STUDIES OF TRAPPING AND LUMINESCENCE PHENOMENA IN YTTRIUM
ALUMINUM GARNETS
By
CHRISTOPHER RICHARD VARNEY
A thesis submitted in partial fulfillment of the requirements for the degree of
DOCTOR OF PHILOSOPHY
WASHINGTON STATE UNIVERSITY Department of Physics and Astronomy
December 2012
© Copyright by CHRISTOPHER RICHARD VARNEY, 2012 All Rights Reserved
© Copyright by CHRISTOPHER RICHARD VARNEY, 2012 All Rights Reserved
To the Faculty of Washington State University:
The members of the Committee appointed to examine the thesis of CHRISTOPHER RICHARD VARNEY find it satisfactory and recommend that it be accepted.
______Farida A. Selim, Ph.D., Co-Chair
______Matthew D. McCluskey, Ph.D., Co-Chair
______Gary S. Collins, Ph.D.
ii ACKNOWLEDGEMENT
I first would like to thank my parents for their love and support. I am grateful for all the years they spent pushing me to work hard and prioritize my studies. Without their guidance, I could have never become the physicist I am today. I am fortunate to have had such an upbringing.
I am grateful for the efforts of my advisor, Dr. Farida Selim, under whose guidance I was able to find myself as a physicist and a researcher. I am indebted to the time and patience she afforded me during this dissertation.
My committee: Drs. Farida Selim, Matt McCluskey, and Gary Collins, for their time and trouble in reviewing this manuscript. Each member played a crucial role in my development during my graduate career and it is fitting that they comprise my committee.
My co-workers I worked with in this lab: Autumn Pratt, Jianfeng Ji, David
Mackay, Sherif Reda, Mohammad Khamenchi, John Buscher, Frederick Chen, and
Kimberly Heiner. I am grateful for their willingness to help and often do the more mundane tasks. I also thank Mike Rowe and Marianne Tarun for the measurements they conducted.
There were countless members of the faculty and staff of the physics and astronomy department at Washington State University who have helped me along the way and I could thank every one of them for the unique way they have affected my graduate experience. I would like to thank Dr. Fred Gittes, for taking me on to work on the bloodstain pattern analysis research project I did under him and helping me prepare for the preliminary examination; Dr. Gordon Johnson, for helping me teach physics to others which in turn made me a more capable physicist; Dr. Yogendra Gupta, for
iii developing me into the scientist I am today and helping me realize my goals; Tom
Johnson, for all the assistance he provided and all the hours lost discussing Cougar basketball; Dr. Mark Kuzyk, for providing floor hockey as an escape from work and studies; Dr. Mike Allen, for being a good friend; and the wonderful ladies in the physics office, in particular Sabreen Dodson, Laura Krueger, and Mary Guenther, who kept me on track throughout my graduate career.
I am grateful for the friends I made along the way while working toward my doctorate. Time spent with them seemed to make all the hard work worth it, especially when we could all commiserate together about struggles with research or classes.
I wish to thank my wonderful girlfriend, Elora DeGreef, and her family for all the support and encouragement they have shown me for as long as I have known them. Elora is my love and inspiration, and it is to her that this thesis is dedicated.
This work was supported by the National Science Foundation, who funded me and my work under Grant No. DMR 10-06772.
iv STUDIES OF TRAPPING AND LUMINESCENCE PHENOMENA IN YTTRIUM
ALUMINUM GARNETS
Abstract
by Christopher Richard Varney, Ph.D. Washington State University December 2012
Chair: Farida A. Selim
Rare-earth-doped yttrium aluminum garnet (YAG) crystals are important photonic materials with a wide range of applications. The optical properties and performance of these crystals are largely governed by exciton dynamics, which is greatly affected by the presence of defects. In this work, the optical and scintillation properties of undoped and rare-earth-doped YAG crystals were studied in conjunction with thorough characterization of defects. First, a comprehensive study of optical properties, including absorption and luminescence, was carried out. Color centers were found to be present in both undoped and doped crystals. A new x-ray luminescence spectrometer was developed and installed to investigate both luminescence and scintillation properties in a new way. Defects that trap excitons were studied by thermoluminescence spectroscopy.
