(11.2)

The F-parameter is shown on the left side of Fig. 11.3 for a photonic crystal fiber having a core formed from a single missing hole. The condition for higher-order mode cut-off can be formulated as VPCF = 7T.[18] In contrast to step-index fibers, the effective index of the cladding of a PCF is strongly wavelength dependent. If the ratio of the wavelength of the guided mode to the hole-to-hole distance, X/A, approaches zero, then the effective cladding index approaches the effective core index. 1.451.44I 1.43-

^•^^_^

£ 1.42at £ 1.41 •

V

0.5 d*.= 0.3

0.4

TJ

^ - ^ d 4 l = 04

1? 1.40TS £ 1.39-

^VtfA^O.S

V. 1.38% 1.37-

< 0.3 0.2 0.1

1.36-

endlessly single-mode 'PCF-

multi-mode

0.0 04

0.6

VA

Fig. 11.3 (A) Effective index of photonic crystal cladding and (B) single-mode and multimode parameter space. After Reference [17].

380

Glass Fibers for Photonic Crystals

These unusual dispersion properties of the cladding, taken together with the single-mode condition of Fig. 11.3(a), lead to the behaviors as shown in Fig. 11.3(b). Here, the curved line across the lower-right side of the plot indicates where the single-mode condition is met. Above the line, only single-mode propagation is possible. For any fiber whose d/A ratio is less than 0.45, light with all wavelength will propagate in a single mode. Such fibers are termed endlessly single mode. The unusual guiding property is due to that the air holes (diameter d, spacing A) are considered as a modal filter or modal sieve.[6] Due to the evanescence of light in air, the holes are strong barriers and can be analogous to the wire mesh of a sieve. The fundamental transverse mode with a lobe dimension of ~2A fits into the core and cannot escape through the too-narrow silica gaps between the air holes. However, the scaling of the core size is limited by the increasing propagation loss. If the F-parameter of a PCF is smaller than 1, confinement of the mode is too weak and leakage occurs through the finite cladding. On the other hand, if X/A becomes too small (<0.1), scattering loss due to longitudinal non-uniformity (e.g. microbending, macro-bending and dielectric imperfections) plays an important role. Taking all these design considerations into account, the realization of "one missing-hole" photonic crystal fibers with a mode-field-diameter of about 26 um in the 1.5 (am wavelength region with low bending loss has been demonstrated.^91 11.1.2 Photonic band-gap guidance As for hollow core PCF, the effective cladding index, nodding, is greater than the index of the core, ncore, so TIR is no longer satisfied. However, if the holes are big enough, photonic band-gap exists and light is guided by photonic band-gap effects.11'~14] In a hollow core PCF, a large central air hole is surrounded by a number of periods of silica/air photonic crystal cladding, formed by a regular array of air holes in a glass matrix as shown in Fig. 11.1. It can exhibit photonic band-gaps, prohibiting propagation of light with frequency located within the band-gap. By adding an extra air hole to form a low-index defect, it is possible to introduce a localized mode in the band gap. Such a defect can act as the core, and guide light within a well-defined frequency window. In the

11.1 Light Guidance in PCF

381

most common design, air holes are arranged in a close-packed triangular lattice, and the core is composed of seven missing elements and surrounded by 10-12 rings of holes as shown in Fig. 11.1. Fig. 11.4(a) shows[14] the projected band diagram centered on the dispersion curve of the vacuum (vertical line) for such a triangular lattice with different air hole diameters D, while keeping the pitch A fixed.

Fig. 11.4 Band structure with different the air-filling fraction, (a) band structure for a triangular lattice with different d/A. (b) the evolution of the edges and the center of the lowest-frequency band-gap as d/A increases, (c) the evolution of the bandwidth with d/A, After Reference [14].

The gray scale indicates the density of states (DOS) of the modes supported by the structure. The DOS values are normalized to that of silica bulk material (n=1.44). The band-gap corresponds to DOS=0, and is depicted in white. For D/A=90% and below, the band structure

382

Glass Fibers for Photonic Crystals

exhibits several narrow band gaps. J As D/A increases, the band-gaps located in the higher frequency region (relatively to KA = 9 for D/A = 90%), tend to shrink and eventually disappear, while the band-gap at KA~9 gets wider and wider and its location shifts to higher frequency (see Fig. 11.4(b)). Fig. 11.4 (c) shows the evolution of the bandwidth of this band-gap with D/A, and indicates sort of a cut-off value for D/A above which the bandwidth widens at a very fast rate as D/A increases, e.g. an increase of D/A from 94% to 96% doubles the bandwidth. Moreover, the band diagram for D/A=100%, shown for solely academic purpose as its fabrication is impossible, indicates that the band-gap is due to the glass features located at the interstitials of the stacked air holes. The sensitivity of the width of this band-gap to the air-filling fraction is consistent with the experimental result of dramatic drops in the transmission loss of hollow core-PCF based on this structure. Therefore a high air filling factor (D/A>90%) is usually required to achieve a broad bandwidth and low-loss in silica hollow core-PCFs. This band-gap structure of the fiber is measurable, angularly-resolved scattering measurement enables a direct experimental visualization of the photonic band gap and guided modes in a hollow-core PCF.[16] Yan et al. proposed an improved photonic crystal cladding structure for hollow core-PCFs whose cladding air-holes are arranged in a triangular lattice pattern.[20] A larger degree of freedom in controlling the cladding band-gap regions can be obtained by increasing the size of concentrated silica region in the cladding. A fiber with this type of cladding would perform better in terms of the PBG-guiding wavelength range, radiation loss owing to finite cladding size, and the ability to avoid surface mode problems. On the other hand, moderately low-loss guidance can be obtained in certain hollow core-PCFs even without a band gap. One of the most intriguing hollow core-PCFs is one with a Kagome lattice (Kagome means meshes of a bamboo cage) in the cladding.[21] The lattice consists of fine silica webs arranged in a Kagome lattice and surrounded by air. It has a relatively higher transmission loss in comparison with a band-gap hollow core PCFs but with an extremely large bandwidth. The calculated band diagram of the Kagome lattice does not show any significant band-gap. However, the structure exhibits low density of states over the normalized frequency range.[14]

11.2 Fabrication

383

11.2 Fabrication There are many procedures, such as stacking, extrusion, drilling and solgel, to fabricate photonic crystal fiber performs, independently of their cladding structure.111"141 Stacking is the principal technique that has been used to make all fibers shown in Fig. 11.5. The typical method for an index guiding PCF is an array of capillary glass tubes bundled around a glass rod replacing the center capillary. For hollow core fibers, one or more capillary tubes in the center part are removed in order to create a hollow 'defect' core. A sleeved tube surrounds the entire assembly that forms the preform. Then the preform is drawn to a fiber by using a conventional drawing tower for glass fibers. Some typical hollow core PCFs and index guided PCFs as shown in Fig. 11.6 and Fig. 11.7, respectively. Chen and Hou et al. provided these images. [22~25]

frzhnit

KssSftsg &**«

^?^%^>l

Fig. 11.5 Schematic of the fabrication procedure of hollow core-PCF by stacking techniques.

Fig. 11.6 Cross-sectional profiles of hollow core PCFs.

384

Glass Fibers for Photonic Crystals

Fig. 11.7 Cross-sectional profiles of index guided PCFs.

However, PCFs with good uniformity have proven to be extremely difficult to manufacture. During air/glass holey fibers drawing, some deformations of the holey cladding occur, the holes become non-circular because the surface tension varies with the temperature gradient around each individual hole. The changes of cladding microstructure during the drawing process are also relative to the effects of pressure inside the holes, the structural diameter of fiber perform and the draw/down ratio. By using a classic principles of mechanical method, Zhou and Hou et al. analyzed the shape transformation of a fiber preform microstructure during of the fiber drawing process at high temperature. They provided a mathematical model for the fabrication procedure/24' Some drawing technical parameter and physical constants of the microstracture fiber material can be predicted. One example of the fabricated fiber based on the model is shown in Fig, 11.7 (right side). On the other hand, Zhou proposed a method for designing beam-shaping optical fibers, and applied the method to design photonic crystalfibers.*251A shaped single mode field, such as flattop, concave or convex intensity distribution can be formed in these fibers v/hen the refractive index profile matches with structure parameters. Based on the design, an Nd3+ doped silicate glass holey fiber was fabricated. When the Nd3+ doped holey fiber was excited. by an 808 nm Laser diode, it behaved a shaped single mode field with concave intensity distribution as shown in Fig. 11.8. Silica holey fibers have been made by using sol-gel casting techniques.1261 A mold containing an array of mandrel elements was assembled and then filled with colloidal silica dispersed at high pH with an average particle size of 40 nanometers. The pH was lowered causing

385

11.2 Fabrication

the sol to gel. At the wet gel stage, the mandrel elements were removed, leaving air columns within the gel body. The gel body was then treated thermo-chemically to remove water, organic and transition metal contaminants. The dried porous gel body was then sintered near 1600 °C into viscous glass and subsequently draws into fiber. During removal of the mandrels, the gel body with air-fill fractions >25% would easily crack. Therefore, high air-fill fraction glass fiber can be obtained from hydrofluoric acid etched low high air-fill fraction preforms.

Pic* 1 H i

if*,'

A

4 A .O ^ i

' '

!

* •'

> »i • i< -I '

i3iiliL/»rfVi v311Jl&I*^ iiAWV*** 1 1 V 1 U

i I. V^t U I V

\ ' i^U

, *»•»

.'

.

»^VJJJ*^VI U I I 1 V U I W

.1

'.>( u ^ v H t ^ d H v a n

gylfctHk} * V j l

ViVVItWti

U V

CHI

808nm laser diode, (a) 3D intensity distribution and (b) 2D intensity distribution.

Extrusion technique suits for fabricating the structured perform of compound glass with low-softening temperatures."' J In this process, a glass billet is forced by using a die at elevated temperature near the glass softening point. The perform geometry is determined by the die orifice. Furthermore, drilling method also is used to fabricate the perform. [29'30] The advantage of the extrusion and drilling techniques are allowance of fiber drawing directly from bulk glass. It works for many materials, including chalcogenides, polymers, and compound glasses. By comparison, stack and draw methods are limited to closest-packed geometries such as triangular or honeycomb lattices and cannot easily generate peculiar circular patterns. Extrusion technique provides design freedom, but is typically limited to the glasses with low glass softening point. Drilling method allows adjustment of both the hole size and spacing, but is generally limited to a small number of holes and restricted to circular shapes. Furthermore, drilling of preforms leads to roughened surfaces along the air hole so that extra steps of etching and polishing of

386

Glass Fibers for Photonic Crystals

the inner surfaces are desired. Sol-gel method accompanies large weight loss during drying, moreover, attenuation due to absorption of hydroxy bonds and roughened surfaces is large. Several designs such as fibers for low-bend loss, dispersion flattened designs or birefringent fibers require independent spacing, hole size or even noncircular holes. Different methods provide additional design flexibility that will be necessary for different types of fibers. 11.3 Properties of PCFs and Device Applications The presence of wavelength-scale holes in the transverse profile of a PCF can lead to novel optical properties that cannot be achieved in conventional optical fibers. It opens new prospects in various fields such as spectroscopy, metrology, telecommunication, nonlinear optics, laserinduced guidance, quantum optics. One important property of optical fiber is it extremely low propagation loss. The complex structure of PCFs generates several new loss mechanisms. The losses of PCFs depend strongly on their structures, the number of rings, and local variations of structure and surface roughness of the fiber. Early results yielded propagation loss of PCFs in the order of 100 dB/km, but the loss has been reduced dramatically. The loss of index-guiding PCF reached to the current record level of 0.28 dB/km [31] and the hollow core-PCF loss was ofl.2dB/km. [32] 11.3.1 Index-guiding PCFs Due to an operation based on TIR, the properties of index-guiding PCFs in many respects resemble those of step-index fibers. However, very important differences occur as a result of the complex geometry of the cladding structure and the large refractive index contrast between silica core and the air-filled cladding. Hence, depending on PCF design, the presence of wavelength-scale holes in the transverse profile of a PCF can lead to novel optical properties that cannot be achieved in more conventional forms of optical fiber.133"381 Examples of such properties include very-large-core (up to 25 um or larger) with endless single-mode

11.3 Properties of PCFs and Device Applications

387

guidance, ultra-small effective areas (down to approx. 1 um) and the extremes of fiber non-linearity, and a range of remarkable dispersion properties, including broadband flattened dispersion, anomalous dispersion below 1.3 um, and large normal dispersion values at 1.55 urn, polarization maintenance and high birefringence and very high numerical apertures (up to 0.9). Non-silica based PCFs [27"30] have been identified as particularly promising candidates for highly nonlinear applications due to the high nonlinear index of the materials. PCFs are ideally suited for applications requiring large nonlinearity, broadband operation range with single-mode guidance, large mode areas. 11.3.1.1 Dispersion Characteristics Dispersion is usually defined as a parameter proportional to the second derivative of the effective refractive index of a waveguide, and is determined both by the material properties and by the geometry. Since PCF has greater freedom in fiber design as d and A can be independently decided, a dispersion shifted and dispersion flattened PCF can be achieved with a proper design. By increasing the relative air-hole size d/A, the zero dispersion wavelength can be shifted to shorter wavelength down to 560 nm.[36'38] Shifting the zero-dispersion wavelength to regimes where there are convenient high-power femtosecond laser source (Ti: sapphire at 800nm, Yb-fiber and Nd:YAG at 1060nm) also allows the development of efficient supercontinuum sources. On the other hand, wavelength converters based on modulation of cross-phase and optical threshold devices based on self-phase-modulation require small normal dispersion to minimize coherence degradation, In addition, parametric oscillation and wavelength converters based on four-wave-mixing require small normal dispersion to achieve efficient phase matching. For dispersion compensation of standard fibers, a large normal dispersion is needed at 1550nm. The slop of the dispersion is also important and ultraflat dispersion at the wavelength range of interest enhances the useful spectral bandwidth of nonlinear devices. Several methods have been developed to design PCFs with ultra-flat dispersion of a certain value and for a certain wavelength range. For example, Saitoh et al. proposed a PCF structure as shown in Fig. 11.9,[39] the central core is perturbed with

Glass Fibers for Photonic Crystals

388

an extra air-hole with diameter dc. By a judicious choice of the geometrical parameters d/A and dJA, this PCF structure can exhibit ultraflattened dispersion characteristics from 1100 to 1800 nm with low confinement losses and small effective area. Rarity et al. used a singlemode photonic crystal fiber and pumped in the normal dispersion regime,1401 they obtained a PCF source of correlated photon pairs at 839 nm and 1392 nm. This single-mode source of pair-photons will have wide application in quantum communications.

