Nanoporous glass-ceramics transparent in infrared range to be used as optical sensor-Mechanical and viscoelastic properties of the TAS (Te-As-Se) glass
Item Type text; Electronic Dissertation
Authors Delaizir, Gaelle
Publisher The University of Arizona.
Rights Copyright © is held by the author. Digital access to this material is made possible by the University Libraries, University of Arizona. Further transmission, reproduction or presentation (such as public display or performance) of protected items is prohibited except with permission of the author.
Download date 26/09/2021 09:42:09
Link to Item http://hdl.handle.net/10150/195636
NANOPOROUS GLASS-CERAMICS TRANSPARENT IN
INFRARED RANGE TO BE USED AS OPTICAL SENSOR
______
MECHANICAL AND VISCOELASTIC PROPERTIES
OF THE TAS (TE-AS-SE) GLASS
by
Gaelle Delaizir
A Dissertation Submitted to the Faculty of the DEPARTMENT OF MATERIALS SCIENCE & ENGINEERING In Partial Fulfillment of the Requirements For the Degree of
DOCTOR OF PHILOSOPHY
In the Graduate College THE UNIVERSITY OF ARIZONA
2007 2
THE UNIVERSITY OF ARIZONA GRADUATE COLLEGE
As members of the Dissertation Committee, we certify that we have read the dissertation prepared by Gaelle Delaizir entitled : Nanoporous glass-ceramics transparent in infrared range to be used as optical sensor – Mechanical and viscoelastic properties of the TAS (Te-As-Se) glass
and recommend that it be accepted as fulfilling the dissertation requirement for the Degree of Doctor of Philosophy ______Date: 11/30/07 (Pierre Lucas)
______Date: 11/30/07 (B. G. Potter)
______Date: 11/30/07 (Donald R. Uhlmann)
______Date: 11/30/07 (Xiang-Hua Zhang)
______Date: 11/30/07 (Jean-Christophe Sangleboeuf)
______Date: 11/30/07 (Bruno Bureau)
Final approval and acceptance of this dissertation is contingent upon the candidate’s submission of the final copies of the dissertation to the Graduate College.
I hereby certify that I have read this dissertation prepared under my direction and recommend that it be accepted as fulfilling the dissertation requirement.
______Date: 11/30/07 Dissertation Director: Pierre Lucas
3
STATEMENT BY AUTHOR
This dissertation has been submitted in partial fulfillment of requirements for an advanced degree at the University of Arizona and is deposited in the University Library to be made available to borrowers under rules of the Library.
Brief quotations from this dissertation are allowable without special permission, provided that accurate acknowledgment of source is made. Requests for permission for extended quotation from or reproduction of this manuscript in whole or in part may be granted by the head of the major department or the Dean of the Graduate College when in his or her judgment the proposed use of the material is in the interests of scholarship. In all other instances, however, permission must be obtained from the author.
SIGNED : Gaelle Delaizir
4
ACKNOWLEDGEMENTS
I would like to express my gratitude to all those who gave me the possibility to complete this degree. I want to thank the Department of Materials Science & Engineering and more particularly the head department, Dr J. H. Simmons for giving me the permission to commence this joint Ph.D. degree in his department. I want to gratefully acknowledge Profs. P. Lucas (USA), X. H. Zhang, J. C. Sangleboeuf and B. Bruno (France), my “American-French” and French wonderful advisors, for their dedication to this joint project. This dissertation would not have been possible without their expert guidance. I have enjoyed every moment that we have worked together including those around a drink at Gentle Ben’s. To all of you, thank you. I would like to thank Professor B. G. Potter and Professor D. R. Uhlmann who served on my Ph.D. committee as well as Professor P. A. Deymier who served in the different committees during the different stages of my degree. I am deeply indebted to Professor J. Lucas for his encouragement, advice and mentoring throughout my doctoral studies. I also want to thank his wife for her kindness. My gratitude goes out to Dr. C. Juncker (l’Alsacien) for teaching me how to use and operate on the different equipments in the lab. Ellyn, Allison, Christophe, Ping, Dave, Jessica, Sidd……You were amazing and I enjoyed all these unforgettable moments that we have spent together for partys, poker and shopping. Many thanks. Since I spent time in France, I can not forget to thank Marie-Laure and Virginie (the “girls” team), Thierry P., Thierry J., Didier, Catherine, Johann, Corinne, Seb, Mathieu, Fred C., Fred D., Laurent C., Patrick, Laurent B., Yannick, Quentin, Eric, Sylvain, Erwan, Hong-Li and Professor J. L. Adam, director of this wonderful “Glass & Ceramic” team. Finally, I extend special thanks to my family and friends for their encouragement. I am very happy to have been the “Guinea pig” of this new joint Ph.D. program. It has been an exceptional experience that I will never forget.
5
TABLE OF CONTENTS
LIST OF FIGURES ...... 12 LIST OF TABLES...... 17 ABSTRACT...... 18
INTRODUCTION...... 20
PART 1 : GLASSES AND GLASS-CERAMICS IN THE SYSTEM
GES2-SB2S3-CSCL ...... 24
CHAPTER 1 : BACKGROUND ON CHALCOGENIDE GLASSES AND GLASS-CERAMICS ...... 24 1.1 Chalcogenide glasses...... 24 1.1.1 Introduction...... 24 1.1.2 Optical properties...... 26 1.1.2.1 Band-gap...... 26 1.1.2.2 Multi-phonon absorption ...... 29 1.1.3 Thermal properties...... 31 1.1.4 Viscosity ...... 32 1.1.5 Linear index of refraction ...... 34 1.2 Glass-ceramics...... 35 1.2.1 Introduction...... 35 1.2.2 Thermodynamic and kinetic approach...... 39 1.2.3 Glass ceramics from oxide glassy matrix ...... 42 1.2.4 Glass ceramics from chalcogenide glassy matrix ...... 47
6
TABLE OF CONTENTS - Continued
CHAPTER 2 : NANOPOROUS GLASS-CERAMICS USED AS BIOSENSOR ...... 49 2.1 Introduction...... 49
2.2 Ternary system GeS2-Sb2S3-CsCl...... 50 2.3 Glass preparation...... 51 2.4 Physical properties...... 53 2.4.1 Thermal properties...... 53 2.4.2 Thermal expansion coefficient...... 54 2.4.3 Viscosity ...... 55 2.4.4 Optical properties...... 56 2.5 Glass ceramic...... 57 2.5.1 Thermal treatment...... 57 2.5.2 Optical properties...... 58 2.5.3 Influence of crystallization on the linear index of refraction...... 60 2.5.4 X-Ray diffraction...... 60 2.5.5 SEM (Scanning Electron Microscopy) analysis of crystals...... 61 2.5.6 Structural arrangement in the base glass and the associated glass ceramic...... 62 2.6 Nanoporous glass ceramic...... 66 2.6.1 Etching with water ...... 66 2.6.2 Etching with an acid treatment ...... 68 2.6.3 Control of the pore size...... 70 2.6.4 Optical properties...... 72 2.6.5 Specific surface gain...... 74 2.6.6 Surface composition analysis...... 75 2.6.7 ATR (Attenuated Total Reflections) measurements...... 75 2.6.7.1 Principle of the method...... 75 2.6.7.2 ATR glass ceramic plate ...... 76 2.6.8 Applications ...... 79
7
TABLE OF CONTENTS - Continued
2.6.8.1 Experiments with gaseous samples...... 79 2.6.8.2 Experiments with liquid samples ...... 81 2.6.8.3 Experiments with a spray...... 82 2.6.8.4 Experiments with APTS (Aminopropyltriethoxysilane) ...... 83 2.7 Porous glass ceramic in the fiber configuration...... 85 2.8 Discussion...... 88 2.9 Conclusion ...... 90
CHAPTER 3 : PHOTOINDUCED EFFECTS IN THE SYSTEM GES2-SB2S3-CSCL...... 91 3.1 Introduction...... 91 3.2 Electronic structure of chalcogenide glasses ...... 92 3.3 Previous studies on photodarkening effect...... 93 3.4 Photodarkening of chalcogenide and chalco-halide glasses...... 96 3.4.1 Set-up ...... 96 3.4.2 Results...... 97 3.5 Discussion and conclusion ...... 100
8
TABLE OF CONTENTS – Continued
PART 2 : MECHANICAL AND VISCOELASTIC PROPERTIES OF
THE TAS (TE-AS-SE) GLASS...... 101
CHAPTER 4 : MECHANICAL PROPERTIES OF THE TAS (TE2AS3SE5) GLASS ...... 101 4.1 Introduction...... 101 4.2 TAS synthesis ...... 103 4.3 Drawing process...... 106 4.3.1 Drawing process inducing residual stresses...... 107 4.3.2 Fiber annealing during the drawing process ...... 109 4.4 Effect of annealing on the glass structure...... 109 4.5 Physical properties...... 110 4.5.1 Thermal properties...... 110 4.5.2 Optical properties...... 111 4.5.2.1 Bulk ...... 111 4.5.2.2 Refractive index ...... 112 4.5.2.3 Fiber...... 112 4.5.2.3.1 Set-up and principle of attenuation measurement ...... 112 4.5.2.3.2 Attenuation spectrum...... 113 4.5.3 Density ...... 114 4.5.4 Viscosity ...... 114 4.6 Mechanical properties of TAS bulks...... 115 4.6.1 Elastic moduli ...... 115 4.6.1.1 Ultrasonic echography method...... 118 4.6.1.2 Results...... 119 4.6.2 Hardness...... 119 4.6.3 Toughness ...... 120 4.7 Mechanical properties of TAS fibers ...... 122
9
TABLE OF CONTENTS - Continued
4.7.1 Statistical analysis of fracture ...... 122 4.7.2 Experimental determination of the crack resistance ...... 123 4.7.2.1 In tension...... 123 4.7.2.2 In two-point bending...... 124 4.7.3 Fractographic analysis ...... 125 4.8 Effect of environment on mechanical properties...... 126 4.8.1 Previous studies ...... 126 4.8.2 Air ...... 127 4.8.2.1 Results...... 127 4.8.2.2 Surface analysis ...... 129 4.8.3 Vacuum...... 130 4.8.3.1 Results...... 131 4.8.3.2 Surface analysis ...... 132 4.8.4 Under loading...... 134 4.8.4.1 Experiments...... 134 4.8.4.2 Results...... 135 4.8.4.3 Surface analysis ...... 138 4.8.5 Discussion...... 138 4.9 Effect of annealing on mechanical properties ...... 142 4.10 Influence of the coordination number
10
TABLE OF CONTENTS - Continued
CHAPTER 5 : VISCOELASTIC PROPERTIES OF THE TAS (TE2AS3SE5) GLASS ...... 149 5.1 Introduction...... 149 5.2 Viscoelasticity ...... 150 5.2.1 Boltzmann superposition principle ...... 151 5.2.2 Viscoelastic moduli ...... 151 5.2.3 Mechanical models for linear viscoelastic response...... 152 5.2.3.1 Simple Models...... 152 5.2.3.1.1 Maxwell ...... 152 5.2.3.1.2 Kelvin-Voigt...... 154 5.2.3.1.3 Burger...... 155 5.2.3.2 Complex models...... 156 5.2.3.2.1 Generalized Maxwell model...... 156 5.2.3.2.2 Kohlrausch-Williams-Watt (KWW)...... 157 5.2.3.2.3 Wiechert...... 158 5.2.4 Creep...... 159 5.2.5 Stress relaxation...... 163 5.3 Delayed elasticity in the TAS fibers ...... 164 5.3.1 Stress relaxation experiment...... 164 5.3.1.1 Relaxation and unrolling ...... 165 5.3.1.2 Recovery...... 169 5.3.2 Results...... 170 5.3.2.1 Relaxation ...... 170 5.3.2.2 Steady stress...... 174 5.3.2.3 Recovery...... 175 5.4 Effect of annealing on viscoelasticity...... 180 5.4.1 Relaxation ...... 180
11
TABLE OF CONTENTS - Continued
5.4.2 Recovery ...... 183 5.5 Effect of light on viscoelasticity ...... 184 5.6 Conclusion ...... 187
CONCLUSION ...... 189
APPENDIX A : RAMAN Spectroscopy...... 192 APPENDIX B : BET (Brunauer, Emmet, Teller) method : principle ...... 196 APPENDIX C : Bending stress calculations...... 201 APPENDIX D : Calculation of the stress relaxation in a Burger cell...... 204
REFERENCES...... 206
12
LIST OF FIGURES
FIGURE 1.1 : Spectral radiance of a perfect blackbody at different temperatures ...... 25 FIGURE 1.2 : Civilian applications of thermal camera...... 26 FIGURE 1.3 : Band-gap ...... 26 FIGURE 1.4 : Additional states contributing to the band-gap...... 27 FIGURE 1.5 : Optical window of 3 chalcogenide glasses made of sulfur, selenium or tellurium ...... 30 FIGURE 1.6 : Typical DSC curve for a glass...... 31 FIGURE 1.7 : Principle of a viscometer with parallel plates ...... 33 FIGURE 1.8 : Glass fragility model according to Angell ...... 34 FIGURE 1.9 : Stookey at the origin of the commercialized CorningWare® products ...... 36 FIGURE 1.10 : Young’s modulus for different materials ...... 37 FIGURE 1.11 : Coefficient of thermal expansion for different materials ...... 38 FIGURE 1.12 : Example of TTT curve ...... 41 FIGURE 1.13 : Nucleation and growth phenomena as a function of temperature...... 41 FIGURE 1.14 : Temperature dependence as a function of time for three different thermal treatment ...... 42 FIGURE 1.15 : Examples of Neoceram products...... 44 FIGURE 1.16 : Examples of Fotoceram® products...... 45 FIGURE 1.17 : Examples of Zerodur® products ...... 46 FIGURE 1.18 : Example of Macor® products ...... 47 FIGURE 1.19 : Machinability of glass-ceramics compared with metals ...... 47 FIGURE 2.1 : Glass formation domain in the ternary diagram GeS2-Sb2S3-CsCl...... 50 FIGURE 2.2 : Experimental set-up...... 52 FIGURE 2.3 : Thermal treatment profile...... 53 FIGURE 2.4 : GSSCC glass rod...... 53 FIGURE 2.5 : DSC curve of the glass composition 62.5GeS2-12.5Sb2S3-25CsCl (GSSCC) ...... 54 FIGURE 2.6 : TMA curve of the glass composition 62.5GeS2-12.5Sb2S3-25CsCl (GSSCC) ...... 54 FIGURE 2.7 : Viscosity as a function of the inverse of temperature for the GSSCC glass ...... 55 FIGURE 2.8 : Transmission spectrum of the 62.5GeS2-12.5Sb2S3-25CsCl glass...... 56 FIGURE 2.9 : SEM pictures for GSSCC glass ceramics obtained by different thermal treatment...... 57 FIGURE 2.10 : Glass composition 62.5GeS2-12.5Sb2S3-25CsCl heated at 290°C for different crystallization times...... 58 FIGURE 2.11 : GSSCC transmission spectra for short and long wavelengths heated at 290°C for different thermal treatment times...... 59 FIGURE 2.12 : Diffraction pattern of GSSCC glass ceramics heated at 290°C ...... 61 FIGURE 2.13 : SEM pictures of the glass ceramics (heated at 320°C for 10 hours)...... 61
13
LIST OF FIGURES - Continued
FIGURE 2.14 : Raman spectra of the glass composition 50GeS2-50Sb2S3 with different amount of CsCl ...... 63 FIGURE 2.15 : Decomposition of the main peak of the glass composition 50GeS2-50Sb2S3 into 5 peaks...... 63 FIGURE 2.16 : Evolution of the peak surface area for both 5 peaks of deconvolution ...... 64 FIGURE 2.17 : Structural arrangement in the glass GeS2-Sb2S3-CsCl ...... 65 FIGURE 2.18 : Mechanisms of the glass ceramic formation ...... 66 FIGURE 2.19 : Glass ceramic after 2h immersion in water...... 67 FIGURE 2.20 : Glass ceramic after 5h immersion in water...... 67 FIGURE 2.21 : Glass ceramic after 24h immersion in water...... 67 FIGURE 2.22 : Glass ceramics bulk immersed during 1h30 in the piranha solution (SEM images) ...... 69 FIGURE 2.23 : Glass ceramics with various pore morphologies (SEM images)...... 69 FIGURE 2.24 : Surface comparison of the base glass (a) and a glass-ceramic obtained by heating the base glass during 40h at 290°C (b), both etched by the piranha solution for 30 min ...... 70 FIGURE 2.25 : SEM micrograph of a glass ceramic etched in the “piranha” solution for different times...... 71 FIGURE 2.26 : SEM micrograph of different glass ceramics etched in the “piranha” solution for 1 min ...... 72 FIGURE 2.27 : Transmission spectra of the GSSCC glass ceramic before and after etching with the “piranha” solution...... 73 FIGURE 2.28 : Cross section for a bulk treated for 10 min with the “piranha” solution ...... 74 FIGURE 2.29 : AFM measurements on the surface of a porous sample ...... 74 FIGURE 2.30 : Evanescent field at the surface of an ATR plate ...... 76 FIGURE 2.31 : ATR glass ceramics plate...... 77 FIGURE 2.32 : Optical set-up for measurements...... 78 FIGURE 2.33 : Single beam comparison for the flat ATR plate and the porous ATR plate (heated at 290°C for 32h) ...... 78 FIGURE 2.34 : Experimental set-up for gaseous samples ...... 80 FIGURE 2.35 : Absorbance spectra of ethanol (gas) for the GSSCC ATR plate before etching and after 2h etching...... 80 FIGURE 2.36 : Experimental set-up for testing liquids ...... 81 FIGURE 2.37 : Absorbance spectrum of ethanol (liquid) for the GSCC ATR plate before etching and after 2h etching...... 82 FIGURE 2.38 : Absorbance spectrum of an organic analyte sprayed at the surface of a GSSCC ATR plate, before etching, after 1h and 2h etching...... 83 FIGURE 2.39 : 3 Aminopropyltriethoxysilane molecule...... 84 FIGURE 2.40 : Absorbance spectra in the region of Si-O-Si and amino group vibrations of silanol (aminopropyltriethoxysilane) adsorbed at the
14
LIST OF FIGURES - Continued
surface of a porous and a flat glass-ceramic sample...... 84 FIGURE 2.41 : DSC curve for the composition 65GeS2-15Sb2S3-20CsCl...... 85 FIGURE 2.42 : Transmission spectra for the 65GeS2-15Sb2S3-20CsCl base glass and the corresponding glass ceramic heated at the rate of 1°C/min to 400°C ...... 86 FIGURE 2.43 : SEM picture of the 65GeS2-15Sb2S3-20CsCl fiber composition ...... 87 FIGURE 2.44 : SEM picture of the 65GeS2-15Sb2S3-20CsCl fiber composition heated at 270°C for 12h in a tubular furnace ...... 87 FIGURE 3.1 : Bonding in selenium (atomic states, molecular states and broadening of states into bands in the solid). Kastner ...... 93 FIGURE 3.2 : Schematic composition dependence of the photodarkening effect as a function of the average coordination number
15
LIST OF FIGURES - Continued
FIGURE 4.21 : Different modes for crack opening...... 121 FIGURE 4.22 : Vickers mark seen under microscope ...... 121 FIGURE 4.23 : Tensile test...... 124 FIGURE 4.24 : Two-point bending test ...... 124 FIGURE 4.25 : Rupture facies...... 125 FIGURE 4.26 : Facies after fracture of a TAS fiber ...... 126 FIGURE 4.27 : TAS fibers ageing in air ...... 127 FIGURE 4.28 : Attenuation spectra for a TAS fiber : common synthesis (a) synthesis based on dynamic distillation (b) ...... 129 FIGURE 4.29 : TAS surface after the drawing process (a), after15 days (b) and after 1 month ageing in air (c) ...... 130 FIGURE 4.30 : AFM performed on a 1 week-old TAS fiber ageing in air...... 130 FIGURE 4.31 : TAS fibers ageing under vacuum...... 131 FIGURE 4.32 : TAS surface after the drawing process (a) and after 1 month ageing under vacuum (b) ...... 132 FIGURE 4.33 : DSC traces of Te2As3Se5 (TAS) fibers ageing at room temperature for up to 5 years ...... 133 FIGURE 4.34 : Structure of the fiber after drawing (a) after relaxation (b) / Analogy with polymers ...... 134 FIGURE 4.35 : TAS fibers under loading ...... 135 FIGURE 4.36 : Structure of the fiber after drawing (a) / Effect of loading on the fiber structure (b) ...... 136 FIGURE 4.37 : TAS fiber enthalpy change under loading...... 137 FIGURE 4.38 : TAS surface after the drawing process (a) and after 15 days under loading (b)...... 138 FIGURE 4.39 : Summary of the different drawing processes...... 139 FIGURE 4.40 : Spectrum of the neon lighting ...... 140 FIGURE 4.41 : SEM images of TAS fibers ageing under vacuum and exposed to light (a) stored in the dark (b)...... 141 FIGURE 4.42 : SEM images of TAS fibers ageing in air and exposed to light (a), stored in the dark (b) ...... 141 FIGURE 4.43 : SEM images of TAS fibers ageing in air and exposed to light (a), stored in the dark (b) ...... 142 FIGURE 4.44 : Bending strength as a function of time for the TAS
16
LIST OF FIGURES - Continued
FIGURE 5.5 : Generalized Maxwell model ...... 156 FIGURE 5.6 : Comparison of the relaxation function G(t) simulated by (a) KWW function and (b) generalized Maxwell model ...... 158 FIGURE 5.7 : Wiechert model ...... 159 FIGURE 5.8 : Creep experiment ...... 160 FIGURE 5.9 : Creep-Recovery experiment...... 162 FIGURE 5.10 : Stress relaxation – recovery experiment ...... 164 FIGURE 5.11 : Rolling and unrolling experiment...... 165 FIGURE 5.12 : Deformation of the fiber...... 166 FIGURE 5.13 : Stress relaxation-Recovery experiment...... 168 FIGURE 5.14 : Evolution of the radius curvature as a function of the unrolling time, tu ...... 170 FIGURE 5.15 : Experimental data for the stress relaxation experiment ...... 170 FIGURE 5.16 : Burger model...... 171 FIGURE 5.17 : Experimental data and Burger fitted model for the first 32 days of relaxation ...... 172 FIGURE 5.18 : Burger fitted model for two sets of parameters...... 172 FIGURE 5.19 : KWW fitted function ...... 174 FIGURE 5.20 : Experimental data for the recovery part...... 176 FIGURE 5.21 : Different fitting models for experimental data...... 177 FIGURE 5.22 : KWW fitting function with different parameters for different relaxation times ...... 178 FIGURE 5.23 : Evolution of the KWW parameters (b and τKWW) for different relaxation times ...... 179 FIGURE 5.24 : Stress relaxation curves for both annealed and non-annealed TAS fibers...... 180 FIGURE 5.25 : Wiechert and KWW fitting models for the stress relaxation curves for both annealed and non-annealed TAS fibers ...... 181 FIGURE 5.26 : σ(t) and ε(t) for both annealed and non-annealed TAS fibers...... 183 FIGURE 5.27 : Evolution of the KWW function parameters as a function of relaxation time for the recovery ...... 184 FIGURE 5.28 : Stress relaxation experiment in torsion for a TAS fiber...... 185 FIGURE 5.29 : Stress relaxation as a function of time for the TAS fibers tested in torsion in the dark or exposed continuously to light ...... 186 FIGURE A-1 : Raman scattering mechanisms ...... 194 FIGURE A-2 : Anti-Stokes and Stokes peak for silicon ...... 195 FIGURE B-1 : Different adsorption isotherms...... 197 FIGURE B-2 : BET plot ...... 199 FIGURE C-1 : Two-point bending geometry ...... 201 FIGURE D-1 : Burger model...... 204
17
LIST OF TABLES
TABLE 1.1 : Fixed points...... 32 TABLE 1.2 : Example of glass ceramics composition and properties ...... 43 TABLE 2.1 : Physical properties of glasses in the vitreous domain ...... 51 TABLE 2.2 : Assignment of the absorption bands due to impurities in chalcogenide glasses ...... 56 TABLE 4.1 : Characteristics of the TAS drawing process...... 106 TABLE 4.2 :Assignment of the absorption bands due to impurities...... 113 TABLE 4.3 : Tensile strength (in Newton N) of TAS fibers ageing in air for four drawing processes ...... 128 TABLE 4.4 : Tensile strength concerning the fibers put under vacuum for the different drawing processes ...... 131 TABLE 4.5 : Comparison in the tensile strength between the fibers ageing in air and under vacuum...... 132 TABLE 4.6 : Tensile strength concerning the fibers ageing under static stress for the different drawing processes ...... 135 TABLE 4.7 : Comparison in tensile strength concerning the fibers annealed or not annealed TAS fibers ...... 143 TABLE 5.1 : Wiechert parameters for the stress relaxation for annealed and non-annealed TAS fibers...... 182 TABLE 5.2 : KWW parameters for the stress relaxation for annealed and non-annealed TAS fibers ...... 182
18
ABSTRACT
GeS2–Sb2S3–CsCl glass-ceramics with nanoporous surfaces were synthesized and tested
as optical elements. The nanoporosity is obtained through a two-step process, including
controlled nucleation of CsCl nuclei in the glass matrix followed by selective etching of
the nuclei with an acid solution. The porous surface is several hundred nanometers thick
and results in a surface area increase of almost four orders of magnitudes. The pores size
is approximately 150 nm and can be tailored by controlling the nucleation process and the
etching time. It is shown that the creation of the nanoporous surface does not critically
affect the optical transmission of these infrared transparent glass-ceramics. These
materials can therefore be used for the design of optical elements and an ATR
(Attenuated Total Reflections) plate with nanoporous surface was fabricated and tested as
an optical infrared sensor. The porous element shows higher detection sensitivity in
initial experiments with a coating of silane molecules.
The TAS (Te2As3Se5) infrared glass, used as optical sensor in many fields of applications
(medicine, environment, etc), exhibits poor mechanical properties rapidly that enable it to be used. Its mechanical properties have been investigated as a function of time and
environment. From a general observation, air and vacuum have dramatic effects on TAS fibers tensile strength. When ageing under static stress, they exhibit an increase of tensile strength. The structural relaxation phenomenon is hypothesized to explain these results.
The coordination number,
19
It is shown that the TAS glass exhibits photosensitive effects. This effect seems to be only a surface effect, not a volume effect in the sense that light has no influence on the kinetic of a stress relaxation experiment.