Deep and shallow traps were identified by high and low thermoluminescence measurements and their energy levels in the band gap were calculated. Positron annihilation lifetime spectroscopy was carried out for the first time on YAG to provide
v information about defect types, structures, and concentrations. The positron measurements revealed the presence of isolated aluminum vacancies and defect complexes of aluminum and oxygen vacancies. They also revealed the dependence of defect structure on growth atmosphere and post-growth treatments. This knowledge gained from thermoluminescence and positron lifetime measurements elucidated the effects various defects have on scintillation properties and suggested ways to control them. Lastly, a new fast scintillator with good energy resolution is discussed.
vi TABLE OF CONTENTS
ACKNOWLEDGEMENTS…………………………………………………………...iii-iv
ABSTRACT……………………………………………………………………………v-vi
TABLE OF CONTENTS……………………………………………………………vii-viii
LIST OF TABLES……………………………………………………………………..…ix
LIST OF FIGURES……………………………………………………………………...xii
CHAPTER
1. INTRODUCTION……………………….………………………………………..1
1.1. REFERENCES….……………………………………………………………8
2. EXPERIMENTAL METHODS……………………………...…………………..15
2.1. SAMPLES…………………………………...……………………………...15
2.2. NEWLY DEVELOPED EXPERIMENTAL TECHNIQUE TO STUDY
LUMINESCENCE…………………………....…………………………….21
2.3. REFERENCES……………………………………………………………...23
3. OPTICAL PROPERTIES OF YAG…………………………………………..…24
3.1. ABSORPTION……………………………………………………………...25
3.1.1. RESULTS……………………………………………………………...29
3.2. PHOTOLUMINESCENCE…………………………………………………42
3.2.1. RESULTS……………………………………………………………...46
3.3. COLOR CENTERS…………………………………………………………66
3.3.1. RESULTS……………………………………………………………...73
3.4. RADIOLUMINESCENCE.…………………………………………………83
vii 3.4.1. RESULTS……………………………………………………………...87
3.5. REFERENCES…………………………………………………………….102
4. DEFECT CHARACTERIZATION BY TRAP LEVEL MEASUREMENTS....117
4.1. THEORY…………………..….…………………………………………...123
4.2. RESULTS………………………………………………………………….135
4.3. REFERENCES…………………………………………………………….165
5. DEFECT IDENTIFICATION BY POSITRON LIFETIME
MEASUREMENTS…………………………………………………………….170
5.1. METHODS………………………………………………………………...174
5.2. RESULTS………………………………………………………………….185
5.3. REFERENCES…………………………………………………………….210
6. SCINTILLATION PROPERTIES……………………………………………...216
6.1. METHODS………………………………………………………………...218
6.2. RESULTS………………………………………………………………….224
6.3. REFERENCES…………………………………………………………….249
7. CONCLUSION…………………………………………………………………253
8. APPENDIX A…………………………………………………………………..258
viii LIST OF TABLES
2.1.1 List of samples studied in this thesis…………..…...………………………………17
4.2.1 Thermoluminescence activation energies of Ar-grown undoped YAG (integration range: 340-570 nm)……..………………………………………………………………151
4.2.2 Thermoluminescence activation energies of Ar-grown undoped YAG (integration range: 570-800 nm)…………………….…………………….…………………………151
4.2.3 Thermoluminescence activation energies of O 2-grown undoped YAG (integration range: 340-570 nm)……………………………………………………………………..152
4.2.4 Thermoluminescence activation energies of O 2-grown undoped YAG (integration range: 570-800 nm)……………………………………………………………………..153
4.2.5 Thermoluminescence activation energies of Ce:YAG 0.1%, 0.15%, and 0.2%.....162
4.2.6 Thermoluminescence activation energies of Ce:YAG 0.2% calculated by the initial
rise and corrected initial rise methods………………………………………………….162
4.2.7 Thermoluminescence activation energies of Ce:YAG 0.3% ………………….....163
4.2.8 Thermoluminescence activation energies of Ce:YAG 0.14%...... 163
5.1.1 Summary of equipment settings used for PALS experiments……………………178
5.2.1 Positron lifetimes in polished Ar-grown undoped YAG 5x5x1………………….186
5.2.2 Positron lifetimes in unpolished (fine ground) Ar-grown undoped YAG