(a)

(b)

Fig. 11.9 (a) Cross-sectional profiles of the PCF with ultra-flattened dispersion characteristics and (b) dispersion as a function of wavelength and diameter of cladding air-holes. After Reference [39].

11.3.1.2 Polarization characteristics Polarization maintaining (PM) fibers are optical fibers that preserve the polarization state of the light. PCFs take advantage that is able to fabricate anisotropic claddings, cores with high ellipticity and the high index contrast between silica and air. PM-PCFs have been fabricated based on structures of circular air holes with different diameters along two orthogonal axes near the core region,[33,34,381 or on asymmetric cores obtained by placing two adjacent rods. The values for birefringence reported for these structures range from 9.3x10"* to 3.7 xlQ"3, and the configuration using two large air holes near the core region has become the standard for the kind of structures'411 as shown in Fig. 11.10(a). Fig.

11.3 Properties ofPCFs and Device Applications

389

11.10(b) shows the wavelength dependence of the fiber birefringence. Its modal birefringence is 1.4xl0"3 at 1550 nm and three times larger than that of existing PN- fiber.

b

Fig. 11.10 (a) Cross-sectional profiles of the polarization maintaining-PCF. (b) wavelength dependence of the birefringence. After Reference [41].

A PM large mode area photonic crystal fiber has been demonstrated.1423 The birefringence was introduced using stress-applying parts. The fibers with mode field diameters from about 4 to 6.5 um exhibited a typical birefringence of 1.5x10"4 and were both single mode at any wavelength and had a practically constant birefringence for any wavelength. Another interesting application of PM-PCF is absolutely single polarization PCF.I43] Single-polarization fibers are optical fibers that guide only one polarization state in a specific wavelength range. Compared with PM-fibers that guide both polarization states, singlepolarization fibers typically have a higher polarization extinction ratio, which is independent of the length. Single-polarization fibers have various applications in gyroscopes, fiber polarizers, and fiber lasers and amplifiers to ensure linearly polarized output. Single-polarization fibers based on fibers with air-holes have been demonstrated. For example, Folkenberg et al. reported a single-mode photonic crystal fiber with mode-field diameter of 15.5 um,[43] it supported only one polarization state in a 220-nm broad spectral region centered at 727 nm.

390

Glass Fibers for Photonic Crystals

11.3.1.3 Nonlinear characteristics ofPCF The high index difference between the silica core and the air-filled microstructure enables tight mode confinement and a significantly larger numerical aperture (NA) resulting in a low effective area and thereby a high nonlinear coefficient.133'34,36"38] The air-filled region also results in strong wavelength dependence and is responsible for the large waveguide dispersion possible in such fibers as mentioned above. Small core PCFs can obtain short zero dispersion wavelengths in the infrared and visible range and such fibers are ideally suited for pumping with Ti: Sapphire femtosecond pulse sources since the low fiber dispersion provides excellent phase matching for creating wideband supercontinuum.[50] Some nonlinear PCFs were designed with a small core to get a high nonlinear coefficient as shown in Fig. 11.7. By choosing the dispersion profile carefully, the fibers can be tailored to facilitate different nonlinear processes. Nonlinear effects can be used for a wide range of optical processing applications including optical data regeneration, wavelength conversion, optical demultiplexing and Raman amplification. Specially, broad phase-stabilized frequency combs can be generated by launching femtosecond pulses from a Ti:Sapphire laser into nonlinear PCF and phase-locking the resulting output frequency comb.[44] It has led to technologically important applications in the field of precision metrology. Hu and Wang et al. reported that a random array of fused silica waveguide wires in a microstructure fiber could frequency converts unamplified ultrashort laser pulses in a broad frequency range with high efficiency.145^71 A cross-sectional view of the fiber is shown in Fig. 11.11.[46] There are two sections in the microstructured part of the fiber, inner microstructure part and outer microstructured part with radically different sizes of air holes (Fig. 11.11(a)). The size of air holes in the outer section is typically about 20 um. This outer microstructured part serves as a cladding, confining light to the inner microstructured part of the fiber (Fig. 11.11(a)). The sizes of air holes and fused silica channels in the inner section vary from 0.3 up to 2 (am and from 0.6 up to 1.5 |am, respectively. Each of an array of fused silica channels can be viewed as a fiber core surrounded by a random holey cladding, providing waveguide

11.3 Properties qfPCFs and Device Applications

391

due to the TIR. There are two typical geometries of waveguide channels in the fiber as shown in Fig. 11.11(b). Thefirsttype is bounded by a triad of air holes and has a triangular shape, such as channels 2, 3 in Fig. 11.11(b). The second type is bounded by four air holes, such as channels 1, 4 in Fig. 11.11(b). Dispersion can be switched in such waveguide arrays by coupling the pump field into waveguide wires with different diameters.

Fig. 11,11. Cross-sectional profiles of the random-hole microstracture fiber: (a) general view and (b) a close-up view. After Reference [46].

Fig. 11.12 Output beam patterns of anti-Stokes-shifted emission from channels 1 - 6 in Fig. 11.11(b) of the random microstructurefiber.After Reference [46].

This microstructured fiber integrated random arrays of waveguides with different diameters can frequency-convert unamplified subnanojoule Ti: sapphire laser pulses to any wavelength within a broad

392

Glass Fibers for Photonic Crystals

spectral range from 400 up to 700 nm. As shown in Fig. 11.12, when a Ti: sapphire laser pump pulses were coupled into submicron channels of the 3-cm-long microstructure fiber, the generation of supercontinuum was observed. Fig. 11.12 (l)-(6) are output beam patterns of anti-Stokesshifted emission and display the basic colors of palette ((1) blue, (2) orange, (3) celadon, (4) red, (5) green, (6) Cambridge blue) produced by waveguide channels 1 - 6 in Fig. 11.11(b). Zheng and Hou et al. reported that holey fibers with random cladding distribution have special ability of localizing light and controlling the group velocity dispersion.[48] A supercontinuum extending from 350 nm to more than 1700 nm was observed in the fiber by using a Ti: sapphire femtosecond laser. The maximum total power of the supercontinuum was 63 mW with 288 mW pump power. The wavelength and power, polarization states and waveguide modes of the visible light ranging in the supercontinuum could be tuned by adjusting the pump incident point or incident angle. Due to the small cores of the above described dispersion engineered/phasematched nonlinear fibers, output pulse energy is still limited and may be too low for applications like hyperspectral laser radar, hyperspectral imaging and speckle-free illumination. Such limitations are overcome in the recently demonstrated supercontinuum generation in a 25|j.m core large mode area-type fiber.[34] 11.3.2 Laser active PCFs One of the most promising applications for PCFs is in high-power fiber lasers. Doping of the core of PCFs with rare-earth elements such as Nd, Yb or Er produces fibers for laser and amplifier applications.118'34'35' 37,38,49, so] Traditionally, fiber lasers are constructed using dual-clad, stepindex fibers with a polymer outer cladding and a core doped with such rare-earth ions. [18] Unfortunately, it is difficult to extend this design to produce single-mode output at higher pump and output powers, as was discussed in Chapter 7 of this book, high power generated in small core size fiber will lead to detrimental nonlinear effects. The special and outstanding characteristics of PCFs are: very high numerical aperture (NA) [15] with double-clad structures for efficient pump light use and large core sizes improving the destruction limit for high power

11.3 Properties ofPCFs and Device Applications

393

applications. The combination of very large core size and high NA makes it possible to fabricate lasers and amplifiers with very high power and short fiber lengths. For example as shown in Fig. 11.13,t50] it is a rod-type photonic crystal fiber, the inner cladding of the fiber is surrounded by an air-cladding region, which produces a significantly greater index difference between the inner and surrounding region, and therefore a higher NA than that can be achieved by conventional polymer coated fibers, and effectively confines the pump light to a silica multimode pump core, commonly referred to as the inner cladding.

(A)

(B)

Fig. 11.13 (A). Microscope image of the rod-type photonic crystal fiber with a diameter of 1.7mm and (B) Close-up of the hexagonal inner cladding with a diameter of 117 \xm and core region with a diameter of 35 |im. After Reference [50].

The higher NA permits efficient pumping with relatively inexpensive high-power large-emitting-area pump diodes. Within the inner cladding, another microstructured rare-earth-doped core defines the laser-beam output parameters. The single-mode core can be expanded to a large mode area to facilitate high power levels in a single mode while avoiding nonlinearities and providing a good overlap between the pump mode and the laser mode. 11.3.2.1 CW high powerfiberlasers Continuous wave (CW) high power fiber lasers based on Yb-doped large mode area PCF (LMA PCF) have been the focus of considerable research. Li et al. obtained a 50 W fiber laser with single transverse mode operation using an Yb-doped LMA PCF and a Fabry-Perot (F-P)

394

Glass Fibers for Photonic Crystals

configuration. J Output powers up to 1.53 kW based on a LMA PCF has been obtained.118'34] The rod-type photonic crystal fiber as shown in Fig. 11.13(A) has a large outer diameter of the fiber (1.7 mm), it provides enough mechanical stability and the need for protective coating is eliminated, thereby improves the thermal properties of the fiber. Fig. 11.13(B) shows a close-up of the inner cladding and core region. The hexagonal inner cladding has a diameter of 117 um (face to face) and a numerical aperture of around 0.6. The signal core has a diameter of 35 urn and the hole-diameter to pitch ratio is ~ 0.33. The fiber with length of 48cm was pumped with 165 W at 976 nm resulting in 120 W laser output at 1035 nm with a slope efficiency of 74%. This is equivalent to a power extraction of 250 W/m.[50] An analysis based on a single-mode core with a mode-field diameter of 35 um reveals that a power level of «10kW with diffraction-limited beam quality is possible with established fiber technology.149'53] The first step in this direction has been done with a 1.53 kW emission out of an ytterbium-doped photonic crystal fiber.[18' 34] PCFs have been developed in which the cores have also been co-doped with photosensitive materials, enabling fiber-Bragg gratings to be inscribed right into the core using UV laser light, thus further simplifying the architecture.[53] 11.3.2.2 Pulsed high-power fiber lasers Due to the short fiber length, large mode area (reduced nonlinearity) and relatively high-energy storage capacity, PCF is also very suitable for Qswitched laser operation. Roser et al. reported a LMA Yb doped-fiber based chirped-pulse-amplification system (CPA) generating up to 131 W of average power of 220 femtosecond pulses at 1040 nm center wavelength in a diffraction limited beam (M2 < 1.2).[51] The pulse repetition rate was 73 MHz, corresponding to a pulse energy of 1.8 uJ and a peak power as high as 8.2 MW. The high average power femtosecond laser system consisted of a passively mode-locked, diodepumped solid-state laser system, a gold-coated diffraction grating stretcher, a two-stage Yb-doped single-mode PCF amplifier, and a fusedsilica transmission grating compressor. The experimental setup is shown in Fig. 11.14.

11.3 Properties ofPCFs and Device Applications

395

*as &i, rasas

Fig. 11.14 Schematic setup of the high average power fiber CPA system. After Reference [51].

The femtosecond seed source was a passively mode-locked Yb:KGd (WO^ (Yb:KGW) laser system producing 150 femtosecond pulses at a 73 MHz repetition rate and a 1040 nm center wavelength. These pulses were stretched to 120 ps using a conventional gold-coated 1200 line/mm diffraction grating. The preamplifier and the power amplifier were constructed with identical PCFs. The PCFs possessed an active core diameter of 40 um (NA, 0.03) and an inner cladding diameter of 170 urn (NA=0.62). This structure had a pump light absorption of 15 dB/m at 976 nm. Therefore, the fiber length used v/as just 1.2 m. The core parameters leaded to a calculated mode-field diameter of 35 um, resulting in an effective mode-field area of -1000 um2. The output power of the fiber laser system was limited by available pump power and not by degradation in pulse quality owing to nonlinearity or even by fiber damage. Thus, further scaling of the output power is possible with the present system. 11.3.3 Properties of hollow core-PCFs Hallow core photonic crystal fibers have become the most advanced manifestation of two dimensional photonic band-gap structures. Due to the hollow core guidance and the resulting small overlap between the propagating mode and the silica fiber material, these fibers exhibit lower Rayleigh scattering and absorption in the glass, lower nonlinearity and a range of interesting prospects such asiWMWSfl high damage threshold,

396

Glass Fibers for Photonic Crystals

radiation insensitive, easy access to rilling the core or cladding with gasses or liquids, gas/liquid sensors with long interaction lengths, low Fresnel reflection from the fiber facets, large span of dispersion values achievable: from highly negative to highly positive, bend insensitive. It is possible to guide light in a hollow core with low attenuation on kilometer length scales and provide excellent potential for applications such as power delivery and gas lasing or gas-based nonlinear, laser-induced particle guidance, stimulated Raman scattering (SRS) and laser frequency metrology.

Fig. 11.15 Optical micrograph of a hollow core-PCF. The other end of the fiber is illuminated with white light. The fiber showed guides blue and green light in the lowindex core. After Reference [10].