Due to their low glass transition temperature, TAS fibers exhibit viscoelastic behavior at room temperature. The study of the change of radius curvature allows for the determination of constitutive laws both for the stress relaxation kinetics and the delayed elasticity process which are well described by a stretched exponential function KWW
(Kolraush-Williams-Watt).
20
INTRODUCTION
Glasses are non-crystalline solids and are defined as frozen liquids. There are three main types of optical glassy matrix : oxide, fluoride and chalcogenide. The difference between these three types of glass is basically the difference in their chemical composition which leads to changes in optical transmission.
Chalcogenide glasses are interesting materials due to their large transparency in the mid and far infrared including the two atmospheric windows, 3-5 µm and 8-12 µm. Moreover everybody and every object, near room temperature, has its maximum of radiation in the
8-12 µm infrared range. This particularity is at the origin of many applications such as lenses for night vision systems in the military domain or the driving assistance in the new
BMW cars and medical imagery to detect tumor in the civilian domain. Infrared technology using chalcogenide fibers is also very promising and of paramount interest in spatial applications, for example, to detect gas traces on exo platets.
Because all living cells and chemical compounds in general have fingerprint absorptions in infrared range, chalcogenide glasses are interesting for medical applications or for
chemical analysis [1-3]. For example, glasses from the Te-As-Se (TAS) system are very
resistant to devitrification and can be drawn into optical fibers which offer exceptional
transparency in the mid infrared range. These fibers are used as optical sensors to carry
out Fiber Evanescent Wave Spectroscopy (FEWS) to investigate, at molecular scale,
several problems encountered in microbiology, for example the distinction between
21
healthy and cirrhotic liver, or in environmental protection to monitor pollutant in waste
water [4-5].
Chalcogenide glasses are also well known for their poor mechanical properties compared
to oxide glasses. The Young’s modulus of most chalcogenide systems is within 10-20
GPa [6-8] compared to 73 GPa for silica. This can lead to drastic limitations for
applications in devices requiring rough operating conditions such as field deployment. In
order to improve on this problem, several efforts have been focused, in the last decades,
on the development of chalcogenide glass-ceramics which are defined by the controlled
crystallisation of the glassy matrix. Mecholsky, in 1973, synthesized the first transparent
glass ceramics in the range 8-12 µm in the glassy system Ge-Se-As-Pb. He showed that
the glass-ceramic modulus of rupture was increased to as much as twice that of the glass
and the Vickers hardness increased by 30% [9]. Since Mecholsky, few glass ceramics
from chalcogenide glasses have been produced, the reproducibility concerning the glass
ceramics synthesis remaining the main problem. Recently, new reproducible glass
ceramics, combining both chalcogenide and halide chemical elements, has been
synthesized in our laboratory in the systems GeS2-Sb2S3-CsCl and Ga2Se3-GeSe2-CsCl ;
Cesium chloride is the nucleating agent. An appropriate choice of glass composition and heat treatment allows control of the crystallisation process, namely the nucleation rate and the crystal growth. It was shown that these glass-ceramics exhibit better mechanical properties without significantly affecting the optical properties of the material [10-11].
22
The objective of this PhD work is double and constitutes the two parts of this dissertation.
First we have investigated the way to create a new class of infrared material based on nano-porous glass ceramics in the GeS2-Sb2S3-CsCl system to be used as optical sensor.
Due to the soluble nature of alkali halide, cesium chloride, the nuclei embodied in the chalcogenide matrix can be selectively removed, hence these systems have interesting potential for the development of nano-porous structures. This new material is aimed at designing infrared sensors with increased surface area. It is hypothesized that a porous surface can help trap analyte molecules and can improve the sensitivity of detection. In order to investigate this property, measurements with a nano-porous ATR (Attenuated
Total Spectroscopy) plate were performed (Chapter 2).
The first part also deals with photosensitive effects in this glassy system (Chapter 3).
Indeed, chalcogenide glasses are of technological importance because of their photosensitivity, especially when illuminated with sub bang gap light. They present numerous applications in the areas of real-time optical information storage, switching devices, waveguides, holographic optical element and more recently in the area of micro- lenses. Among photosensitive effects, photodarkening was tested and more particularly we studied the influence of alkali halide addition, CsCl, in the GeS2-Sb2S3 glassy system on this phenomenon.
The second part (Chapter 4 and Chapter 5) concerns the better understanding of the mechanical properties of the TAS (Te2As3Se5) fibers.
23
As previously mentioned, the TAS fiber is currently used in our laboratory as a bio-
sensor in several fields of applications including medical and environmental applications.
Experimentally, it is shown that TAS fibers age rapidly in air and their poor mechanical properties observed after a short period of time prevent them to be used for long term application. Therefore, the characterisation of their mechanical properties is essential to have a better understanding of their ageing and to define the best conditions of storage. In this dissertation, we report on the effect of environment and static stress on the TAS fibers mechanical properties. We also deal with TAS glass photosensitivity and viscoelastic behavior at room temperature by studying stress relaxation phenomena inside the fiber when constrained.
24
PART 1 : GLASSES AND GLASS-CERAMICS IN THE SYSTEM GES2-SB2S3-CSCL
CHAPTER 1 : BACKGROUND ON CHALCOGENIDE GLASSES AND GLASS-CERAMICS
1.1 Chalcogenide glasses
1.1.1 Introduction
Chalcogenide glasses are, by definition, glasses which contain at least one of the three elements : sulphur, selenium and tellurium. Depending on the application, we usually add other chemical elements such as arsenic, antimony, germanium, etc to improve mechanical and optical properties and also to increase the stability against devitrification.
The most interesting property of these glasses is associated with their transparency in the mid and far infrared (IR).
Chalcogenide glasses have been in the last decades, of paramount interest for night vision devices because of their remarkable transparency in the two atmospheric windows (3-
5µm and 8-12µm). Chalcogenide glasses tend to replace, at least partially, the expensive mono-crystalline germanium for IR lenses. The ease of processing and the lower cost of chalcogenide glasses compared to mono-crystalline germanium have made them one of the best candidates for lenses. Thermal cameras are based on the fact that everybody and every object, near room temperature, radiates in the IR range. This radiation is more important when the temperature of the body or one given object increases (Figure 1.1).
25
Figure 1.1 : Spectral radiance of a perfect blackbody at different temperatures
Thermal cameras are very useful in many fields of civilian applications as well as in military applications. It can be used by mechanics or electricians to detect overheated assemblies in electric installations (Figure 1.2) or power transfer lines. It can also be used by firemen, for example to see if there is people in a house on fire and to save them.
Another important application is the medical imagery. In this case, thermal camera is used as a body thermometer to detect temperature increase which could mean vascular problem, infection or tumor. And finally, the most recent application concerns systems for driving assistance in the new BMW cars (Figure 1.2).
26
(a) (b)
Figure 1.2 : Civilian applications of thermal camera (a) driving assistance (b) electric installation
1.1.2 Optical properties
The optical transmission of a glass is characterized by its optical window. At shorter wavelengths, the band gap limits the optical window while at longer wavelengths, the optical window is limited by the multi-phonon absorption.
1.1.2.1 Band-gap
The band gap results from electronic transition inside the glass. Photons with sufficient energy are absorbed by exciting electrons across the forbidden band-gap. The electrons are excited from the top of the valence band to the bottom of the conduction band (Figure
1.3).
Figure 1.3 : Band-gap
27
In glasses, additional states exist just above the valence band and just below the
conduction band (Figure 1.4). These states are present because the disorder creates localized electronic states. These localized states participate in the absorption process.
Extended states
Localized states
Figure 1.4 : Additional states contributing to the band-gap
The electronic absorption edge consists mainly of two spectral curves, the “Tauc region”
and the “Urbach region” [12]:
- The electronic transitions between the extended valence band and the conduction
band are very similar to the ones known from ideal crystalline materials as shown
by arrow in Figure 1.3. This leads to the “Tauc Region”. They determine the
absorption of light at high energies. The absorption spectrum in this region can be
approximated by the following equation :
1 2 α(E) = ()E − E where E0 is termed the optical energy gap. E 0
- Below the gap, electronic transitions between a localized band tail and an
extended band as shows by arrow in Figure 1.4 may occur. The so-called band tail
states lead to an extension of the absorption into the band-gap. The absorption
coefficient exhibits an exponential dependence on the energy of the light; hence,
28
it is often referred to as the exponential absorption tail with the form
⎛ E ⎞ α(E) ∝ exp⎜ ⎟ where Eu is the Urbach energy. The exponential band tail is also ⎝ Eu ⎠
known as an Urbach tail and is temperature dependant.
In chalcogenide glasses, absorption phenomena are due to the excitation of non bonding electrons of chalcogen chemical element : S, Se or Te. Because non bonding electrons of selenium (4s2p4) or tellurium (5s2p4) are higher in energy than non-bonding electrons of sulphur (3s2p4), they are more excitable. Therefore, the band gap shifts from visible with sulphur-based glass to near infrared for selenium or tellurium-based glass.
In fact the absorbance is a measure of the amount of light absorbed by the sample under specified conditions. The Beer–Lambert law is the basis of the quantitative of UV/visible spectroscopy. That is why the absorption coefficient, α, can also be determined by using this law.
The most common and easier way to get the band gap width is usually taken for an absorption coefficient equal to 10 cm-1 [13].
−αx 1 ⎛ I 0 ⎞ I = I 0 × e or α = ln⎜ ⎟ (Beer-Lambert law) x ⎝ I ⎠ with I, the transmitted beam
I0, the incident beam
α, the absorption coefficient
x, the thickness of the material
29
To be more accurate, we have to consider Fresnel losses (reflexion of light at the 2 surfaces of the material). These losses are more important for glasses with high refractive index.
2 −αx 2 I (1− R) e (n −1) 1− T0 T = = 2 −2αx with R = 2 = I 0 1− R e (n +1) 1+ T0
T0 is the maximal transmission, α the absorption coefficient, R the reflectivity and n, the refractive index of the material.
Therefore the absorption coefficient is given by the equation :
1 ⎛ 2TR 2 ⎞ α = ln⎜ ⎟ x ⎜ 2 4 2 2 ⎟ ⎝ − ()()1− R + 1− R + 4T R ⎠
1.1.2.2 Multi-phonon absorption
The multi-phonon absorption at longer wavelengths deals with interaction between light and vibration modes of the chemical bonds inside the glass. The phonon energy, E, is directly linked to the atoms weight [14]. The bonding between two atoms of mass m1 and mass m2 is often seen as a spring with the force constant, k :
1 k 1 1 1 E = hυ υ = with = + 2π μ μ m1 m2
With h the Planck’s constant, υ the vibration frequency, k the force constant and μ the reduced mass of the atoms.
30
The large atomic mass of chalcogen elements causes the phonon vibrations to have low energies. Materials with high phonon energies have multi-phonon absorption in the mid- and near-infrared region. This is especially true for materials with lightweight, strongly bound atoms such as silica glasses and polymers and it limits their usefulness for infrared applications. Chalcogenide glasses typically have optical windows that extend into the far infrared beyond 10 μm [15-17].
The following transmission spectra (Figure 1.5) illustrates the multi-phonon absorption for three representative glasses containing both germanium and gallium but different chalcogen element. The molecular weight (MW), MW(Sulfur) < MW(Selenium) <
MW(Tellurium), therefore the multi-phonon absorption for tellurium-based glass is shifted toward longer wavelengths compared to sulphur-based glass.
100
90
Ga5Ge25S70 (1.00mm) 80
Ga5Ge15Se80 (t = 1.65mm) 70
60 Ga10Ge15Te75 (t = 1.60mm) 50
40 %Transmission 30
20
10
0 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28
Wavelength (µm) Figure 1.5 : Optical window of 3 chalcogenide glasses made of sulfur, selenium or tellurium
31
1.1.3 Thermal properties
Thermal characteristics of a glass, such as glass transition temperature Tg and crystallization temperature Tc, are determined using Differential Scanning Calorimetry
(DSC). This technique is based on difference in heat flow between one sample and a reference. Figure 1.6 represents the thermogram heat flow versus temperature for one given glass undergoing crystallization phenomenon (exothermic peak). Glasses which are stable against devitrification (ex : Te2As3Se5 glass) don’t have any crystallization peak.
The glass transition temperature, Tg, is the main characteristic of a glass. Before Tg, the viscosity is infinite (solid state), at Tg, the viscosity is equal to 1013 poises and after Tg, the viscosity decreases as the temperature increases, therefore, the material can be easily shaped. The crystallization phenomenon is characterized by the rearrangement of atoms in organized lattice due to the change of viscosity. Crystallization is at the origin of the loss of the viscoplastic properties as well as the optical transparency. The exothermic crystallization peak, when close to Tg, is in general catastrophic for drawing process.
The stability against devitrification is associated with the difference Tc-Tg. The higher the difference in Tc-Tg, the better the stability against devitrification.
exo
Figure 1.6 : Typical DSC curve for a glass
32
1.1.4 Viscosity
Viscosity is defined as the measurement of the internal friction of a fluid. Viscosity knowledge for one given glass is essential for the working and shaping [18-20].
Viscosity, η, is expressed in Pascal second (Pa.s) in the international system or in Poise
(P) (1P=0.1 Pa.s). From a general observation, viscosity decreases as the temperature increases because of bonds breaking. On the contrary, viscosity increases as the temperature decreases because of bonds formation.
Viscoplastic behavior of glasses is characterized by different precise temperatures called fixed points [21] (Table 1.1).
Table 1.1 : Fixed points
Temperature Viscosity (Poise) Melting Point 102 Working Temperature 104 Softening Temperature 4.2.107 Annealing Temperature 1013 Glass transition Temperature 1013 to 1013.3 Strength Temperature 1014.5
The viscosity, η, describes the flow of an ideal fluid given by the following equation :
Shear stress η = Shear rate
For low values, the viscosity should be measured with a viscometer with parallel plates
(Figure 1.7) as depicted by Fontana [22].
33
The principle is based on two parallel plates of fluid of equal area, A, which are separated by a distance dx and are moving in the same direction at different velocities V1 and V2. F is the force required to maintain this difference in speed and it is proportional to the difference in speed through the liquid also called the velocity gradient.
This velocity gradient, dv/dx, is a measurement of the change in speed when the intermediate layers move with respect to each other. It describes the shearing the liquid experiences and is thus called shear rate. This will be symbolized as S.
The term F/A indicates the force per unit area required to produce the shearing action. It is referred to as shear stress and will be symbolized by F′.
Therefore, the viscosity coefficient is given by the following equation :
Shear stress dx.F F' η = = = with η : the viscosity coefficient Shear rate dv.A S
Figure 1.7 : Principle of a viscometer with parallel plates
Viscosity, η, is a function of temperature and follows the Arrhenius law :
B logη = A + T
Where A, B are some constants and T, the temperature in °K.
34
C. A Angell proposed a method [23] to determine the glass fragility by examining the dependence of the viscosity toward temperature. “Strong” glasses like SiO2 follow the
Arrhenius law which is represented by a line in Figure 1.8 whereas what we call “fragile” glasses do not. Indeed, log η = f (Tg/T) curve for fragile glasses shows a deviation from the Arrhenius law.
Figure 1.8 : Glass fragility model according to Angell
1.1.5 Linear index of refraction
The linear index of refraction is estimated from the maximum optical transmission according to the equation :
2 (n −1) 1− T0 R = 2 = (n +1) 1+ T0
With T0, the maximum percentage of transmission,
R the reflectivity (light is reflected by the 2 surfaces of the glass)
n, the refractive index of the glass
For more precise measurements, the classic method is the method of minimum deviation by using a prism.
35
1.2 Glass-ceramics
1.2.1 Introduction
Glass ceramics are defined as polycrystalline solids prepared by the controlled crystallization of glasses. They can also be seen as composite materials made of a glassy matrix containing crystals as fillers [24]. These kinds of materials may have exceptional properties that can be optimized according to the targeted applications.
In the mid-18th century, Réaumur, a French chemist, is the one who first converted a glass bottle, packed into a mixture of sand and gypsium and heated for several days, to a porcelain polycrystalline ceramic. But Réaumur had no control regarding the crystallization process which is necessary for the production of reproducible glass- ceramics [24].
In the literature, Stookey from Corning (Figure 1.9) is the first who discovered accidentally this new material, called glass-ceramics, in the middle of the 1950’s. The story tells us that he was trying to precipitate silver particles in a lithium silicate glass to achieve a permanent photographic image [25-26] while the furnace programmed to heat at 450°C accidentally over heated to 850°C and surprisingly to Stookey, the bulk had not change its shape. Secondly, Stookey found an unexpected strength when he dropped the sample and it didn’t break. Therefore, glass ceramics are very recent materials. Two years after Stookey’s discovery, the first commercialized glass ceramics, FotoForm® was produced. Stookey is also at the origin of the famous CorningWare® glass ceramics used mainly for cooking ware.
36
Figure 1.9 : Stookey at the origin of the commercialized CorningWare® products
Properties of glasses and glass ceramics strongly depend on their chemical composition and their complex non crystalline structure. Oxide glasses are by far the most studied glasses but we can synthesize glass-ceramics from fluoride glassy matrix [27-29] as well as chalcogenide glassy matrix. Today, we can find glass ceramics in many field of applications [30] :
- Domestic : cooking ware, cooking stove, tableware, microwave oven shelves, etc
- Industrial : windows for oven, furnaces, fireplaces, aircraft glazing, heat
exchangers, pipe tubing for chemical and petroleum industries, electronic devices,
etc
- Construction : building block, floor covering, wall covering, stair treads, paving,
water and sewer pipe, sanitary ware, etc
- Institutional & Research : laboratory equipment, telescope mirror blanks, optics for
observatories in space, cooking surfaces and working tops for restaurants, laser
structure, etc
- Military : radomes, infrared devices, leading edges of supersonic aircraft, etc
- Biological : implants, artificial tooth, tooth fillings, etc
37
Glass ceramics are preferred to glasses because of their better mechanical properties and are usually preferred to ceramics because of the ease of processing. Indeed, moulding is easier, faster and cheaper than solid state sintering which is the common way to produce crystalline ceramics.
General observation regarding glass ceramics compared to the derivative base glass shows an increase of toughness, stiffness and hardness as well as an interruption of the crack propagation due to the crystals [31]. Mechanical properties of glass-ceramics are influenced by several factors such as : particle size and volume fraction of crystalline phase (which can be up to 90%), interfacial bond strength, differences in elastic modulus and thermal expansion between the glassy matrix and the crystals. Figure 1.10 compares the Young’s modulus E of glass ceramics, which is a measure of stiffness, to other materials.
Figure 1.10 : Young’s modulus for different materials [32]
38
The increase in mechanical properties is maybe the most important advantage of glass ceramics over glasses but glass-ceramics present other great advantages. Because of their adjustable coefficient of thermal expansion, glass-ceramics are resistant to thermal shock and permit the sealing to a variety of metals. Regarding this property, Figure 1.11 compares the coefficient of thermal expansion of glass ceramics to other materials. Glass ceramics present the largest range of possible coefficient of thermal expansion from negative to positive values.
Figure 1.11 : Coefficient of thermal expansion for different materials [32]
Then depending on the crystal size, glass-ceramics can be totally transparent. Glass ceramics are also used in electronic for their wide range of dielectric constants and are chosen instead of glasses for their lower dielectric losses.
Glass-ceramics are also corrosion resistant against weathering and a wide range of chemicals, particularly under hot and abrasive environments.
In conclusion, because of their wide and excellent properties (mechanical, thermal, electric), glass-ceramic material can not only compete with traditional materials but are
39
often much better. This wide variety of possible combinations enables use of these materials in practically all industrial fields, technology and household uses where they can replace deficient materials or lead to the introduction of new and better products.
These advantages are also increased by their very good optical properties.
1.2.2 Thermodynamic and kinetic approach
Glass ceramics synthesis, or the devitrification process of the glassy matrix, implies a two-step procedure : nucleation and crystal growth.
Nucleation represents the first step of glass devitrification. It consists in inducing a germinate from which the growth can start. It is based on kinetic parameters. Nucleation may be homogeneous or heterogeneous. In homogeneous nucleation, the first tiny seeds are of the same constitution as the crystals which grow upon them, whereas, in the case of heterogeneous nucleation, the nuclei can be quite different chemically from the crystals which are deposited.
Substances enabling or hastening bulk nucleation are termed nucleating agent. We can distinguish two types of nucleating agents. Metallic nucleating agent such as Au, Cu, Pt, etc, are added to the glass in very small amounts (0.01 to 1% mass). The mechanism of the effect of these nucleation agents in increasing the nucleation rate of the principal crystalline phase is quite complex but is based on heterogeneous nucleation. A second group of nucleating agents including TiO2, ZrO2, SnO2, P2O5 or metallic sulphide can be added in greater amounts (mostly up to 20%). They are part of the glassy composition
40
and they are found to be effective nucleating agents in specific initiation of bulk nucleation.
The so called TTT curve (Time, Transformation rate, Temperature) can predict the time needed to crystallize a fraction of glass at one given temperature. The advantage of the
TTT curve (Figure 1.12) lies in the fact that it permits the determination of a critical point for which the time needed to crystallization is minimal and the temperature for instability is maximal. Avrami’s equation permits to build TTT curve [33] :
x = 1− e− f with f = (kt) n
Where n is the Avrami exponent. This equation is valid if the nucleation is monotonous.
It is assumed that k, the rate constant, varies with time according to the Arrhenius law :
⎛ − E ⎞ k = k0 × exp⎜ ⎟ ⎝ RT ⎠
Where E is the activation energy and k0, a frequency factor.
Without nucleation, crystal growth cannot happen and without growth, no crystals can appear; These processes must take place within a certain range of temperatures which is critical for the devitrification (Figure 1.13).
Above the melting temperature, Tm, the liquid constitutes the stable phase. When the liquid cools down, the crystal growth can theoretically happen between Ta and Tb.
However, the initial nuclei needed before the crystal growth can theoretically happen between Tc and Ta. The critical range is therefore between Ta and Tb (in the case of non addition of nucleating agents in the glass).
41
Figure 1.12 : Example of TTT curve Figure 1.13 : Nucleation and growth phenomena as a function of temperature
The most common treatment to obtain glass ceramics is the thermal treatment. The common thermal treatment used in industry to obtain glass ceramics is treatment
(Figure 1.14). It consists of heating the glassy matrix (base glass) at a temperature above
Tg (glass transition temperature) in order to induce nuclei in the glass. The temperature is then increased to a second plateau to induce the growth of these nuclei.
A second technique consists in a single plateau (Figure 1.14). The glass is heated at a temperature above the glass transition temperature, Tg but below the crystallization temperature, Tc. This technique allows the nucleation phenomenon and avoids excessive growth.
The third treatment is based on heating the glass with a low heating rate of 1°C/min to a temperature above the glass transition temperature Tg+140°C.
42
Several books mention the difficulty of adapting some of these processes to industrial production [34-35].
Temperature
Ta Tb
Tc
Time Figure 1.14 : Temperature dependence as a function of time for three different thermal treatments : two plateaux , single plateau, without plateau
1.2.3 Glass ceramics from oxide glassy matrix
As previously mentioned, oxide glasses are by far the most studied glasses to obtain glass ceramics. They are generally made of several compounds containing oxygen. Every compound has a specific role: glass forming compounds (SiO2, GeO2, P2O5, B2O3, etc), compounds that decrease the glass transition temperature (CaO, Na2O, Li2O, K2O, etc), nucleating agents (TiO2, ZrO2, etc), fining compounds, etc.
Several systems were investigated to synthesize glass ceramics. The following systems
(Table 1.2) from which glass ceramics are derived are some examples but this list is by no means exhaustive [34][36].
43
Table 1.2 : Examples of glass ceramics composition and properties
Li2O Al2O3 SiO2 Na2O MgO ZnO B2O3 K2O TiO2 P2O5 F Properties Low coefficient of 4.3 16.2 73.5 - - - - - 6.0 - - thermal expansion, Photosensitivity - 30.2 42.8 - 14.0 - - - 13.0 - - Low dielectric losses Unusual high 13.6 - 67.9 - 15.5 - - - - 3.0 - coefficient of thermal expansion High mechanical 23.1 - 58.1 - - 15.8 - - - 3.0 - strength High electrical resistivity, good - 14.6 48.5 - - 34.0 - - 2.9 - - chemical resistance, low coefficient of thermal expansion - 16.7 47.2 - 14.5 - 8.5 9.5 - - 6.3 Good machinability
Some of these systems lead to commercial glass-ceramics materials with the trade marks:
Pyroceram®, CorningWare®, Cercor®, Macor®, Zerodur®, Fotoform®, Fotoceram®, etc.
● Pyroceram (Corning)
Pyroceram was originally developed for military applications in the middle of the 1950’s because this material is transparent to radars. Therefore pyroceram is ideal for nosecones and anti-aircraft missiles.
Pyroceram composition is based on the SiO2-Al2O3-Li2O system. It exists different types of pyroceram which differ in their mechanical, thermal and electric properties.
44
Depending on the type of Pyroceram, the main crystalline phase is a solid solution of β- spodumene or β-quartz. Pyroceram is very durable, corrosion resistant and has a very low coefficient of thermal expansion.
Today, the most popular pyroceram is called CorningWare® mainly used for cooking or kitchen ware. It is a white glass ceramic with high mechanical strength, low coefficient of thermal expansion and good chemical resistance. Its composition is 69.7SiO2-17.8Al2O3-
2.8 Li2O-2.6MgO-1ZnO-0.4Na2O-0.2K2O-4.7TiO2-0.1ZrO2-0.2As2O3-0.1Fe2O3 [37].
The crystalline phase is a mix between β-quartz and β-Spodumene.
● Neoceram (Nippon electric glass)
It exists 3 types of Neoceram based on the SiO2-Al2O3-Li2O system :
- Transparent with a low coefficient of thermal expansion : -4×10-7 K-1 - 0 K-1 in the range 30-380°C (Neo-0) [36]
- Semi-transparent (Neo-15)
- Opaque with high strength and breakage resistance. (Neo-11)
Neoceram glass ceramics are especially used for high temperature applications.
(a) (b)
Figure 1.15 : Examples of Neoceram products (a) Neo-0 (b) Neo-11
45
● Fotoceram® (Corning)
Fotoceram is the first glass ceramics discovered by Stookey in the 1950’s. The glasses used are lithia-alumina-silica types (Li2O-SiO2–Al2O3 system). In addition the glasses contain a photosensitive metal : Ag, Cu or Au. UV irradiation followed by heat treatment
(near the annealing point) of the glass induces the precipitation of lithium silicate crystals, Li2SiO3, upon the metallic nuclei. At this step, it is an opacified glass ceramics.