10x10x1…………………………………………………………………………………186
5.2.3 Positron lifetimes in unpolished (fine ground) Ar-grown undoped YAG 5x5x1...186
5.2.4 Positron lifetimes in H 2-grown undoped YAG 10 mm dia. x ~1 mm……………189
5.2.5 Positron lifetimes in O 2-grown undoped YAG 10 mm dia. x ~1 mm……………189
ix 5.2.6 Positron lifetimes in Ce:YAG 0.1%...... 196
5.2.7 Positron lifetimes in Ce:YAG 0.15%...... 199
5.2.8 Positron lifetimes in Ce:YAG 0.2%...... 200
5.2.9 Positron lifetimes in Ce:YAG 0.14% 10 mm dia. x ~1 mm……………………...202
5.2.10 Positron lifetimes in polished Ce:YAG 0.3%...... 204
5.2.11 Positron lifetimes in unpolished (fine ground) Ce:YAG 0.3%...... 204
5.2.12 Positron lifetimes in Nd:YAG 1%...... 206
5.2.13 Positron lifetimes in Tm:YAG 0.8%...... 206
5.2.14 Positron lifetimes in Yb:YAG 5%...... 208
5.2.15 Positron lifetimes in Yb:YAG 10%...... 208
5.2.16 Positron lifetimes in the undoped YAG sample mislabeled as Yb:YAG 5%...... 209
6.1.1 Key specifications for R329 photomultiplier tube………………………………..221
6.2.1 Energy resolution and decay time for 0.662 MeV photopeak of 137 Cs using Ce:YAG
samples as scintillator crystals………………………………………………………….230
6.2.2 Energy resolution and decay time for 0.662 MeV photopeak of 137 Cs using undoped
YAG samples as scintillator crystals…………………………………………………...242
A.1 Positron lifetimes in polished Ar-grown undoped YAG 5x5x1………………..….258
A.2 Positron lifetimes in unpolished (fine ground) Ar-grown undoped YAG 10x10x1.259
A.3 Positron lifetimes in unpolished (fine ground) Ar-grown undoped YAG 5x5x1.....260
A.4 Positron lifetimes in H 2-grown undoped YAG 10 mm dia. x ~1 mm……………..261
A.5 Positron lifetimes in O 2-grown undoped YAG 10 mm dia. x ~1 mm……………..262
A.6 Positron lifetimes in Ce:YAG 0.1%...... 265
A.7 Positron lifetimes in Ce:YAG 0.15%...... 268
x A.8 Positron lifetimes in Ce:YAG 0.2%...... 268
A.9 Positron lifetimes in Ce:YAG 0.14% 10 mm dia. x ~1 mm…………………….....270
A.10 Positron lifetimes in polished Ce:YAG 0.3%...... 271
A.11 Positron lifetimes in unpolished (fine ground) Ce:YAG 0.3%...... 272
A.12 Positron lifetimes in Nd:YAG 1%...... 273
A.13 Positron lifetimes in Tm:YAG 0.8%...... 273
A.14 Positron lifetimes in Yb:YAG 5%...... 274
A.15 Positron lifetimes in Yb:YAG 10%...... 275
A.16 Positron lifetimes in the undoped YAG sample mislabeled as Yb:YAG 5%...... 276
xi LIST OF FIGURES
2.2.1 Schematic diagram of x-ray-stimulated luminescence and some photoluminescence measurements…………………………………………………………………………….21
3.1.1 Absorption spectra of undoped YAG...... 30
3.1.2 Absorption spectra of annealed Ar-grown undoped YAG…….…………………..31
3.1.3 Absorption spectra of annealed O2-grown undoped YAG………...... 32
3.1.4 Absorption spectra of Ce:YAG…………………………………………………….33
3.1.5 Absorption spectra of unpolished Ce:YAG 0.3%...... 35
3.1.6 Absorption spectra of polished Ce:YAG 0.2%...... 36
3.1.7 Absorption spectra of unpolished (as cut) Ce:YAG 0.14%...... 37
3.1.8 Absorption spectra of Nd:YAG 1% with transitions marked……………………...38
3.1.9 Absorption spectra of Tm:YAG 0.8% with transitions marked...... 39
3.1.10 Absorption spectra of Yb:YAG…………………………………………………..40
3.1.11 Absorption spectra of Yb:YAG 10%...... 41
3.2.1 Photoluminescence emission spectra of undoped YAG for 270 nm excitation..…..47
3.2.2 Photoluminescence excitation spectra of undoped YAG for 385 nm emission...... 47
3.2.3 Photoluminescence emission spectra of undoped YAG for 270 nm LED
excitation...... 49
3.2.4 Photoluminescence emission spectra of Ar-grown undoped YAG for 190 nm
excitation………………………………………………………………...... 50
3.2.5 Photoluminescence excitation spectra of undoped YAG for 325 nm emission…....50
3.2.6 Photoluminescence emission spectra of undoped YAG for 195 nm excitation…....51