The first PCF in which light guidance was by a true full twodimensional photonic band gap was reported in 1998.[10] This PCF had a honeycomb lattice of air-holes, with an additional air-hole at the center, and light was guided in narrow glass regions adjacent to the central airhole as shown in Fig. 11.15. Theory predicts that the guided mode in this photonic band-gap fiber is highly dispersive and exists only for narrow wavelength ranges. This was confirmed by experiments, when illuminated with white light, a bright colored mode appears in the core the modes of this fiber came in yellow, blue and green. A calculation revealed that there is also photonic gap in two-dimensional amorphous photonic materials, which does not possess any long-range order but have only short-range order.[55] Yang and Chen et al. fabricated a hollow core holey fiber with a random distribution of air holes in the cladding, and this fiber also demonstrated many features that previously attributed

11.3 Properties ofPCFs and Device Applications

397

to photonic crystal fibers with perfect arrangement of air holes, J such as the fiber behaving a second guided mode with two-lobe pattern and different colors when a white light was launched. 11.3.3.1 Loss The minimum optical attenuation of-0.15 dB/km in conventional fibers is determined by fundamental scattering and absorption processes in the high-purity glass. However, over 99% of the light in hollow core PCFs can propagate in air and avoid these loss mechanisms, making hollow core PCFs promising candidates as a new generation ultra-low loss telecommunication fibers.[32] Nevertheless the lowest loss reported in hollow core PCFs is 1.2 dB/km. Direct leakage of light from the core is easily suppressed by incorporating enough holes in the cladding. Fiber non-uniformity then becomes the key loss mechanism. The minimum loss in hollow-core photonic crystal fibers is limited by scattering due to surface roughness from frozen-in surface capillary waves. It can be mitigated through fiber design, and attenuation of the order of 0.1 dB/km is plausible. Design improvements to eliminate surface modes, material processes to increase surface tension or alternative glasses transparent at longer wavelengths may further reduce the attenuation of hollow core PCFs. Silica based hollow core photonic crystal fibers could be useful for many mid-IR applications.'571 The spectral region of 3 to 5 urn, in the mid-infrared (mid-IR), is currently of growing interest because the development of a new generation of laser sources promises to open this spectral window for applications in the near future. Many gases exhibit strong molecular absorption at these wavelengths, especially in the wavelength range of 3 to 3.5 um. For example, CH4 has a strong absorption band around 3.3 um. In bulk silica the material loss above 3 um is greater than 60 dB m"1. For practical purposes this means that standard silica fibers are unusable in this wavelength range. Due to the low overlap of the guided light with glass in hollow core photonic crystal fibers, which can be less than 1%,[32] the effect of the relatively high material loss of silica at these wavelengths is minimized, giving a loss of 2.6 dB m"1 in the wavelength range 3.1-3.2 um in an effectively single-

398

Glass Fibers for Photonic Crystals

mode fiber. This level of loss means that the current fiber is suitable for applications where short lengths of fiber are required. However, with improvements in fiber design lower losses around 0.5 dB/m were predicted and such fibers will be suitable for a much wider range of applications. 11.3.3.2 Gas in PGBfibers Gas phase materials are used in a variety of laser-based applications,'541 for example, in high-precision frequency measurement, quantum optics and nonlinear optics. Their full potential has however not been realized because of the lack of a suitable technology for creating gas cells that can guide light over long distance in a single transverse mode while still offering a high level of integration in a practical and compact set-up or device. Gas-filled hollow-core photonic crystal fibers demonstrated substantially enhanced stimulated Raman scattering[21'54] and exhibited high performance, excellent long-term pressure stability and ease of use. There are two different devices: a hydrogen-filled cell for low threshold stimulated Raman scattering (SRS); and acetylene-filled cells for absolute frequency-locking of diode lasers with very high signal to noise ratios. The generation of vibrational SRS in hydrogen using a Kagome hollow core-PCF with a pump threshold 100 times lower than any previously reported in single-pass or multi-pass cells, and rotational SRS using a band-gap hollow core-PCF with thresholds some 1 million times lower than its corresponding experiments and with photon conversion reaching almost the quantum limit.'211 Acetylene represents an excellent frequency standards source for the optical communications wavelength range, and the long interaction length offered by hollow core-PCF leaded to unprecedented enhancement in the signal to noise ratio, it offered a comb of stable and regularly spaced ro-vibrational overtone transitions around 1.55 um. With the two isotopes of carbon 12C and 13C, the spectrum spans a range between -1.510 um and 1.560 um and exhibits more than 50 strong lines, thereby providing a wide grid of frequency references. The stable performance of these compact gas-phase devices could permit, for example, gas-phase laser devices incorporated in a 'credit card' or even in a laser pointer.

11.3 Properties ofPCFs and Device Applications

399

11.3.3.3 Particles guidance Small particles can be propelled and suspended against gravity using only the force of radiation pressure.[14'58] The use of the radiation pressure principle provides a useful means for non-intrusive manipulation of microscopic objects such as biological objects, particles etc, it has been employed in different areas such as biology, chemistry, atomic physics and engineering. However, these applications were intrinsically limited by the diffraction of the laser beam to micrometer length scales, as strong lateral confinement requires tight beam focusing. Overcoming this limitation is of particular interests in many areas where transportation of micro-sized objects over longer distance is required. For stable guidance, including cornering, it requires constant beam intensity focused to a small spot over many Rayleigh lengths. Up until now, the only possibility is to use fiber capillaries as the waveguide. Unfortunately, two inherent light guiding properties of capillaries set a limit to the guidance length: fundamentally leaky and power loss. Unlike hollow fiber capillaries, hollow core-PCF guides light without leakage using a photonic band gap, enabling a much longer guidance length to be allied with stable strong transverse particle confinement by using a small hollow core. Benabid et al. demonstrated particle guidance in a hollow core-PCF.[581 The hollow core diameter of the fiber is 20 um and the loss was measured to be 10 dB/m at 514 nm. An Argon ion laser beam operating at a wavelength of 514 nm with a power of 80 mW was sufficient to levitate a 5 urn diameter polystyrene sphere and guided it through a -150 mm long hollow-core crystal photonic fiber. The speed of the guided particle was measured to be around 1 cm/s. Such a strong gradient force with a comparable laser power would only be attainable over a distance of 0.6 mm using a focused beam in free space, or over 12 mm using a standard fiber capillary. With loss below 10 dB/km, the possible guidance lengths in a hollow core-PCF would increase to a few 100 m.

400

Glass Fibers for Photonic Crystals

11.4. Non-Silica Glasses for PCFs Non-silica glasses, often referred to as compound or soft glass, such as silicate glasses, [28 ' 29 ' 60] phosphate glass, [59] chalcogenide glasses, [27] tellurite glasses,[60] and other heavy metal oxide glasses. It offers unique material properties, which cannot be provided by SiC>2 glass. Typically, they have special properties like[60] highly linear refractive index, high transparency from the near-infrared (near-IR) to the mid-IR region, high rare-earth solubility, low phonon energy, low melting temperature, etc. Thus, non-silica glass PCFs were proposed as noteworthy candidates for highly nonlinear applications, novel devices for mid-IR laser transmission and short active fiber devices. Capillary-stacking techniques and the extrusion techniques have already been utilized for fabricating non-silica glass PCFs. The highest nonlinearity in optical fibers, 640 W"1 km"1,[28] which is more than 600 times that of standard single mode silica fiber, was reported in the extruded index-guiding PCFs based on Schott SF57 glass (Si02-PbO glass). Supercontinuum generation, i.e., the broadening of the ultrashort pulse signal after passing through a nonlinear media, was also demonstrated in an extruded nonsilica glass index-guiding PCFs based on Schott SF6 glass.[61] Because non-silica glasses typically possess higher refractive indexes (n=1.5~2.8) than that of pure silica glass (n=1.44), photonic band-gaps can be easily observed in band-gap PCFs based on high-index nonsilica glass.'62'631 It has been demonstrated that a robust photonic band-gap exists in high-index glass («>2.0) when the air-filling fraction is only about 60%. As mentioned in section 11.1.2, however, a high air filling factor (D/A>90%) is required to achieve a broad bandwidth and low-loss in silica hollow core PCFs. Using chalcogenide glasses, which have high index (2.0-2.8) and high transparency in mid-IR regions (2-1 Oum), very broad band-gaps in the mid-IR region was predicted in hollow core-PCFs for the potential applications for C0 2 laser transmission.'621 Large mode area index-guiding PCFs with mode areas as large as 430 urn2 have been developed using phosphate glass, which is of particular interest for Er-Yb-codoped fiber laser and amplifiers at 1550 nm.[59] For an only 11 cm long cladding-pumped fiber laser, more than 3 W of continuous wave output power was demonstrated, and near single-mode

References

401

beam quality was obtained for an active core area larger than 400 um 2 . An Nd3+ doped silicate glass photonic crystal fiber with a large fiber core of 58 um as shown in Fig. 11.16(a) has been demonstrated.164,651 Optical measurement demonstrated that the fiber sustain a single mode at least over wavelength range from 660 nm to 1550 nm as shown in Fig. 11.16 (b). By pumping with an 808 nm laser diode, the fiber exhibited an amplified spontaneous emission.

Fig. 11.16 (a) Cross-sectional profiles of the Nd3+ doped silicate glass PCF and (b) a 3D intensity distribution of the guided mode from the output end of the PCF excited by an 808nm laser diode.

PCFs are expected to have various applications, such as higher-power fiber lasers, power delivery, creating supercontinuum, wavelength tunable light source, and distribution compensation. Much more works remain to be explored. There is no doubt that the impact of PCF will spread to numerous applications in research, biomedicine, sensing, and material process.

References 1. J. Joannopoulos, R. Meade, and J. Winn, "Photonic Crystals: Molding the Flow of Light", (Princeton Press, Princeton, New Jersey, 1995). 2. E. Yablonovitch, Phys. Rev. Lett., 58 (1987) 2059. 3. S. John, Phys. Rev. Lett., 58 (1987) 2486.

402

Glass Fibers for Photonic Crystals

4. P. Russell, D. Atkin, and T. Birks, "Bound modes of two-dimensional photonic crystal waveguides," Quantum Optics in Wavelength Scale Structures, 1996. 5. S. John, "Photonic Band Gap Materials", (C. M. Soukoulis, Ed., NATO ASI Series: Series E, Applied Sciences, vol. 315, NATO Scientific Affairs Division, Kluwer, Dordrecht, 1996) pp. 563-666. 6. P.Russell, Science, 299 (2003) 358. 7. T. Birks, P. Roberts, P. Russell, D. Atkin and T. Shepherd, Elect. Lett., 31 (1995) 1941. 8. J. Knight, T. Birks, P. Russell and D. Atkin, postdeadline paper at OFC'96; Opt. Lett., 21 (1996) 1547. 9. T. Birks, J. Knight, and P. Russell, Opt. Lett., 22 (1997) 961. 10. J. Knight, J. Broeng , T. Birks T, et al., Science, 282 (1998) 1476. 11. J. Knight, Nature, 424 (2003) 847. 12. D. Ouzounov, F. Ahmad, D. Muller, et al., Science, 301 (2003) 1702. 13. A. Bjarklev, J. Broeng, and A. Bjarklev., "Photonic Crystal Fibers." (Dordrecht: Kluwer; 2003). 14. F. Benabid and P. Russell, Proc. of SPIE, 5733 (2005) 176. 15. W. Wadsworth, J. Knight, T. Birks, and P. Russell, Proc. of SPIE, 5246 (2003) 362. 16. F. Couny, H. Sabert, P. Roberts, et al., Opt Express, 13 (2005) 534. 17. N. Mortensen, J. Folkenberg, M. Nielsen, and K. Hansen, Opt. Lett., 28 (2003) 1879. 18. A. Tunnermann, T. Schreiber, F. Roser, et al., J. Phys. B: 38 (2005) S681. 19. M. Nielsen, J. Folkenberg, and N. Mortensen, Electron. Lett., 39 (2003) 1802. 20. M. Yan, and P. Shum, Opt. Lett., 30(2005) 1920. 21. F. Benabid, J. Knight, G. Antonopoulos, and P. Russell, Science 298 (2002) 399. 22. W. Chen, J. Li, S. Li, et al., Study on Opt. Commun., 129 (2005) 47 (In Chinese). 23. S. Li, L. Hou, Y. Ji, and G. Zhou, Chin. Phys. Lett., 20 (2003) 1300. 24. G. Zhou, Z. Hou, S. Li, and L. Hou, Chin. Phys. Lett., 22 (2004) 1162. 25. Q. Zhou, X. Lu, J. Qiu, D. Chen, X. Jiang, and C. Zhu, Chin. Opt. Lett., 3 (2005) 686. 26. R. Bise, W. Monberg, F. DiMarcello, et al., IEEE, 2 (2004) 643. 27. T. Monro, Y. West, D. Hewak, et al., Electr. Lett., 36 (2000) 1998. 28. P. Petropoulos, H. Heidepriem, V. Finazzi, et al, Opt. Express, 11 (2003) 3568. 29. X. Feng, T. Monro, P. Petropoulos, et al., Opt. Express, 11(2003) 2225. 30. J. Canning, E. Buckley, K. Lyttikainen, and T. Ryan, Optics Comm., 205 (2002) 95. 31. J. Zhou, K. Tajima, K. Nakajima, et al., Optical Fiber Technology, 11 (2005) 101. 32. P. Roberts, F. Couny, H. Sabert, et al., Opt. Express, 13 (2005) 236. 33. S.Kawanishi, Proc. of SPIE, 5596 (2004), 280. 34. R.Kristiansen, K.Hansen, J. Broeng, et al., 4a Reunion Espanola de Optoelectronica, OPTOEL'05, CI-5(2005), 37. 35. K. Mori, H. Muller, J. Kirchhof, et al., Proc. of SPIE, 5595 (2004) 66. 36. J. Eichenholz, http://newport.com/Support/Magazine_Features, Optoelectron. World (2004). 37. T. Monro, and David J. Richardson, C. R. Physique, 4 (2003) 175. 38. H. Ebendorff-Heidepriem, K. Furusawa, et al., Proc. of SPIE, 5350 (2004) 35. 39. K.Saitoh, N. Florous, and M. Koshiba, Opt. Express, 13 (2005) 8365. 40. J. Rarity, J. Fulconis, J. Duligall, et al, Opt Express, 13 (2005) 534.

References 41. 42. 43. 44. 45. 46. 47. 48. 49.

50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65.

403

K. Suzuki, H. Kubota, S. Kawanishi, et al., Opt. Express, 37 (2001) 1399. J. Folkenberg, M. Nielsen, N. Mortensen, et al., Opt. Express, 12 (2004) 956. J. Folkenberg, M. Nielsen, and C. Jakobsen, Opt. Lett., 30 (2005) 1446. J. Stenger, H. Schnatz, C. Tamm, and H. Telle, Phys. Rev. Lett., 88 (2002) 073601. M. Hu, C. Wang, L. Chai, and A. Zheltikov, Opt. Express, 12(2004) 1932. M. Hu, C. Wang, Y. Li, et al., Opt. Express, 12 (2004) 6129. M. Hu, C. Wang, Y. Li, L. Chai, and A. Zheltikov, Opt. Express, 13 (2005) 5974. Y. Zheng, Y. Zhang, X. Huang, et al, Chin. Phys. Lett., 21 (2004) 750. J. Limpert, T. Schreiber, and A. Tunnermann, "Fiber Based High Power Laser Systems", (RP Photonics), http://www.highpowerlasers.rp-photonics.com/ fibersy stems. Html. J. Limpert, N. Deguil-Robin, I. Manek-Honninger, et al, Opt. Express, 13 (2005) 1055. F. Roser, J. Rothhard, B. Ortac, A. Liem, et al., Opt. Lett., 30 (2005) 2754. K. Li, Y. Wang, W. Zhao, et al, Chin. Opt. Lett., 3 (2005) 457. J. Canning, Optics and Lasers in Engineering, in press (2006). F. Benabid, F. Couny, J. Knight, T. Birks, and P. Russell, Nature, 434 (2005) 488. C. Jin, X. Meng, B. Cheng, Z. Li, and D. Zhang, Phys. Rev. B, 63 (2001) 195107. L. Yang, D. Chen, N. Da, et al., Chin. Phys. Lett., 22(2005) 2592. J. Shephard, W. MacPherson, R. Maier, et al, Opt. Express 13 (2005) 7139. F. Benabid, J. Knight, and P. Russell, Opt. Express, 10 (2002) 1195. L. Li, A. Schiilzgen, V. Temyanko, et al., Opt. Lett., 30 (2005) 1141. X. Feng, A. Mairaj, D. Hewak, and T. Monro, J. Lightwave Tech., 23 (2005) 2046. V. Kumar, A. George, W. Reeves, et al., Opt. Express, 10 (2002) 1520. B. Temelkuran, S. Hart, G. Benoit, et al., Nature, 420 (2002) 650. J. Pottage, D. Bird, T. Hedley, et al., Opt. Express, 11 (2003) 2854. L.Yang, D. Chen, L. Da, X. Jiang, C. Zhu, and J. Qiu, Opt. Mater., in press (2006). L.Yang, D. Chen, J. Xia, et al, Chin. Phys. Lett., 22 (2005) 388.