These crystals can be easily and selectively etched from an HF treatment. It is therefore possible to obtain a fine grinding using a photographic mask. The material is then heated a second time. The final glass ceramics has good mechanical strength, good chemical resistance and excellent dielectric properties. They are used for electronic devices (very fine grids) as well as for magnetic recording heads and optical coding disks [36].
Figure 1.16 : Examples of Fotoceram® products
● Zerodur® (Schott)
Zerodur glass ceramics are one of the most famous commercialized glass-ceramics. They are made from the SiO2-Al2O3-LiO2 system with addition of MgO, P2O5 and ZnO. The main crystalline phase is a solid solution of β-quartz (SiO2-AlPO4, LiAlO2, MgAl2O4,
ZnAl2O4) and the volume fraction of the crystalline phase is about 70%. The main
46
characteristic of this glass-ceramics is its low coefficient of thermal expansion in a wide temperature interval.
Zerodur glass ceramics are transparent in the range 0.6-2µm. They are useful for optical applications such as the production of telescope mirrors because they have an excellent temperature stability due to their low coefficient of thermal expansion [36].
Figure 1.17 : Examples of Zerodur® products
● Macor® (Corning)
Macor is a glass ceramic which combines good mechanical properties (high resistance to crack propagation) with the possibility of mechanical turning. It has a low thermal conductivity and is a useful high temperature insulator as well as an excellent electric insulator. This glass ceramic can be cut, ground, drilled and milled. It is based on the
SiO2-Al2O3-MgO-B2O3-K2O-F system. The volume fraction of crystals KMg3AlSi3O10F2
(fluorphlogopite mica) is about 55% [38-39]. Macor is used in electro-technology and technology for space research, in equipment for work under ultra high vacuum (windows in microwave tubes), where high specific resistance or impermeability for gases are required (for hermetic joints). This material is also used in nuclear technology as reference materials in study of the effects of radiation, as they are not changed or affected
47
by radiation [36]. Figure 1.19 shows the comparison between machinable glass ceramics such as Macor® and metals.
Figure 1.18 : Example of Macor® products
Figure 1.19 : Machinability of glass- ceramics compared with metals [36]
1.2.4 Glass ceramics from chalcogenide glassy matrix
Oxide glass ceramics have been widely investigated since 1950 and the research associated with this area is now slowing down. Today, research is more focused on the nucleation and growth phenomena to have a better understanding. However chalcogenide glass ceramics still remain of great interest because of their transparency in the infrared range associated with better mechanical properties. Potential applications are infrared lenses for thermal camera.
48
Chalcogenide glass-ceramics transparent in the range 8-12 µm were first synthesized in
1973 by Mecholsky in the system 0.3 PbSe-0.7 Ge1.5As0.5Se3. The crystalline volume fraction is about 60%. The crystallized phases are PbSe, PbSe2 and GeSe2 with a crystal size of about 0.5µm. He showed that the glass-ceramic modulus of rupture was increased to as much as twice that of the base glass and the Vickers hardness increased by 30% [9].
Other people worked on other system such as As-Ge-Se-Sn [38-41], Ga-Ge-Sb-Se [42-
43] or Ge-Te-Se [44] but the reproducibility of the glass ceramics synthesis remained difficult.
Dong et al. recently reported the generation of nonlinear optical crystals in the GeS2-
Ga2S3-CdI2 α-CdGa2S4 system to enhance the second-order nonlinear optical properties
[45].
First chalco-halide glass ceramics, transparent in the far infrared (10 µm) was obtained in
2003 within the system GeS2-Sb2S3-CsCl in the “Glass and Ceramic” laboratory in
Rennes (France) [10]. The simultaneous presence of ionic and covalent compounds prevent from the rapid and uncontrollable crystallization. Three years later, glass ceramics transparent until 14µm were synthesized in the system GeS2-Ga2Se3-CsCl [11].
Chalcogenide glasses are also good matrix for rare earth because of their low phonon energy. It was recently shown that nanocrystals containing Nd3+ ions can be generated from a Ge-Ga-Sb-S-CsCl system. Emission of this ion is more intense in the case of the glass ceramic compared to the base glass [46].
49
CHAPTER 2 : NANOPOROUS GLASS-CERAMICS USED AS BIOSENSOR
2.1 Introduction
This project was conducted in order to find an alternative to the Te2As3Se5 (TAS) fiber currently used in our laboratory as an optical biosensor and especially we aimed at developing a new material with higher sensitivity of detection.
The currently developed bio-sensor consists of a locally tapered TAS fiber and is proven to be highly interesting for many applications such as in environmental applications to monitor polluted water as well as in medical domain to detect mice’s liver diseases for example or to detect biochemical changes in human lung cells [47]. The principle of detection is based on remote infrared spectroscopy, known as Fiber Evanescent Wave
Spectroscopy (FEWS).
The glass chosen for this study belongs to the GeS2-Sb2S3-CsCl ternary system. The creation of nano-pores at the surface of this new material would increase the contact surface between the material and the substance to be analysed and therefore may increase the sensitivity of detection. The porosity is obtained through a two-step process, including controlled nucleation of cesium chloride, CsCl, nuclei in the glassy matrix followed by selective etching of the nuclei with an acid solution.
50
2.2 Ternary system GeS2-Sb2S3-CsCl
The glassy domain of the GeS2-Sb2S3-CsCl ternary system was previously determined by synthesizing 7 g of glass for different compositions in the diagram (Figure 2.1). All the glasses were quenched into water. The color of glasses varies from orange to red depending on the amount of antimony and cesium chloride. We can notice that the maximum amount of cesium chloride (CsCl) introduced into the matrix is about 25%. For greater alkali halide content, CsCl precipitates out during normal water-cooling.
The physical properties (the glass transition temperature: Tg, the crystallization temperature: Tc, the electronic band gap, the density and the absorption coefficient α) of different glasses in the vitreous domain were determined and are presented in Table 2.1.
2 GeS2 0 Base glass 100 Crystallized domain 10 62.5GeS2-12.5Sb2S3-25CsCl 90
20 80 Vitreous domain
30 70 40 60
50 50
60 40 70 30
80 20
90 10 100 0 Sb2S3 0 102030405060708090100 CsCl
Figure 2.1 : Glass formation domain in the ternary diagram GeS2-Sb2S3-CsCl
51
Table 2.1 : Physical properties of glasses in the vitreous domain
Properties Density Tg (°C) Tx (°C) Tx-Tg α (10-7 K-1) Band Gap (nm) Composition (g/cm3)
GeS2 Sb2S3 CsCl 80 10 10 294 375 81 3.09 217 545 70 20 10 278 349 71 3.28 180 556 70 10 20 293 355 62 3.16 217 556 60 30 10 256 328 72 3.39 197 576 50 30 20 237 355 118 3.48 223 619 40 40 20 236 337 101 3.61 208 630 40 50 10 238 315 77 3.70 208 652 10 80 10 225 260 35 4.05 211 730
2.3 Glass preparation
The glass composition 62.5GeS2-12.5Sb2S3-25CsCl, called GSSCC, was chosen to perform nucleation and growth experiments because this composition is situated at the limit of the glassy domain (Figure 2.1). This composition is stable enough to obtain glass rods but sufficiently unstable to allow nucleation and growth phenomena when heated at appropriate temperatures.
The procedure for the glass preparation is the following : the set-up (Figure 2.2) is put under vacuum during about 4 hours in order to dry the silica tube previously washed with solvents. A trap is cooled down in liquid nitrogen to retain these washing liquids. Then, the chemical compounds (polycrystalline germanium 99.999%, antimony 99.99%, sulphur 99.999%, cesium chloride 99.9%), weighed in predetermined quantities are introduced in the 10 mm diameter silica tube. They are put under primary vacuum during few minutes then, once, the primary vacuum established, a turbo molecular pump permits
52
to have a better vacuum during about 8 hours. The reactional tube is obtained by sealing the tube as shown in Figure 2.2.
Vacuum
Trap
Sealing Dewar with liquid nitrogen
Reactional Chemical Products tube Ge, S, Sb + CsCl
Figure 2.2 : Experimental set-up
Figure 2.3 represents the different thermal treatment steps for the reactional tube, which is first heated (at the very low rate of 2°C/min to avoid the rapid formation of vapor and therefore the explosion of the tube) until the homogenization temperature (800°C). The glassy composition stays about 8 hours at this temperature. In order to decrease vapor pressure above the melted bath and also to improve the quenching, the furnace temperature is decreased to 700°C. The tube containing the melt is quenched into water.
The glass rod is then annealed in an annealing furnace previously heated at a temperature
Tannealing near the glass transition temperature Tg.
The GSSCC glass is annealed at 260°C during 10 minutes to avoid the beginning of crystallization and is cooled down to room temperature very slowly (10 hours). We therefore obtain a glass rod (Figure 2.4).
(c) 53
900 800 700 (b) (d) 600
500 (e) 400 (a) 300 (f)
Temperature (°C) 200 (g) 100 0 0 5 10 15 20 25 30 hours
Figure 2.3 : Thermal treatment profile : (a) heating and reaction of the elements (b) Refining and homogenization of the melting bath (c) vapor condensation in equilibrium with the molten bath (d) plateau before quenching the glass, the glass is kept in vertical position (e) quenching, (f) annealing (g) slow cooling down until room temperature
Figure 2.4 : GSSCC glass rod
2.4 Physical properties
2.4.1 Thermal properties
The DSC curve (Figure 2.5) is obtained with a heating rate of 10°C/min on an equipment
DSC 2010 from T.A Instruments and for a batch of about 5 mg. The glass transition
temperature, Tg, is 260°C and the crystallization temperature, Tc, is 350°C.
54
Figure 2.5 : DSC curve of the glass composition 62.5GeS2-12.5Sb2S3-25CsCl (GSSCC)
2.4.2 Thermal expansion coefficient
The thermal expansion coefficient for the GSSCC glass is 21.6×10-6 K-1 between 27°C and 225°C and was determined with a T.A Instruments equipment.
Figure 2.6 : TMA curve of the glass composition 62.5GeS2-12.5Sb2S3-25CsCl (GSSCC)
55
2.4.3 Viscosity
In the range 105-108 Poises, the viscosity is measured with a viscometer with parallel plates used in compression (Rheotronic, Theta Industries, inc.).
11,00
10,00
) 9,00 η 8,00
7,00 log viscosity (log (log viscosity log 6,00
5,00
4,00 1,57 1,59 1,61 1,63 1,65 1,67 1,69 1,71 1,73 1,75 1,77 1000/T(K)
Figure 2.7 : Viscosity as a function of the inverse of temperature for the GSSCC glass
Viscosity, η, is a function of temperature and follows the Arrhenius law :
B logη = A + T
With A, B some constants and T, the temperature in °K.
The GSSCC viscosity was modelled by an Arrhenius law (black line) with the parameters
:
25800 logη = −35.8 + T
56
2.4.4 Optical properties
The UV – visible and near IR transmission spectrum of the 62.5GeS2-12.5Sb2S3-25CsCl glass (GSSCC) (Figure 2.8) is obtained from a CARY 5 VARIAN spectrophotometer
(200nm<λ<3000nm) and from a BRUKER Tensor 27 FTIR (Fourier Transformed
InfraRed spectroscopy) for the infrared range (λ>1µm). Due to heavy component elements, the transmission window of the 62.5 GeS2–12.5 Sb2S3–25 CsCl glassy sample extends from the visible (red) down to almost 12 µm.
90 80 70 60 50 40 30 % Transmission Transmission % 20 10 0 012345678910111213 µm
Figure 2.8 : Transmission spectrum of the 62.5GeS2-12.5Sb2S3-25CsCl glass (1.15 mm thickness)
Different absorption bands are present on this spectrum at different wavelengths due to the absence of starting chemical compounds purification. The impurities are listed in the
Table 2.2.
Table 2.2 : Assignment of the absorption bands due to impurities in chalcogenide glasses [48]
Impurities S-H Ge-O H20 O-H λ (µm) 3.6 ; 4 7.9; 12.5; 12.7; 13 6.3 2.9
57
2.5 Glass ceramic
2.5.1 Thermal treatment
The GSSCC glass rod is cut into 2 mm thick samples. Several crystallization tests were done on these samples. They are put in a ventilated furnace heated from room temperature to the final temperature in 1 hour. They are kept at this temperature during different times [10]. Figure 2.9 shows SEM (Scanning Electron Microscopy) pictures of samples for different crystallization times and temperatures.
The base glass 62.5GeS2-12.5Sb2S3-25CsCl which is initially red and transparent (due to sulphur chemical element) becomes an orange and opaque glass ceramics due to the nucleation and crystal growth induced by the thermal treatment (Figure 2.10).
Figure 2.9 : SEM pictures for GSSCC glass ceramics obtained by different thermal treatment. (a), (b), (c) and (d) : 7, 31, 144 and 487h at 290°C, (e) : 10h at 320°C, (f) : 50h at 300°C
58
(a) (b)
(c) (d)
Figure 2.10 : Glass composition 62.5GeS2-12.5Sb2S3-25CsCl heated at 290°C for different crystallization times (a) No thermal treatment (b) 7h (c) 73h and (d) 144h
2.5.2 Optical properties
The presence of nanocrystals induces scattering especially in the visible region but the optical transmission in the mid infrared, which is of interest for optical sensing, is still excellent for different glass ceramics (Figure 2.11). The potential for using these materials as optical elements for remote or in-situ vibrational spectroscopy is therefore maintained. At one given temperature, the scattering phenomena become more and more important as the crystals become bigger and bigger when the thermal treatment continues.
Therefore, it is possible to control the nucleation and growth phenomena.
59
) % ( Transmission
Wavelength (µm)
) % (
Transmission
Wavelength (µm)
Figure 2.11 : GSSCC transmission spectra for short and long wavelengths heated at 290°C for different thermal treatment times
60
2.5.3 Influence of crystallization on the linear index of refraction
The approximate linear index of refraction is determined by the maximum percentage of transmission using a double beam Perkin Elmer infrared 882 instrument. The maximum of transmission percentage for the base glass, GSSCC and the corresponding glass ceramics heated at 290°C for 40h, at 9 µm, are respectively 74% and 75% (1.2 mm thick samples). The index of refraction is calculated to be 2.25 for the base glass and 2.22 for the glass ceramics (290°C for 40h). As a consequence, crystallization does not induce a significant change in the refractive index. Precise measurements for the refractive index would necessitate the preparation of a prism with excellent optical polished surfaces.
2.5.4 X-Ray diffraction
X-Ray diffraction was performed on glass ceramic samples (bulk or powder) using an
INEL diffractometer. The nature of the phase which crystallises remains difficult to determine but the RX diagram indexation lets suppose the presence of cesium chloride,
CsCl. Indeed the indexation of diagrams (Figure 2.12) shows a cubic unit cell with a lattice parameter of 4.12 Å which is very close to that of cesium chloride crystals (a = 4
Å). The higher lattice parameter may be explained by the substitution of few Cl- by S2- or
Ge2+ by Cs+. After 73h at 290°C, the cubic phase is still present but two other peaks appear which let suppose the apparition of a second phase which is still undetermined
(Figure 2.12).
61
Figure 2.12 : Diffraction pattern of GSSCC glass ceramics heated at 290°C
2.5.5 SEM (Scanning Electron Microscopy) analysis of crystals
A crack on the surface of the GSSCC glass ceramics obtained by a thermal treatment at
320°C during 10 hours was made in order to visualize the crystals inside the bulk. SEM
(Scanning Electron Microscopy) pictures show the presence of two types of crystals : stick and round crystals (Figure 2.13). The samples were first coated with a very thin gold film due to the insulating character of this composition. Again, the exact chemical composition of these crystals remains uncertain due to the difficulty of RX diagram indexation but SEM analysis confirms the presence of two phases.
Figure 2.13 : SEM pictures of the glass ceramics (heated at 320°C for 10 hours)
62
2.5.6 Structural arrangement in the base glass and the associated glass ceramic
A structural assumption approach is needed to understand the crystallization mechanisms of the base glass GSSCC and more particularly we have to know where the Cl- ions are located in the structure when cesium chloride, CsCl, is introduced. Raman spectroscopy
(Appendix A) provides information about molecular vibrations that can be used for sample identification, quantification or in our case for describing and for understanding the glass structure.
Raman spectroscopy, using a fiber optic Raman probe QE65000 from Ocean Optics, was performed on the glass composition 50GeS2-50Sb2S3 with different amounts of CsCl : 0,
5, 10, 15 and 20%. The results are shown in Figure 2.14.
As the amount of cesium chloride increases, the intensity of the peak centered at 300 cm-1 decreases.
This complex peak can be deconvoluted into five peaks (Figure 2.15) [49-51]. Each peak corresponds to a different vibration : S3Ge-S-GeS3, GeS4 (F2), GeS4 (A1), SbS3 (A1) and
SbS3 (E). The different vibration modes are detailed in reference 49.
63
0% CsCl
5% CsCl
10% CsCl
15% CsCl
Intensity 20% CsCl
cm-1
Figure 2.14 : Raman spectra of the glass composition 50GeS2-50Sb2S3 with different amount of CsCl
Vibrations :
5 : S3Ge-S-GeS3 3 4: GeS4 (F2) 3 : GeS4 (A1) 2 : SbS3 (A1) 1 : SbS (E) 2 3
Intensity 4 1
5
cm-1
Figure 2.15 : Decomposition of the main peak of the glass composition 50GeS2-50Sb2S3 into 5 peaks
64
The peak at about 175 cm-1 (Figure 2.15) is not assigned due to the limit of detection of the equipment which is 180 cm-1. The intensity of the five peaks have been integrated using OPUS software for different glasses containing from 0% to 20% cesium chloride.
The glass composition with 0% CsCl is chosen as the reference.
The most important decrease of the intensity is observed for the GeS4 (A1) and SbS3 (A1) vibrations (Figure 2.16), indicating that Cl- ions are attached to both antimony and germanium. It is noteworthy that the decrease in surface is exactly proportional to the amount of CsCl inserted into the structure.
This can be explained by looking at the bond energy of Sb-Cl (in SbCl3 315 kJ/mol) and
Ge-Cl (349 kJ/mol) which are quite similar therefore there is no preferential site for the
Cl- anion to bond to.
S3Ge-S-GeS3 2500000 GeS4 (F2) GeS4 (A1) SbS3 (A1) 2000000 SbS3 (E)
1500000
1000000 Peak surface area
500000
0 0 5 10 15 20 % CsCl Figure 2.16 : Evolution of the peak surface area for the 5 peaks of deconvolution
65
A structural model (Figure 2.17) based on Raman measurements is proposed to describe the GSSCC glass structure.
Figure 2.17 : Structural arrangement in the glass GeS2-Sb2S3-CsCl
As previously described, Cl- ions are attached to both Ge and Sb and form non bridging atom. Some S- ions form non bridging atoms as well. The structure is therefore made of
+ GeS3Cl, GeS4, SbS3 and SbClS2 polyhedrons connected by sulphur atoms. Cs ions are bonded to the matrix through ionic bonds. At some point, the amount of added CsCl compound is critical and permits the release of Cl- ions from an appropriate thermal treatment. The thermal agitation due to this thermal treatment allows the diffusion of Cl- ion to Cs+ ions to form the ionic compound cesium chloride, CsCl. The proposed structural arrangement and the mechanism of crystal growth are depicted in Figure 2.18.
66
S
S Ge S
S S S
S- Ge S Ge S Ge Cl δ-
+ S Cs S Cs+ S S
Sb Sb Ge S- S Ge S S S δ-- 290°C Cl S S S S S S
S Ge S Ge S Ge S
+ - + - Cs Cl Cs Cl S S S
Sb Sb S S Ge S
S S S
Figure 2.18 : Mechanisms of the glass ceramic formation
2.6 Nanoporous glass ceramic
In order to produce a nanoporous layer at the surface of the glass-ceramic samples, the
CsCl crystallites are selectively etched out of the glass matrix. Initial tests were performed with simple aqueous solution as well as with acid or basic solutions.
2.6.1 Etching with water
Cesium chloride, CsCl, is an ionic compound so that water with a high dielectric constant may easily dissolve these crystals.
Glass ceramic bulks were immersed in water during different times ranging from 2 to 24 hours. The crystal size is about 150 nm for a thermal treatment of 10 hours at 320°C.
67
SEM analysis were performed to monitor the evolution of the etching. The results are presented in Figure 2.19, 2.20 and 2.21.
Figure 2.19 : Glass ceramic after 2h immersion in water
Figure 2.20 : Glass ceramic after 5h immersion in water
Figure 2.21 : Glass ceramic after 24h immersion in water
68
Initial tests with water resulted in superficial removal of few crystallites at the sample surface. Depending on the immersion time, pores appears but we can notice the presence of a thin white film which cover the pores. These results are not satisfactory and other tries have to be performed to improve crystals etching.
All the samples are polished prior to the etching tests. The presence of pores lines in
Figure 2.20 is due to the polishing disk which removes preferentially some crystals according to parallel lines.
2.6.2 Etching with an acid or a base
Most nuclei embodied beneath the glass surface were not affected by the aqueous solution. Tests with a more aggressive etching solution produced much better results. The
“piranha” solution which consists of 70% concentrated H2SO4 – 30% H2O2 (35% weight) in volume percentage has been used previously for chemical polishing of chalcogenide glass from the Te-As-Se system [52]. This solution was shown to dissolve the glass homogeneously and produce smooth and flat surface. The “piranha” solution was therefore applied to the GeSe2-Sb2S3-CsCl glass ceramics with the aim of dissolving the superficial glass layer and exposing a large number of CsCl nuclei. The different etching speeds between the glass matrix and the crystallite should then result in preferential etching of the nuclei leading to the desired nanoporous structure. The experimental procedure is the following : after an appropriate ceramisation, the GSSCC sample is immersed in the “piranha” solution which permits to etch the crystals at the surface of the material and to obtain nano-porous glass-ceramics. The surface morphologies are shown
69
in Figure 2.22. It can be seen that with the appropriate heat treatment and etching time, a highly porous surface can be obtained. The etching time can range from several minutes to several hours, resulting in various pore morphologies (Figure 2.23).
Surface
Crack
Figure 2.22 : Glass ceramics bulk immersed during 1h30 in the piranha solution (SEM images)
Figure 2.23 : Glass ceramics with various pore morphologies (SEM images)
Other tests, made with different acids and bases including sulfuric acid (H2SO4, pH = 4-
5) and soda (NaOH, pH = 9-10), were also tested to create nanopores. The same unsatisfactory surface, seen with water, was observed with acids and bases. The best results have been obtained with the “piranha” solution as etching agent. The base glass
GSSCC (no crystals) has been also etched with the piranha solution (Figure 2.24) in order to see the influence of the crystals on the etching phenomena. The behavior toward the
70
acid treatment is totally different. The base glass doesn’t show any porosity whereas the
glass ceramics treated for 40h at 290°C is clearly porous. Therefore, the crystals play a
key role in the creation of the porosity by chemical etching.
(a) (b)
Figure 2.24 : Surface comparison of the base glass (a) and a glass-ceramic obtained by heating the base glass during 40h at 290°C (b), both etched by the piranha solution for 30 min
2.6.3 Control of the pore size
Figure 2.25 describes the effect of etching time on the glass surface. It can be seen that
the pore size increases as the matrix dissolution follow its course. For these tries, the
durations for chemical etching vary from 15 seconds to 1 hour. The surface attack starts
with the surface crystals etching. As the time of acid treatment increases, the size of the
pores increases from about 100 nm for a 15 seconds acid treatment time to 500 nm after 1
hour, acid leading to the damage of the glassy matrix (Figure 2.25). We observe the
creation of cavities when the time of acid treatment is more than few minutes.
The heat treatment procedure, used for the glass ceramic fabrication, also plays a role in
shaping the surface morphology as it allows the control of the size and number of
crystallites. Figure 2.26 shows the effect of these heat treatments on the pore
morphology. Etching of a glass containing no crystal shows no significant porosity as
71
described before. As the number of crystals increases with ceramisation time, the porosity develops and eventually reaches a level where the porous structure starts to collapse and creates larger voids.
(e)
Figure 2.25 : SEM micrograph of a glass ceramic etched in the “piranha” solution for different times (a): no acid treatment, (b): 15 s , (c): 1 min , (d): 10 min, (e): 1h
72
(a) (b)
(c) (d
Figure 2.26 : SEM micrograph of different glass ceramics etched in the “piranha” solution for 1 min. (a): base glass, (b): 20h at 300°C, (c): 40h at 300°C, (d): 60h at 300°C
2.6.4 Optical properties
The effect of the nano-porous layer on the optical transmission was measured by UV-VIS spectroscopy on a 1.5 mm thick sample etched for up to 1 hour. Figure 2.27 shows that the thin porous layer induces slight scattering in the short wavelength region but does not have any influence beyond 5 µm. It can be concluded that the acid treatment of the surface and the creation of the pores are not critical for optical transmission even for a 1 hour acid treatment time.
73
In order to estimate the thickness of the porous layer, the cross section of an etched sample was analyzed by SEM. Figure 2.28 shows that the typical depth after a 10 minute- acid treatment is only few hundred nanometers.
AFM (Atomic Force Microscopy) measurements also permit to determine the roughness of the surface and the typical depth of the pores on the surface (Figure 2.29). After a 5 minute-acid treatment, the depth is about 70 nm.
90 80
70
60 50 GSSCC glass ceramic 40 30 After 1h30 acid treatment
% Transmission 20
10 0 0246810 µm Figure 2.27 : Transmission spectra of the GSSCC glass ceramic before and after etching
with the “piranha” solution (1.15 mm thick)
74
200 nm
Figure 2.28 : Cross section for a bulk treated for 10 min with the “piranha” solution
Figure 2.29 : AFM measurements on the surface of a porous sample (5 min acid treatment time)
2.6.5 Specific surface
The specific surface of the porous glass ceramics was determined using the Brunauer,
Emmett, and Teller (BET) gas adsorption method and measured on a Flowsurb analyser
2300 (Micromeritics). For this measurement, nitrogen gas adsorbed at the boiling
75
temperature of liquid nitrogen was used. The adsorption was measured on a porous bulk
(3-4 g) that had been heated under vacuum at about 120°C for a few hours prior to the measurements. The specific surface was calculated using the BET equation (Appendix
B). The surface area of a non porous glass ceramic with the dimension 24×10×5 mm was measured to be around 10 cm2. The same sample was then treated for 30 minutes with the
“piranha” solution and the new surface area was found to be about 7 m2. So the surface area is multiplied by a factor of 7000 for the porous surface compared to the non-porous surface.