xii 3.2.7 Photoluminescence excitation spectra of undoped YAG for 700 nm emission…....53
3.2.8 Photoluminescence emission spectra of undoped YAG for 450 nm excitation..…..53
3.2.9 Photoluminescence emission spectra of undoped YAG for 425 nm excitation...... 54
3.2.10 Photoluminescence emission spectrum of undoped YAG (labeled as Yb:YAG 5%) for 230 nm excitation………………………………………………………………….....56
3.2.11 Photoluminescence emission spectrum of undoped YAG (labeled as Yb:YAG 5%) for 270 nm LED excitation……………………………………...…………………….....57
3.2.12 Photoluminescence of undoped YAG (labeled as Yb:YAG 5%): emission spectrum for 460 nm excitation and excitation spectrum for 590 nm emission…………58
3.2.13 Photoluminescence excitation spectra of Ce:YAG for 550 nm excitation…….....60
3.2.14 Photoluminescence emission spectra of as grown Ce:YAG for 455 nm LED excitation…………………………………………………………………………………60
3.2.15 Photoluminescence emission spectra of annealed Ce:YAG for 455 nm LED excitation…………………………………………………………………………………61
3.2.16 Photoluminescence emission spectra of Ce:YAG for 270 nm excitation………...63
3.2.17 Photoluminescence emission spectra of Ce:YAG for 270 nm LED excitation…..63
3.2.18 Photoluminescence emission spectra of Yb:YAG 5% for 273 nm excitation……65
3.3.1 Absorption spectra shift of Ar-grown YAG after UV excitation and heating…...... 73
3.3.2 Absorption spectra shift of air annealed Ar-grown YAG after UV excitation and heating……………………………………………………………………………………74
3.3.3 Absorption spectra shift of undoped YAG after UV excitation…………………...74
3.3.4 Absorption spectra shift of undoped YAG (labeled as Yb:YAG 5%) after UV excitation…………………………………………………………………………………77
xiii 3.3.5 Absorption spectra shift of Nd:YAG 1% and Tm:YAG 0.8% after UV excitation…………………………………………………………………………………78
3.3.6 Absorption spectra shift of Yb:YAG after UV excitation…………………………78
3.3.7 Absorption spectra shift of Yb:YAG 10% after UV excitation……………………79
3.4.1 Schematic diagram of radioluminescence measurements conducted with the 60 Co source…………………………………………………………………………………….84
3.4.2 Schematic diagram of x-ray-stimulated luminescence and some photoluminescence measurements…………………………………………………………………………….85
3.4.3 X-ray-stimulated luminescence spectra of undoped YAG………...………………87
3.4.4 X-ray-stimulated luminescence spectra of Ce:YAG (65 second integration)...... 91
3.4.5 X-ray-stimulated luminescence spectra of Ce:YAG (20 second integration)……..92
3.4.6 Radioluminescence spectra of Ce:YAG…………………………………………...93
3.4.7 X-ray-stimulated luminescence spectra of Al sputtered H2-grown undoped YAG..95
3.4.8 X-ray-stimulated luminescence spectra of Al sputtered O2-grown undoped YAG..95
3.4.9 X-ray-stimulated luminescence spectra of Al sputtered Ce:YAG 0.2%...... 96
3.4.10 X-ray-stimulated luminescence spectra of vacuum annealed Ce:YAG 0.14%...... 97
3.4.11 X-ray-stimulated luminescence spectra of annealed Nd:YAG 1% and Tm:YAG
0.8%...... 98
3.4.12 X-ray-stimulated luminescence spectra of Yb:YAG…………………………...... 98
3.4.13 Radioluminescence spectrum of Nd:YAG 1%...... 100
3.4.14 X-ray-stimulated luminescence spectra of O 2-grown undoped YAG before and
after UV excitation……………………………………………………………….……..101
xiv 4.0.1 Schematic diagram of the thermoluminescence process after thermal activation of trapped charge carriers………………………………………………………………….118
4.0.2 Glow curves of Ce:YAG 0.2% at multiple heating rates…………………………120
4.0.3 Schematic diagram of thermoluminescence measurements……………………...121
4.1.1 Glow curve of Ce:YAG 0.2% demonstrating corrected initial rise parameters….127
4.2.1 Contour plot of Ar-grown undoped YAG thermoluminescence………………....136
4.2.2 Glow curves of undoped YAG…………………………………………………...138