Chapter 12

Functional Microstructures in Glass Induced by a Femtosecond Laser

12.1. Introduction In 1994, Prof. Hirao of Kyoto University, Japan proposed a basic research idea of "induced structure".1'1 He paid attention to the fact that glass is metastable from the viewpoint of thermodynamics. A metastable state of glass is easily changed to other states in an intensive external electromagnetic field. If we can control the external electromagnetic field induced structure and the concentration, variety and valence state of active ions in glass; in particular, if we can space-selectively control the induced microstructures or 3-dimensionally distributed electronic structure in glass, we expect that novel optical functions of the glass will be achieved. From the viewpoint of practical applications, we expect to obtain a glass with properties superior to corresponding single crystal. Based on this idea, we applied various electromagnetic fields such as X-ray, ultra-violet light, electron beam and laser to make microscopic modifications to glass structure, and observed many interesting phenomena and discussed the promising applications of the observed phenomena.'2"81 We selected a femtosecond laser as a powerful tool to make microscopic modifications to glass structure. Compared with CW and long pulsed lasers, femtosecond laser has two apparent features: (1) elimination of the thermal effect due to extremely short energy deposition time, and (2) participation of various nonlinear processes enabled by highly localization of laser photons in both time and spatial 405

406

Functional Microstructures in Glass Induced by a Femtosecond Laser

domains. Due to the ultra-short light-matter interaction time and the high peak power, material processing with the femtosecond laser is generally characterized by the absence of heat diffusion and, consequently molten layers.[9] The nature of ultra-short light-matter interaction permits femtosecond laser to overcome the diffraction limit.[10] The first important discovery of femtosecond laser induced structure inside a transparent material was made by Dr. Misawa of Masuhara MicroChemistry Project, Japan Science and Technology Corporation. He occasionally observed the formation of femtosecond laser-induced micro-spot in a slide glass in 1994, and suggested that the observed phenomenon can be used for the fabrication of 3-dimensional optical memory with ultrahigh storage density.[U] Mazur group at Harvard University, Gaeta group at Cornell University, Misawa group at Tokushima University and Hokkaido University, Hosono group at Tokyo Institute of Technology, Itoh group at Osaka University and Tunnermann group at the Friedrich Schiller University have made great contributions to the application of femtosecond laser micro-processing of glass. We started the systematic investigations on the femtosecond laser induced microstructures in glasses and applications in micro-optics at the end of 1994. The reason for using this laser is that the strength of its electric field in the focal point of the laser beam can reach lOTW/cm2, which is sufficient for inducing various nonlinear physicochemical reactions in glasses by using a focusing lens, when the pulse width is lOOfs and the pulse energy is 1 u J. The photo-induced reactions are expected to occur only near the focused part of the laser beam due to multiphoton processes. In the past several couple of years, a lot of research efforts have been devoted to the field of 3 dimensional microscopic modifications to transparent materials by using femtosecond laser. Promising applications have been demonstrated for the formation of 3dimensional optical memory[11"14] and multicolor image,[15] and fabrication of optical waveguide,[1617] coupler [18"19] and photonic crystal.[20]

12.2 Micro-Structural Changes Induced by Femtosecond Lasers

407

12.2 Micro-Structural Changes Induced by Femtosecond Lasers 12.2.1 Various femtosecond laser induced localized microstructures in glasses Considerable research has been carried out on the writing of Bragg gratings inside optical fibers since the first observation of photo-induced refractive index change in glass fiber by Hill et al. in 1978.[21] The reaction between light and glass is usually induced by irradiating an area in a glass to achieve various types of light-induced structural changes. It is difficult to produce an interaction effect between glass and light by a one-photon process when the wavelength of excitation light differs from the resonant absorption wavelength of the glass. But, due to the ultrashort laser pulse and ultrahigh light intensity, femtosecond laser may induce different microstructures from those induced by CW, nano- and picosecond pulsed laser in glass. A Typical optical setup for irradiation of the femtosecond laser is shown in Fig. 12.1. 200 kHz

A I CCD[—

V

<0

/ o ••

/0

v

MHz 1 kHz

Ar* laser LD-risp Nd:YVO.

IMeUchal

•V

1i:n|phirela!

—-y Ri'Eiwi din

1

Aapkbr

NI:YLF1

/ 1

-0-v.4> 120fe - 250fc

™

™

^

NJDLF.

Fig. 12.1 Optical setup for irradiation of the femtosecond laser.

Regeneratively amplified Ti: sapphire lasers are used in the experiments of femtosecond laser irradiation. The lasers emit 1Hz to 200kHz mode-locked femtosecond pulses. Usually, a microscope system is used to focus the laser pulses inside various types of glasses. The laser

408

Functional Microstructwes in Glass Induced by a Femtosecond Laser

beam is focused through microscope objectives with various numerical aperture and injected onto the polished glass samples. As shown in Fig. 12.2, various localized structures can be produced inside a glass sample by using pulsed laser operating at the non-resonant wavelength with pulse widths of the order of femtoseconds: colored line due to the formation of color center and valence state change of active ions such as rare-earth and transition metal ions, refractive index changed spot due to local densification and defect formation, micro-void due to re-melting and shock wave, micro-crack due to destructive breakdown etc.'22'

Fig. 12.2 Various structures induced by the femtosecond laser pulses.

We know that usually glass has no intrinsic absorption at 800nm (the wavelength of the used Ti:sapphire laser). Linear absorption of the laser radiation does not occur when the glass is irradiated by the laser beam. This is because the energy gap between the valence and conduction bands is larger than the energy of the single photon. Thus the energy of a single photon is not sufficient to excite an electron at the ground state. At sufficiently high laser intensities, however, an electron can simultaneously absorb the energy from the multiple photons to exceed the band gap. This nonlinear process is called multiphoton absorption. It is a highly intensity-dependent process, with the rate P(I)=o\Ik, where o\ is the multiphoton absorption coefficient for k-photon absorption. Once an electron is promoted to the conduction band, it serves as a seed to a process called avalanche ionization.'23' Seed electrons can also arise from the other processes such as electron tunneling and thermal excitation from impurity states. An electron in the conduction band can absorbs sufficient energy nh v >E (band gap), where n is the number of photons absorbed sequentially, it can then use the excess energy to ionize

12.2 Micro-Structural Changes Induced by Femtosecond Lasers

409

another electron via direct collision, known as impact ionization. The resulting two electrons in the conduction band can then continue the process of the linear absorption and impact ionization to achieve an exponential growth of the free electrons. Such avalanche ionization produces a highly absorptive and dense plasma, facilitating the transfer of energy from the laser to the glass. Many phenomena occur during the irradiation of femtosecond laser due to the various nonlinear interaction between laser and glass. 12.2.2 Three-dimensional optical data storage using the localized induced microstructures As we enter the information era, it is necessary to develop a data storage technique with ultrahigh recording capacity. Three-dimensional optical data storage may offer the potential for solving this problem. There have been a number of papers on the pointlike or bitwise binary 3-dimensional optical storage in photopolymers[24'251 and photorefractive materials.1261 By using a nonlinear-optical process, the photo-induced reaction can be confined all three dimensions to a micrometer-sized focal volume. Strickler et al. demonstrated a 3-dimensional optical recording with densities of up to 1.6 X1012 bits/cm3 in a photopolymer by using twophoton absorption.[25] The stored information was read out serially with a Nomarski differential interference contrast laser microscope. Kawata et al. suggested that using two-photon absorption could increase the recording density. The stored data were read out with a phase-contrast microscope.[26] Misawa observed the femtosecond laser induced micro-spot in glass, and suggested the application for 3-dimensional data recording with ultrahigh storage density.[11] Glezer et al. reported a novel method for creating submicrometer-sized bits that have a large contrast in index of refraction in various transparent materials including fused silica, fused quartz, sapphire, and various glasses and plastics, and that can be read out with transmitted or scattered light under a standard microscope.[12] They tightly focused ultrashort laser pulses inside a transparent material to create localized structural changes, and wrote a pre-designed pattern. The pattern was read out in parallel by use of transmitted light in

410

Functional Microstructures in Glass Induced by a Femtosecond Laser

microscope. The written spots can be viewed as dark or bright points, depending on the focusing of the readout objective. The longitudinal extent of the structurally altered regions can be ~2.5um. They recorded 10 layers spaced by 15 urn, using a standard 0.65-N.A. refractive objective. With an objective with a large working distance and an adjustment for aberrations caused by focusing into the material it should be possible to record more than 100 layers spaced by 10 um. The diameters of the cavity induced by the laser irradiation was about 200 nm, implying a recording density limit of ~1013 bits/cm3 with a 0.65N.A. objective. The mechanism for the creation of these structures is considered to be a microexplosion that occurs inside the material. The submicrometer bit size is not due to a simple threshold effect, which would occur with an excitation that is only a few percent above the threshold. The small size is caused by the nonlinearity of the absorption, which creates an excited region significantly smaller than the linear intensity distribution. Furthermore, self-focusing and the dynamics of the microexplosion may confine the extent of the structurally altered region. The femtosecond laser pulse driven micro-explosions provide a unique method for internal microstructuring of transparent materials. This method can be used for three-dimensional data storage with ultrahigh storage density. They stressed that the ability to write three-dimensional objects with sub-micrometer precision by using focused femtosecond laser may be useful for creating periodic structures such as diffractive optical elements, photonic band gap materials, and patterned gratings in fibers. The technique is not limited to optically transparent materials, and can also be applied to semiconductors by using a laser frequency below the band gap. 12.2.3 Direct writing of optical waveguide The research of photoinduced effects, such as the Bragg grating and three dimensional memory, have continued into both the mechanism of the phenomena and the potential applications. Most of the previous researches have concentrated on refractive index change written into germanium-doped silica or silica glass. As described in previous sections, visible damages can be formed only in the focused region of the

12.2 Micro-Structural Changes Induced by Femtosecond Lasers

411

glasses, because the nonlinear optical processes such as multiphoton absorption occur in a region with high optical intensity above damage threshold. Davis et al. discovered that refractive index changes of the order from 10"2 to 10"3 can be induced within various types of glass by irradiating the glass with focused femtosecond laser pulses.[16] By using a femtosecond laser with a high repetition rate, permanent optical waveguides can be successfully written in various glasses, where refractive index changes are continuously induced along a path traversed by the focal point of the laser beam.[17] Waveguides could be written inside various glasses such as fused and synthetic silica, Ge-doped silica, borosilicate, borate, phosphate, fluorophosphate, fluoride, and chalcogenide glasses.'171 In our study, the waveguides were written by focusing the laser pulses through a microscope objective and translating the glass sample parallel or perpendicular to axis of the laser beam. The refractive index changes were induced along the path traversed by the focal point, and the waveguides were written inside various glasses. On the formation of waveguides, initial laser-induced damage thresholds differ with each glass. It is noted that multiphoton absorption coefficient, band-gap, bond dissociation energy, and thermal properties such as thermal conductivity, thermal expansion coefficient, and melting temperature of each glass were essentially different. Therefore, we can explain the difference of damage thresholds by assuming that laserinduced damage results from multiphoton ionization, Joule heating, and/or conventional heating by multiphoton transition and/or plasma formation. Atomic force microscope observation shows that there was a concave on the surface of a core end on a silica glass. We once attributed this phenomenon to the densification of a glass, which occurred in the laser irradiation region. It seems to be the surface laser ablation based on our recent experiments. However, it was confirmed there is an increase in the refractive index after the laser irradiation. The increases of refractive index can be associated with the local densification which was finally occurred inside the glasses, though initial process of the optical waveguide formation may be accompanied by various phenomena like formation of color centers or lattice defects, or melting of very small area.

412

Functional Microstructures in Glass Induced by a Femtosecond Laser

We examined the effects of the average power, the pulse width, and the number of laser passes on the refractive index and core size of optical waveguides formed by translating the sample parallel to the axis of the laser beam. For laser-induced optical waveguide written by translating the sample parallel to the axis of the laser beam, experimental results indicated that the refractive-index difference and the core diameter could be controlled by adjusting the writing conditions. Typical photo-written waveguides formed in a fluoride glass are shown in Fig. 12.3.

Fig. 123 Photo-written waveguides inside fluoride glass formed using 800-nm 200-kHz mode-locked pulses. The waveguides were written by translating the sample (a) parallel or (b) perpendicular to the axis of the laser beam at a rate of 20 p,m/s and focusing the laser pulses through a 10X or SOX microscope objective, respectively.

They were created using 800 nm, 120 fs, 200 kHz mode-locked pulses. The waveguides were written by translating the sample (a) parallel or (b) perpendicular to the axis of the laser beam at a rate of 20 |im/s and focusing the laser pulses through an objective, respectively. Optical microscope observation shows that the cross sections are almost circular for the waveguides written by translating the sample parallel to the axis of the laser beam. When the samples were translated parallel to the laser beam, the shapes of the cross sections remained unchanged despite changes in the average laser power or the magnification of the objective, with diameters of approximately 7 to 30 p,m. On the other hand, the core of the waveguides produced when the samples were translated perpendicular to the laser beam became more

12.2 Micro-Structural Changes Induced by Femtosecond Lasers

413

elliptical as the average laser power was increased because of the selffocusing. Similar phenomena were also observed with the decrease of the numerical aperture of the objective. Therefore, a high-numerical-aperture objective should be used to form waveguides with a circular core. Writing waveguides perpendicular to the incident laser beam provides the most flexibility for writing planer patterns and allows one to create multiple-pattern layers by simply changing the focus depth of the beam. Fig. 12.4 shows the fabrication of curved waveguides with curved radius of 30 to 40 mm by translating the silica glass perpendicular to the axis of the laser beam.