2.6.6 Surface composition analysis
Because the material is treated with a strong acid, we may suppose that a thin oxide film may create at the surface of the material. Surface composition analysis was performed using a micro-electronic probe associated with the scanning electronic microscope. The surface composition concerning a sample treated 1 hour with the “piranha” solution confirmed the absence of an oxide film. The same amount of oxygen was found at the surface and inside a crack made on the sample (about few hundreds nanometers under the surface). This amount of oxygen is less than 1% and is therefore negligible.
2.6.7 ATR (Attenuated Total Reflections) measurements
2.6.7.1 Principle of the method
ATR (Attenuated Total Reflections) spectroscopy works along the same principle as
Fiber Evanescent Wave Spectroscopy (FEWS). The IR beam enters the ATR plate
76
perpendicularly to the surface therefore the refractive index of the material doesn’t influence the direction of the beam. The light is reflected N times inside the ATR plate.
The number N depends on the θ angle of incidence between the light and the surface of the ATR plate, the thickness of the plate W and the length of the plate L according to the equation : N = L/Wcosθ. In our case θ=90°.
These internal total reflections result in the creation of an evanescent field at the surface of the ATR plate, penetrating in air on few micrometers (Figure 2.30). When the surrounding medium absorbs, the evanescent field is absorbed at each reflection. Then the light collected at the output is affected by the characteristic absorptions of the substance to be analyzed.
Evanescent field Sample
ATR plate
Figure 2.30 : Evanescent field at the surface of an ATR plate
2.6.7.2 ATR glass ceramic plate
In order to produce the ATR element, a batch of 110g was produced in a silica tube
(internal diameter 16 mm and length of 250 mm). The nucleation and growth
phenomena versus temperature and the time of thermal treatment for 1.5 mm disks of
this composition have already been studied [10]. As a consequence, one can easily
77
adapt the time of thermal treatment from the 1.5 mm disk to the 110g glass rod in order to obtain the same transmission properties. The ATR glass ceramic plate was then obtained by heating the glass rod at 290°C for 32 hours and by cutting and polishing the resulting glass-ceramic rod to the proper ATR plate geometry
80×10×4mm (Figure 2.31). This glass-ceramic ATR plate was then etched to develop the nanoporous surface. The resulting optical element was mounted on the ATR attachment of a Bruker tensor 27 FTIR (Fourier Transformed InfraRed) spectrometer
(Figure 2.32). The overall transmitted intensity through the ATR plate decreased by about 30% as a result of the pore formation most likely due to scattering (Figure 2.33).
This observation is not in contradiction with Figure 2.27, obtained with a 1.15 mm thick sample.
80 mm
4mm
10 mm
Figure 2.31 : ATR glass ceramics plate
78
ATR glass ceramic plate
IR light
Figure 2.32 : Optical set-up for measurements
0,25
Non porous ATR plate
0,2 Porous ATR plate (1h acid treatment time)
0,15
Single beam 0,1
0,05
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 -1 cm
Figure 2.33 : Single beam comparison for the flat ATR plate and the porous ATR plate (heated at 290°C for 32h)
79
2.6.8 Applications
2.6.8.1 Experiments with gaseous samples
Several tries were made with different solvents (acetone, ethanol, etc) in order to test the sensitivity of the nano-porous sensor ATR plate with gaseous molecules. For these experiments, two different ATR plates were synthesized. The first one is acid-treated during 2 hours to obtain a porous plate. The second one is not treated and as a consequence is not porous. The set-up used to perform the measurements is described in
Figure 2.34. A silica piece is stuck on both ATR plates for the airtightness. A 2-way tap permits to vacuum the ATR plate. The recipient containing the solvent is first frozen in liquid nitrogen and vacuumed. Then the recipient which contains the solvent is put in contact with the ATR plate thanks to the tap. As the solvent warms up, a vapor pressure equilibrium establishes in the set-up. Then the ATR plate is removed from the set-up by closing the 1-way tap and put in the FTIR chamber for measurements. The result obtained with ethanol is presented in Figure 2.35. We see a small increase of the absorbance in the case of the porous plate compared to the smooth plate. For these first tries, we encountered the problem of set-up airtightness, therefore the results may not be consistent.
80
2-way Tap
Vacuum 1-way Tap
Silica piece to ensure airtightness
ATR plate
Trap Solvent
Figure 2.34 : Experimental set-up for gaseous samples
Non porous ATR plate 0,14
Porous ATR plate (2h acid treatment 0,12 time)
0,1
0,08
Absorbance 0,06
0,04
0,02
0 6 6,5 7 7,5 8 8,5 9 9,5 10 µm
Figure 2.35 : Absorbance spectra of ethanol (gas) for the GSSCC ATR plate before etching and after 2h etching
81
2.6.8.2 Experiments with liquid samples
To avoid airtightness problem observed in the case of gaseous samples, we performed experiments with liquids and especially with ethanol. In this case, the same ATR plate was used for both measurements. A little silica recipient was stuck as a container on the surface of the ATR plate and some ethanol was introduced inside (Figure 2.36). We measured the absorbance with the FTIR equipment. To create pores at the surface of the
ATR plate, some piranha solution is then introduced in the little container. The surface is then washed several times with water to remove all traces of acid and the same amount of ethanol is introduced into the recipient for both measurements. The results show a significant increase of the ethanol absorbance for the porous ATR plate compared to the non-porous ATR plate (Figure 2.37). The difference in absorbance can be explained by an increase in surface area due to the presence of the pores for the porous ATR plate compared to the smooth plate.
ATR glass ceramic plate Container ATR attachment of the Bruker FTIR spectrometer
Figure 2.36 : Experimental set-up for testing liquids
82
3
2,5 Non porous ATR plate
2 Porous ATR plate (2h acid treatment time)
1,5 Absorbance 1
0,5
0 8 8,2 8,4 8,6 8,8 9 9,2 9,4 9,6 9,8 10 µm
Figure 2.37 : Absorbance spectrum of ethanol (liquid) for the GSCC ATR plate before etching and after 2h etching
2.6.8.3 Experiments with a spray
In order to test again the porous sensor, an analyte was sprayed on the sensor to force the spray molecules enter into the pores. The spray exhibits several absorption bonds between 3 and 8 µm. This experience was performed with the non porous ATR plate as well as with the porous ATR plate with a 1 and a 2-hour acid treatment time. The same
ATR plate area was covered by the spray for both measurements. Figure 2.38 shows that the signal with the porous ATR plate (1 or 2-hour treatment time) is enhanced compared to the smooth plate.
83
3
2,8 ATR without acid treatment (reference) 2,6 ATR treated 1 hour with the acid solution 2,4 ATR treated 2 hours with the acid solution 2,2
2
Absorbance 1,8
1,6
1,4
1,2
1 33,544,555,566,577,5 µm
Figure 2.38 : Absorbance spectrum of an organic analyte sprayed at the surface of a GSSCC ATR plate, before etching, after 1h and 2h etching
2.6.8.4 Experiments with APTS (Aminopropyltriethoxysilane)
The GSSCC glass ceramic surface was functionalized with aminopropyltriethoxysilane
(APTS) (Figure 2.39) by immersing the sample in a solution of toluene containing 8.6
W% APTS for a duration of 5 min to 3h. Silane molecules are commonly used to functionalize surfaces and permit to attach layers of molecules with specific recognition sites such as enzymes [53-54]. The silane molecules bind to the glass ceramic surface as described in Taga et al. [53]. Figure 2.40 shows the transmission spectra of a porous and a flat glass-ceramic sample immersed in a silane solution for 5 min. The porous surface shows four to fivefold increase in surface silane molecules. Similar results were obtained
84
after rinsing with toluene. This demonstrates the possibility of functionalizing the glass- ceramic surface and consequently the potential of applying these materials to the design of sensors with specific binding capabilities. It is noteworthy that the four to fivefold increase observed in this case is still far from the 7000 times increase expected from the gain in surface area. This limitation is likely due to the difficulty in generating confluent monolayers within the pores.
Figure 2.39 : 3 Aminopropyltriethoxysilane molecule
2,5
APTS 5 mns non porous sample 2 APTS 5 mns porous sample
1,5
1 Absorbance
0,5
0 67891011 µm
Figure 2.40 : Absorbance spectra in the region of Si-O-Si and amino group vibrations of silanol (aminopropyltriethoxysilane) adsorbed at the surface of a porous and a flat glass-ceramic sample
85
2.7 Porous glass ceramic in the fiber configuration
For some applications, sensor shaped as a fiber is more convenient than ATR plate.
Therefore, efforts have been focused on synthesizing a locally porous fiber through a three-step process :
- First a glass composition with a large difference between the glass transition
temperature Tg and the crystallization temperature Tc was chosen to be drawn
into fiber. A large value for Tc-Tg is a stability criterion.
- Second, the fiber is heated locally (1 cm) using a tubular furnace to allow
nucleation and growth phenomena.
- Third, the fiber is etched with the “piranha” solution to obtain a locally porous
fiber.
The glass composition 65GeS2-15Sb2S3-20CsCl was preferred for these promising tries instead of the base glass composition, 62.5GeS2-12.5Sb2S3-25CsCl, especially for its large Tc-Tg which is not critical for the drawing process (Figure 2.41).
Figure 2.41 : DSC curve for the composition 65GeS2-15Sb2S3-20CsCl
86
Ceramisation tests have been performed on the 65GeS2-15Sb2S3-20CsCl bulks. The heating rate of 1°C/min from room temperature to 400°C induces the scattering of the light by crystals (Figure 2.42).
A 6.6 mm diameter glass rod (65GeS2-15Sb2S3-20CsCl) was drawn into 400 microns diameter fiber and SEM analysis performed on this fiber did not show any crystals
(Figure 2.43). First tests of ceramisation (from room temperature to 270°C in 6h and plateau at this temperature for 12h) using a tubular furnace (1 cm heating zone) do not induce any deformation of the fiber. SEM analysis performed on the local heated part indicates the presence of crystals with a size of about 500nm (Figure 2.44). This size is critical for optical properties.
These first tests are promising. We have now to focus on the ceramisation parameters
(time, temperature) in order to get smaller crystals. This experiment is still in progress.
80
70
60
50
40
30 % Transmission 65GeS2-15Sb2S3-20CsCl 20 65-15-20 heated at 1°C/min to 400°C 10
0 500 550 600 650 700 750 800 850 900 950 1000 nm
Figure 2.42 : Transmission spectra for the 65GeS2-15Sb2S3-20CsCl base glass and the corresponding glass ceramic heated at the rate of 1°C/min to 400°C
87
Figure 2.43 : SEM picture of the 65GeS2-15Sb2S3-20CsCl fiber composition
Figure 2.44 : SEM picture of the 65GeS2-15Sb2S3-20CsCl fiber composition heated at 270°C for 12h in a tubular furnace
88
2.8 Discussion
In a conventional ATR experiment the analyte is placed on the plate surface and it can interact with the probing light through the evanescent wave which typically extends several hundred nanometers above the surface [55-56]. This estimation assumes an interface with a well-defined change of refractive index between the ATR plate and the analyte. In the current set-up, it is not clear how the porous layer affect the refractive index at the plate surface and consequently the nature of the evanescent wave since the refractive index measurements were not accurate enough. However the results of different experiments demonstrate that the porous surface does not seem to affect the total internal reflections negatively despite a possible gradient of index at the surface. The
IR beam is still guided properly through the ATR element and the detection sensitivity is at least conserved and often significantly improved. The increase in absorbance observed with the porous plate is likely the result of a greater concentration of analyte molecules in the region where the evanescent wave is most intense. In a standard set-up, the intensity of the evanescent electric field decreases exponentially with distance from the surface. In the porous set-up, the analyte is somewhat incorporated within the surface where the evanescent filed is most intense. This could explain the greater sensitivity of the porous plate in comparison with the flat plate. Noteworthy, for every results performed with gas, liquid or solid, the increase in absorbance is far from the 7000 times found from the specific surface gain measurements. This is still fully understood. In the case of APTS experiment, the limitation may be due to the difficulty in generating confluent monolayers within the pores.
89
The greater interest of porous sensor might reside in the potential benefit of high surface area for adsorption phenomenon or functionalization of surfaces. It was shown that chalcogenide glasses have a hydrophobic surface that preferentially interact with non- polar species and results in increasing sensitivity of organic molecules in aqueous solution [57]. It can then be expected that a higher surface area would enhance this phenomenon and somewhat help concentrate non-polar species in the region where the evanescent wave is the strongest. This could greatly improve detection limits of analytes in aqueous solutions where the water signal is usually overwhelming.
Another potential application resides in the fact that chalcogenide glasses can be functionalized with active surface coatings that lead to highly selective biosensors by forming either a silane molecule monolayer or a thiol monolayers attached to a gold- coated surface [58-59]. These layers act as anchor for molecules with specific recognition sites. Linker molecules such as antibodies or enzymes can provide active biological selectivity in capturing specific target analytes. The efficiency of such sensors should be improved significantly with increasing surface area and a porous layer could therefore result in considerable sensitivity improvement.
This novel optical element may also be used for trapping and immobilizing macromolecules within the pores. Control of the pores size should then allow size selectivity for the detection of large molecules.
90
2.9 Conclusion
In this chapter, 62.5GeS2-12.5Sb2S3-25CsCl glass ceramics composition with nanoporous surfaces were synthesized by surface etching of CsCl nanocrystals. The pore morphology can be tailored by controlling the ceramisation time and temperature as well as the etching time. The etching agent is the “piranha” solution made of a mix of sulphuric acid and oxygen peroxide. It is demonstrated that porous glass ceramic can be used as optical elements in an ATR configuration. These infrared transparent porous elements have many potentials for sensing application due to their high surface area. Experiments performed with different chemical substances show that the sensitivity is enhanced by a four to fivefold factor in the case of the porous sample compared to the smooth sample.
Total internal reflections are not affected negatively and it is therefore expected that such material could also be applied to the fiber configuration.
91
CHAPTER 3 : PHOTOINDUCED EFFECTS IN THE SYSTEM GES2-SB2S3-CSCL
3.1 Introduction
This chapter deals with photo-induced effects and mainly photodarkening effects in the
GeS2-Sb2S3 system and special attention will be paid to the influence of addition of alkali halide (cesium chloride) on the phodarkening phenomenon.
Photo-induced changes of structure and properties have been intensively studied since the discovery of an optical phase-change phenomenon by Ovshinsky et al. in 1971 [60] because of their potential applications in optoelectronics (manufacture of optical components, optical storage, optical waveguides, optical gratings, micro-lenses, etc). In spite of the fact that large effort was given to this subject and several models for the description of individual photoinduced changes have been proposed, the mechanism of many photo-induced processes in chalcogenide glasses is still unclear and has been the subject of many debates. Explanations range from the production of defects created by the optically induced breaking of bonds [61], the tunnelling of twofold coordinated chalcogen atoms in double-well potentials [62] and changes in the overlap of neighboring non-bonding chalcogen electronic levels due to subtle rearrangement of the chalcogen atoms [63].
Photo-induced effects typically include photodarkening (red-shift), photobleaching, photoexpansion, photo-crystallization, photoanisotropy and photorelaxation.
92
Photosensitive processes only occur in amorphous samples [64] and glasses are irradiated with sub band gap light therefore it is assumed that these phenomena inside the glasses are athermal if the energy level is moderate [65].
Most photo-induced phenomena have been studied with glasses from As-S, As-Se, As-S-
Se, Ge-S or Ge-Se systems by Andriesh, Fritzsche, Elliott, Tanaka, Lucas, etc [66-70].
3.2 Electronic structure of chalcogenide glasses
Electric conductivity of semiconductors at room temperature ranges from 10-9 to 102 Ω-
1.cm-1 [14]. Chalcogenide glasses have electric conductivity in this range and accordingly they can be regarded as amorphous semiconductors [71-73].
Their energy band gap is sensitive to composition and varies from 0.7 eV in GeTe [71] to
3.24 in GeS2 [72]. Chalcogenide-rich glasses appear to share a common electronic band structure. All chalcogen atoms have six valence electrons in an s2p4 configuration (sulfur
: 3s2p4, selenium : 4s2p4, tellurium : 5s2p4). Tellurium’s lone pairs are therefore higher in energy than sulphur’s lone pairs. As a consequence, the gap decreases in the sequence of sulphur, selenium and tellurium-based composition. The s shell and one p shell are completely full. The full p shell is known as a lone-pair (LP) orbital. The other two half- filled p shells participate in the formation of covalent bonds, so the chalcogen atoms are normally twofold coordinated. The valence band of chalcogenide glasses consists of states from the p bonding (σ) and LP orbitals. The LP electrons have higher energy than the bonding electrons, so the full LP band forms the top of the valence band. The conduction band is formed by the antibonding band (σ*). The LP band falls between the
93
σ and σ* bands, so the band gap is about half of the bonding-antibonding splitting energy
[74]. Figure 3.1 shows an example of band structure.
Figure 3.1 : Bonding in selenium (atomic states, molecular states and broadening of states into bands in the solid). Kastner [74]
3.3 Previous studies on photodarkening
The photodarkening process involves a shift of the optical absorption edge to lower energy and an increase in the band tail absorption. The photodarkening can be induced with above-bandgap or below-bandgap light, as long as the light has sufficient energy to excite electrons from the LP (lone-pair) band [75]. Some structural studies have demonstrated that the amorphous structure becomes more disordered with illumination.
However it is difficult to identify explicitly the structural change in amorphous phases and the mechanism is still not elucidated.
94
Measurements of photodarkening in various compositions of chalcogenide glass bear out the significance of the chalcogen atom in determining the darkening behavior [76]. It is well known that As2Se3 and As2S3 undergo photo-induced phenomena such as photodarkening [66]. With the addition of 1 and 5 at.% copper to glassy As2Se3 and
As2S3, respectively, the photodarkening is essentially destroyed [77]. In this case, it is believed that the metal atom add new levels above the lone-pair band [78]. Therefore, it seems that the LP orbitals of chalcogen atoms are involved in the photo-induced phenomena.
Photodarkening can be recovered with annealing near glass transition temperature whereas what we call “photobleaching” can not. Photobleaching, only observed in thin films, is an irreversible change in optical properties but it appears to be a different process compared to photodarkening [79-81].
The index of refraction of the glass also changes with photodarkening [82]. The refractive index change associated with the photodarkening is expected from the Kramers-Kronig relations. The associated index change may prove useful for the fabrication of optical structures such as optical waveguides in bulk glasses and in thin films.
The stability of photodarkened films or bulks can also be influenced by the average coordination number
84].
< r >= ∑ xi N i with x the atomic fraction of element i and N, the atomic coordination number (N=2 in the case of a chalcogen atom). In a glass with
95
exists the so-called zero-frequency vibrational modes (Phillips and Thorpe). At
Figure 3.2 : Schematic composition dependence of the photodarkening effect as a function of the average coordination number
Photoinduced effects in the Ge-S-Sb system have been previously studied by M. Frumar and M. Vclek [86]. Most of their studies concern fully reversible photodarkening effects on thin films. They also demonstrated the analogous reversible photodarkening occurred in the bulk sample too, but the change of transmission of bulks is very small compared to thin films. They assume that the origin of photoinduced effects is the same in both glasses and thin layers, that is the redistribution of chemical bonds. They also tried to characterize the structural change under exposure using Raman spectroscopy [87].
96
The maximum photodarkened effect performed on several compositions GexSb40-xS60 was found to be for x =25-30 at.% corresponding to an average coordination number
(Figure 3.3).
Figure 3.3 : Compositional dependences of optically induced ΔT change of thin layers in the GexSb40-xS60 compositions
3.4 Photodarkening of chalcogenide and chalco-halide glasses
3.4.1 Set-up
Two glasses with compositions 50GeS2-50Sb2S3 and 50GeS2-50Sb2S3+10%CsCl have been selected for this study. Samples were irradiated with light from a Ti:Sapphire laser pumped by a doubled-frequency solid state Nd:YAG laser emitting at 532 nm. At the output of the Ti:Sapphire the light can be tuned between 740 nm and 850 nm. The experimental conditions are indicated in Figure 3.4.
Samples were prepared by the conventional bulk-melting technique described in Chapter
2 in a 10mm diameter silica tube. The samples were polished and the thickness was typically 0.7 mm.
97
Divergent Lens
Millenia Vs Sample Solid-state, frequency- doubled laser Tsunami 532 nm Ti:Sapphire laser CW, 5 W 800 nm Mode-locked, 82 MHz, 80 fs, 10 nJ Tunable 740-850nm
Figure 3.4 : Experimental set-up
3.4.2 Results
The 50GeS2-50Sb2S3 and the 50GeS2-50Sb2S3+10%CsCl glasses were chosen to perform the experiments because it was shown that photodarkening effects would be enhanced in this glass compared to the 20GeS2-80Sb2S3 glass composition [86]. Glasses richer in
GeS2 would exhibit more important effects but the wavelength limitation of the laser does not allow us to use them. Indeed, glasses rich in GeS2 have a band gap less than 700 nm which is out of the laser range.
Optical spectra were measured for both glasses (Figure 3.5). The addition of cesium chloride leads to a shift of the band-gap toward shorter wavelengths.
98
0,6
50GeS2-50Sb2S3
0,5 50GeS -50Sb S + 10%CsCl 2 2 3
0,4
0,3
% Transmission 0,2
0,1
0 550 600 650 700 750 800 nm
Figure 3.5 : Band-gap comparison for the 50GeS2-50Sb2S3 and the 50GeS2-50Sb2S3+10%CsCl (thickness : 0.7 mm)
Figure 3.6 shows the result of photodarkening experiments performed on a GeSe2-Sb2S3 glass and the same glass containing 10% CsCl. The irradiation wavelength, typically in the Urbach region, was adjusted for each glass in order to account for the different band- gap and retain the same quantum efficiency. The 50GeS2-50Sb2S3 and the 50GeS2-
50Sb2S3 + 10% CsCl glass compositions were respectively irradiated at 760 nm and 740 nm for 3 hours. These wavelengths correspond to the so called band tail of the absorption edge. The power density was 2W/cm2 for both glasses.
The photodarkening effect for the 50GeS2-50Sb2S3 glass composition is very small but detectable whereas the effect for the 50GeS2-50Sb2S3 + 10% CsCl glass composition is
99
negligeable (Figure 3.6). Therefore, it appears that the introduction of electronegative elements such as chlorine ions Cl- in the glass structure has severely detrimental consequences on the photostructural changes.
0,8 50GeS2-50Sb2S3 0,7 760 nm 3h 400mW 50GeS2-50Sb2S3+10%CsCl 740 nm 3h 400mW 0,6
0,5
0,4
% Transmission % 0,3
0,2
0,1
0 600 620 640 660 680 700 720 740 nm
Figure 3.6 : Comparison of photodarkening in a GeS2-Sb2S3 glass irradiated at 760nm and a GeS2-Sb2S3-10%CsCl glass irradiated at 740nm. Both glasses were irradiated for 3 hours at 400mW (Power density 2W/cm2)
100
3.5 Discussion and conclusion
It is shown that the GeS2-Sb2S3 glass undergoes photodarkening while the glass containing CsCl does not display any changes. The presence of CsCl in the glass structure appears to suppress the photostructural effect entirely. It is believed that Cs remains a Cs+ ion while Cl is covalently bonded to Ge and Sb and acts as a non-bridging atom. This theory is confirmed by Raman spectroscopy (Chapter 2). This suggests that the presence of an electronegative atom within the covalent network might affect the lone pair states at the top of the valence band sufficiently to prevent the photostructural process.
101
PART 2 : MECHANICAL AND VISCOELASTIC PROPERTIES OF THE TAS
(TE-AS-SE) GLASS
CHAPTER 4 : MECHANICAL PROPERTIES OF THE TAS (TE2AS3SE5) GLASS
4.1 Introduction
The Te2As3Se5 (TAS) fiber composition is currently used in our laboratory as a bio- sensor in several fields of applications including medical and environmental applications
[1][2][4]. One of these applications is shown below (Figure 4.1). In this case biochemical changes in human lung cells can be detected using a TAS fiber coupled with an infrared spectrophotometer. A monolayer of these cells is able to attach to the surface of the TAS fiber. Changes in human lung cells signatures can be monitored upon TritonX-100 exposure and we clearly see the decrease of absorption bonds indicating the damage to cells membrane [3]. 5.0 4.5 Live cells in NaCl a) 4.0 No Triton 3.5 3.0 2.5 2.0 1.5 Absorbance (u.a.) 1.0 0.5 0.0 3000 2980 2960 2940 2920 2900 2880 2860 2840 2820 2800 Wavenumber (cm-1)
5.0 Asym CH2 4.5 Live cells in NaCl b) Sym CH2 4.0 1 mM Triton 3.5 3.0 Sym Asym 2.5 CH3 CH3 2.0
Absorbance Absorbance 1.5 1.0 0.5 0.0 3000 2980 2960 2940 2920 2900 2880 2860 2840 2820 2800 -1 Wavenumber (cm )
Figure 4.1 : One of the several applications of the TAS fiber [3]
102
The mechanical properties of TAS glasses have been little studied so far [5][88].
Preliminary experiments on TAS fibers have shown that ageing treatments in air below
Tg induce a dramatic decrease of the tensile strength of the fiber [88].
This chapter will focus on different effects influencing the mechanical properties of the
TAS fibers and particularly, we will study the influence of environment (ageing in air, under vacuum and under static stress), the effect of an annealing treatment and the coordination number,
Glasses from the Te-As-Se ternary system (Figure 4.2) are of great interest because of their large transparency in the infrared range as well as for their thermo-mechanical properties which allow the glass to be drawn into fiber [89-91]. The glass composition
Te2As3Se5 was chosen because of its stability against devitrification.
103
Te2As3Se5
Figure 4.2 : Te-As-Se ternary diagram. Zone I : glasses stable against devitrification, Zone II : glasses which can easily crystallize
So far, no detailed study on the Te2As3Se5 (TAS) network structure has been reported, only some proposed structural models have been suggested which suppose a two- dimensional glass network by the cross-linkages between Se-Te-As chains through As-As bonds [92-93]. A Raman study involving several compositions in the As-Se and Te-As-
Se systems is still in progress to get information about the structure of the Te2As3Se5 glass composition.
4.2 TAS synthesis
TAS synthesis like other chalcogenide glasses is made under vacuum to avoid oxygen contamination. The set-up (Figure 4.3) allows the purification of the three chemical
104
elements : selenium, arsenic and tellurium. This set-up is first washed with a strong acid
(1/3 HNO3 - 2/3 HCl) then with distilled water and finally it is dried under vacuum.
Tellurium, arsenic and selenium are bought with high purity but they easily oxidize in contact with air. Oxide layers on the selenium and arsenic are removed by heating these chemical elements at 240°C and 290°C respectively because the vapor pressure associated with the corresponding oxide are higher than that of the selenium or arsenic
[94-95]. The tellurium is purified by a chemical reaction with hydrobromic acid [96].