4.2.3 Thermoluminescence emission spectrum of Ar-grown undoped YAG at 252°C..139
4.2.4 Glow curves of annealed Ar-grown undoped YAG……………………………...140
4.2.5 Glow curves of annealed O2-grown undoped YAG……………………….……..142
4.2.6 Glow curves of undoped YAG demonstrating shift of high temperature peak after
oxidative or reductive annealing………………………………………………………..144
4.2.7 Contour plot of O 2-grown undoped YAG thermoluminescence………………….145
4.2.8 Thermoluminescence emission spectrum of annealed O 2-grown undoped YAG at -
138°C…………………………………………………………………………………...145
4.2.9 Low temperature glow curves of H 2-grown (integration range: 191-450 nm) and O 2- grown (integration range: 340-570 nm) undoped YAG………………………..……...147
4.2.10 Low temperature glow curves of H 2-grown (integration range 450-700 nm) and
O2-grown (integration range: 570-800 nm) undoped YAG……………………...……..148
4.2.11 Glow curve of undoped YAG (labeled as Yb:YAG 5%) (integration range 470-720 nm)……………………………………………………………………………………...149
4.2.12 Low temperature glow curves of Ar-grown undoped YAG and undoped “Yb:YAG
5%”-labeled undoped YAG…………………………………………………………….150
xv 4.2.13 Contour plot Ce:YAG 0.14% thermoluminescence……………………………..155
4.2.14 Thermoluminescence emission spectrum of Ce:YAG 0.14% at 262°C………...155
4.2.15 Glow curves of Ce:YAG………………………………………………………...156
4.2.16 Glow curves of annealed Ce:YAG 0.3%...... 157
4.2.17 Glow curves of annealed Ce:YAG 0.15%...... 158
4.2.18 Glow curves of annealed Ce:YAG 0.2%...... 158
4.2.19 Low temperature glow curves of annealed Ce:YAG 0.14%...... 160
5.1.1 Schematic diagram of the sandwich configuration setup used in positron annihilation lifetime spectroscopy measurements……………………………………...175
5.1.2 Schematic diagram of the electronics setup used for positron annihilation lifetime spectroscopy measurements……………………………………………………………177
5.2.1 Positron lifetime fit for a measurement conducted on polished Ar-grown undoped
YAG 5x5x1 fit to two lifetimes………………………………………………………..188
5.2.2 Positron lifetime fit for a measurement conducted on as grown and annealed H 2-
grown undoped YAG fit to one and two lifetimes, respectively……………...………..194
5.2.3 Comparison of PALS measurements conducted on undoped YAG samples…….195
5.2.4 Comparison of raw data from positron annihilation lifetime spectroscopy
measurements conducted on undoped and Ce-, Nd-, and Tm-doped YAG…………….207
6.1.1 Schematic diagram of scintillation measurements conducted using γ rays………219
6.1.2 Representative decay measurement on Ce:YAG 0.2% Al sputtered and Ar annealed using 137 Cs source………………………………………………………………………223
6.2.1 X-ray stimulated luminescence of BaF 2 and Tl:NaI reference crystals…………..226
xvi 6.2.2 Pulse height spectra of 137 Cs using Ce:YAG samples of various sizes as scintillator crystals………………………………………………………………………………….227
6.2.3 Pulse height spectra of 137 Cs using various 10 mm diameter x 1 mm thick Ce:YAG samples as scintillator crystals………………………………………………………….228
6.2.4 Pulse height spectra of 137 Cs using various 10 mm diameter x 1 mm thick Ce:YAG samples as scintillator crystals, focused on the photopeak……………………………..228
6.2.5 X ray stimulated luminescence spectra of Ce:YAG to demonstrate UV emission and reabsorption by Ce 3+ ions…………………………………………….………………....232
6.2.6 Decay measurement on Ce:YAG 0.3% and Ce:YAG 0.1% using 137 Cs source….233
6.2.7 Luminescence spectra of Ce:YAG 0.15% at various x-ray powers………………234
6.2.8 Light yield calculated by plotting the integrated 530 nm peak (characteristic of
Ce:YAG emission) versus x-ray tube power for Ce:YAG 0.15%...... 234
6.2.9 Luminescence decay measurement of Ce:YAG 0.14% at 530 nm...... 235
6.2.10 Decay measurement on Ar-grown undoped YAG using 137 Cs source………….237