LI Wm#$j0

%$Fig. 12.4 Curved waveguides with curved radius of 30 to 40 mm by translating the silica glass perpendicular to the axis of the laser beam. Bright spot in (b) is the guided light at the another end face of the curved waveguides.

We then coupled a He-Ne laser beam into one end face of the waveguides (Fig. 12.4(a)). The guided light was observed at another end face of the curved waveguides (Fig. 12.4(b)). For the waveguide formed at pulse energy of 0.6 ^J? scanning rate of 500 p,m/s, repetition rate of 250kHz and pulse width of 600 fs? minimum internal loss measured by cut-back method was 0.1 dB/cm or less at near L4 ^im. Recently, some research groups have discussed the fabrication of rare-earth-doped waveguide by using the direct writing of the femtosecond laser and a femtosecond laser oscillator with nanojoule energy.[27*28] Watanabe et al. show that 2-dimensional translation of a 40-p,m»long filament leads to bending of the filament. They demonstrated the fabrication of directional couplers to split the coupled beam into 1:1 at a wavelength of 632.8nm by using their observation.[29]

414

Functional Microstructures in Glass Induced by a Femtosecond Laser

12.2.4 Fabrication of Dammann micro-grating and micro-lens Dammann grating is one of the diffractive optical elements, which can generate regular one- or two-dimensional beam patterns of equal intensity spots. Practically it is used as optical splitter in information technology. Other use is for parallel digital optics and laser fabrication, etc. The structure of Dammann gratings is designed under the condition that each intensity of the high order diffractive lights is equal to the intensity of the zero order light. Usually Dammann grating is fabricated through complicated processes, i.e. formation of a photoresist film on a glass substrate via spin coating, ultraviolet exposure and etching, etc. Usually it takes over one month to complete all these processes. Using the writing technique by the femtosecond laser, however, it can shorten the processing time for fabrication of the Dammann grating. We used a regeneratively amplified Ti:sapphirc femtosecond laser to write the Dammann grating structure inside a silica glass sample. '30'

Fig. 12.5 Microstructure of a Dammann grating.

Fig. 12.5 shows a photograph of the microstructure of a 6 X 6 Dammann grating. The Dammann grating consists of many rectangle refractive index change patterns. In order to form single rectangle pattern, several lines were written in specific pitch. Fig. 12.6 shows a diffraction pattern of the 6X6 Dammann grating. The light source was a He-Ne laser (633 nm). It is shown that the laser beam was split to 6 X 6 beams. The efficiency and the uniformity of the 6X6 Dammann grating are 7.7% and 29%, respectively. The theoretical value of the diffraction efficiency is 71%, and the uniformity of the grating produced by etching process is approximately 10%. Thus, the diffraction efficiency and the uniformity are not sufficient. We have

12.2 Micro-Structural Changes Induced by Femtosecond Lasers

415

shown that the diffraction efficiency can be improved by the way of a multilayer process. By using the same technique, we have also fabricated micro-lens inside the silica glass. The focal length of the fabricated lens was estimated to be 2mm as designed. It is easy to create threedimensional structures in a glass by the femtosecond laser. Using this technique, we expect that the novel optical components with composite functions can be developed.

Fig. 12.6 Diffraction pattern of the 6X6 Dammann grating.

12.2.5 Fabrication of fiber attenuator and grating Optical attenuators are widely used to control optical power of light signals in the optical communication systems. Co2+ -doped fiber (CoDF) attenuators with optical absorption of Co2* ions at the optical communication wavelengths have been utilized due to their simple fiber structure and high input power endurance. However, the fabrication of the CoDF attenuators with desired attenuation property needs fiber performs with different doping concentration of Co2+. We thought that light scattering due to the femtosecond laser-induced localized structures can be used for the fabrication of optical fiber attenuator of light scattering type. Photo-induced optical attenuations in the optical fibers were observed after the irradiation of femtosecond pulse laser.'3'1 Fig. 12.7 shows the relationship between optical attenuation at 1.55 urn and number of irradiation point formed inside the fiber. The optical attenuations

416

Functional Microstructures in Glass Induced by a Femtosecond Laser

proportionally increased with the increase in the number of irradiation point and were controllable up to about 20 dB. Larger optical attenuations per one-irradiation point were observed by the irradiation of higher average power of the laser. From the analysis of near-field patterns at 1.55 urn, it was confirmed that single-mode waveguide properties and mode field diameters of the fabricated optical attenuation fibers were identical to those of non-irradiated optical fibers. We suggest that the optical attenuations result from the scattering of propagating light by the photo-induced refractive index changed regions.

0

5

10 Point number

15

20

Fig. 12.7 Relationship between optical attenuation and numbers of irradiation point formed inside optical fibers by using 200kHz- 120fs pulse laser.

Fig. 12.8 shows the induced microstructures inside the silica glass fiber.

Fig. 12.8 Femtosecond laser induced structures in the silica glass fiber.

12.3 Valence State Manipulation of Active Ions

417

Fig. 12.9 shows the picture of a fabricated fiber attenuator by using our newly developed method. In addition, Kondo et al. have proposed and demonstrated a novel technique for fabrication of long period fiber grating by focused irradiation of femtosecond laser pulses.'321 They induced a periodic structure of the refractive index in the core of a singlemode fiber by use of light that is not absorbed by core and clad glasses and a polymer coated upon the clad. The thermal stability of the long period fiber grating was also discussed.

Fig. 12.9 Photograph of the fabricated fiber attenuator.

12.3 Valence State Manipulation of Active Ions 12.3.1 Transition metal ions Materials with 3-dimensionally modulated microstructures have potential applications in optical field. Up to now, there have been a lot of investigations on the 3-dimensional micro-fabrication.[33"35] Braun and Witzius have successfully fabricated 3-dimensional structures of semiconductors by template-directed electrochemical deposition.1331 Cumpston et al. have succeeded in the fabrication of micro-optical elements with two-photon photopolymerization.t34] Holographic lithography has been used to fabricate 3-dimensional photonic crystals, which have periodical dielectric stractures and can manipulate light in much the same way that a superconductor manipulates electrons.1351 We realized space-selective valence state change of active ions and

418

Functional Microstructures in Glass Induced by a Femtosecond Laser

demonstrated the promising application in three-dimensional optical memory with ultrahigh storage density. A 4 urn spot was formed in the focused area of the laser beam in the Mn and Fe ions co-doped silicate glass sample. In addition, an purple colored area with diameter of about 30 um was observed around the spot. Fig. 12.10 shows the absorption spectra of the glass sample before (a) and after (b) the femtosecond laser irradiation. No apparent absorption was observed for the unirradiated glass sample in the wavelength region from 400 to 1000 nm, while there was an apparent increase in the absorbance in the wavelength region from 300 to 1000 nm in the irradiated region. The difference absorption spectrum of the glass sample after and before the femtosecond laser irradiation shows that there is a peak ranging from 400 to 800 nm, peaking at 520 nm. This peak can be ascribed to the absorption of Mn3+ ions. In addition, a peak was observed at 320 nm, which can be assigned to the absorption of hole-trapped centers as observed in the X-ray irradiated silicate glasses. 1 0.8 B0.6

•s 10.4 < 0.2 "0

1000 Wavelength /nm

2000

Fig. 12.10 Absorption spectra of the Mn2+-Fe3+-doped silicate glass sample before (a) and after (b) the femtosecond laser irradiation.

Absorption spectra show there was no apparent decrease in the laserinduced absorption at annealing temperature below 300°C. We did not observe any variation for the absorption of the induced Mn3+ ions at room temperature even after one month. Therefore, the induced Mn3+ ions are thermally stable. Electron spin resonance spectra of the glass

12.3 Valence State Manipulation of Active Ions

419

sample before and after the femtosecond laser irradiation at room temperature show that the unirradiated glass exhibits a resolved hyperfine structures of six lines spread over a range of about 500 Gauss in width centered at 3350 Gauss (splitting coefficient g~2.0). The spectrum showed a pattern similar to those observed for various glasses containing Mn2+ ions. The low resolution of the hyperfine lines is due to the dipolar broadening. No apparent absorption due to Mn3+ was observed in unirradiated glass. Therefore, most of Mn ions are present as divalent state in glass. Two new signals at g of 2.010 and 2.000 were observed in the glass sample after the femtosecond laser irradiation. The signals can be assigned to hole-trapped centers in the glass matrix. From the above results, a part of Mn2+ was oxidized to Mn3+ after the femtosecond laser irradiation. Mn and Fe co-doped silicate glass sample has no absorption in the wavelength region near 800 nm. Therefore, photo-oxidation of Mn2+ to Mn3+ should be a nonlinear optical process. We suggest that multiphoton absorption be one of the mechanism of the observed phenomenon. Free electrons are generated by the multiphoton absorption of the incident photon and consequent avalanche ionization. Mn2+ captures a hole to form Mn3+, while Fe3+ as well as active sites in glass matrix may act as electron trapping centers, resulting in the formation of Mn3+. Since focused area becomes purple after the laser irradiation, it is possible to write a 3-dimensional colored image inside the transparent and colorless glass as shown in the inset of Fig. 12.10.[15] 12.3.2 Rare-earth ions After irradiation by the focused infrared femtosecond laser on each spot for 1/125 s (i.e. 8 pulses), a 5 urn spot was formed in the focused area of the laser beam in the Eu3+-doped fluoroaluminate glass sample via the observation of optical microscope. We measured the absorption spectra of the glass sample before and after the femtosecond laser irradiation. The difference absorption spectrum of the glass sample before and after the femtosecond laser irradiation showed a peak ranging from 200 to 380 nm, peaking at 250 nm. This peak can be ascribed to the 4f7—4f65d1 of Eu2+. Fig. 12.11(a) shows the photograph when the glass sample was excited by a UV light

420

Functional Microstructures in Glass Induced by a Femtosecond Laser

at 365 nm. Unirradiated part shows red emission, while femtosecond laser-irradiated part shows blue emission. Fig. 12.11(b) shows the emission and excitation spectra of the glass sample before and after the femtosecond laser irradiation. All emission and excitation peaks observed in the unirradiated glass sample can be ascribed to the 4f6-*-4r transitions of Eu3+, no emission due to Eu2+ was detected. Eu ions are present in the trivalent state in the unirradiated glass sample. The emission peaks at 360 nm and 400 nm in the irradiated glass sample can be assigned to the 6P1/2 — 8 S 7/2 and 4f55dl — 4f7 transitions of Eu2+ , respectively. Therefore, Eu3+ near the laser-focused area in the glass sample was reduced to Eu + after the femtosecond laser irradiation.

Fig. 12.11 (a) Photograph of images written inside the Eu -doped fluoroaluminate glass sample using the femtosecond laser during excitation of UV light at 365 nm. (b) Emission and excitation spectra of the glass sample before and after the femtosecond laser irradiation, a: emission spectrum of unirradiated glass sample, b: excitation spectrum of unirradiated glass sample. The emission at 593 nm is monitored, c: emission spectrum of irradiated glass sample, d: excitation spectrum of irradiated glass sample. The emission at 400 nm is monitored.

The electron spin resonance spectra of the glass sample before and after the femtosecond laser irradiation show that there is no apparent signals in the unirradiated glass sample, while apparent signal at 3480 Gauss (splitting coefficient g~2.0) and broad signals were observed in the glass sample after the femtosecond laser irradiation. The broad signals were similar to those observed in various Eu2+-doped glass samples. The signal at 3480 Gauss and broad signals can be ascribed to

12.3 Valence Stale Manipulation of Active Ions

421

hole-trapped centers such as hole trapped by nonbridging fluorine observed in radiation-irradiated fluoride glass samples and Eu2+, respectively. The intensity of photoinduced signals decreased with increasing temperature, and almost disappeared after the annealing at 400°C for 30 min. The glass transition temperature (Tg) of Eu3+-doped glass sample is 430°C, therefore, electron-trapped by Eu3+ was excited by thermal energy and moved to recombine with the trapped hole at temperature below Tg. It is interesting that we observed the luminescence due to the 4f-5d transition of photo-reduced Eu2+ in the present AlF3-based glass sample, but no apparent luminescence was observed in ZrFV and ZnCl2-based glass systems when excited by the UV light, though ESR and absorption spectra also confirmed the photo-reduced formation of Eu in these systems.38 We suggest that may be due to the simultaneous formation of luminescence killer centers e.g. Zr3+ and Zn+ near photo-reduced Eu2+. Further study is necessary to clarify this phenomenon. We have also observed space-selective room-temperature permanent photoreduction of Sm3+ to Sm2+ in glasses and crystals by the femtosecond laser.137'381

Fig. 12.12 Photoluminescence image written inside a Sm -doped glass before erasure (a), after Ar+ laser irradiation to bit I (b) and Ar+ laser irradiation to bit II (c).