After all steps of purification, arsenic and tellurium move into the filter tube. After sealing, the tube is put into a furnace to undergo distillation. Chemical elements are distilled at 900°C in order to eliminate all impurities, especially carbon impurities, which have a boiling temperature higher than that of the glass.
After distillation, the reactional tube is sealed and put in a rocking furnace during 12 hours at 650°C to homogenize the melt. The furnace is then put in the vertical position and the temperature is decreased to 500°C to increase the viscosity and to avoid the formation of vapor pressure above the melt and the explosion of the tube.
The melt is quenched into water from 500°C to room temperature in few seconds. The quenching allows the glass formation.
The glass rod is annealed near the glass transition temperature, Tg, to relax local mechanical stresses due to the quenching.
105
Vacuum
Trap Te Sealing As Furnace Dewar with Reactional Se liquid Tube nitrogen
SiO2 tube
Figure 4.3 : Set-up for TAS synthesis
Because the TAS glass is dark, the homogeneity of the rod and the absence of crystallization and bubbles, which are critical for the drawing process, are controlled by infrared imagery using a thermal camera ThermaCAM E300 from FLIR systems working in the 8-12µm range (Figure 4.4).
Schrinkage due to the quenching
Figure 4.4 : Infrared imagery of a TAS glass rod
106
4.3 Drawing process
Elaboration of optical fibers is made possible from a drawing tower which was entirely conceived in our laboratory (Figure 4.5). The glass rod is fixed at the top of the tower in a chamber with an helium flow. This inert gas permits to eliminate all impurities in the chamber as well as to avoid the preform surface crystallization due to remaining water.
Moreover, helium gas has a good thermal conductivity which allows the temperature homogenization around the glass rod during the drawing process.
The drop formation constitutes the first step of the process. This is made possible with a tubular furnace (Figure 4.5). In the case of the TAS glass, the furnace is heated at about
290°C. The glass drop, because of its weight, leads to a glassy wire or fiber. The fiber is drawn out up to the drum in rotation where it is fixed. The fiber diameter is a function of the drum speed and the preform descent speed. Captors allow the control of the fiber diameter.
The typical parameters for the TAS glass drawing process are listed in the following table.
Table 4.1 : Characteristics of the TAS drawing process
Helium gas output 2 L/min Preform diameter 10 mm Preform length 10 cm Fiber diameter 400 µm Tension 15 to 20g Preform descent speed 2 mm/min Drum rotation speed 1.23m/min Furnace temperature 290°C
107
Preform descent
10 cm Helium flow TAS glass rod Ring shaped furnace Diameter measure 1 cm
Pulley coupled with a tensiometer Drum D
Figure 4.5 : Drawing tower
4.3.1 Drawing process inducing residual stresses
Drawing process induces high residual stresses inside the fiber which is attributed to the applied pulling force due to the drum speed. The residual stress increases as the drawing speed increases as well. Y. Park et al reports that the mean axial stress in the fiber
Corning SMF-28 (silica fiber) ranges from 4.5 MPa for a 18m/min drawing speed to 6.5
MPa for a 50 m/min drawing speed [97]. In addition to mechanical stresses, the drawing process imposes large thermal stresses onto fibers while they are being cooled through
Tg. Glass fibers undergo stringent thermal quenching from the drawing temperature
108
(290°C in the case of TAS fibers) to room temperature very quickly as shown by the thermal imagery taken during a TAS drawing procedure (Figure 4.6).
The quenching generates stresses due to the local variation of the viscosity with temperature and the gradient of dilatation. During early cooling stages, when the external surface of the fiber cools and starts to shrink, the hot core is still molten and free to contract. However, as the internal core cools, local thermal contraction is constrained by the already-rigid external layers. This results in a typical state of stress distribution with tension in the core balanced by compression in the outer layers (Figure 4.7). These stresses are at equilibrium since no external effort is needed to maintain this state of constraints. The sum of stresses on a fiber section verifies : ∫σ (r,θ )rdrdθ = 0 S
It is believed that the measured stress profile of an optical fiber is quite different from that of the associated fiber preform [98].
Tubular furnace (290°C)
Scale : 70 cm
Figure 4.7 : Thermal stress distribution TAS fiber along the cross section of a fiber[99] (room temperature)
Figure 4.6 : Thermal imagery of the drawing process
109
4.3.2 Fiber annealing during the drawing process
Some tests were made to anneal the fiber online and therefore to relax the residual stresses induced during the drawing process. This is to evaluate the effect of an online thermal treatment on the fibers mechanical properties. The thermal treatment process was made possible using heating wires enrolled on a 1 meter length silica tube
(Figure 4.8). The drum rotation speed is 1.23m/min, therefore the thermal treatment time for the TAS fiber is about 50 seconds.
Heating wires were connected to an electric power. By varying the rolling up of the wire and the power, it was easy to adapt the temperature to the wanted annealing temperature. We can Figure 4.8 : Fiber distinguish two different heating parts. The first one corresponds to annealing set-up a temperature near the Te2As3Se5 glass transition temperature
(137°C) and the second one corresponds to a progressive decrease from 130°C to room temperature.
4.4 Effect of annealing on the glass structure
A glass is by definition a frozen liquid and is therefore in a meta-stable state. The cooling rate or the quenching needed to the glass formation is fast enough not to allow the atoms to rearrange and to form a crystal. A crystal corresponds to a stable state. Glasses that are quenched rapidly don’t have an homogeneous structure. This induces residual stresses in the glass (Figure 4.9) which can be reduced by annealing the glass near its glass transition
110
temperature, Tg. Annealing the glass near Tg allows for slow rearrangement of the atoms which leads to a decrease of entropy or the specific volume. The green arrow on Figure
4.10 simulates the effect of annealing.
Figure 4.9 : Stress in the glass
Figure 4.10 : Effect of annealing on the specific volume
4.5 Physical properties
4.5.1 Thermal properties
The DSC curve was obtained for a 5 mg batch with a DSC 2010 from TA Instruments.
The heating rate is 10°C/min. The Te2As3Se5 (TAS) glass transition temperature, Tg, is
137°C. No crystallization peak is observed (Figure 4.11).
111
Tg = 137°C
Figure 4.11 : DSC curve for the TAS glass
4.5.2 Optical properties
4.5.2.1 Bulk
TAS bulks have a transmission percentage of 60% between 2 and 12 µm (Figure 4.12).
70 60 50 40 30 20 Transmission (%) 10 0 0 2 4 6 8 10 12 14 16 18 20 Longueur d'onde (µm)
Figure 4.12 : Optical transmission for the TAS glass (thickness : 2mm)
112
4.5.2.2 Refractive index
The refractive index can be determined from the maximum of transmission percentage,
T0.
2 (n −1) 1− T0 From the relation 2 = we calculate the refractive index to be close to 2.8. (n +1) 1+ T0
4.5.2.3 Fiber
4.5.2.3.1 Set-up and principle of attenuation measurement
Attenuation of the light inside a fiber is due to several phenomena :
- Light scattering
- Absorption due to impurities
- Absorption due to intrinsic electronic transitions
- Multiphonon cut-off absorption of the matrix
The recording of attenuation spectrum is based on the “cut back” method [100]. The set- up consists of an FTIR (Fourier Transformed InfraRed) spectrophotometer (Bruker 22) coupled with a Mercury-Cadmium-Tellurium (MCT) detector. Measurements are taken with the same fiber but with different lengths, L1 and L2 (L1>L2). Attenuation α is measured in dB/m and is calculated from the following equation :
10 I α = Log 2 L1 − L2 I1
α is recording in the transparency region from 2 to 12 µm.
113
4.5.2.3.2 Attenuation spectrum
Figure 4.13 represents the attenuation spectrum of a mono index TAS fiber with a 400µm diameter and a difference of length of 1.26 m.
The spectrum shows different absorption bands (Table 4.2) even if the glass has been purified.
Table 4.2 :Assignment of the absorption bands due to impurities [48]
Wavelength (µm) Impurities
6.3 H2O
4.2 CO2
2.8 O-H
3.5 and 4.6 Se-H
The minimum of optical losses is situated between 6.5 and 9.5µm and are less than 1 dB/m.
Figure 4.13 : Attenuation spectrum for a TAS fiber
114
4.5.3 Density
The glass density is measured using Archimedean’s displacement technique in carbon tetrachloride, CCl4. This method consists in weighing the sample into the air, mair and then in CCl4, mCCl4. The density, ρ, is calculated according to the equation :
mair ρ = × ρ with ρ , the density of CCl4 m − m CCl4 CCl4 air CCl4
-3 The Te2As3Se5 (TAS) has at room temperature a density of 4.80 ± 0.01 g.cm .
4.5.4 Viscosity
In the range 106-1010 Poises, the viscosity is measured in compression from a parallel plates viscometer (Rheotronic, Theta Industries, inc.).
The plot, log η = f (Tg/T) for the TAS shows a deviation from the Arrhenius law,
B logη = A + , simulated by a black line in Figure 4.14, therefore, the TAS glass is T
“fragile” according to Angell [23].
115
Arrhenius law
Measured with
a heating rate of 2°C/min
(Poise) viscosity Log
Tg/T (°K)
Figure 4.14 : Curve log η = f(Tg/T) of the TAS glass
4.6 Mechanical properties of TAS bulks
4.6.1 Elastic moduli
We generally find three main elastic moduli, characteristic of the mechanical properties of one given material including glasses :
- The Young’s modulus, E
- The Poisson’s ratio, ν
- The shear modulus, G
Glasses in general, including the TAS glass, have an elastic and brittle behavior under the glass transition temperature, Tg. It means that when a glass test piece is tested in tension or in bending, its deformation is perfectly reversible until it breaks when the stress is too high.
Glasses can show different behaviors toward applied stress (Figure 4.15) :
116
- Glass type A has a high strength and very little deformation prior to failure.
- Glass type B has intermediate strength and some deformation prior to failure.
- Glass type C has low strength and bends easily before failure.
Figure 4.15 : Different behaviors for glasses
The Young’s modulus, E, is defined as the slope of the line σ=f(ε) in the elastic domain.
For example, the glasses A, B and C have 3 different Young’s modulus. Glass A has the highest one.
F The stress, σ, is equal to σ = with F, the force applied in the longitudinal direction and S
S, the surface of the material submitted to the force. The corresponding strain is
Δl ε = with L, the initial length of the piece testing and Δl, the length expansion due to L the applied force (Figure 4.16). E is expressed in Pascal (Pa). [21]
Δl S
Figure 4.16 : Young’s modulus, E
117
When the length expansion Δl occurs in the case of a longitudinal stress, a breadth contraction, Δd occurs as well in the perpendicular direction. The Poisson’s ratio, ν, is
Δd equal to ν = − D . ν has no unit [21]. Δl L
Δd D
Figure 4.17 : Poisson’s ratio, ν
In the case of a shear stress, the material is submitted to an angular distortion, θ, which is
F proportional to the applied force, F. The shear modulus, G, is defined as : G = S with θ
S, the surface area where the force, F, is applied. G is expressed in Pascal (Pa).
S
Figure 4.18 : Shear modulus, G
These three elastic moduli are linked by the equation : E = 2(1+ν )× G
118
4.6.1.1 Ultrasonic echography method
We usually measure the Young’s modulus, E, and Poisson’s ratio, ν, by means of the ultrasonic echography method on a simple bulk. This method deals with the measurements of acoustic longitudinal and transversal waves velocities, VL and VT into the material. When a potential difference is applied to a piezoelectric transductor, a wave is generated. This wave propagates into the material and is reflected by the opposite surface according to the Snell-Descartes laws.
The time interval measurement between two successive echos permits to determine the
2e wave propagation velocities from the relation V (m / s) = where e represents the Δt thickness of the sample and Δt, the time interval between two echos.
In an isotropic material like the glass, the wave propagation velocity allows us to determine the elastic moduli (Young’s modulus: E, the shear modulus: G and their combination gives the Poisson ratio: ν) inside the material from the following equations
[101] :
⎛ ⎞ ⎜ ⎟ 2 2 ⎜ 3VL − 4VT ⎟ 2 E E(Pa) = ρ⎜ ⎟ G(Pa) = ρVT ν = −1 ⎛V 2 ⎞ 2G ⎜ ⎜ L ⎟ −1 ⎟ ⎜ ⎜ 2 ⎟ ⎟ ⎝ ⎝VT ⎠ ⎠
ρ is the density (in g.cm-3)
VL and VT are respectively the longitudinal and transversal wave propagation velocities.
119
4.6.1.2 Results
The Young’s modulus, E, is equal to 16.9 GPa for the Te2As3Se5 glass (
17.2 GPa for the Te2As4Se4 glass (
The Poisson’s ratio, ν, is equal to 0.29 for both
4.6.2 Hardness
Hardness measurements correspond to the resistance to localized plastic deformation.
The methods to determine glass hardness (Vickers, Brinell, Knoop, etc) are based on the application of a constant load on an indenter (which can have different geometries : pyramidal, spherical, etc) at the surface of a well-polished material. The determination of the hardness value is made possible by the dimensions of the mark made by the indenter which is submitted to controlled loads and velocities.
The TAS hardness was realized by a Vickers type indenter durometer (Matsuzawa
VMT.75). The load (2.943 Newtons) was applied during 15 seconds.
The indenter employed in the Vickers test is a square-based pyramid whose opposite sides meet at the apex at an angle of 136º (Figure 4.19). The diamond is pressed into the surface of the material and the size of the impression is measured with the aid of a calibrated microscope or by Atomic Force Microscopy (AFM) (Figure 4.20). The Vickers number (HV) is calculated using the following equation : HV = 1.854(F/d2), with F being the applied load (measured in kilograms-force) and d corresponds to the arithmetic mean of the two diagonals, d1 and d2 (mm).
The hardness for the TAS glass is equal to 1.4 GPa.
120
Figure 4.19 : Vickers indenter
Figure 4.20 : AFM of the mark of a Vickers indenter on the TAS glass
4.6.3 Toughness
Toughness is a property which describes the ability of a material containing a crack to resist to fracture. It is denoted KIc. The subscript “I” denotes mode I which is the crack opening in ordinary stress (Figure 4.21) and the subscript “c” stands for critical. Modes II and III correspond respectively to shear opening and tear opening. The toughness for glasses is generally in the range 0.1-1 MPa..m
121
Mode I Mode II Mode III
Figure 4.21 : Different modes for crack opening
Indentation experiment is an easy method to have an approximation of the toughness,
Kind. The Kind determination needs a sample with a well polished surface. This surface is indented, in general with a Vickers indenter. The knowledge of the Young’s modulus : E, the indenter load : P, the hardness H and the crack length, 2c which extends radially from the edges of the mark (Figure 4.22) allows us to calculate KIc. In the literature, we can find different equations to determine KIc, for example :
3 2 / 3 2 K ind = 0.01(E / H ) (P / c ) ± 0.0025 if c/a>2.5 [102]
2c
2a
Figure 4.22 : Vickers mark seen under microscope
The toughness for the TAS is 0.18 MPa..m
122
4.7 Mechanical properties of TAS fibers
4.7.1 Statistical analysis of fracture
Griffith is the first one who supposed that the glass has a multitude of microscopic cracks. Typically, the glass surface is covered by more than 10000 defects or scratches per cm2 which are less than 1/100 mm [103].
These surface defects are responsible for the critical breaking of the fiber. These defects, nominally identical, can have different natures, sizes, geometries, orientations and localizations. This disparity is at the origin of the wide variation of the tensile strength.
The Weibull statistical theory [104] defines the breaking probability Pf. Pf represents the fraction of the fibers which will break under the stress σ according to the equation :
⎡ ⎛σ −σ ⎞ ⎤ P = 1− exp − ⎜ u ⎟dV f ⎢ ∫⎜ ⎟ ⎥ ⎣ V ⎝ σ 0 ⎠ ⎦
With σ, the applied stress
σu, the stress under that the breaking probability is zero (we usually take σu=0)
σ0 is a constant
The Weibull modulus, m, characterizes the defect size dispersion. m does not describe the geometry or the spatial arrangement of the defects. A high m value indicates that along the fiber the defects sizes are averaged. A low m value reveals that the defects found at the surface of the fiber have varying sizes, which result in different values of the tensile or bending strength.
If we consider that all the glass testing samples or the fibers have the same volume V0, then we end with the following equation :
123
m ⎡ ⎛ σ ⎞ ⎤ P (V ) = 1− exp⎢− ⎜ ⎟ ×V ⎥ with m, the Weibull modulus f 0 ⎜σ ⎟ 0 ⎣⎢ ⎝ 0 ⎠ ⎦⎥
By taking the double neperien logarithm, we have :
⎛ ⎛ 1 ⎞⎞ ⎛ σ ⎞ ln⎜ln⎜ ⎟⎟ = mln⎜ ⎟ + lnV ⎜ ⎜1− P ⎟⎟ ⎜σ ⎟ 0 ⎝ ⎝ f ⎠⎠ ⎝ 0 ⎠
⎛ ⎛ 1 ⎞⎞ Therefore, m represents the slope of the plot ln⎜ln⎜ ⎟⎟ = f ()lnσ ⎜ ⎜1− P ⎟⎟ ⎝ ⎝ f ⎠⎠
Practically, we classify all the stresses at fracture of the testing series by ascending order.
σ 1 ≤ σ 2 ≤ ...... ≤ σ i ≤ σ i+1 ≤ ...... ≤ σ N
At each stress σi, we attributes a fracture probability Pfi calculated from the equation :
i − 0.5 P = fi N
Where N is the number of samples and i represents the sample (i=1 for the lowest stress at fracture)
In some cases, the plot on the Weibull diagram is not a single straight line which corresponds to a multimodal behavior expressing the fact that fracture occurs from different families of defects.
4.7.2 Experimental determination of the crack resistance
4.7.2.1 In tension
The determination of the tensile strength σr is realised by uniaxial tension testings with a traction equipment Adamel-Lomarghy DY (Figure 4.23). It is generally common to
124
perform these tests on a bunch of fibers (typically more than 20 fibers). Fiber length is about 13 cm and fiber diameter is 400 µm. Before testing, we have to make sure that the fiber is in the axis right in the direction of applied load to valid the test.
The testing rate is 10 mm/min. The load cell can be used in the range 0-100 N, with a precision of 0.1 N on the measurement. The distance between the jaws is 10 cm for all measurements. The tensile strength is expressed in Newton (N).
Figure 4.23 : Tensile test
4.7.2.2 In two-point bending
TAS fibers during manipulations are usually used in bending due to the spectrophotometer configuration.
Therefore, the breaking mode is observed in bending, not in tension, that is why, it is important to test these fibers by means of a two-point bending test Figure 4.24 : Two-point bending test to determine the critical radius curvature.
The bending stress is determined from the equation (Appendix C) :
125
d σ = 1.198× E × max D − d
Where E is the Young’s modulus, d is the fiber diameter and D is the difference between the jaws after fracture.
4.7.3 Fractographic analysis
The fragile fracture implies a two-step procedure :
- crack or defects production
- crack propagation
As previously said, Griffith is the first Figure 4.25 : Rupture facies one who supposed that the glass has a multitude of microscopic cracks at the origin of the fracture. Typically, the glass surface is covered by more than 10000 defects or scratches which are less than 1/100 mm per cm2
[103]. The glass breaking mechanism under tensile strength is brutal since the glass is a fragile material and has no plastic behavior below its glass transition temperature. The fracture surface has different characteristic regions which allow us to obtain information about the crack propagation and in some cases, it allows us to guess the origin of fracture.
The fracture is conchoidal type [105]. Figure 4.25 shows the detailed facies of the surface after fracture and especially the source of failure. The propagation of the fracture leads to first a smooth mirror region followed by a mist and hackle regions.
The frature surface observed in the case of the TAS fibers is representative of what we can find generally on glass fibers. A SEM picture permits to visualize the TAS fracture
126
surface. During this study, only surface defects were found, no volume defects. The surface defects come from the presence of defects at the surface of the preform or are induced by the manipulation of the fibers (Figure 4.26).
Figure 4.26 : Fracture surface of a TAS fiber
4.8 Effect of environment on mechanical properties
4.8.1 Previous studies
Few studies have been reported regarding the effect of environment on the mechanical properties of the TAS fibers.
K. Michel [106] showed that both the mechanical and chemical polishings of the preform, prior to the drawing process, have an influence on the TAS fiber tensile strength. Indeed the tensile strength is increased by about 20% in the case of the preform polishing. This demonstrates the importance of removing the defects, at the surface of the preform, which are at the origin of the critical fracture. She also showed that avoiding friction at the surface of the fiber during the drawing process, by removing all the pulleys needed to guide the fiber on the drum, leads to the same percentage of increased tensile strength.
127
Concerning the effect of environment, the prolonged exposure to water is critical for the mechanical properties of the fibers. A decreased of 57% of the original tensile strength was observed after two hours spent in water at room temperature. Therefore, the effect of environment on the fibers mechanical properties is determinant.
For the following study, several drawings were performed. A series of fibers, freshly drawn, were put in different environments (in air, under vacuum and under loading) and tested in tension after different periods of time.
4.8.2 Air
The TAS fibers of 13cm length and 400µm diameter, freshly drawn, were stored in air for different periods of time : 15 days, 1 month and 2 months and then tested in tension.
Figure 4.27 : TAS fibers ageing in air
4.8.2.1 Results
The average tensile strengths σr, from four different drawings, just after the drawing process at t0 and after different ageing periods, are presented in the following table. Each test was performed on a series of 20 fibers. The dispersion, d, which represents the range between the lowest and the highest tensile strength value is also presented.
128
Table 4.3 : Tensile strength of TAS fibers ageing in air for 4 drawing processes Drawing N°1 Drawing N°2 Drawing N°3 Drawing N°4 17.7 MPa 12.7 MPa 9.3 MPa 8.7 MPa t0 d : 20.5MPa d : 18.3MPa d : 12.9MPa d : 9.5MPa 7.4 MPa t+15days - - - d : 9.5MPa 7.0 MPa 9.9 MPa 7.2 MPa 5.6 MPa t+1 month d : 5.6MPa d : 13.1MPa d : 9.1MPa d : 6.4MPa 3.8 MPa t+2 months - - - d : 6.2MPa
From a general observation, as the time spent in air increases, the mechanical properties and especially the tensile strength of the TAS fibers decreases dramatically.
The wide range of tensile strengths σr, just after the drawing (t0), can be explained by the step of purification during the TAS synthesis which could be at the origin of remaining impurities into the structure. Typically the fibers with a tensile strength, σr<12MPa (t0), were obtained according to the common synthesis method explained in 4.2. Recently a new method of purification based on the addition of Al and TeCl4 to the glass composition followed by dynamic TAS distillation permits to obtain lower optical losses compared to the common synthesis (Figure 4.28). The fibers obtained this way have typically a tensile strength, σr>12MPa (t0).
129
14 (a) 12 (b) ) 10 dB/m ( 8
6
4 A ttenuation
2
0 2345678910 Wavelengths (µm)
Figure 4.28 : Attenuation spectra for a TAS fiber : common synthesis (a) synthesis based on dynamic distillation (b)
4.8.2.2 Surface analysis
Because the surface defects are in general at the origin of the fracture [103], it is important to study the fibers surface in order to correlate the loss of mechanical properties to the surface.
SEM analysis was performed on both freshly drawn fibers and aged-fibers stored in air.
After the drawing process, the fibers surface is smooth and presents few superficial defects or scratches (Figure 4.29-a). The scratches on surface were used to focus the SEM picture (Figure 4.29-a).
After 15 days in air, this surface is very different from the freshly drawn fibers and presents some defects or “bumps” (Figure 4.29-b-c). These defects become more and more important after 1 month. AFM (Atomic Force Microscopy) was performed on a 1 week-old fiber to determine the roughness (Figure 4.30). The defect depth is less than 3 nm. These defects which may result from a reaction between the TAS fiber and water or
130
O2 from atmosphere increase the stress concentration at the surface of the fibers and can be at the origin of the loss of mechanical properties.
(a) (b) (c)
Figure 4.29 : TAS surface after the drawing process (a), after15 days (b) and after 1 month ageing in air (c)
Figure 4.30 : AFM performed on a 1 week-old TAS fiber ageing in air
4.8.3 Vacuum
Freshly drawn fibers were put in a pyrex tube and put under vacuum overnight and then sealed as shown on Figure 4.31. This is to study the effect of environment on the fibers mechanical properties. The fibers were then tested in tension after 15 days, 1 month or 2 months.
131
Figure 4.31 : TAS fibers ageing under vacuum
4.8.3.1 Results
Like the fibers ageing in air, the tensile strength of vacuumed fibers is less than that of the freshly drawn fibers (Table 4.4).
Table 4.4 : Tensile strength of fibers put under vacuum for the different drawing processes
Drawing N°1 Drawing N°4 Drawing N°5 Drawing N°6
17.7MPa 8.7MPa 16.1MPa 16.1MPa t0 d : 20.5MPa d : 9.5MPa d : 21.9MPa d : 16.1MPa t+15days - - - -
8.3MPa 6.4MPa t+1 month - - d : 5.0MPa d : 6.0MPa 6.2MPa 13.5MPa 8.7MPa t+2 months - d : 5.6MPa d : 5.8MPa d : 6.2MPa
We can notice that the tensile strength of the fibers ageing in air is less than that of the fibers ageing under vacuum (Table 4.5).
132
Table 4.5 : Comparison of tensile strength between the fibers ageing in air and under
vacuum Air Vacuum 8.7MPa t0 d : 9.5MPa 5.6MPa 6.4MPa t+1 month d : 6.4MPa d : 6.0MPa 3.8MPa 6.2MPa t+2 months d : 6.2MPa d : 5.6MPa
4.8.3.2 Surface analysis
The surface of the fibers ageing under vacuum, after 1 month, is similar to the fibers freshly drawn. The defects or the “bubbles” observed in the case of the fibers ageing in air are not detectable (Figure 4.32-b).
(a) (b)
Figure 4.32 : TAS surface after the drawing process (a) and after 1 month ageing under vacuum (b)
In this case, the loss of mechanical properties can not be associated with the creation of surface defects but it can be explained by a structural relaxation as a function of time. At
133
room temperature, it is already shown that the TAS glass ageing in air undergoes structural relaxation (Figure 4.33) [107].
Figure 4.33 : DSC traces of Te2As3Se5 (TAS) fibers ageing at room temperature for up to 5 years
We have now to understand how this structural relaxation affects negatively the tensile strength.
Glasses are sometimes compared to amorphous polymers. Indeed, amorphous selenium has been widely studied as a model in polymer science because it is made of linear chain without cross-linking. Therefore it constitutes the simplest possible “polymer” [108]. We can take this example to try to explain the mechanism at the origin of this decrease of tensile strength for the TAS fibers ageing in air or under vacuum.