6.2.11 X ray stimulated luminescence of undoped YAG to demonstrate UV emission and
compare UV emission to BaF 2………………………………………………………….238
6.2.12 Pulse height spectra of 137 Cs using various 10 mm diameter x 5 mm thick undoped
YAG samples as scintillator crystals, compared to Ce:YAG 0.14%...... 239
6.2.13 Pulse height spectra of 137 Cs using various 10 mm diameter x 5 mm thick undoped
YAG samples as scintillator crystals, compared to Ce:YAG 0.14%, focused on the photopeak…………………………………………………………...... 239
6.2.14 Pulse height spectra of 137 Cs using various 10 mm diameter x 1 mm thick undoped
YAG samples as scintillator crystals, compared to Ce:YAG 0.15%...... 240
xvii 6.2.15 Pulse height spectra of 137 Cs using various 10 mm diameter x 1 mm thick undoped
YAG samples as scintillator crystals, compared to Ce:YAG 0.15%, focused on the photopeak……………………………………………………………………………….241
6.2.16 Luminescence decay measurement of Ar-grown undoped YAG at 300 nm...... 244
6.2.17 Luminescence decay measurement of Ar-grown undoped YAG at 320 and 385 nm………………………………………………………………………………………245
60 6.2.18 Pulse height spectra of Co using O 2-grown YAG………...... 247
xviii CHAPTER ONE
INTRODUCTION
Garnet crystals are complex oxides of the form A 3B2C3O12 , where A, B, and C are cations, typically metals, of different sizes and B and C can be the same element. The cations are organized with dodecahedral (A site), octahedral (B site), and tetrahedral (B site) coordination of oxygen ions. The unit cell of a garnet is a cubic structure containing of 8 characteristic A 3B2C3O12 molecules, or 160 individual atoms, and belongs to the space group Ia3d. Garnets can possess properties attractive to many optical applications such as a phosphor, scintillator, or laser host material. One garnet crystal that is widely used in optical applications is yttrium aluminum garnet (Y 3Al 5O12 , YAG). In YAG, the yttrium ion sits at the dodecahedral site while the aluminum ion sits at the octahedral and tetrahedral sites and it has a lattice constant of 12 Å [Geller-1967].
Upon the discovery of YAG as a photonic crystal, credited to Geusic et al. in
1964 [Geusic-1964], it became immediately clear that YAG possesses many qualities advantageous to several applications in optics. Mechanically, it is a hard crystal with a
Mohs hardness around 8 [Sirdeshmukh-2001] and a density of 4.55 g/cm 3, has a low thermal expansion, and has stability against chemical and mechanical changes. The YAG crystal is optically isotropic, has high optical transparency and low acoustic loss, and can accept substitutionally trivalent ions of rare-earth (RE) and iron group elements. RE- doped single crystal (SC) YAG is useful as a laser material [Geusic-1964], scintillator
[Moszynski-1994], and phosphor [Blasse-1967(a)], to name a few of the more common applications. YAG is widely used in many photonic applications, but its use as a
1 scintillator has not yet achieved its predicted potential due to intrinsic defects in the crystal lattice.
In our work, we characterize the defect structure of YAG to eliminate defects and optimize luminescent output. The nature and density of defects in the YAG lattice can have a substantial impact on the energy transfer processes within the crystal [Gibbons-
1973, Nikl-2005, Pankove-1971, Robbins-1979(a-d), Rotman-1985,Rotman-1989,
Stanek-2008, Vedda-2004, Weber-2004, Wong-1984, Zorenko-2011, Zych-2000] via phenomena which will be further discussed in later chapters. The most important example of defects affecting crystal performance in YAG is their influence on the scintillation properties of Ce:YAG, in which they both decrease luminescence output and increase luminescence decay time, both of which degrade scintillation performance.