422

Functional Microstructures in Glass Induced by a Femtosecond Laser

The results demonstrated the possibility of selectively inducing a change of valence state of Eu3+ (Sm3+) ions on the sub-micrometer scale inside a glass sample by use of a focused nonresonant femtosecond pulsed laser. Therefore, the present technique will be useful in the fabrication of three-dimensional optical memory devices with high storage density. Optical memory using a valence-state change of rareearth-ions in a spot may allow one to read out data in the form of luminescence, thus providing the advantage of a high signal-to-noise ratio. Recently, we observed that photoreduction (Sm3+ - - Sm2+) bit written by using the femtosecond laser can be erased by irradiation with an Ar+ laser at 514.5nm (Fig. 12.12).[14]This result indicated the possibility of achieving a three dimensional optical memory with rewrite capability. 12.3.3 Three-dimensional hole-drilling in glasses Stookey et al. developed photosensitive glasses early in the 1950s.[391 These glasses contain noble metal photosensitive ions such as Ag+ and Au+ together with Ce3+, which acts as sensitizer. After the irradiation by ultraviolet (UV) light, Ce3+ releases an electron to form Ce4+, while Ag+ or Au+ captures the electron to form an Ag or Au atom. After subsequent heat treatment, crystallite, e.g., LiF and Li2Si05 crystals, precipitates in the UV-irradiated area due to the crystal nucleus effect of the metal cluster or colloids. The crystals can be etched away in diluted hydrofluoric acid. Therefore, it is possible to fabricate a two-dimensional designed hole structure when a mask is used. However, it is impossible to fabricate a three-dimensional modulated structure inside glasses since the UV light resonates with the absorption band of Ce3+. Recently, drilling with femtosecond laser was found to be much more controllable and reproducible than drilling with CW or longer pulsed lasers. Kondo et al. produced a Y-branched hole by moving the photomachinable glass to be perpendicular to the incident femtosecond laser beam, followed by heat treatment and subsequent etching in a diluted aqueous HF solution.[40] Marcinkevicius et al demonstrated femtosecond laserassisted three-dimensional micro-fabrication of an H-shaped channel in silica glass. They first wrote a preprogrammed pattern to the silica

12.3 Valence State Manipulation of Active Ions

423

volume by using laser pulses to produce uniformly distributed damage spots and then etching the glass in a 5% aqueous solution of HF acid.[41] Li et al. demonstrated the three-dimensional hole drilling of silica glass form from the rear surface with femtosecond laser pulses.[42] When diluted water was introduced into a hole drilled from the rear surface of the silica glass, the effects of blocking and redeposition of ablated materials were greately reduced and the aspect ratio of the depth of the hole was increased. They also demonstrated three-dimensional channels by drilling a square-wave-shaped hole inside silica glass. 12.3.4 Fabrication of photonic crystal Sun et al. reported multiphoton microfabrication of photonic crystal structure in Ge-doped silica glass.'201 Femtosecond laser pulses of 400nm wavelength at repetition rates from 1 to 1000Hz were focused into a 10GeO2.90SiO2 glass by an objective lens (lOx, NA-1.35). With stage translation synchronized to the laser pulse output, the designed 3D dot patterns were replicated to glass structures. Nearly spherical voxels with outer diameters of ~1.0um are recorded by a single shot per voxel. The voxel, as revealed by atomic-force microscopy, is hollow, with a core diameter of approximately 200-300nm. Commonly, no wavelengthdependent transmission dip is observable in as-fabricated samples. It was observed that the local explosion at focal points leads to intense structural modifications and the formation of materials defects, thusinduced metafeatures can result in strong light scattering, which may obscure any potential PBG effects. It has been observed that point defects about the voxels were reduced by annealing and diminished at 700 °C. The disappearance of defects should be accompanied with a variation of internal morphology of voids, changed to a smoother state. They annealed the sample containing PhC structure at 900 °C for 30min, and observed an apparent attenuation of frequency gaps as shown in Fig. 12.13. The same thermal treatment was performed on samples consisting of randomly distributed dots; no similar band appeared, which shows that the band is not caused by absorption. They consider that in future, two problems need to be solved for the realization of photonic crystal with

424

Functional Microstructures in Glass Induced by a Femtosecond Laser

full band gap: (i) low filling ratio, that is, a volume percentage of holes that is smaller than that necessary for opening a full bandgap. This problem may be solved by wet etching, (ii) Low dielectric ratio. A utilization of dielectrics with high refractive indices, for example, use of amorphous Ti0 2 (n=2.4-2.6; transparent at a =lcm"' in 0.43-5.3 urn) and GaP (n = -3.1; transparent at a =10cm_1 in 0.52-7.2 am), will solve this problem.

3000

3500

4000

Wave number (cm" Fig. 12.13 Measured and simulated transmission spectra. Squares, upward-pointing triangles, and circles denote spectra from annealed PhC's, as-fabricated PhC's, and annealed randomly distributed dots, respectively. The solid curve is from a simulation.

12.4 Precipitation of Functional Crystals 12.4.1 Silver nanoparticles Nanoparticles have a wide range of electrical and optical properties due to the quantum size effect, surface effect and conjoint effect of the nanostructures. Materials doped with noble metal nanoparticles exhibit large third-order nonlinear susceptibility and ultrafast nonlinear response. They are expected to be promising materials for ultrafast all-optical switches in the THz region. For the applications in integrated optoelectronics, a well-defined assembly and spatial distribution of nanoparticles in materials are essential. Many studies have been carried

12.4 Precipitation of Functional Crystals

425

out on fabrications of nanoparticle-doped materials, but there are no effective methods to control the spatial distribution of nanoparticles in materials. Here, we introduce the space-selective precipitation and control of silver nanoparticles in glasses by using femtosecond laser irradiation and subsequent annealing.[43] After irradiation by the focused infrared femtosecond laser, a 10 urn spot was formed in the focused area of the laser beam in the Ag+-doped glass sample. In addition, a gray-colored area with a diameter of about 40 u m was observed around the spot. The glass sample was further annealed at 550 °C for 10 min. The laser-irradiated part became yellow after the heat treatment. No apparent absorption was observed for the unirradiated glass sample in the wavelength region from 600 to 800 nm, while there was an apparent increase in the absorbance in the wavelength region from 220 to 600 nm in the irradiated region. The difference in absorption spectra of the glass sample before and after the femtosecond laser irradiation shows that there are absorption peaks at about 240 nm and 350 nm, which can be assigned to the atomic silver and hole trap centers at nonbridging oxygen near Ag+ ions, respectively. Therefore, an electron was driven out from the 2p orbital of a nonbridging atom near the Ag+ ions after femtosecond laser irradiation, while Ag+ captured the electron to form an Ag atom. A new peak appeared at 450 nm in the absorption of the glass sample after further annealing at 550 °C. The peak can be assigned to the absorption due to the surface plasmon of the silver nanoparticle. Preliminary observation with a JEM-2010FEF transmission electron microscope also showed that spherical particles with sizes ranging from 1 to 8 nm precipitated in the sample. No apparent signal was detected in the electron spin resonance spectrum of the unirradiated glass sample, while the spectrum of the glass sample after femtosecond laser irradiation showed a broad signal at g~2.10 and two signals at g~2.00. The broad signal at 2.10 may be due to the Ag atom, while two signals at g~2.00 can be assigned to hole trap centers (HC), e.g., HCi and HC2. The HCi and HC2 are holes trapped at the nonbridging oxygen in the Si0 4 polyhedron with two and three nonbridging oxygen, respectively. Therefore, the photon-reduced Ag atoms aggregated to form nanoparticles during the heat treatment. An unirradiated glass sample precipitates nanoparticles only at temperatures

426

Functional Microstructures in Glass Induced by a Femtosecond Laser

above 600 °C. We suggest that the neutralized Ag acts as crystal nucleus. Femtosecond laser irradiation can be used to separate and control the nucleation and growth processes. 12.4.2 Gold nanoparticles Here, we report a method that can control the precipitation of Au nanoparticles in three-dimension inside transparent materials using focused femtosecond laser irradiation.'44'451 In brief, the precipitation involves two processes: photoreduction of Au ions to atoms induced by multiphoton process and precipitation of Au particles driven by heat treatment. The size of nanoparticles and their spatial distribution can be controlled by the laser irradiation conditions. Interestingly, the precipitated nanoparticles by this technique can be also space-selectively "dissolved" by the femtosecond laser irradiation, and re-precipitated by annealing processes. This implies that the focused femtosecond laser irradiation can be used not only in the practical applications, such as the three-dimensional optical memory and the fabrication of integrated alloptical switches, but also in the study of controlling processes of nucleation and crystal growth. Gray-colored spots of about 40 (xm in diameter were then observed in the focused area through an optical microscope after the femtosecond laser irradiation. No micro-crack was observed in the sample. Annealing at 550 °C for 30 min, these gray-colored spots became red. Using this technique, we first wrote a gray-colored butterfly using the focused femtosecond laser beam, and then annealed the sample at 550 °C for 30 min. The gray-colored butterfly became red. After the sample cooled down to room temperature, we wrote a gray-colored image in the different area of the sample. The images are shown in Fig. 12.14. Optical absorption spectra of the 0.01mol%Au2O3-doped glass sample before and after femtosecond laser irradiation show that there is an apparent increase in absorbance in the wavelength region from 300 to 800 nm in the irradiated area. The peaks at 245, 306, 430, and 620 nm can be assigned to E' centers, which include an electron trapped in an sp3 orbital of silicon at the site of oxygen vacancy, hole-trapped by oxygen

12.4 Precipitation of Functional Crystals

All

vacancy neighboring to alkali ions, nonbridging oxygen hole centers HC1 and HC2, respectively.

Fig. 12.14 Photograph of images drawn inside the Au 2 O r doped glass by using the femtosecond laser irradiation: a) gray image (without annealing) b) red butterfly (with annealing).

If the annealing temperature is below 300 °C, the absorption (300-800 nm) intensities induced by irradiation decrease as the annealing temperature increases, and completely disappear when the temperatures reaches 300 °C. Visually, the femtosecond laser irradiation induced gray color disappears at 300 °C and the glass becomes colorless and transparent. Annealing at 450 °C results in the appearance of a new peak at 530 nm, and visually the laser-irradiated areas turns into red color. The wavelength of this absorption peak slightly increases with increasing annealing temperature, while its intensity significantly increases. TEM image showed that nanoparticles precipitated in the femtosecond laser-irradiated 0.01mol%Au2O.rdoped glass after annealing at 550°C for 30 min. Composition analysis using energy dispersive spectroscopy (EDS) in TEM confirms that these spherical nanoparticles are metallic Au. The size of the Au nanoparticles is ranging from 6 to 8 nm. Therefore, we assign the above absorption peak at 530 nm to the surface plasmon resonance absorption of Au nanoparticles. The inset of Fig. 12.15 shows the photograph of a 0.1mol%Au2O3-doped glass sample, which is irradiated using femtosecond laser beams of 6.5xl013, 2.3x1014 to 5.0xl016 W/cm2 in the different areas and then annealed at 550°C for 1 hour. With increasing light intensity, the color of the femtosecond laser-irradiated areas became violet, red and yellow,

428

Functional Microstructures in Glass Induced by a Femtosecond Laser

respectively. Fig. 12.15 shows the extinction spectra from these different colored areas. The extinction peak shifts from 568 to 534 to 422nm with the increase of the light intensity. The peak with the wavelength longer than 500nm observed at spectra a and b, can also be assigned to the surface plasmon resonance absorption of Au nanoparticles. The apparent blue shift of the peak from 568 to 534 nm is due to the decrease in the average size of the Au nanoparticles. An extinction peak is observed at 420 nm in the spectrum c of the Fig. 12.15. I 0.8

V

0.6

0.2 J i l i l i I i I 0 300 400 500 600 700 800

X /nm

-

Fig. 12.15 Extinction spectra of Au2Oj-doped glass irradiated by using different light intensities: a) 6.5x10l3W/cm2; b) 2.3xl014; c) 5.0xl016. All samples were annealed at 550°C for 1 hour. Inset of figure: Photograph of images drawn inside the Au203-doped glass sample.

There are few reports on the observation of such peaks in glasses doped with Au nanoparticles. However, the peak position and shape are very similar to those of an undecagold compound with small Au clusters, for example, [Aun]. The peak can be attributed to interband transitions from 5d to 6sp, that is, originating in the submerged and quasicontinum 5d band and terminating in the lowest unoccupied conduction band of the Au clusters. The average size of Au nanoparticles in area c is much smaller than those in areas a and b. Therefore, the average size of the Au nanoparticles decreases with an increase of the light intensity. This is

12.4 Precipitation of Functional Crystals

429

probably because the high irradiation intensity produces a high concentration of reduced Au atoms per unit volume, and thus a high concentration of nucleation centers. As a result, under the same annealing process, the higher the light intensity, the smaller but denser the precipitated particles are. The reduction of Au ion to atom by femtosecond laser irradiation is the key process of this method. Au ions capture the "free" electrons created by multi-photon process, and are reduced to atoms, which aggregate to form nanoparticles during annealing. During the femtosecond laser irradiation, electrons are driven out of the valence states via the multiphoton absorption of the incident photon. A part of Au ions capture free electrons to form Au atoms. At temperatures below 300 °C, only some trapped electrons and holes are excited by thermal energy, and recombine with each other. When annealing at temperatures above 400 °C, Au atoms get sufficient energy to overcome the interaction between Au atoms and the glass network structure and start to move apparently. Au nanoparticles form due to the aggregation of Au atoms. a

!

b

I

c

i

Fig. 12.16 Photographs of images drawn inside the Au203-doped glass: a) by using femtosecond laser irradiation and annealing at 550 "C for 30 min; b) further irradiation at the center part of the image in (a) by focused femtosecond laser by using a 20X objective lens; c) then the glass was annealed at 300 °C for 30 min.

Fig. 12.16 shows the changes of the Au nanoparticle-precipitated O.OlmoP/oAuiOj-doped glass sample after further femtosecond laser irradiation. Glass sample was first irradiated by the focused femtosecond laser with a light intensity of 5.8x1014 W/cm2 and a scanning rate of 1000 Lim/s, and then was annealed at 550 °C for 30 min. The laser-irradiated area became red as shown in Fig. 12.16(a), which has been discussed in

430

Functional Microstructures in Glass Induced by a Femtosecond Laser

the previous paragraphs. Then the laser beam was focused on the center of the nanoparticle-precipitated region and wrote lines with length slightly longer than that of the nanoparticle-precipitated region (Fig. 12.16(b)). The light intensity and scanning rate were 3.9x1014W/cm2 and 1000 um/s, respectively. It is seen that there is a slight change between Fig. 12.16(a) and (b) due to the formation of color-centers. After the annealing at 300 °C for 30 min, the second femtosecond laser-irradiated part became transparent, which is shown in Fig. 12.16(c). Interestingly, the transparent part in the center became red after further annealing at 550°C for 30 min. The absorption due to the surface plasmon resonance absorption decreases while the absorption due to the nonbridging oxygen hole centers HC 1(430 nm) and HC2 (620 nm) increases after further laser irradiation. Therefore, we suggest that some precipitated nanoparticles were broken into small size or atoms due to the strong interaction between the Au nanoparticles and ultrashort laser pulses e.g. dramatic heating of nanoparticles due to the linear and nonlinear absorption of laser energy during the further femtosecond laser irradiation. A femtosecond optical Kerr shutter (OKS) experiment was carried out in the Au nanoparticle-precipitated sample using the output pulse from an optical parametric amplifier (OPA) with wide tunability from 450 to 700 nm. The typical energy density of the pump pulse was 33 uJ/cm2 at the power density of 80 MW/cm2. The output beam from the OPA was split into probe and pump beams. A polarizer and an analyzer were set in a cross-Nicol configuration in the path of the probe beam. The glass sample was placed between the polarizer and the analyzer. The pump and probe beams overlapped spatially and temporally on the sample and induced the third-order nonlinearity. Because of the crossNicol configuration, only the Kerr signal generated through the thirdorder nonlinearity could pass through the analyzer. The Kerr signal was detected by a photomultiplier tube. The polarization plane of the pump beam was rotated by A/4 from that of the probe beam to optimize the intensity of the Kerr signal. All of the experiments were carried out at room temperature. A 2-mm thick Au nanoparticle-precipitated 0.1 mol%Au203-doped glass sample annealed at 550 °C for 30 min was used to measure the all-