The TAS glass is made of covalent bonds as well as weaker bonds called Van Der Walls
(VDW) bonds. During the drawing process, the chains constituting the fiber tend to align in the pulling direction (Figure 4.34-a). As the TAS fibers relax, the covalent and the
VDW bonds rearrange in order to minimize energy (Figure 4.34-b). As a consequence, the chains constituting the fiber have a random distribution as contrary to a freshly drawn fiber (Figure 4.34-a). When the fibers are tested in tension after a certain period of time,
134
the rearrangement of the chains is unfavorable to the direction of stress (tension) and the fiber breaks.
VDW bonds
Covalent bonds (b) (a) Direction of stress
Figure 4.34 : Structure of the fiber after drawing (a) after relaxation (b) / Analogy with polymers
4.8.4 Under loading
In order to corroborate our assumption about structural relaxation that affects negatively the tensile strength, we impose a static stress to the fiber to prevent the chains constituting the TAS fiber from rearrangement.
4.8.4.1 Experiments
Fibers freshly drawn were attached to a 250g (2.45 N) load. The tensile strength, at t=0,
σr0 permits to determine the critical load they can support. The fibers were then tested in tension after 15 days, 1 month or 2 months.
135
Figure 4.35 : TAS fibers under loading
4.8.4.2 Results
Globally, for several ageing times, fibers appear stronger under static stress than the as- received fibers. Their tensile strength after 1 month or 2 months is higher than that at t0
(Table 4.6).
Table 4.6 : Tensile strength of fibers ageing under static stress for the different drawing processes
Drawing Drawing Drawing Drawing Drawing Drawing
N°2 N°4 N°5 N°7 N°8 N°9 12.7MPa 8.7MPa 16.1MPa 11.1MPa 10.3MPa 9.5MPa t0 d : 18.3MPa d : 9.5MPa d : 21.9MPa d : 18.3MPa d : 19.3MPa d : 13.3MPa 16.9MPa 14.1MPa 15.1MPa 11.9MPa 15.1MPa t+1 month - d : 11.3MPa d : 7.7MPa d : 7.4MPa d : 7.0MPa d : 7.9MPa 12.5MPa 22.9MPa t+2 months - - - - d : 16.5MPa d : 8.1MPa
Experiments performed on silicate glasses with large cracks and in double torsion (DT) loading, showed that crack closure takes place before complete unloading of the crack. In this case, the crack is considered “healed”. A possible explanation lies in the reaction of
136
water in atmosphere which can cure the defects at the surface of the fibers and increase
the strength of the fiber [109].
In the case of TAS fibers, the mechanism at the origin of the increased tensile strength
would be explained by a structural change rather than cured defects, as explained above
for silica fibers. Again, we can take the analogy with polymer to try to explain the
mechanism at the origin of the increase of tensile strength for the TAS fibers. As
previously mentioned, the TAS glass is made of covalent bonds as well as weaker bonds
called Van Der Walls (VDW) bonds. Under loading, weak VDW bonds allow the
structure to align in the direction of stress. The tension during the drawing process ranges
from 15 to 20g. The load we apply to the fiber for the static stress experiment is 250g.
Therefore, we can assume that the structure of the TAS fiber ageing under loading is
more aligned than the as-received fiber (load of 250g>>20g) (Figure 4.36). Then the
fibers are tested in tension in the direction of the covalent bonds alignment. Therefore,
the structure tends to be stronger in the direction of testing and the tensile strength is
increased. (a) (b) VDW bonds
Covalent bonds
Direction of stress
Figure 4.36 : Structure of the fiber after drawing (a) / Effect of loading on the fiber structure (b)
137
To demonstrate the change in structure (chains alignment) induced by the static stress, calorimetry technique was performed for studying the change in relaxation enthalpy.
Enthalpy measurements are obtained by taking the difference in surface area between a reference curve and the curves showing an overshoot in heat capacity, Cp. The overshoot of Cp is representative of enthalpy change on the glass network.
The change in enthalpy for fibers ageing under static stress is represented in Figure 4.37.
A TAS fiber which does not age under static stress is used as a reference state for measuring change in relaxation enthalpy. This change of enthalpy as the fiber ages under static stress is a proof of the difference in the TAS structure between the fiber ageing under static stress and a fiber free from static stress (reference).
-1,5
-1,7
-1,9
-2,1
-2,3
Enthalpy Change (J/mole) -2,5
-2,7 0 1020304050 Time Under Loading (Days) Figure 4.37 : TAS fiber enthalpy change under loading
4.8.4.3 Surface analysis
The surface of the fibers ageing under loading is different from the surface of the fibers freshly drawn (Figure 4.38). The surface after 15 days spent under loading presents less
138
defects compared to the surface of the fibers ageing in air solely. Even with the presence of these defects, the tensile strength is increased. The effect of structural change due to the applied static stress is therefore the predominant influence on the mechanical properties compared to the surface defects.
(a) (b)
Figure 4.38 : TAS surface after the drawing process (a) and after 15 days under loading (b)
4.8.5 Discussion
In summary, the average tensile strength of the fibers ageing in air or under vacuum is less than that of the freshly drawn fibers whereas that of the fibers put under static stress is higher.
Figure 4.39 presents the variation of the tensile strength compared to the freshly drawn
(t=0) tensile strength as a function of time. It summarizes some of the results obtained for several drawings. The decrease of the tensile strength for the fibers under tensile stress, at t=60 days, for the drawing 5 is not yet explained.
139
Figure 4.39 : Summary of the different drawing processes
Even if the tendency is always the same (for example a general decrease of the tensile strength for fibers ageing in air), the relative variation in tensile strength is different after the same period of time for each drawing process.
These variations can result from the TAS photosensitivity. It is noteworthy that all these previous experiments were performed without taking into consideration the effect of light, as a consequence, the fibers were exposed continuously or partially continuously to the light (neon lighting) of the experiment room depending on the period of the year the fibers aged (during summer and winter vacations, the light of the experiment room is switched off) . The spectrum of the neon lighting is presented in Figure 4.40. The fine peaks correspond to mercury.
140
16000
14000
12000
10000
8000
Intensity 6000
4000
2000 0 200 300 400 500 600 700 800 Wavelength (nm)
Figure 4.40 : Spectrum of the neon lighting
It is well known that chalcogenide glasses exhibit some photosensitive effects
(photodarkening, photoexpansion, etc) when exposed to light (Chapter 3).
The following experiment was performed to see if the TAS glass is photosensitive to neon lighting and the surfaces were analyzed by SEM :
- Two batches of fibers were put under vacuum in a pyrex tube as previously
described. One was continuously exposed to neon lighting, the other one was
stored in the dark. The surface analysis after one year spent in these conditions
show no differences (Figure 4.41). The surface is smooth in both cases. At this
stage, the effect of light on the TAS surface is not significant.
141
(a) (b)
Figure 4.41 : SEM images of TAS fibers ageing under vacuum and exposed to light (a) stored in the dark (b)
Then two fibers of each batch were stored :
- one in air and in the dark
- the other one in air and exposed continuously under neon lighting
The surface was analyzed after 2 months spent in these conditions. The results are presented below. Fibers under VACUUM LIGHT 1 year Fiber Fiber AIR/LIGHT 2 months AIR/DARK 2 months
(a) (b)
Figure 4.42 : SEM images of TAS fibers ageing in air and exposed to light (a), stored in the dark (b)
142
Fibers under VACUUM, DARK, 1 year
Fibers Fibers AIR/LIGHT 2 months AIR/DARK 2 months
(a) (b)
Figure 4.43 : SEM images of TAS fibers ageing in air and exposed to light (a), stored in the dark (b)
The results show clearly a combined effect of the light and the environment (water, O2) on the TAS fibers surface. In air, when exposed to light, the surface tends to present a granular aspect whereas for those stored in the dark, the same smooth aspect is observed even if these fibers aged in air. These phenomena are not well yet explained.
4.9 Effect of annealing on mechanical properties
As previously described, the annealing process allows structural relaxation phenomenon.
It permits to reduce the residual stresses inside the glass by the rearrangement of bonds.
The fibers were annealed according to the process described in 4.3.2.
143
At t0, just after the drawing, the structure of the non annealed fibers is more aligned than that of the annealed fibers because of structural relaxation undergone by annealed fibers which allows the rearrangement of bonds. Therefore the bonds rearrangement associated with the structural relaxation phenomenon is unfavorable to the direction of stress
(tension). Therefore, the tensile strength for the annealed fiber is lower than that of the fibers which are not annealed (Table 4.7). The kinetic of relaxation is more rapid for the non-annealed fibers. It means that in the same period of time, the structure of the non annealed fiber will undergoe more rearrangement than in the case of the annealed fiber because these fibers are already in a relaxed state. This explains the larger decrease in tensile strength after 2 months for the non-annealed fibers.
Table 4.7 : Comparison in tensile strength for annealed and not annealed TAS fibers
Non-annealed fibers Annealed fibers
t0 8.1 N 6 N t + 2 months under vacuum 4.4 N 4.5 N
4.10 Influence of the coordination number
As previously described in chapter 3, the coordination number
< r >= ∑ xi N i
144
with x the atomic fraction of element i and N, the atomic coordination number (N=2 in the case of a chalcogen atom : S, Se, Te and N=3 in the case of As). In a glass with
1,2
r=2.3 1 r=2.4
0,8
0b 0,6 σ / b σ
0,4
0,2
0 0 102030405060708090100 Days
Figure 4.44 : Variation in bending strength as a function of time for the TAS
145
The bending strength of the fibers freshly drawn is σ0b. The bending strength when fibers age, σb, decreases dramatically after 15 days for the fibers
A TAS fiber with a coordination number
4.11 Statistical analysis of fracture
The Weibull modulus, m, which corresponds to the slope of the plot ln(ln(1/(1-
Pf)))=f(lnσ) was calculated for one drawing (Figure 4.45). The fibers were tested just after the drawing (t0), after 1 month ageing in air and after 1 month ageing under loading.
The results show a Weibull modulus, m, of 4.05 for the freshly drawn fibers and 3.71 for the fibers stored in air for 1 month. This means that the lower m value observed in the case of the fibers stored in the air reveals that the defects found at the fiber surface have higher varying sizes compared to the fibers freshly drawn. The shape of the plot on the
Weibull diagram concerning the fibers put under static stress is not a single straight line which corresponds to a multimodal behavior expressing the fact that breaking occurs from different families of defects.
146
2 t0 y = 3.71x - 4.61 t+1 month under 1 loading t+1month in air y = 5.49x - 10.96 0 00,511,522,5
-1
-2 y = 20.14x - 38 ln(ln(1/(1-Pf)))
y = 4.05x - 6.32 -3
-4
-5 ln(sigma)
Figure 4.45 : Weibull diagram in the case of a freshly drawn fiber, a fiber ageing under static stress for 1 month and a fiber ageing in air for 1 month
147
4.12 Conclusion
The mechanical properties of TAS fibers have been investigated as a function of time and environment. The fibers were tested in tension. From a general observation, air has a dramatic effect on TAS fibers tensile strength. This loss of mechanical properties can be correlated to the defects observed at the surface of the fibers ageing in air. The same loss of mechanical properties has been observed in the case of the fibers ageing under vacuum. However, in this case, the surface of the fibers even after 1 month is similar to that of the fibers freshly drawn.
The structural relaxation phenomenon is suggested as a possible explanation of this loss of mechanical properties.
TAS fibers were then tested under static stress. The increase of tensile strength observed for the fibers ageing under loading for 1 month is related to a structural change. This experiment corroborates the structural relaxation assumption as a predominant influence on the tensile strength.
The effect of annealing and coordination number
A large decrease of tensile strength compared to annealed fiber has been observed after 2 months in the case of non-annealed fibers. These results are consistent with the structural relaxation assumption.
The coordination number,
148
Finally it is shown that TAS fibers exhibit photosensitivity. Experimentally, the surface of the fibers when ageing in air and exposed continuously to the neon lighting (400-
700nm), becomes granular. This surface aspect was not seen in the case of the fibers ageing in the dark.
149
CHAPTER 5 : VISCOELASTIC PROPERTIES OF THE TAS (TE2AS3SE5) GLASS
5.1 Introduction
Near the glass transition temperature, Tg, we observe a progressive change from the viscous liquid (Newton’s liquid) to the elastic solid (Hooke’s solid). In this range of temperatures, the response of the glass to the mechanical stress is a function of time
[110]. Its response is a combination of viscous fluidity and elastic solidity. That is viscoelasticity. The study of this phenomenon is of great importance because it permits to predict the dimensional changes which will appear during usage on the elements of one given structure. It is also important to have a good knowledge of viscoelasticity and its parameters for one given glass since most of the technological operations (annealing, drawing process, etc) take place in this range of temperatures.
Because of its relatively low Tg ranges (~130°C), the Te2As3Se5 (TAS) glass composition exhibits some viscoelastic effects at room temperature. Preliminary experiments on TAS fibers have shown that indentation creep occurs at room temperature
[111]. In this chapter, the viscoelastic behavior of a TAS glass fiber is investigated mainly by means of fiber bending tests. This kind of test was already used by M. Koide et al. to characterize mechanical relaxation and recovery in silicate glass fibers during an annealing below the glass transition temperature Tg [112]. The effect of online annealing during the drawing process on stress relaxation for the TAS fibers was also investigated to compare the different relaxation times.
150
5.2 Viscoelasticity
Viscoelasticity corresponds to the study of materials which present mechanical behavior close to that of elastic solids and viscous liquids. The main characteristics of viscoelastic materials are:
- the response of the material when stressed is a function of time
- the variation of the response is a function of temperature
Viscoelasticity is correlated to the glass transition temperature, Tg, since these two phenomena are linked to the evolution of the viscosity as a function of temperature.
Different experiments such as creep or stress relaxation can permit to describe viscoelastic behavior of one given material. These two experiments will be defined later.
Viscoelasticity is generally linear if the stress and the deformation applied to the material are low. It means that the coefficients attributed to the phenomenon do not change with time and do not depend on the stress intensity (if the temperature is kept constant).
For example, the three following curves, corresponding to a creep experiment, are the strains measured at three different stress levels, each one twice the magnitude of the previous one (Figure 5.1).
Note that when the stress is doubled, the resulting strain is doubled over its full range of time.
This occurs if the material is linear in its response. The same linearity can be shown with a stress relaxation experiment as well.
151
4×σ0
Strain 2×σ0
σ0
Time
Figure 5.1 : Creep strain at various constant stresses [113]
5.2.1 Boltzmann superposition principle
The Boltzmann superposition principle applies to linear viscoelasticity. It states that strain is a linear function of stress therefore the total effect of applying several stresses is the sum of the effect of applying each one separately. Mathematically, if the stress due to a strain ε1(t) is σ(ε1) and that due to a different strain ε2(t) is σ(ε2), then the stress due to both strains is σ(ε1+ε2)= σ(ε1) + σ(ε2).
5.2.2 Viscoelastic moduli
A viscoelastic material submitted to a constant stress σ0, undergoes a strain ε(t) which varies as a function of time, that is the creep phenomenon. This variation is called creep compliance and is defined by : J (t) = ε (t) for t>0. σ 0
In the same way, when a viscoelastic material is submitted to a constant strain, ε0, the corresponding stress σ(t) decreases as a function of time, that is the relaxation
152
phenomenon. This variation is described by the relaxation modulus G(t) defined by :
G(t) = σ (t) for t>0. ε 0
5.2.3 Mechanical models for linear viscoelastic response
This part deals with several common mechanical models which can describe the glass viscoelastic behavior.
5.2.3.1 Simple Models
5.2.3.1.1 Maxwell
Every viscoelastic behavior can be described by the combination of two simple mechanical models :
- the “Hookean spring” which represents the elastic behavior
σ e = GM ×ε e
- the “Newtonian dashpot” which represents the viscous behavior
σ η = η M ×ε&η
With σ : the stress, ε : the strain, ηM : the viscosity, GM : the shear modulus and ε& : the strain rate
The Maxwell model is a mechanical model in which one spring and one Figure 5.2 : Maxwell model dashpot are connected in series (Figure 5.2). It is the simplest linear viscoelastic model.
153
In a series connection, the stress on each element is the same and equal to the imposed stress, while the total strain is the sum of the strain in each element.
σ = σ = σ e η ε = εη +ε e
The derivative of strain with respect to time leads to the total strain kinetic of the
Maxwell element. The total strain rate of the Maxwell element or its rheological equation is :
dε(t) dεη (t) dεe (t) 1 dσ(t) σ(t) = + = + dt dt dt GM dt ηM
σ& σ Or in dot notation : ε& =ε&e +ε&η = + GM ηM
σ η Therefore, σ = G ε − where τ = M is the “relaxation time” for the Maxwell & M & M G τ M M model. τM is physically the time needed for the stress to fall to 1/e of its initial value
[113]. The relaxation time τ is strongly dependent on temperature that affects the mobility of atoms constituting the material.
dσ (t) dε (t) σ (t) By integration we obtain : = G − ∫ M ∫ ∫ dt dt τ M
t t σ (t) = GM ε (0)exp(− ) = σ (0)exp(− ) τ M τ M
t The corresponding relaxation modulus is G(t) = GM × exp(− ) τ M
1 t and the creep compliance is J (t) = + GM η M
154
5.2.3.1.2 Kelvin-Voigt
The Kelvin-Voigt model is also a simple linear viscoelastic model. It can be represented by a purely viscous dashpot and purely elastic spring connected in parallel as shown in Figure 5.3.
Since the two components of the model are arranged in parallel, the strains in each component Figure 5.3 : Kelvin-Voigt model are identical :
ε = ε e = εη
Similarly, the total stress will be the sum of the stress in each element:
σ = σ e + σ η
The rheological equation associated with the Kelvin-Voigt model is :
dε (t) σ (t) = G ε (t) +η η K e K dt
1 The corresponding creep compliance is : J (t) = ⎡1− exp(− t )⎤ ⎢ τ ⎥ GK ⎣ K ⎦
η K τ K = is the characteristic time of the Kelvin-Voigt model. GK
155
5.2.3.1.3 Burger
The Burger model is made of a Maxwell and a Kelvin-Voigt element in series (Figure 5.4). Four parameters are needed to describe this model : G, GK, η and ηK.
This model has two characteristic times :
τ = η is the characteristic time of the Maxwell part M G
(instantaneous elasticity and viscosity)
η K τ K = is the characteristic time of the Kelvin-Voigt GK Figure 5.4 : Burger model model (delayed elasticity)
The corresponding relaxation function is (Appendix D) :
G ⎡⎛ GK ⎞ ⎛ GK ⎞ ⎤ G(t) = ⎢⎜α − ⎟exp()−α.t − ⎜ β − ⎟exp()− β.t ⎥ α − β ⎣⎝ η K ⎠ ⎝ η K ⎠ ⎦
Where α and β are the roots of the following equation :
2 ⎛ G G GK ⎞ G.GK x − ⎜ + + ⎟x + = 0 ⎝ η η K η K ⎠ η.η K
−t 1 t 1 ⎡ τ ⎤ The creep compliance is : J(t) = + + . 1− e K G η G ⎣⎢ ⎦⎥
156
5.2.3.2 Complex models
In practice, we observe that simple models are not able to fit suitably the experimental data. Therefore, we usually use more complex models such as the generalized Maxwell model, the Wiechert model or the KWW (Kohlrausch-Williams-Watt) function.
5.2.3.2.1 Generalized Maxwell model
This model takes into account that the relaxation phenomenon does not occur at a single time, but at a distribution of times. This model is constituted by n models of Maxwell cells (Gi, ηi) arranged in parallel (Figure 5.5). The main advantage of this model resides in the fact that we can use as many couples of parameters as we need to well describe the experimental data. This model allows to describe the elasticity, anelasticity and the inelasticity.
The corresponding relaxation function is defined by :
n G(t) = w × exp⎛− t ⎞ where each couple (w , τ ) defines a Maxwell cell. ∑ i ⎜ τ ⎟ i i i=1 ⎝ i ⎠
Figure 5.5 : Generalized Maxwell model
157
This model presents some problems such as the choice of the number of Maxwell’s cells and the parameters wi and τi we have to choose to fit perfectly the experimental data.
This model allows a better fit than the other simple models but does not have any concrete meaning. Indeed it is difficult to attribute a meaning of the different τi to the mechanisms implied in the viscoelastic phenomenon.
5.2.3.2.2 Kohlrausch-Williams-Watt (KWW)
Some studies on inorganic glasses showed that the KWW function also known as the extended exponential function is able to describe the viscoelastic phenomenon for uniaxial or shear stresss near the glass transition temperature, Tg [114].