Although studies into the defects inherent to YAG crystals have been conducted for several decades, our understanding about defects in YAG is still very limited. F- centers (oxygen vacancies with two trapped electrons), F +-centers (oxygen vacancies with one trapped electron) [Springis-1991, Zorenko-2010(a)], and even F --centers
(oxygen vacancies with three trapped electrons) have been suggested to exist in YAG and
play a role in trapping and color center dynamics [Pujats-2001]. The self-trapped exciton
(STE) [Bernhardt-1978, Murk-1995, Zorenko-2004(b), Zorenko-2007, Zorenko-2011,
Zorenko-2012] and self-shrinking exciton (SSE) are proposed defect-related excitonic
phenomena present in YAG [Murk-1997]. An STE can occur when one charge carrier
(electron or hole, the positively-charged quasiparticle indicating the absence of an
electron) in a bound electron and hole pair, called an exciton, becomes trapped by a
lattice distortion, be it created by thermodynamic vibrations or otherwise, and the exciton
2 lacks the energy necessary for recombination [Williams-1990]. In YAG, it is believed that an Al vacancy can act as a trap for STE formation [Hayes-1980, Kirm-2000]. An
SSE is similar to an STE except that neither the electron nor hole can self-trap separately
[Murk-1997]. Potential defects resulting from nonstoichiometry include cation or anion vacancies, accidental doping of a trace impurity, and antisite defects [Hewitt-2011,
Kukllja-1999, Kuklja-2000, Landron-1996, Liu-2009, Milanese-2004, Rotman-1990,
Stanek-2008]. In YAG, especially in crystals grown from melts at high temperatures, nonstoichiometry arises from an aluminum deficiency in the crystal, which occurs even when the initial melt has stoichiometric composition [Ashurov -1977]. The resulting
3+ surplus of yttrium is most often accounted for in the form of antisite defects Y Al , where a yttrium ion occupies the location in the lattice that would normally contain an aluminum ion, which have been theoretically calculated to prefer the octahedral position over the tetrahedral position of aluminum [Kuklja-1999, Kuklja-2000, Liu-2009,
Milanese-2004, Patel-2008] and proved to exist in YAG samples due to precision composition and lattice constant measurements [Geusic-1972]. Trace amounts of unintentional dopants, such as Cr, Fe and H are determined to be present in some of our samples. The antisite defect can also be considered an unintentional trace dopant in a way, as one type of ion is located in a position that would under normal circumstances be occupied by a different ion. The different lattice structure surrounding the Y ion in the octahedral Al site combined with the lattice distortion caused by the incorporation of the much larger yttrium ion is hypothesized to cause a perturbation in energy levels of oxygen or yttrium ions, which creates energy levels within the forbidden band gap of
YAG [Zorenko-2007]. Thus, the antisite defect can effectively act the similarly to a
3 dopant. All of the defects mentioned above have the potential to trap charge carriers, thus lowering the efficiency and increasing the luminescence decay lifetime of luminescence centers in YAG, which could prove detrimental to such applications as
Ce:YAG as a scintillator. The effects these defects have on the optical properties of
YAG crystals and possible post-growth annealing treatments to treat some of them will be explored in the scope of this thesis.
This work focuses on YAG undoped and doped with various REs such as Ce, Nd,
Tm, and Yb grown under various conditions from different distributors, both before and after post-growth annealing treatments. Though undoped YAG does not have very many direct applications, studying these crystals will help to distinguish which defects are affected by the dopants in the doped samples and what sort of defects are inherent to the
YAG lattice itself in our samples. Ce:YAG has widespread use as a phosphor and has potential to be a very good scintillator material. It has actually been suggested that the search for a good scintillator material has so thoroughly exhausted the periodic table that there is little hope of finding a revolutionary new scintillator material, emphasizing the importance of defect characterization and elimination in existing scintillator materials to maximize performance [Weber-2004]. However, Ce 3+ doping has been shown to
increase the concentration of lattive defects [Jung-2007]. Nd:YAG has been the most
impactful laser crystal since Geusic et al. demonstrated its quality in 1964 [Geusic-1964].
Tm:YAG is also used in lasers [Kubo-1992]. Yb:YAG has seen increased use as a laser
crystal in recent years and also has been suggested to have promise in neutrino detection
[Guerassimova-2001, Guerassimova-2002]. Studying the effects that various dopants
have on YAG serves to further some of the knowledge gained from studying undoped
4 YAG. Through these studies, we can understand the defects intrinsic to the YAG lattice itself, how dopant size and electronic configuration affect and are affected by those defects and if the dopant itself can introduce new defects.