12.4 Precipitation of Functional Crystals

431

optical response. The full-width at half-maximum of the incident pulse was estimated as 500 fs at the position of the sample. The photon energy of the pulse was set to the surface-plasmon resonance peak (2.3 eV) of Au nanoparticles. The Kerr signal rose and decayed suddenly at around the time was equal to 0. The FWHM of the signal is 240 fs and no slow decay component was observed. This is a very fast nonlinear response time for an Au nanoparticle system. The ultra-fast response of the Kensignal may be due to the self-diffraction of the pump pulse by the generated transient index grating. From the diffraction efficiency of Kerr signal in comparison with the efficiency of CS2, the x<3) of the Au nanoparticle-precipitated part is estimated to be 0.93 x 10~H, which is 300 times the magnitude of that for Si0 2 glass (2.8 x 10"14 esu). Therefore, the present technique will be useful in the fabrication of threedimensional multicolored industrial art objects, optical memory with ultrahigh storage density and ultrahigh recording speed, and integrative waveguide-type all optical switches with ultrafast nonlinear response. 12.4.3 Functional crystals When we irradiated the interior of the material with ultrashort pulses with a high peak energy, energy was accumulated at a minute area for a very short time. Various nonlinear optical effects such as multiphoton ionization and plasma vibration were induced in the focused area of the laser beam, resulting in a dramatic rise in the temperature and internal pressure in the region of focus. In taking advantage of this phenomenon, we could form spherical melting regions at arbitrary sites within a bulk glass by using a femtosecond laser with a high repetition rate, and for the first time we have succeeded in growing crystal with single-crystal-like structure from a glass. Here, we introduce the growth of a secondharmonic-generation p-BaB204(BBO) crystal using a nonresonant ultrashort pulse laser. Though there are still a lot of subjects to be clarified including the identification of orientation of the crystal, conversion efficiency of the second-harmonic generation, etc., the grown BBO crystal is expected to be applied to waveguide-type frequency conversion devices, especially in the UV region.[46]

432

Functional Microstructures in Glass Induced by a Femtosecond Laser

Microscopic photographs taken near the focal point right after focused irradiation and several ten minutes after irradiation of the laser (wavelength: 800 nm, repetition rate: 200 KHz, average output: 600 mW) show that crystals precipitated in the glass sample. The measurement of X-ray diffraction patterns confirmed that only BBO crystals were grown in the laser focused area. In a similar way, we have confirmed that it is possible to generate other functional crystals such as LiNb0 3 crystal from corresponding glasses using a nonresonant femtosecond laser. When the laser pulse width is shorter than 10 ps, the high peak power creates a short burst of ionization in the focused area, which subsequently become localized heat. Therefore, during the focused laser irradiation with a high repetition rate, there is a temperature field in the laser focused area. The glass transition temperature and crystallization temperature were 510°C and 565 °C, respectively, determined by the differential thermal analysis. Those parts in the focused area where the temperature exceeds crystallization temperature forms crystal nuclei at first, due to the atomic diffusion and microstructure rearrangement driven by the heat energy. Then, BBO crystals grow on the nuclei with the passage of time. Further study is needed to clarify the mechanism of the precipitation of BBO crystals. The emission spectra obtained from moving the laser focal point accompanies with the growth of frequency conversion crystals. Although only broadening of 800 nm femtosecond laser pulses due to self-phase modulation was observed immediately after laser irradiation, with the passage of time, it was confirmed that the intensity of second harmonic generation increases with the growth of frequency conversion crystals. In addition, after frequency conversion crystals have been created by femtosecond laser irradiation, second harmonic generation of blue beam was observed by the incidence of an infrared nanosecond laser pulse. Moving the position of the focal point of the laser beam relative to the glass sample enables the heated region to be moved freely. We attempted to generate frequency conversion crystals continuously inside the glass sample by moving the heated zone. The moving speed of the heated zone for region A and B was 100 (im/s and 10 um/s, respectively. The heated zone was moved continuously from region A to region B. The glass layer

12.4 Precipitation of Functional Crystals

433

situated at the periphery of the grown crystal constitutes an area whose structure has changed due to the crystallization of specific components in the glass. The crystal of region A is clearly a poly crystal due to the existence of grain boundaries revealed by a change in interference color. On the other hand, the crystal of region B exhibits hardly any grain boundaries indicating a crystal having a fixed orientation. The above phenomenon indicates that by decreasing the moving speed of the heated zone, a stable structure having an accommodation layer was formed at the solid-liquid interface between a part of the polycrystal in region A and the heated zone. It means that this technique gives the growth of a single crystal or a crystal with a single-crystal-like structure. By using seed-crystals in the phase matching direction, it is possible to grow crystals adequate for use in frequency conversion devices. 12.4.4 Interfered femtosecond laser field induced microstructures The interfered femtosecond laser field can be used for the formation of periodical microstructures in glasses. Kawamura et al. were the first to report the formation of periodic nanostructures encoded on the surface of a silica glass by a double-exposure technique.[47] The superposed gratings, rotated 90° with respect to each other, are encoded by two successive fs laser pulses. The resulting structures are arrays of holes with diameters of 20-300 nm, rectangular islands with a dimension of ~600X600 nm or periodic grooves with the smallest width of-15 nm, depending on the laser pulse energy and incident angles. Field emission-scanning electron microscope (FE-SEM) image of a typical grating structure encoded on a silica glass by an interfered femtosecond pulse shows clearly visible periodic line grooves with a constant spacing. The spacing agrees with the equation given by d=X /[2 sin(0/2)], where X is the wavelength of the laser (800 nm). After the first grating was encoded, the sample was rotated by 90° followed by the recording of the second grating. When the sample position was readjusted, with the help of an optical microscope installed at the irradiation stage to superpose the second pulse at the center of the first grating, a cross-oriented double grating was recorded on the superposed area. A two-dimensional lattice pattern with constant spacing

434

Functional Microstructures in Glass Induced by a Femtosecond Laser

is clearly formed in the superposed area. The possible mechanism for the formation of the present nanostructure array is considered to be microexplosion. We also demonstrated the formation of one- dimensional microarrays constituted with Au nanoparticles in AuaCVdoped glasses induced by two interfered femtosecond laser pulses followed by successive heat treatment/483 The spacing and width of the microarrays can be controlled by changing the incident angle between the two interfered pulses and the laser pulse energy. The two incident beams were first focused onto the front surface of Au203-doped glass to optimize the incident pulse energy. In the case of sufficiently high energy, the two coherent beams can induce periodic ablation, forming a grating in the glass. Herein, we reduced the energy to a certain lower level at which the two coherent beams cannot directly induce periodic ablation, and the grating can only be formed after heat treating the sample at 55 °C for 1 h. Such grating was constituted by the laser-heating induced Au nanoparticle precipitation in the AuaOj-doped glass. In our experiments, this lower pulse energy was selected to be ~30 and 38 uJ for comparison, and the colliding angle 8 between the two incident beams was fixed at ~ 45°.

Fig. 12.17 Optical microscope photographs of Au nanoparticles precipitation in periodic arrays in silica glass, taken by a lOOx transilluminated optical microscope, (a) Energy is 30fiJ per pulse, (b) Magnified view of (a), (c) Energy is 38uJ per pulse, (d) Part of a group of microgratings.

12.5 Novel Phenomena Induced by Femtosecond Lasers

435

Fig. 12.17 shows a part of the new grating. The absorption spectrum of the grating was measured by a spectrophotometer. A weak peak occurs around 508 nm, which is induced by the surface plasmon resonance of An nanoparticles in the glass. The Au nanoparticles with ~3 nm average size in the glass were observed from a transmitted electronic microscopy image. However, we observed neither absorption peak in the range of 500-600 nm as seen, nor Au nanoparticles constituted micrograting in the glass sample irradiated only by the interfered pulses but not heat treated. This indicates that the Au nanoparticles can be precipitated in the periodic one-dimensional arrays in the glass through the irradiation of two coherent beams with the aid of heat treatment. This method should be useful not only for practical applications in the fabrication of integrated all-optical devices, but also for controlling nucleation process. 12.5 Novel Phenomena Induced by Femtosecond Lasers 12.5.1 Formation of polarization-dependent nanograting After the observation of memorized polarization-dependent light scattering, [49' 50] we suspected that there should be some permanent polarization-dependent micvostructure in glass after the femtosecond laser irradiation. We started the examination on the laser induced structure and observed the polarization-dependent nanograting.[51] In our experiments, we used commercially available synthetic silica glass. The laser radiation in Gaussian mode produced by regenerative amplified mode-locked Ti:Sapphire laser (150 fs pulse duration, 200 kHz repetition rate) operating at a wavelength of 800 nm was focused via lOOx (NA=0.95) microscope objective into the silica glass samples placed on the XYZ piezo-translation stage. After laser irradiation, we observed that a micro-spot with diameter of about 2um was formed at the focal point of the laser beam through the optical microscope (Fig. 12.18(a)). Then, the sample was polished to the depth of the beam waist location. Secondary electron images of the polished silica sample

436

Functional Microstructures in Glass Induced by a Femtosecond Laser

indicate that the morphology of an irradiated sample in the examined cross-section almost does not change, namely, a void does not exist. On the other hand, the backscattering electron images as shown in Fig. 12.18(b) reveal a periodic structure of stripe-like dark regions with low density of material and of ~ 20 nm width which are aligned perpendicular to the writing laser polarization direction.

200nm

a

b

Fig. 12.18 (a) Optical microscope photograph of the micro-spot induced in silica glass. The irradiation duration is Is for each spot, (b) Backscattering electron image (X30000) of silica glass surface polished close to the depth of focal spot.

We speculated, based on the fact that the elements constituting the sample are silicon and oxygen (average molecular weight of SiC>2 glass -60.1), that the oxygen defects were formed in the regions corresponding to dark domains of the BE image, which reduce the average molecular weight in these regions (S1O2-X ~60.1-16x). To test this suggestion, we carried out Auger spectra mapping of silicon and oxygen on the same surface with 10 nm spatial resolution. The Auger signal of the oxygen in the regions corresponding to dark domains in the BE image is lower compared to other regions as shown in Fig. 12.19, indicating low oxygen concentration in these domains. Furthermore, the intensity of the oxygen signal is stronger in the regions between the dark domains of the BE image. On the other hand, the intensity of the silicon signal is the same in the whole imaged region. These results indicate that the periodic structure observed in the BE image consists of periodically distributed oxygen-deficient regions (Si02-x). The Auger signal intensity is

12.5 Novel Phenomena Induced by Femtosecond Lasers

437

proportional to the concentration of element constituting the surface, which gives an estimate to the value x ~ 0.4.

Fig. 12.19 Auger spectra maps and corresponding line scans of oxygen (a) and silicon (b) on the silica glass surface polished close to the depth of focal spot.

We observed the decrease of the grating period with an increase of the exposure time. The grating periods were about 240 nm, 180 nm and 140 nm for the number of light pulses of 5 x 104, 20 x 104 and 80 x 104 respectively and for the pulse energy of 1 uJ, corresponding to intensity of 2x1014 W/cm2. This indicated a logarithmic dependence of the grating period A on the number of light pulses. The dependence of the observed periodic nanostructures on pulse energy for a fixed exposure time was also investigated and an increase of the period with the pulse energy was observed. Grating periods of 180 nm, 240 nm and 320 nm were measured at pulse energies of 1 uJ, 2 uJ and 2.8 uJ respectively and for the number of light pulses of 20 x 104. The following explanation of the observed phenomenon is proposed. Once a high free electron density is produced by multiphoton ionization the material has the properties of plasma and will absorb the laser energy via one-photon absorption mechanism of inverse Bremsstrahlung (Joule)

438

Functional Microstructures in Glass Induced by a Femtosecond Laser

heating. The light absorption in the electron plasma will excite bulk electron plasma density waves. These are longitudinal waves with the electric field component parallel to the direction of propagation. Such electron plasma wave could couple with the incident light wave only if it propagates in the plane of light polarization. Initial coupling is produced by inhomogeneities induced by electrons moving in the plane of light polarization. The coupling is increased by a periodic structure created via a pattern of interference between the incident light field and the electric field of the bulk electron plasma wave, resulting in the periodic modulation of the electron plasma concentration and the structural changes in glass. A positive gain coefficient for the plasma wave will lead to an exponential growth of the periodic structures oriented perpendicular to the light polarization, which become frozen within the material. The observed increase of the grating period with the pulse energy is in agreement with this theoretical prediction. The observed formation of stripe-like regions with low oxygen concentration can be also explained as follows. The plasma electrons are created in the process of breaking of Si-O-Si bonds via multi-photon absorption of light which is accompanied by the generation of a Si-Si bonds, non-bridging oxygen-hole centers (NBOHC, ^Si-O") and interstitial oxygen atoms (Oi). Such oxygen atoms are mobile and can diffuse from the regions of high concentration. Negatively charged oxygen ions can be also repelled from the regions of high electron concentration. The photoluminescence and electron spin resonance spectra confirmed the presence of non-bridging oxygen defects and E' centers in the irradiated samples. The small thickness of these regions, compared to the period of the grating, could be explained by a highly nonlinear dependence of the structural changes on the electron concentration. Major changes in composition take place after the attainment of thermal equilibrium, involving formation and decay of defect states, such as oxygen vacancies. Detailed mechanism of the structural changes responsible for the nano-grating formation is under investigation.