b ⎛ t ⎞ The KWW function can be described by : FKWW = K × exp⎜ ⎟ with 0 Its relaxation kinetic differs from the generalized Maxwell model at the beginning of the relaxation process (Figure 5.6). However, it is shown that this model cannot describe the viscoelastic behavior for both the beginning and the end of the relaxation phenomenon [115]. The b coefficient cannot be constant to well describe the phenomenon. Its empirical character makes this function well criticized mainly because dF lim KWW → ∞ which is impossible regarding the phenomenon, the relaxation kinetic t→0 dt cannot be infinite at t=0. The main advantages of the KWW exponential function are : 158 - the good approximation regarding the material behavior on a large interval of times - only two parameters, b and τKWW have to be considered. Figure 5.6 : Comparison of the relaxation function G(t) simulated by (a) KWW function and (b) generalized Maxwell model 5.2.3.2.3 Wiechert The Wiechert model is derived from the generalized Maxwell model with the addition of a pure elastic component in parallel which allows the description of a steady stress. The Wiechert model can have as many spring-dashpot Maxwell elements as are needed to approximate the distribution satisfactorily. This model has been widely used to describe the cross-linked polymers [116-118]. However the Wiechert model cannot describe the inelastic deformation because of the spring G0 which makes the Maxwell’s cells recover their deformation after a period of time (Figure 5.7). 159 Figure 5.7 : Wiechert model The total stress σ corresponds to the stress in the isolated spring added to the stress in each of the Maxwell’s spring-dashpot cells : σ = σ 0 + ∑σ i i The corresponding relaxation function is : ⎛ ⎛ t ⎞⎞ G(t) = G + G × exp⎜− ⎜ ⎟⎟ 0 ∑ i ⎜ ⎜ ⎟⎟ i ⎝ ⎝τ i ⎠⎠ 5.2.4 Creep The creep test consists of measuring the time dependent strain ε(t) resulting from the application of a steady uniaxial stress σ0 as illustrated in Figure 5.8. Again, the study of this phenomenon is of great importance because it permits to predict the dimensional differences which will appear during usage on the element of one given structure. 160 When static stress is applied to viscoelastic material, it deforms continuously. The initial strain is predicted by its stress-strain modulus. The material will continue to deform slowly with time indefinitely or until fracture. The primary region is the early stage of loading when the creep rate decreases rapidly with time. Then it reaches a steady state which is called the secondary creep stage followed by a rapid increase (tertiary stage) and fracture (Figure 5.8). 0 Figure 5.8 : Creep experiment If the applied load is released before the creep fracture occurs, an immediate elastic recovery equal to the elastic deformation, followed by a period of slow recovery is observed. The material in most cases does not recover to its original shape and a permanent deformation remains. Creep is a thermally activated phenomenon, it means that this phenomenon is more rapid as the temperature increases. The deformation rate or creep rate becomes more and more 161 important when the temperature is close to the glass transition, Tg, in the case of glasses. Because the TAS glass has a low Tg (137°C), creep phenomenon can theoretically happens at room temperature. The deformation, ε(t), induced by a creep experiment under constant applied stress, σ0, has three components : εe, εd and εη. - ε e is the instantaneous elastic deformation, ε e = σ 0 / E with E, the Young’s modulus of the material - ε d is the delayed elasticity component or anelastic deformation, ε d (t) = σ 0 ()1− exp[]− ()t /τ (σ 0 ) / Ed - εη is the viscous or inelastic deformation, εη (t) = (t /η(σ 0 ))×σ 0 . The inelastic deformation can not be recovered and is therefore permanent. The equations corresponding to εe, εd and εη depend on the mechanical model we choose to describe the creep experiment. In this last case, the equations for εe, εd and εη corresponds to the Burger model. When the applied stress is cancelled at t1 (Figure 5.9), a part of all the deformations is recovered instantaneously and the other part is delayed (anelastic part). The deformation at t=t∞ is permanent and corresponds to the inelastic component. The total deformationε (t ) can be described by the sum of all the components : ε (t) = ε e + ε d (t) + ε η (t) 162 σ ⎛ ⎡ ⎛ ⎞⎤⎞ ⎛ ⎞ 0 1 ⎜ t ⎟ t ε (t) = + 1− exp⎢− ⎜ ⎟⎥ σ 0 + ⎜ ⎟σ 0 (Burger model) E E ⎜ ⎜τ (σ ) ⎟ ⎟ ⎜η σ ⎟ d ⎝ ⎣ ⎝ 0 ⎠⎦⎠ ⎝ ()0 ⎠ The corresponding creep compliance, J(t), is given by : ε (t) 1 1 ⎛ ⎡ ⎛ t ⎞⎤⎞ t J (t) = = + ⎜1− exp⎢− ⎜ ⎟⎥⎟ + σ E E ⎜ ⎜τ σ ⎟ ⎟ η σ 0 a ⎝ ⎣ ⎝ ()0 ⎠⎦⎠ ()0 Figure 5.9 : Creep-Recovery experiment 163 5.2.5 Stress relaxation Stress relaxation experiment consists of monitoring the time-dependent stress resulting from a steady strain. This experiment which can also characterize the viscoelastic behavior of one given material is studied by applying a constant deformation ε0 to the glass test piece and measuring the stress σ(t) required to maintain that constant strain as a function of time. In the case of a constant deformation ε0, the stress, σ(t), decreases from σ0 at t=0 to σu when the deformation ε0 is released at tu (Figure 5.10). At t=0, we can define σ0 which is equal to σ 0 = E ×ε 0 where E corresponds to the Young’s modulus of the material. The evolution of the stress follows the Hookes’s law : σ (t) = E ×ε e (t) Where ε e (t) corresponds to the elastic deformation that the material recovers when the constant deformation ε0 is released. Like for the creep experiment, the stress relaxation experiment is followed by the recovery part when the applied deformation ε0 is released. Therefore we study the strain evolution for the material. The recovery part is characterized by three components : - ε e is the instantaneous elastic deformation - ε d is the delayed elasticity or anelastic deformation, - εη is the viscous or inelastic deformation. The inelastic deformation can not be recovered and is therefore permanent 164 σ0 σu Figure 5.10 : Stress relaxation – recovery experiment 5.3 Delayed elasticity in the TAS fibers 5.3.1 Stress relaxation experiment TAS fibers viscoelasticity was investigated by means of a stress relaxation experiment. This experiment can be divided into three steps : - The relaxation : the fiber is fixed on a mandrel. A constant deformation is imposed. The stress decreases continuously as a function of time. The radius curvature corresponds to the mandrel radius, R0. 165 - The unrolling : After a certain period of time spent on the mandrel, the fiber is unrolled at the time, tu. The fiber recovers instantaneously the elastic part. The radius curvature is R(tu) (R(tu)>R0). - The recovery : the fiber is let on a plate surface, free from stress. The radius curvature increases until the asymptotic value R∞. This recovery corresponds to the delayed elasticity and is a function of time. It is also called anelasticity. 5.3.1.1 Relaxation and unrolling The Te2As3Se5 (TAS) fibers were synthesized and drawn according to the process explained in Chapter 4. The 150 mm length and 400µm diameter fibers were then rolled on a 10 cm diameter mandrel (R0=5cm). At tu, a fiber is unrolled from the mandrel and we measure immediately its radius curvature R(tu) (Figure 5.11). Figure 5.11 : Rolling and unrolling experiment A constant deformation, ε0, is applied to these fibers when they are rolled on the mandrel. In every point M of the fiber at a distance r of the revolution axis (Figure 5.12), the 166 r r deformation is equal to ε = ≈ because R0>>Rf (Rf is the fiber radius and R0 R0 + R f R0 is the curvature radius of the mandrel). Arbitrarily, we chose the maximum of deformation for the measurements which is situated at the periphery of the fiber at the point M’ on Figure 5.12 where y=r=Rf. R f Therefore the constant deformation, ε0 at t=0 is equal toε 0 = when the fiber is on the R0 mandrel. R f When the fiber is unrolled, the deformation is equal to ε (tu ) = with R(tu) the radius R(tu ) curvature measured when the fiber is unrolled at the time tu. M’ f Figure 5.12 : Deformation of the fiber The curvature radius R(tu) when the fiber is unrolled is calculated according to the following equation : 167 4a 2 + L2 4a 2 + L2 4(a ± Δa) 2 + L2 R = and ΔR = − with Δa = 1mm tu 8a 8a 8(a ± Δa) As illustrated in Figure 5.13, as the deformation is imposed, the stress relaxes continuously from σ0 to σu. When the fiber is unrolled at tu, the stress drops from σu to 0. In the meantime, the deformation drops from an amplitude εe which corresponds to the elastic part that contributes immediately to the recovery (at tu, we assume that there is no anelastic recovery). This elastic deformation is instantaneous and linear since the deformation is small. Therefore, the Hooke’s law can be applied : σ (tu ) = E ×ε e (tu ) with E, the Young’s modulus of the TAS glass. 168 σ0 σu Figure 5.13 : Stress relaxation-recovery experiment For each fiber unrolled at different times tu, the plot stress σ versus time (tu) can be determined since σ is directly proportional to the elastic recovery, εe as previously shown. εe can be easily determined from the measurements of the curvature radius R(tu) R f R f according toε e (tu ) = ε 0 − ε (tu ) = − R0 R(tu ) 169 5.3.1.2 Recovery Once a fiber is unrolled, it is left on a plate surface for the recovery step. The evolution of the curvature radius R(t), t>tu, is measured regularly and is calculated, as previously 4a 2 + L2 explained, from the following equation R = . 8a The deformation ε(t) during the recovery step can be expressed as : ε (t) = ε 0 − ε e (tu ) − ε d (t − tu ) Where εd is the anelastic deformation or the delayed elasticity and εe is the instantaneous elasticity. After a certain period of time, the deformation tends to a constant deformation, ε∞ or R∞ (Figure 5.14-a). It is noteworthy that the longer the time spent on the mandrel (long tu), the smaller the radius of curvature just after unrolling, R(tu) (Figure 5.14-b). This is related to the stress relaxation as the fiber is fixed on the mandrel. (a) 170 (b) Figure 5.14 : Evolution of the radius curvature as a function of the unrolling time, tu (a), (b) 5.3.2 Results 5.3.2.1 Relaxation The following experimental data were obtained for the stress relaxation part (Figure 5.15). The time a fiber is unrolled, tu, ranges from few hours to 170 days. Figure 5.15 : Experimental data for the stress relaxation experiment 171 First, the Burger model (Figure 5.16) was used to characterize the TAS fiber behavior during stress relaxation. When we impose a constant deformation ε0 to this model, the stress σ induced by this deformation is given by (Appendix D) : σ 0 ⎡⎛ GK ⎞ ⎛ GK ⎞ ⎤ σ (t) = ⎢⎜α − ⎟ × exp()−α.t − ⎜ β − ⎟ × exp()− β.t ⎥ α − β ⎣⎝ η K ⎠ ⎝ η K ⎠ ⎦ G corresponds to the Young’s modulus of the TAS glass and Figure 5.16 : Burger model has already been measured from ultrasonic echography method. GK, ηK and η were determined by using the solver tool of Microsoft® Excel 2000 solver tool. These parameters were optimized to fit the experimental data. The following values corresponding to the different Burger model parameters allow a good approximation for the experimental data but only for the 30 first days (Figure 5.17). 17 15 {G, η, GK, ηK} = {16.9 GPa, 1.39.10 Pa.s, 5.51 GPa, 6.4.10 Pa.s} This leads to the different characteristic times : η η K τ M = ≈ 95 days andτ K = ≈ 1.2 days G GK 172 Figure 5.17 : Experimental data and Burger fitted model for the first 32 days of relaxation However, the Burger model with these last parameters is not able to fit perfectly the low stress relaxation kinetic observed after 40 days. Another set of parameters with longer characteristic times can fit the experimental data for longer relaxation time but can not fit the rapid stress relaxation kinetic observed in the first 40 days (Figure 5.18). BURGER MODEL τM = 95 days ; τK = 1.2 days τM = 390 days ; τK = 10 days Figure 5.18 : Burger fitted model for two sets of parameters 173 In conclusion, the Burger model with two characteristic times (τM and τK) is insufficient to fit both the rapid stress relaxation kinetic in the early stage and the lower relaxation after 40 days. Therefore we have to find another model. The KWW function is well known to characterize the high kinetic during the beginning of the relaxation for inorganic glasses. As previously described the KWW function is more an adjustment function with no physical meaning rather than a real model. However, some authors tried to find a physical understanding for this particular behavior [118-121]. For example, Trachenko discusses the relationship between b and τ parameters and the glass structure in order to understand the relaxation process. He discusses the relaxation in glasses in terms of local relaxation events (LRE) which glass uses to adjust to external perturbations [122]. A LRE is identified as a re-bonding event that involves a sudden jump of an atom in silicate glasses. Therefore, this involves old bonds, forming new ones and subsequent relaxation of the local structure. He also found a broad correlation between the fragility and the b parameter, b increasing from about 0.2 to 1 as the system becomes stronger [23]. Again the Microsoft® Excel solver tool was used to determine the best parameters to fit the experimental data. These parameters are : {τKWW, b} = {69 days, 0.2}. As we can see on Figure 4.19, the KWW function well describe the relaxation for both shorter and longer relaxation times. The low b value corresponds to the rapid kinetic of the relaxation phenomenon during the early stage. The time constant, τKWW corresponds to the time needed to decrease the initial stress by a 2/3 factor. 174 Figure 5.19 : KWW fitted function 5.3.2.2 Steady modulus Surprisingly, even after 150 days, the stress is not totally relaxed and tends toward an asymptotic value, σ∞. We define the steady modulus, E∞, from the steady stress σ∞ : σ ∞ E∞ = ε 0 Concerning the TAS fiber, σ∞ is estimated to be 20 MPa and ε0 which corresponds to the rfiber -3 constant deformation (ε 0 = ) is equal to 4.10 . Therefore E∞ is about 5 GPa. Rmandrel The presence of the steady stress for the TAS glass could be explained by its structure. As previously described in Chapter 4, the TAS structure is made of a two-dimensional glass network by the cross-linkages between Se-Te-As chains through As-As bonds [92- 93]. 175 Angell and Böhmer show that amorphous selenium has a relaxation behavior with a zero steady stress [123]. The amorphous selenium is well known to have a simple and linear structure without any cross-linkings. Moreover, McEnroe and LaCourse [124] show that in the GexSe1-x glasses, the steady stress increases as the percentage of germanium increases. Because the cross-linkings are higher in high germanium content glasses, we can easily make a correlation between cross-linking degree and steady modulus. The same phenomenon is observed in polymers : the more reticulated is the structure, the higher is the steady modulus [125]. 5.3.2.3 Recovery After unrolling, the fibers undergo the recovery step. Six recovery curves (strain versus time) are represented in Figure 5.20. They correspond to relaxation times ranging from 1 day to 35 days. As the time spent on the mandrel increases, the recovery amplitude is smaller for the fiber and the kinetic is slower in the beginning of the recovery. Two different models, the Kelvin-Voigt model that well describes the delayed elasticity and the KWW function were tested in order to have a good match between the experimental data and the model but none of them matches perfectly the experimental data with only one set of parameters (Figure 5.21). Then, different sets of parameter for the Kelvin-Voigt model have been tested for the different recovery curves. Recall that the Kelvin-Voigt model is characterized by its only characteristic time, τK. Again, the change in this parameter is not able to fit perfectly the rapid kinetic relaxation observed for short relaxation times and the slower one observed after longer relaxation times. But these 176 different sets of characteristic times have a better match than only one set of parameter for all the curves (Figure 5.21). 0,0025 0,0023 Experimental data 0,0021 0,0019 0,0017 (t) ε 0,0015 Strain Strain 0,0013 0,0011 0,0009 0,0007 0,0005 0 102030405060708090100 Time (in days) Figure 5.20 : Experimental data for the recovery part 177 Experimental values Kelvin model (one set of parameters for all the recovery curves) Kelvin model (one set of parameters for each recovery curve) KWW function (one set of parameters for all the recovery curves) Figure 5.21 : Different fitting models for experimental data By looking at the previous curves (Figure 5.21) and especially the first curves, the KWW function even with only one set of parameters for all the curves seems to be the best fit for the different recovery behaviors. This non linear stretched exponential function shows a very fast kinetic in the beginning of the recovery and its corresponding equation is: b ⎛ ⎛ t ⎞ ⎞ ε (t) = ε × exp⎜− ⎜ ⎟ ⎟ ∞ ⎜ ⎜τ ⎟ ⎟ ⎝ ⎝ KWW ⎠ ⎠ 178 The optimization of b and τKWW with the Microsoft® Excel solver tool for all the recovery curves leads to the following parameters: {τKWW, b}={6.6 days, 0.57}. But as shown in Figure 5.21, this couple of values does not involve a suitable simulation for long relaxation times. Then, one different set of parameters {τKWW, b} was applied for each recovery curve to match perfectly (the relative gap is less than 2%) all the experimental data. The results are presented for only 4 curves (Figure 5.22) and provides a very good description of the recovery behavior whatever the relaxation duration. Indeed, experimental and theoretical curves almost overlap. Figure 5.22 : KWW fitting function with different parameters for different relaxation times 179 Figure 5.23 shows the evolution of the two parameters, τKWW and b related to the KWW function as a function of time spent on the mandrel. The characteristic time, τKWW, increases as the time spent on the mandrel tu increases as well. This means that the longer the fiber spends on the mandrel, the longer is the time to relax 2/3 of its initial stress. b and τKWW values are low for short times spent on the mandrel, which is consistent with the high kinetics observed at the beginning of the recovery. Further, for long relaxation durations, the slow kinetics of the recovery is nicely modelled by high values of b and τKWW. The variation of these parameters means that the material behaves differently as a function of the relaxation time. The TAS fiber behaves first as an intermediate between the instantaneous elasticity and the Maxwell liquid (b=0.5) for short relaxation times to the perfect Maxwell liquid (b=1) for long relaxation times. Figure 5.23 : Evolution of the KWW parameters (b and τKWW) for different relaxation times 180 5.4 Effect of annealing on viscoelasticity 5.4.1 Relaxation The same stress relaxation experiment was performed on both annealed fibers and non- annealed fibers. The fibers were annealed online during the drawing according to the process explained in Chapter 4 and rolled on a 10cm mandrel diameter. Figure 5.24 represents the stress relaxation curves for both fibers. 70 60 Annealed fibers 50 Non-annealed fibers 40 30 StressMpa) (in 20 10 Time (in days) 0 0 102030405060708090100110120130140150160 Figure 5.24 : Stress relaxation curves for both annealed and non-annealed TAS fibers The relaxation kinetic for both fibers (annealed and non-annealed fibers) is characterized by a steady stress, σ∞. The experimental data are well described by a Wiechert model with three Maxwell’s cells or by a KWW function (Figure 5.25). 181 Experimental data (Annealed fibers) Experimental data (Non-annealed fibers) Wiechert (Annealed fibers) 0,004 Wiechert (Non-annealed fibers) 0,0035 KWW (Annealed fibers) 0,003 KWW (Non-annealed fibers) 0,0025 0,002 0,0015 Elastic deformation 0,001 0,0005 0 0 50 100 150 200 Relaxation Time (in days) Figure 5.25 : Wiechert and KWW fitting models for the stress relaxation curves for both annealed and non-annealed TAS fibers All the parameters were optimized with a new solver tool developed by Eric Robin from the LARMAUR (Laboratoire de Recherche en Mécanique Appliquée de l’Université de Rennes) at the University of Rennes 1 (France). That is why the different parameters such as the relaxation time are so different from those found with the Microsoft® Excel solver tool in the previous part. For both models (Wiechert and KWW function), the kinetic relaxation is slower and the steady modulus E∞ is higher for the annealed fibers (Table 5.1, Table 5.2). Therefore, there is an effect related to the annealing process which is well known to reduce residual stresses in the fiber. It is noteworthy that the steady stress σ∞ value is lower in the case of the KWW function compared to the Wiechert model. 182 The evolution of stress relaxation in the case of the annealed fibers is the sum of two components, one related to the annealing process and the other one related to the stress relaxation experiment (fibers rolled on the mandrel) (Figure 5.26). The annealing process induces a more stable structure, therefore, it induces a lower stress relaxation and a higher relaxation time. Table 5.1 : Wiechert parameters for the stress Table 5.2 : KWW parameters for relaxation for annealed and non-annealed TAS the stress relaxation for annealed fibers and non-annealed TAS fibers Non-annealed Annealed Non-annealed Annealed fibers Fibers Fibers fibers E∞ (GPa) 5.23 7.73 b (no unit) 0.26 0.36 E1 (GPa) 4.51 2.48 τKWW (days) 11.55 16.00 τ1 (days) 0.08 0.12 E∞ (GPa) 3.47 6.93 E2 (GPa) 2.82 3.15 τ2 (days) 2.65 5.45 E3 (GPa) 4.33 3.51 τ3 (days) 36.7 57.19 183 Non annealed fibers Same deformation on the ε mandrel for both fibers Annealed fibers σ Time RELAXATION Residual stress relaxation due to annealing Time (t) 0 Fibers are rolled on the mandrel, t r Figure 5.26 : σ(t) and ε(t) for both annealed and non-annealed TAS fibers 5.4.2 Recovery The recovery part for both fibers (annealed and non-annealed) were simulated by the KWW function. One set of parameters (b and τKWW) was chosen for each recovery curve. The evolution of these parameters are represented in Figure 5.27. The recovery kinetic in the case of the annealed fibers is higher compared to the non-annealed fibers. Indeed the characteristic time related to the KWW function, τKWW, is lower for the annealed fibers (Figure 5.27). 184 1 16 0,9 14 0,8 12 0,7 0,6 10 0,5 8 Non-annealed fibers 0,4 6 0,3 Annealed fibers Exponant KWW 4 0,2 Time days KWW in 0,1 2 0 0 0 10203040506070 Time spent on the mandrel (days) Figure 5.27 : Evolution of the KWW function parameters as a function of relaxation time for the recovery 5.5 Effect of light on viscoelasticity In Chapter 4, we demonstrated the effect of light on the TAS surface when ageing in air. The risen question is : does the light only affect the TAS surface or does it affect the volume? To answer to that question, some relaxation experiments in torsion (Figure 5.28) were performed on TAS fibers ( α(t) σ (t) = r × μ × θ L 185 Where α(t) is the angle restored at time t when the fiber is released (at t=0, α(0)=θ), μ is the Coulomb modulus, r is the fiber radius and L is the length of the tested fiber. One fiber tested in torsion was stored in the dark, another one was continuously exposed to neon lighting. These two fibers were periodically released and the angle was measured to build the curve stress=f(time of relaxation). The same kinetic relaxation behavior was observed for both fibers. Indeed, the data for the experiment performed in the dark or under neon lighting overlap (Figure 5.29). Therefore, there is no voluminal effect which is coherent. Considering their bang-gap is around 1µm, the penetration depth of visible light is negligible and this produces only surface effects. Figure 5.28 : Stress relaxation experiment in torsion for a TAS fiber 186 22 20 18 dark 16 light 14 12 10 8 Stress (in Mpa) (in Stress 6 4 2 0 0 1020304050607080 Time (in days) Figure 5.29 : Stress relaxation as a function of time for the TAS fibers tested in torsion in the dark or exposed continuously to light 187 5.6 Conclusion The viscoelastic behavior of TAS glass fibers has been investigated by means of a stress relaxation experiment. Fibers have been rolled during several days on a mandrel and their radius of curvature was measured periodically after unrolling. It was observed that both amplitude and kinetics of the delayed elastic recovery decrease when relaxation time increases. The stress relaxation part can not be simulated by a linear viscoelastic model because it is too fast in kinetics at the beginning and slower after a certain period of time. Optimization of a KWW function (b and τKWW parameters) leads to a good simulation of the relaxation stage of the fiber for both the high kinetics observed in the beginning of the phenomenon and the slower kinetics after 40 days. A Wiechert model with 3 Maxwell’s cells leads to the same good fitting. Similarly, the recovery period can not be simulated by a linear viscoelastic model because, for the same reasons as before, the kinetics are too fast in the beginning and on the other hand, it is too dependent on the previous relaxation duration time. So, a KWW function in which the b and τKWW coefficients change as a function of the relaxation time has been found to be a good means to predict the recovery behavior. The effect of online annealing during the drawing process has also been investigated. The results show a longer stress relaxation time in the case of the annealed fibers. This is related to the previous “history” of the fibers which were already relaxed during the annealing process. 188 Finally, the linear character of the viscoelasticity should be investigated by rolling the fibers on different mandrel diameter to prove the linearity of the phenomenon for the stress relaxation part. 189 CONCLUSION The first goal of this project was motivated by the creation of a new optical sensor based on nanoporous glass ceramics. This project was accomplished by studying new glass ceramics derived from chalcogenide glassy matrix synthesized in the GeS2-Sb2S3-CsCl system. Chalcogenide glasses were chosen because they exhibit unique optical properties including a large transparency optical window extended to the near or mid infrared range that make them applicable to the chemical and biological substances sensing. The base glass, 62.5GeS2-12.5Sb2S3-25CsCl was chosen to perform the ceramisation tests. The crystals, induced by a thermal treatment, have a size ranging from 100 nm to more than 200 nm depending on the ceramisation time and temperature. The CsCl crystals at the surface of the sample can be etched by an acid treatment (“piranha” solution). The pore size can be controlled as a function of time of acid treatment. This porous surface only affect few hundreds nanometers and is not critical for optical transmission in the infrared range. This new porous material capable of trapping molecules with size less than those of pores has been tested as an optical sensor. The glass ceramic was designed as an ATR (Attenuated Total Reflections) plate for testing purpose. First tries were realized with liquid, gaseous and solid samples. Promising results performed with an aminopropyltriethoxysilane (APTS) molecule, usually used as a coupling agent, showed a four to fivefold increase absorbance in the case of the porous material compared to the non porous glass ceramics. 