12.5 Novel Phenomena Induced by Femtosecond Lasers

439

12.5.2 Moving of hole Watanabe et al. observed and demonstrated optical movement of a void and merger of two voids along the optical axis by translation of the focal spot of a focusing lens with femtosecond laser pulses.[52] The laser pulses propagated along the optical axis (+z direction). They focused the femtosecond laser pulses inside silica glass to create local structural changes. They used high-NA objectives to avoid self-trapping effects and create a void. The void can be seized and moved to the negative side of the +z direction. First, they created three voids along the y axis at a depth of 300^m beneath the surface, with a spacing of 5|j,m between successive voids. Then, they moved the focal point and translated the focal point to the -z direction by 0.5^m with one exposure. By repeating this step, they moved the void further. The visible trajectory of the void after translation of the focal point implies structural changes. These changes were considered to have a connection with the refractive index change that is utilized for fabrication of waveguides. They also demonstrated that two voids can be merged to form one bigger void. The mechanism of void generation and its properties are open questions. 12.5.3 Formation of periodic nanovoid structures Recently, Kanehira et al. observed the formation of periodic nanosized voids inside a glass sample along the propagation direction of the femtosecond laser beam.[53] Two hundred and fifty pulses of the femtosecond laser were launched in the interior of a glass sample. The laser beam was focused at a depth of 750um from the entrance surface. Sideview scanning electron microscope photograph shows there is an aligned void structure below the focal point along the propagation direction of the femtosecond laser beam. The formed voids have almost a spherical shape, and the neighboring two voids are independent of each other. No microcracks or catastrophic collapses are observed around the voids or the focal point. An interesting feature is that the aligned void structure contains a region with periodically aligned voids without connected voids or cracks, which are located at a distance of ~90 (im from the bottom surface of the glass sample. A void with a diameter of

440

Functional Microstructures in Glass Induced by a Femtosecond Laser

1.6um is formed at the focal point. The distance between the void at the focal point and the next void is 7.2 ^m. Both the diameter of the voids decrease gradually with the closing of the bottom surface of the glass sample, and approach limiting values at a distance of ~90 um from the bottom surface. Namely, the periodic part exists at a range of ~90 urn from the bottom surface. The void size and intervoid separation in the periodic part are 380nm and 1.7 (im, respectively. They observed that the entire length of the periodic void structure gradually increases with increasing pulse numbers and exhibits a constant value of 130 when the pulse number is increased above 125. The phenomenon is explained as follows. Microexplosion takes place around the bottom surface and create a nanosized void at first. When the next femtosecond laser pulse propagates from the focal point, it is trapped in the heated region around the void formed beforehand, resulting in the production of new high temperature region and the formation of the next void. The size of the voids, and the period and the entire length of the viod structure could be controlled by varying the pulse energy, pulse number, and focal point of the femtosecond laser. Though the mechanism remains an open question and the nature of the phenomenon should be further examined, this discovery is promising for the fabrication of controllable periodic void structure such as three-dimensional photonic crystals. Conclusion We have observed and discussed the mechanisms of the femtosecond laser induced various localized microstructures in glasses. We demonstrated the possibility of direct writing three-dimensional optical circuit by using femtosecond laser-induced refractive index change, space-selective control of valence state of active ions (rare-earth, transition metal ions), precipitation and control of metal and functional crystals, defect manipulation, 3-dimensional micro-drilling, coherent field femtosecond laser induced structures and formation of nano-grating by the femtosecond laser single beam, etc. The 3-dimensional direct writing technique by using the femtsecond laser will be useful in the fabrication of three-dimensional multicolored

References

441

industrial art object, rewriteable optical memory with ultrahigh storage density, integrated optical circuit and micro-optical devices. Due to the page limitation, we did not introduce nonlinear coherent femtosecond laser field induced second harmonic generation.[54] For details, please refer the cited references. We are convinced that femtosecond laser will open new possibilities in micro-optics, material sciences, physics, chemistry and bioscience fields. Acknowledgements The author thanks Professor K. Hirao of Kyoto University for his continuous encouragement and guidance. Drs. T. Mitsuyu, K. Miura, J. Si, Y. Kondo, Y. Himeii, H. Inouye, Y. Shimotsuma, Mrs. T. Nakaya, M. Shirai of Hirao Active Glass Project and Photon Craft, ICORP, JST, Prof. C. Zhu, Drs. X. Jiang, S. Qu, Q. Zhao and H. Zeng of Shanghai Institute of Optics and Fine Mechanics, Chinese Academy of Sciences, and Prof. P. Kazansky of the Southampton University, UK are also appreciated for their kind cooperation. J.Q. would like to acknowledge the financial support provided by National Science Foundation of China (No. 50125208).

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.

K. Hirao, Ceramics Japan 30 (1995) 689. J. Qiu, J. Shimizugawa,, Y. Iwabuchi, K. Hirao, Appl. Phys. Lett. 71 (1997) 43. J. Qiu, Y. Shimizugawa, Y. Iwabuchi, Y. Hirao, Appl. Phys. Lett. 71 (1997) 759. J. Qiu, K. Hirao, Solid State Commun. 106 (1998) 795. N. Jiang, J. Qiu, J. Silcox, Appl. Phys. Lett. 77 (2000) 3956. N. Jiang, J. Qiu, A. Gaeta, J. Silcox, Appl. Phys. Lett. 80 (2002) 2005. N. Jiang, J. Qiu, J. Spence, Phys. Rev. B 60 (2002) 2263. N. Jiang, J. Qiu, A. Ellison, J. Spence, Phys. Rev. B 68 (2003) 64207 B. N. Chickov, C. Momma, S. Nolte.et al, Appl. Phys. A 63 (1996) 109. P. Pronko, S. Dutta, J. Squier, J. Rudd, G. Mourou, Opt. Commun. 114 (1995) 106.

442 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47.

Functional Microstructures in Glass Induced by A Femtosecond Laser

H. Misawa, Japanese Patent Application No. 023614 (1995). E. N. Glezer, M. Milosavljevic, L. Huang, et al, Opt. Lett. 21 (1996) 2023. J. Qiu, K. Miura, K. Hirao, Jpn J. Appl. Phys. 37 (1998) 263. K. Miura, J. Qiu, S. Fujiwara, et al, Appl. Phys. Lett. 80 (2002) 2263. J. Qiu, C. Zhu, T. Nakaya, J. Si, K. Hirao, Appl. Phys. Lett. 79 (2001) 3567. K. M. Davis, K. Miura, N. Sugimoto, K. Hirao, Opt. Lett. 21 (1996) 1729. K. Miura, J. Qiu, H. Inouye, T. Mitsuyu, K. Hirao, Appl. Phys. Lett. 71 (1997) 3329. D. Homoelle, S. Wielandy, A. G. Gaeta, et al, Opt. Lett. 24 (1999) 1311. K. Minoshima, A. M. Kowalevicz, I. Hartl, et al, Opt. Lett. 26 (2001) 1516. H. Sun, Y. Xu, S. Joudkazis, et al, Opt. Lett. 26 (2001) 325. K. O. Hill, Y. Fujii, D. C. Jhonson, B. S. Kawasaki, Appl. Phys. Lett. 32 (1987) 647. J. Qiu, J. Ceram. Soc. Jpn 109 (2001) S25. C. Stuart, M. D. Feit, A. M. Rubenchik, et al, Phys. Rev. Lett. 74 (1995) 2248. D. A. Parthenopoulos, P. M. Retzepis, Science 245 (1989) 843. J. H. Strickler, W. W. Webb, Opt. Lett. 16 (1991) 1780. Y. Kawata, H. Ueki, Y. Hashimoto, S. Kawata, Appl. Opt. 34 (1995) 4105. Y. Shikorski, A. Said, P. Bado, et al, Electron Lett. 36 (2000) 226. A. M. Streltsov, N. F. Borrelli, Opt. Lett. 26 (2001) 42. W. Watanabe, T. Asano, Y. Yamada, K. Itoh, J. Nishii, Opt. Lett. 28 (2003) 2491. T. Nakaya, J. Qiu, C. Zhou, K. Hirao, Chin. Phys. Lett. 21 (2004) 1061. Y. Himeii, J. Qiu, S. Nakajima, A. Sakamoto, K. Hirao, Opt. Lett. 29 (2004) 2728. Y. Kondo, K. Nouchi, T. Mitsuyu, et al, Opt. Lett. 21 (1996) 1729. P. V. Braun, P. Wiltzius, Nature 402 (1999) 603. B. H. Cumpston, S. P. Annanthavel, S. Barlow, et al, Nature 398 (2000) 51. M. Campbell, D. N. Sharp, M. T. Harrison, et al, Nature 404 (2000) 53. J. Qiu, K. Kojima, K. Miura, T. Mitsuyu, K. Hirao, Opt. Lett. 24 (1999) 786. J. Qiu, K. Miura, T. Suzuki, T. Mitsuyu, K.Hirao, Appl. Phys. Lett. 74 (1999) 10. J. Qiu, K. Miura, K. Nouchi, et al, Solid State Comrnun. 113 (2000) 341. W. H. Armistead, S. D. Stookey, Science 114 (1964) 150. Y. Kondo, J. Qiu, T. Mitsuyu, K. Hirao, Jpn J. Appl. Phys. 38 (1999) LI 146. A. Marcinkevicius, S. Juodkazis, M. Watanabe, et al, Opt. Lett. 26 (2001) 277. Y. Li, K. Itoh, W. Watanabe, et al, Opt. Lett. 26 (2001) 192. J. Qiu, M. Shirai, T. Nakaya, et al, Appl. Phys. Lett. 81 (2002) 3040. J. Qiu, X. Jiang, C. Zhu, H. Inouye, J. Si, K. Hirao, Opt. Lett. 29 (2004) 370. J. Qiu, X. Jiang, C. Zhu, et al, Angew Chem. Int. Ed. 43 (2004) 2230. K. Miura, J. Qiu, T. Mitsuyu, K. Hirao, Opt. Lett. 25 (2000) 408. K. Kawamura, N. Sarukura, M. Hirano, H. Hosono, Appl. Phys. Lett. 78 (2001) 1038. 48. S. Qu, J. Qiu, Q. Zhao, et al, Appl. Phys. Lett. 84 (2004) 2046. 49. P. G. Kazansky, H. Inouye, T. Mitsuyu,et al, Phys. Rev. Lett. 82 (1999) 2199. 50. J. Qiu, P. G. Kazansky, J. Si, et al, Appl. Phys. Lett. 77 (2000) 1940. 51. Y. Shimotsuma, P. G. Kazansky, J. Qiu, K. Hirao, Phys. Rev. Lett. 91 (2003) 247405.

References 52. W. Watanabe, T. Toma, K. Yamada, et al, Opt. Lett. 25 (2000) 1669. 53. S. Kanehira, J. Si, J. Qiu, K. Fujita, K. Hirao, Nano Lett. 5 (2005) 1591. 54. J. Si, K. Kitaoka, J. Qiu, T. Mitsuyu, K. Hirao, Opt. Lett. 24 (1999) 911.

443

Index

arrayed waveguide grating, 315 amorphous, 17, 40 amplifier fiber, 191 waveguide, 329

fiber laser, 227, 366 end pumping, 243 high power, 227 RE-doped, 247, 250 fiber grating attenuator, 415 birefringence, 367 fabrication, 261 filter, 363 principles, 355 flame hydrolysis deposition (FHD), 302 fluoride glasses, 112 Ho3+ doped, 107 Nd3+ doped, 108 Tm3+ doped, 107 Yb3+doped, 108 fluoride glass fiber, 112 fullerene, 144

beam quality, 286 Bragg grating, 353 chalcogenide, 53, 126 DFB, 290 Dammann micrograting, 413 degenerate four wave mixing(DFWM), 118 energy transfer between laser dyes, 281 rare earth doped glasses, 194 rare earth ions, 106 activated ions, 97

Ge-doped glass photosensitivity, 341 second order nonlinearity, 170 glass borate, 17, 78, 172 chalcogenide, 53, 126 fluoride, 112 non-oxide, 307 non-silica, 400 organic-inorganic, 131, 261, 310 oxide, 124 phosphate, 77, 193 RE-TM, 40 rare-earth doped, 236, 329

femtosecond laser, 62, 405 fiber amplifier, 191,365 double clad, 232 Er doped, 193 fabrication, 234 fluoride glass, 112 numerical aperture, 230 phosphate glass, 193, 211 photonic crystal, 375 rare-earth doped, 110, 236, 239 silica glass, 110 445

446 silica, 110, 167,315 silicate, 17 soft, 171 tellurate, 185 hybrid dye laser glass, 261 gel, 278 organic-inorganic, 261, 310 waveguide, 310, 318 ions ion-exchange, 304 rare-earth, 239, 419 transition metal, 417 Kerr effect, 122 laser cavity, 286 dye, 262, 263 fiber, 227, 366 glass, 18 photonic crystal, 392 solid state dye, 261 tunable, 263, 286, 326 waveguide, 329 laser spectroscopy Nd doped, 77 Yb doped, 84 light scattering, 3 luminescence avalanche luminescence, 101 cooperative luminescence, 99 nonlinear, 94 super luminescence, 110 metallo-phthalocyanine, 148 microstructure femtosecond laser induced, 405 maganetization, 48 magnetic anisotropy, 48 magneto-optical property, 41 storage, 40 nano-composite fullerence, 144

Index glass, 136 metallic, 144 semiconductor, 137 nano pariticle copper, 144 gold, 426 silver, 424 nano grating,435 nano void, 439 nonlinear susceptibility, 118, 131 nonlinearity second order, 153 third order, 117 non-silica glass, 400 optical dispersion, 3 glass, 1 nonlinearity, 5, 117, 153 switch, 318 optical data storage, 40, 409 optical limiting effect, 146 metallo-phthalocyanines, 148 ormosil, 146 sol-gel, 149 organic-inorganic laser, 261 waveguide, 310 ormosil, 146, 269, 291 phase change, 54, 58, 61,62 mask, 361 matched, 187 phosphate glasses, 77 phosphate glass fiber, 193 photonic crystal, 375 glass, 1 photonic crystal fiber band-gap, 380 Hollow core, 395 Index-guiding, 386 non-silica glass, 400 laser, 392 property, 386 photostability, 273

447 photosensitivity Ge doped glass, 341 Pb doped glass, 353 Sn-doped glass, 344 photodegradation, 273 poling, 167, 168 rare-earth ions, 239, 419 Er3+, 193 Nd3+, 77, 112 Yb3+, 77, 194 rare-earth transition metal(RE-TM), 40 second harmonic generation, 158, 187 second order optical nonlinearity characterization, 161 chalcohalide glass, 168 Ge-doped silica, 170 high refractive index glass, 171 lead borate glass, 172 measurement, 158 mechanism, 164 tellurate glass, 185 semiconductor micro-crystallites, 137 sol gel, 267, 290, 305 tellurate glasses, 185 two photon, 291 thermal dynamics, 54 optical effect, 70 thin film amorphous, 17, 40 chalcogenide, 53, 126 hybrid, 290

metallic, 53 RE-TM, 40 third order nonlinearity, 117 waveguide amplifier, 326 arrayed waveguide grating, 315 fabrication, 302 laser, 329 non-oxide, 307 organic-inorganic, 310 principles, 300 properties, 302 sensor, 332 switch, 318 variable optical attenuator, 322 up-conversion avalanche, 101 cooperative, 91 emission, 97 energy transfer, 97 excited absorption, 84 fluorescence, 94 luminescence, 106

waveguide fabrication chemical vapor deposition (CVD), 302 flame hydrolysis deposition (FHD), 302 ion-exchange, 303 sol-gel, 305 sputtering, 302 zirconium, 291 z-scan, 120

GtfSSfc* daces

,bo*"«*° 3o\a« zaW

ves^

s

oitbea*-

*e^deCat