190 Almost all chalcogenide glasses exhibit photosensitive change in properties, especially when illuminated by sub band gap light. Therefore, one of the photosensitive process, the photodarkening which can be used to fabricate optical structures such as Bragg’s gratings or waveguides, was tested in the GeS2-Sb2S3-CsCl glassy system. The objective was to study the influence of the alkali halide, CsCl, on the photodarkening phenomenon, previously investigated in the Ge-Sb-S glass composition. Results showed the dramatic effect of the alkali halide addition which suppresses the photodarkening phenomenon. The second main goal was to advance the understanding of the mechanical and viscoelastic properties of the TAS glass which is currently widely used in our laboratory as a biosensor in many fields of applications including medicine, food safety or environment. This work permitted to define the optimal conditions in which the TAS fibers have to be stored to prevent their premature ageing. Influence of environment on their mechanical properties have been investigated. The results show that the tensile strength deteriorates faster during air exposure in comparison to vacuum whereas fibers aged under static stress show an increase of tensile strength compared to the freshly drawn fibers. The structural relaxation phenomenon is suggested as an hypothesis to explain these results. The glass composition and especially the average coordination number 191 TAS (Te2As3Se5) fibers show surface photosensitive effect characterized by the creation of surface defects therefore optimal conditions result in storing the fibers in the dark. TAS photosensitivity is only a surface effect, light does not affect the volume since no difference in stress relaxation experiment between fibers exposed to light or fibers kept in the dark was observed. Much more research remains to be done in this area and especially the influence of photosensitive effect on mechanical properties have to be tested for the TAS composition. Finally, stress relaxation experiment showed that TAS fibers exhibit viscoelastic behavior at room temperature. The behavior in stress relaxation on a mandrel and the recovery were both simulated by an exponential KWW function. 192 APPENDIX A RAMAN Spectroscopy Raman spectroscopy is a spectroscopic technique used in condensed matter physics and chemistry to study vibrational, rotational and other low-frequency modes in a system. The probability for Raman scattering to occur is extremely low : several orders of magnitude lower than Rayleigh scattering. It relies on inelastic scattering (or Raman scattering) of monochromatic light, usually from a laser in the visible, near infrared or near ultraviolet range. A small fraction of light is scattered at optical frequencies different from, and usually lower than, the frequency of the incident photons, that is the inelastic phenomenon. Raman spectroscopy is useful for analyzing molecules or in our case for determining the glass structure. For spectroscopic techniques such as infrared spectroscopy it is necessary for the molecule being analyzed to have a permanent electric dipole. This is not the case for Raman spectroscopy, rather it is the polarizability (α) of the molecule which is important. If a molecule has a center of symmetry, vibrations which are Raman-active will be silent in the infrared and vice versa. The oscillating electric field, E, of a photon causes charged particles (electrons and, to a lesser extent, nuclei) in the molecule to oscillate. This leads to an induced electric dipole moment, µind, where µind = α E . This induced dipole moment then emits a photon, leading to either Raman or Rayleigh scattering. The energy of this interaction is also dependent on the polarizability: Energy of interaction = -1/2α E 2 193 A molecular polarizability change, or amount of deformation of the electron cloud, with respect to the vibrational coordinate is required for the molecule to exhibit Raman effect. The amount of the polarizability change will determine the intensity, whereas the Raman shift is equal to the vibrational level that is involved. One of the strengths of Raman spectroscopy is that this will be true for both heteronuclear and homonuclear diatomic molecules. In quantum mechanics, the scattering is described as an excitation to a virtual state lower in energy than a real electronic transition with nearly coincident de-excitation and a change in vibrational energy. Therefore, Ev is a virtual energy level (Figure 1). The incident photon can lose (Stokes) or gain (anti-Stokes) energy to an optical phonon, resulting in a shift in the wavelength of the light. For the spontaneous Raman effect, the molecule will be excited from the ground state to a virtual energy state, and relax into a vibrational excited state, which generates Stokes Raman scattering. If the molecule was already in an elevated vibrational energy state, the Raman scattering is then called anti- Stokes Raman scattering. Most molecules are in the ground state E0 at room temperature, so the Stokes peak is much more intense than the anti-Stokes peak. Measuring the shift Δυ in photon energy of the scattered light gives the energy of the optical phonons, or the vibrational modes, which characterize the structure of the sample (crystal or molecule) By convention, Δυ>0 : Δυ= υlaser - υscattered The Stokes and anti-Stokes peaks are symmetrical (Figure 2). 194 Stokes: Anti-Stokes: emission of a phonon absorption of a phonon Ev Ev ħωi ħωs ħωi ħωs E1 E1 ħΩ ħΩ E0 E0 Conservation ħωi + ħΩ = ħωs ħωi = ħωs + ħΩ of energy ks ki K Conservation k = k + K of i s momentum ki + K = ks i = incident, s = scattered, v = virtual Figure A-1 : Raman scattering mechanisms 195 Intensity (A. U.) - 520 0 520 Wavenumber (cm-1) Figure A-2 : Anti-Stokes and Stokes peak for Silicon 196 APPENDIX B BET (Brunauer, Emmet, Teller) method : principle Adsorption phenomena occurs when vapors are put in contact with a clean and solid surface : a part of the molecules disappears from the gaseous phase and come to fix on the solid. The vapor is defined as a gas below its critical temperature (able to condense). The amount of adsorbed vapor depends on the temperature, the pressure and the interaction potential E between the adsorbate (gas) and the adsorbent (solid). Adsorption can take place because of the presence of this intrinsic surface energy. When a material is exposed to a gas, an attractive force acts between the exposed surface of the solid and the gas molecules. The result of these forces is characterized as physical adsorption (Van Der Waals bonding), in contrast to the stronger chemical attractions associated with chemisorption. This phenomenon is characterized by a large molecule density near the superficial layers of the solids compared to the gaseous phase. The amount of adsorbed gas increases with pressure and with the inverse of temperature. The curve W=f(P) are called adsorption isotherms : volume or weight of adsorbed gas as a function of pressure at one given temperature. All the adsorption isotherms can be described by one of the 5 following curves. These curves differ because of different gas/solid interactions. 197 Figure B-1 : Different adsorption isotherms From the isotherm curve, we determine the amount of adsorbed gas as a gas monolayer at the surface of the solid. P : Adsorbate equilibrium pressure P0 : Adsorbate saturated equilibrium pressure (maximal vapor pressure until condensation) 198 BET theory (Brunauer, Emmet, Teller, 1938) The concept of the theory is an extension of the Langmuir theory, which is a theory for monolayer molecular adsorption, to multilayer adsorption with the following hypotheses : (a) gas molecules physically adsorb on a solid in layers infinitely (b) there is no interaction between each adsorption layer (c) the Langmuir theory can be applied to each layer The BET theory predicts the number of molecules needed to cover the surface of one monolayer. It also determines the covering area of 1 molecule at the surface. The total surface of 1 powder or 1 given material is obtained by multiplying the number of molecules needed to form one monolayer by the covering area of one molecule. The adsorbed molecules are continuously exchanged with those of the gaseous phase. At equilibrium, when the adsorption and desorption rate are equal, the surface of the solid is partially or completely covered by adsorbed molecules, under one given pressure. The resulting BET equation is expressed by : C −1 P 1 1 ()P = + 0 (1) ⎛ P ⎞ W C W C W ⎜ 0 −1⎟ m m ⎝ P ⎠ With W : the adsorbed gas quantity(for example in weight or in volume units) Wm : the monolayer adsorbed gas quantity C : the BET constant 199 P and P0 are the equilibrium and the saturation pressure of adsorbates at the temperature of adsorption The BET equation is an adsorption isotherm and can be plotted as a straight line with P 1 on the y axis and P on the x-axis. This plot is called the BET plot. ⎛ P ⎞ 0 W ⎜ 0 −1⎟ ⎝ P ⎠ 1 ⎛ P ⎞ W ⎜ 0 −1⎟ ⎝ P ⎠ P P0 Figure B-2 : BET plot The linear relationship of this equation is maintained only when the relative pressure P/P0 is between 0.05 and 0.35. The intercept of the line is : i =1/(WmC) The slope of the line is : s = (C-1)/(WmC) The BET method is widely used in surface science for the calculation of surface areas of solids by physical adsorption of gas molecules. A total surface area St is evaluated by the following equation : 200 W × N × A S = m t MW With MW the molecular weight of adsorbent gas N, the Avogadro’s number A, the adsorption cross section area 201 APPENDIX C Bending stress calculations The bending stress σ depends on the curved fiber geometry. The curvature radius R is defined as : Figure C-1 : Two-point bending geometry The limit conditions are : 1 dθ = = 0 pour θ = π R ds From a geometric point of view, we have : 202 π EI b = ∫ dssinθ = 0.847( )1/ 2 (1) π / 2 F With E : The Young’s modulus (in Pa) I : The quadratic momentum (in m4) F : The force (in Pa) Therefore : EF σ (θ ) = r [(2 ) sinθ ]1/ 2 I With r : fiber radius The stress is maximal for θ= π/2, then : EF σ = r (2 )1/ 2 (2) max I At fracture, the distance between the jaws is given by : D = 2b + d − 2e If D' is the curvature diameter of the fiber when breaking : D'= D + 2e = 2b + d (3) Where e (700μm) represents the depth of the lines in the jaws to guide the fiber and d represents the total diameter of the fiber. In our case d= 2r because the fiber is not coated by a polymer. 203 By associating the equations (1), (2) et (3) we obtain the maximal stress equation : 2r σ = 1,198E max D'−d 204 APPENDIX D Calculation of the stress relaxation in a Burger cell The laws concerning the behaviors of all Burger cell components are : εel, σel σ el = G ×ε el (Eq-1) σ η = η ×ε&η (Eq-2) εd, σ d = GK ×ε d +η K ×ε&d (Eq-3) σd εη, ση These components are connected in series, therefore : ε(t) σ(t) ε = ε el + εη + ε d (Eq-4) Figure D-1 : Burger Model σ = σ el = σ η = σ d (Eq-5) By deriving Eq-4 and by replacing with Eq-1, Eq-2 and Eq-3, we obtain : 1 ⎛ 1 1 ⎞ η ⎜ ⎟ K ε& = σ + ⎜ + ⎟σ& − ε&&d (Eq-6) η ⎝ G GK ⎠ GK ε&&d can obtained by deriving two times Eq-1 and Eq-2 and inserting in Eq-3 : 205 1 1 ε&&d = ε&&− σ& − σ&& (Eq-7) η G 1 1 ε&& = σ& + σ&& + ε&&d (Eq-8) η G The law for the Maxwell cell behaviour is obtained by replacing ε&d from Eq-7 in Eq-6 : d ⎡η K ⎤ 1 ⎡ 1 1 η K ⎤ η K ⎢ ε& + ε ⎥ = σ + ⎢ + + ⎥σ& + σ&& (Eq-9) dt ⎣GK ⎦ η ⎣G GK η.GK ⎦ G.GK During relaxation, ε& is kept constant, therefore : 1 ⎡ 1 1 η K ⎤ η K σ + ⎢ + + ⎥σ& + σ&& = 0 (Eq-10) η ⎣G GK η.GK ⎦ G.GK The solution concerning this second order equation is : σ 0 ⎡⎛ GK ⎞ ⎛ GK ⎞ ⎤ σ (t) = ⎢⎜α − ⎟exp(−α.t) − ⎜ β − ⎟exp(−β.t)⎥ (Eq-11) α − β ⎣⎝ η K ⎠ ⎝ η K ⎠ ⎦ Where σ 0 = ε 0 .G and α and β are the roots of the following equation : 2 ⎛ G G GK ⎞ G.GK x − ⎜ + + ⎟x + = 0 (Eq-12) ⎝ η η K η K ⎠ h.η K 206 REFERENCES [1] J. Keirsse, C. Boussard-Plédel, O. Sire, O. Loréal, B. Bureau, P. Leroyer, B. Turlin & J. Lucas, “IR Optical Fiber Sensor for Biomedical Applications”, Vibrational Spectroscopy, 32, 23-32 (2003) [2] S. Hocdé, O. Loréal, O. Sire, O. Boussard-Plédel, B. Bureau, B. Turlin, J. Keirsse, P. Leroyer & J. Lucas, “Metabolic imaging of tissues by infrared fibre-optics spectroscopy : a new approach for medical diagnosis”, J. Biomed. Optic, 9, 404-407 (2004) [3] P. Lucas, D. LeCoq, C. Juncker, J. Collier, D. Boesewetter, C. Boussard, B. Bureau & M. Riley, “Evaluation of toxic agent effects on lung cells by Fiber Evanescent Wave Spectroscopy”, Applied Spectroscopy, 59, 1-9 (2005) [4] K. Michel, B. Bureau, C. Boussard-Plédel, T. Jouan, J. L. Adam, K. Staubmann & T. Baumann, “ Monitoring of pollutant in waste water by infrared spectroscopy using chalcogenide glass optical fibers”, Sensor and Actuators B, 101, 252-259 (2004) [5] K. Michel, B. Bureau, C. Boussard-Plédel, T. Jouan, C. Pouvreau, J. C. Sangleboeuf & J. bundles: L. Adam, “ Development of a Chalcogenide glass fiber device for in-situ pollutant detection”, J. Non-Cryst Solids, 327, 434-438 (2003) [6] E. Lebourhis, et al., Scripta Materialia 45, 317 (2001) [7] J. P Guin, T. Rouxel, J. C Sangleboeuf, I. Melscoet, J. Lucas, J. Am. Ceram. Soc. 85, 1545 (2002) [8] C. Pouvreau, et al., 316, 131 (2003) [9] J. J. Mecholsky Jr, C. T. Moynihan, P. B. Macedo, G. R. Srinivasan, “Microstructure and properties of an infra-red transmitting chalcogenide glass-ceramic”, Journal of Materials Science, 11(10), 1952-60 (1976) [10] X. H. Zhang, H. L. Ma & J. Lucas, “A new class of infrared transmitting glass- ceramic based on controlled nucleation and growth of alkali halide in a sulphide based glass matrix”, J. Non-Cryst. Solids, 337, 130-135 (2004) [11] L. Calvez, H. L. Ma, J. Lucas, X. H. Zhang, “Selenium-based glasses and glass- ceramics transmitting light from the visible to the far IR”, Advanced Materials, 19[1], 129-132 (2007) [12] K. Saito, A. J. Ikushima, “Absorption edge in silica glass”, Phy. Rev. B, 62[13], 8584-8587 (2000) 207 [13] C. Kittel, “Introduction to solid state physics”, Ed. I. John Willey and sons (1976) [14] C. Kittel, « Physique de l’état solide », Dunod, Paris, 1998 [15] P. Klocek, M. Roth & R. D. Rock, “Chalcogenide glass optical fibers and image properties and applications”, Opt. Eng. 26 (2), 88 (1987) [16] C. Blanchetière, K. LeFoulgoc, H. L. Ma, X. H. Zhang, and J. Lucas, “Tellurium halide glass fibers : preparation and applications”, J. Non-Cryst. Solids, 184, 200 (1995) [17] L. Le Neindre, F. Smektala, K. LeFoulgoc, X. H. Zhang, and J. Lucas, “Tellurium halide optical fibers”, J. Non-Cryst. Solids, 242, 99 (1998) [18] R. W. Douglas, “The rheology of glass material – a general survey”, In : [188], 3-22 [19] R. W. Douglas, “A general review of relaxation processes”, J. Non-Cryst. Solids, 14, 1-12 (1974) [20] H. E. Hagy, “Rheological behaviour of glass”. In : [93], 4-27 [21] J. Zarzycki, “Les verres et l’état vitreux”, Masson (1982) [22] E. H. Fontana, J. Am. Ceram. Soc., 76 [5], 1081-1087 (1993) [23] C. A. Angell, J. Non-Cryst. Solids, 102, 205-221 (1988) [24] “Glass ceramics”, P. W. Mc MILLAN, Academic Press, 1979, second edition [25] Stookey, S.D., Nucleation. Technol. Press of Massachusetts Inst. Technol. and John Wiley & Sons, Inc., New York, N.Y., 189-95 (1957) [26] Stookey, S. D, Ind. Eng. Chem., 51 [7] 805 (1959) [27] F.Auzel, D. Pecile, D. Morin, “Rare earth doped vitroceramics : new efficient blue and green emitting materials for infrared up-conversion”, J. of the Electrochem. Soc. 122, 101-107 (1975) [28] A. Edgar, S. Schweizer, S. Assmann, J. M. Spaeth, P. J. Newman and D. R. MacFarlane “Photoluminescence and crystallization in europium-doped fluorobromozirconate glassceramics”, J. Non-Cryst Solids, 284[1-3], 237-242, (2001) 208 [29] Daqin Chen, Yuansheng Wang, Yunlong Yu, En Ma and Lihua Zhou, 3+ “Microstructure and luminescence of transparent glass ceramic containing Er :BaF2 nano-crystals”, J. Sol. Stat. Chem., 179 [2], 532-537 (2006) [30] “Advances in Nucleation and Crystallization in Glasses”, Symposium of the Glass Division of The American Ceramic Society, 1971, edited by L. L. Hench and S. W. Freiman, special publication N°5 [31] S. R. Choi, J. P. Gyekenyesi, Results of mechanical testing for Pyroceram glassceramic. NASA/TM, (NASA/TM-2003-212487), 1-84 (2003) [32] “Glass-Ceramic Materials”, Z. Strnad, Publishers of Technical Literature, Prague, 185 (1986) [33] M. Avrami, “Kinetics of phase change. I, General theory”, J. Chem. Phys., 7(12), 1103-1112 (1939) [34] G.H. Beall, W. Holand, “Glass-ceramic technology”, Westerville, OH, American Ceramic Society, 2002 [35] G.H. Beall, Glass-ceramics, Advances in Ceramics, 18, 157-173 (1986) [36] “Glass-Ceramic Materials”, Z. Strnad, Publishers of Technical Literature, Prague, 1986, p185-200 [37] G. H. Beall, Adv. Ceram. 18, 157 (1986) [38] Cheng, Jijian. “Properties and structure of the infrared-transmitting arsenic- germanium -selenium-tin glass-ceramic system” Huadong Huagong Xueyuan Xuebao, 3, 337-51 (1982) [39] Cheng, Jijian. “Separation of phases and glass-ceramic formation of some chalcogenide and oxide glasses”, Glastechnische Berichte, 56(Int. Glaskongr., 13th, Band 1), 590-5 (1983) [40] Cheng Jijian, “Phase separation and controlled crystallization of non-oxide”, J. Non- Cryst. Sol., 80 [1-3], 52-68 (1986) [41] Cheng Jijian, “Phase separation and crystallization of chalcogenide glass-forming systems”, J. Non-Cryst. Solids, 161, 304-308 (1993) [42] H.L. Ma, X. H. Zhang & J. Lucas, “Infrared transmitting chalcogenide glass ceramics”, J. Non-Cryst. Solids, 317[3], 270-274 (2003) 209 [43] H-L. Ma, J. Troles, X.H. Zhang, “Vitrocéramiques transparentes dans les deux fenêtres atmosphériques de 3-5 et 8-12µm”, J. Phys. IV 12 (2002) [44] S. M. Song, S. Y. Choi, Y. K. Lee, “Crystallization property effects in Ge30Se60Te10”, J. Non-Cryst. Solids, 217, 79-82 (1997) [45] G. Dong, H. Tao, X. Xiao, C. Lin, H. Guo, X. Zhao, “Second harmonic generation in transparent microcrystalline α-CdGa2S4 containing chalcogenide glass ceramics”, Optics Communications, 274, 466-470 (2007) [46] V. Seznec, H. L. Ma, X. H. Zhang, V. Nazabal, J. L. Adam, X.S. Qiao and X.P. Fan, “Preparation and luminescence of new Nd3+ doped chloro-sulphide glass–ceramics”, Optical Materials, 29[4], 371-376 (2005) [47] P. Lucas, M. R. Riley, C. Boussard-Pledel, B. Bureau, Anal. Biochem. 351, 1 (2006) [48] V.F. Kokorina, Glass for infrared optics, Ed C.R.C. Press (1996) [49] B. Frumarova et al, ”Synthesis and optical properties of the Ge-Sb-S:PrCl3 glass system“, J. Non-Cryst. Sol., 256, 266 (1999) [50] L. Petit et al, ”Correlation between physical, optical and structural properties of sulphide glasses in the system Ge-Sb-S”, Mat. Chem. Phys, 97, 64 (2006) [51] L. Koudelka, M. Frumar, “Raman spectra of Ge-Sb-S system glasses in the S-rich region”, J. Non-Cryst. Sol., 41, 171 (1980) [52] D. Le Coq, K. Michel, G. Fonteneau, S. Hocdé, C. Boussard-Plédel, J. Lucas, “Infrared chalcogen glasses: chemical polishing and fibre remote spectroscopy”, Int. Journal of Inorg. Mat., 3(3), 233 (2001) [53] K. Taga, D. Schwarzbach, G. Stingeder, M. Grassbauer, and R. Kellner, ”Surface analysis and characterization of chemically modified chalcogenide fibers“, Anal. Chem., 65, 2288 (1993) [54] K. Taga, C. Weigel, and R. Keller, “In situ Fourier Transform infrared analysis of modified chalcogenide infrared transparent fibre surfaces”, Vibrat. Spectrosc., 1, 125 (1990) [55] F. M. Mirabella, “Internal Reflection Spectroscopy”, Appl. Spectro. Rev., 21, 45 (1985) 210 [56] P. Lucas et al., “Evaluation of toxic agent effects on lung cells by Fiber Evanescent Wave Spectroscopy (FEWS)”, Appl. Spectro., 59, 1 (2005) [57] P. Lucas et al., Sens Actuat. B. in press [58] K. Taga, R. Kellner, U. Kainz, and U. B. Sleytr, “In situ attenuated total reflectance FT- IR analysis of an enzyme-modified mid infrared fiber surface using crystalline bacterial surface proteins”, Anal. Chem, 66, 35 (1994) [59] C. X. Yu, A. Ganjoo, H. Jain, C. G. Pantano, and J. Irudayaraj, “Mid-IR biosensor: detection and fingerprinting of pathogens on gold island functionalized chalcogenide films”, Anal. Chem., 78, 2500 (2006) [60] A. Madan, M. P. Shaw, “The Physics and Applications of Amorphous Semiconductors”, Academic Press, Boston, MA, 1988 [61] R. A. Street, Solid State Commun., 24, 363 (1977) [62] K. Tanaka, J. Non-Cryst. Solids, 59, 925 (1983) [63] D. J. Treacy, U. Strom, P. B. Klein, P. C. Taylor, and T. P. Martin, J. Non-Cryst. Solids, 35+36, 1035 (1980) [64] H. Hamanaka, K. Tanaka, and S. Iizima, “Reversible photostructural change in melt- quenched As2S3 glass”, Solid State Commun., 23, 63 (1977) [65] Ph.D Dissertation by C. Russell, “Photodarkening of germanium-selenium glasses induced by below-bandgap light”, University of Florida (2000) [66] A. M. Andriesh et al., “The mechanism of enhancement of photoinduced absorption in As2S3 glass at sub-bandgap illumination”, J. Non-Cryst. Solids, 189, 147 (1995) [67] H. Fritzsche, “Photo-induced fluidity of chalcogenide glasses”, Solid State Commun., 99, 153 (1996) [68] S. R. Elliott, M. Vlcek, P. Krecmer, “Novel photoinduced anisotropy in amorphous As50Se50 films at near the glass transition temperature”, J. Non-Cryst. Solids, 227, 682 (1998) [69] K. Tanaka, H. Hisakuni, “Giant photo-expansion in As2Se3 glass”, Applied Physics Letters, 65, 2925 (1994) [70] P. Lucas et al., “Photoinduced structural relaxation in chalcogenide glasses”, J. Non- Cryst. Solids, 332, 35 (2003) 211 [71] N. F. Mott and E. A. Davis, “Electronic Processes in Non-Crystalline Materials”, 2nd ed. (Clarendon Press, Oxford, 1979) [72] H. Tichá, L. Tichý, and V. Smrcka, “The temperature dependence of the optical gap in quasibinary chalcogenide glasses and their far infrared spectra”, Mater. Lett. 20, 189 (1994) [73] H. Fritzsche, “Electronic properties of amorphous semiconductors”, In Amorphous and Liquid Semiconductors, edited by J. Tauc (Plenum, New York, 1974), p 221 [74] M. Kastner, “Bonding bands, lone pair bands, and impurity states in chalcogenide semiconductors”, Phys. Rev. Lett. 28 (6), 355 (1972) [75] K. Tanaka and H. Hisakuni, “Photoinduced phenomena in As2S3 glass under subbandgap excitation”, J. Non-Cryst. Solids 198-200, 714 (1996) [76] K. Tanaka, “Mechanisms of photodarkening in amorphous chalcogenides”, J. Non- Cryst. Solids 59 & 60, 925 (1983) [77] J. Z. Liu and P. C. Taylor, “Absence of photodarkening in bulk, glassy As2S3 and As2Se3 alloyed with copper”, Phys. Rev. Lett. 59 (17), 1938 (1987) [78] J. Z. Liu and P. C. Taylor, “Effect of Cu alloying on metastable photoinduced absorption in Cux(As0.4Se0.6)1-x and Cux(As0.4S0.6)1-x glasses”, Phys. Rev. B 41 (5), 3163 (1990) [79] S. Rajagopalan, B. Singh, P. K. Bhat, D. K. Pandya, and K. L. Chopra, “Photoinduced optical effects in obliquely deposited amorphous Se-Ge films”, J. Appl. Phys. 50 (1), 489 (1979) [80] M. Okuda, T. T. Nang, and T. Matsushita, “Photo-induced absorption changes in selenium-based chalcogenide glass films”, Thin Solid Films, 58, 403 (1979) [81] E. Márquez, A. M. Bernal-Oliva, J. M. González-Leal, T. Prieto-Alcón, and R. Jiménez-Garay, “On the irreversible photo-bleaching phenomenon in obliquelyevaporated GeS2 glass films”, J. Non-Cryst. Solids 222, 250 (1997) [82] V. K. Tikhomirov and S. R. Elliott, “The anisotropic photorefractive effect in bulk As2S3 glass induced by polarized subgap laser light”, J. Phys.: Condens. Matter 7, 1737 (1995) [83] M. F. Thorpe, J. Non-Cryst. Solids, 57 355 (1983) 212 [84] M. F. Thorpe, H. He, Phys. Rev. Lett., 54, 2107 (1985) [85] P. Lucas, “Energy landscape and photoinduced structural changes in chalcogenide glasses”, J. Phys.: Condens. Matter, 18, 5629 (2006) [86] M. Vlcek, M. Frumar, A. Vidourek, “Photoinduced effects in Ge-Sb-S glasses and amorphous layers”, J. Non-Cryst. Solids, 90, 513 (1987) [87] I. P. Kotsals, D. Papadimitriou, C. Raptis, M. Vlcek, M. Frumar, “Raman study of photostructural changes in amorphous GexSb0.4-xS0.6”, J. Non-Cryst. Solids, 226, 85 (1998) [88] C. Pouvreau, M. Drissi-Habti, K. Michel, B. Bureau, J. C. Sangleboeuf, C. Boussard-Plédel, T. Rouxel, J. L. Adam, « Mechanical properties of a TAS fiber : a preliminary study », J. Non-Cryst. Solids, 316, 131-137 (2003) [89] Z. U. Borisova, Glassy semiconductors, Ed. Plenum Press, New York, 193 (1981) [90] A. R. Hilton, Proc SPIE Infrared Fiber Optics III, 1591, 34-42 (1991) [91] N. J. Pitt, G. S. Sapsford, T. V. Clapp, R. Worthington, M. G. Scott, Proc. Soc. PhotoOptical Instrum. 618, 124 (1986) [92] X. H. Zhang, H. L. Ma and J. Lucas, SPIE. Vol 153, Glasses for Optoelectronics, 11, 209 (1981) [93] H. L. Ma, X. H. Zhang and J. Lucas, J. Solid State Chem, 96, 181 (1992) [94] A. B. Seddon, Proc. 9th International Symposium on non-oxide glasses, Hangzhou, China, 14-24 (1994) [95] D. Lezal, B. Petrovska, G. Kuncova, M. Pospisilova, J. Gotz, Proc. SPIE, 799, 44-51 (1987) [96] X. H. Zhang, H. L. Ma, C. Blanchetière, J. Lucas, J. Non-Cryst. Solids, 161, 327 (1993) [97] Y. Park, U.-C. Paek, S. Han, B.-H. Kim, C.-S. Kim, D. Y. Kim, “Inelastic frozen-in stress in optical fibers”, Optics Communications, 242, 431 (2004) [98] P. L. Chu, T. Whitbread, “Measurement of Stresses in Optical Fiber and Preform”, Appl. Opt., 21, 4241 (1982) [99] D. J. Green, “Critical parameters in the processing of engineered stress profile glasses”, J. Non-Cryst. Solids, 316, 35-41 (2003) 213 [100] R. D. Driver, G. M. Leskowitz, L. E. Curtiss, D. E. Moynihan, L. B. Vacha, Mater. Res. Soc. Symp. Proc, 172, 169 (1990) [101] L. Landau, E. Lifshitz, “Theory of Elasticity” (Theoretical Physics, Vol 7), 3rd Ed., Butterworth-Heinemann (1986) [102] M. T. Laugier, “The elastic/plastic indentation of ceramics”, J. Mat. Sci. Lett., 4, 1539-1541 (1985) [103] A. A. Griffith, Phil. Trans. Roy. Soc. London, A 221, 163-198 (1920) [104] W. Weibull, A. Statistical Theory of the Strength of Matereials, Ingenior Vetenskaps Akademien Handlingar, Nr 151 (1939) [105] J. J. Mecholsky, Fracture Surface Analysis of optical fibers [106] Thèse Karine Michel, “Capteur optique à fibre infrarouge dédié à la detection et à l’analyse de la pollution de l’eau”, Université Rennes 1, 2003 [107] P. Lucas, E. A. King, Y. Gueguen, J.-C., R. Erdman, G. Delaizir, C. Boussard- Pledel, B. Bureau, X.-H. Zhang, T. Rouxel, “Relaxation processes in chalcogenide glass fibers in relation to variations in mechanical properties”, in press [108] R. Böhmer and C. A. Angell, “Elastic and viscoelastic properties of amorphous selenium and identification of the phase transition between ring and chain structures”, Phys. Rev. B, Vol 48, 9 (1993) [109] T. A. Michalske, S. W. Freiman, “A molecular interpretation of stress corrosion in silica”, Nature, 295, 511-512 (1982) [110] “Le Verre, Science et Technologie”, J. Barton and C. Guillemet, EDP Sciences, p 60 (2005) [111] C. Bernard, G. Delaizir, J.-C. Sangleboeuf, V. Keryvin, P. Lucas, B. Bureau, X.-H. Zhang, T. Rouxel, « Room temperature viscosity and delayed elasticity in infrared glass fiber », J. European Ceram. Soc., 27, 3253 (2007) [112] M. Koide, T. Komatsu, K. Matusita, « Delayed elasticityand viscosity of silicate glasses below their glass transition temperature », Thermochimica Acta, 282, 345-351 (1996) [113] D. Roylance, « Engineering Viscoelasticity », Department of Materials Science and Enginerring, Massachusetts Institute of Technology (2001) 214 [114] C. R. Kurkjian, Phy. Chem Glasses, 4, 128 (1963) [115] S. M. Rekhson, Glass Science and Technology, Vol 3, Viscosity and Relaxation, Ed Ulhmann and Kreidl, Academic Press, New York (1986) [116] R. Mao, J. Tang, B. G. Swanson, « Relaxation Time Spectrum of Hydrogels by CONTIN Analys », J. of Food Science, Vol 65, 3 (2000) [117] C. Machiraju, A. V. Phan, A. W. Pearsall, S. Madanagopal, « Viscoelastic studies of human subscapularis tendon : Relaxation test and a Wiechert model », Computer methods and programs in biomedicine, 83, 29-33 (2006) [118] G. Dimitreli, A. S. Thomareis, « Texture evaluation of block-type processed cheese as a function of chemical composition and in relation to its apparent viscosity », J. Food Enginnering, 79, 1364-1373 (2007) [119] K. L. Ngai, J. Non-Cryst. Solids, 131-133, 80-83 (1991) [120] R. G. Palmer, D. L. Stein, E. Abrahams, P. W. Anderson, Phys. Rev. Lett., 53, 953 (1994) [121] V. Keryvin, M-L. Vaillant, T. Rouxel, M. Huger, T. Gloriant, Y. Kawamura, Intermetallics, 10, 1289-1296 (2002) [122] K. Trachenko, M. T. Dove, Phys. Rev. B, 70, 132202 (2005) [123] R. Böhmer, C. A. Angell, “Elastic and viscoelastic properties of amorphous selenium and identification of the phase transition between ring and chain structures”, Phys. Rev. B, Vol 48, 9 (1993) [124] D. J. McEnroe, C. La Course, “Tensile strength of Se, As2S3, As2Se3 and Ge30As15Se55 glass fibers”, J. Am. Ceram. Soc., 72, 1491-1494 (1989) [125] J. Tang, M. A. Tung, Y. Zeng, “Characterization of gellan gels using stress relaxation”, J. Food Engineering, 38, 279-295 (1998)