The bulk of this thesis will focus on defect characterization and luminescence phenomena studies of single crystal YAG. The first step will be to examine the optical properties of YAG. Absorption, photoluminescence, color center, and radioluminescence measurements were conducted to this end. The allowed electron transitions that largely govern these processes are determined by the band gap of YAG and the energy levels of the dopant and defects. The band gap of YAG lies near 190 nm, or about 6.5 eV. The dopants studied in this work have both ground state and excited states within the forbidden band gap of YAG, making it easy to observe direct transitions between energy levels of the dopant via absorption and photoluminescence measurements. Some defects, especially certain unintentional trace dopants, can also be measured this way. A new spectrometer was developed for radioluminescence (RL) measurements using x rays for excitation [Varney-2012(c)]. In RL measurements, excitons are excited to the conduction band of the crystal, which can then become trapped, decay nonradiatively, or emit a photon to recombine with holes in the valence band and return to a relaxed state.
Whereas photoluminescence measurements rely on directly pumping specific energy level transitions to observe particular luminescence phenomena, radioluminescence allows all luminescence phenomena characteristic of the crystal to be observed simultaneously. A potential process that can occur during excitation instead of prompt recombination with an oppositely-charged carrier is that a charge carrier can become trapped at a defect in the crystal, such as an electron becoming trapped at an oxygen
5 vacancy (F-type center). This can discolor the crystal, which lends itself to the name
“color center.” These optical studies will present an overview of all luminescence phenomena of the crystals.
Next this thesis investigates the potential depth of charge carriers by thermoluminescence (TL), which is a powerful method by which to investigate the presence of charge carrier trapping centers and their depth and abundance. TL typically initiates with photon absorption during a period of ionizing radiation. This leads to exciton formation and potential trapping at defects and trapping sites. When the sample is heated, a trapped charge carrier can acquire enough thermal energy to escape the trap and it can recombine with a charge carrier of the opposite sign, be it electron or hole, which can result in photon emission if the recombination process is radiative.
Information regarding the depth and abundance of traps can be discerned from the dependence of luminescence on temperature. TL is used often in dating and dosimetry due to its reliable and reproducible results and those same qualities make TL very useful in crystal defect investigations.
Defects in our YAG crystals are identified by positron annihilation lifetime spectroscopy (PALS). This is the first set of PALS measurements conducted on the garnet structure and one of the first conducted on complex oxides [Mackie-2009 and references therein]. PALS is a unique method that can unambiguously identify negatively charged and neutral vacancies and vacancy clusters and their concentrations.
A lifetime measurement begins with injection of a positron, which quickly thermalizes within a few ps in complex oxides [Mackie-2009]. After thermalization, the lifetime of the positron in the sample is determined by bulk and defect characteristics of the sample.
6 After a period of time, the lifetime of the positron terminates with annihilation with an electron, resulting in the creation of two 511 keV γ rays. The lifetime is defined as the duration of time between the creation and annihilation of the positron. Lifetimes can be lengthened by trapping at vacancy-like defects, where the local charge distribution is changed such that a positron does not feel the Coulombic repulsion of atomic nuclei.
PALS is a very powerful technique for unambiguous identification of defect structures present in a crystal.
The studies contained within this thesis conclude with investigation into
scintillation applications of undoped and Ce-doped YAG. Scintillation is the response of
luminescence by a crystal to ionizing radiation and has applications to particle and high
energy photon detection. Ce:YAG is well-known to have strong emission in the visible
wavelength range and high potential as a scintillator, but its potential has yet to be
realized due to defects that decrease efficiency and speed. On the other hand, undoped
YAG has thus far not been included in scintillator discussions. Our measurements show
undoped YAG to have very fast and high resolution scintillation, demonstrating high
potential as a fast scintillator, which is crucial to experiments requiring high time
resolution such as PALS. Undoped YAG scintillation arises from UV emission that
arises from defects, ironically the same defects that hinder Ce:YAG scintillation.
This work explores the effects of dopants, growth condition, and post-growth
treatments on optical processes and defects of YAG crystals. A wide variety of
measurements are conducted to complement each other and create a better understanding
of results. With this information, some conclusions can be made regarding the effects of
7 certain defects on optical processes in YAG and possible techniques to minimize the defects.
1.1 REFERENCES
[Ashurov-1977] M.Kh. Ashurov, Yu.K. Voronko, V.V. Osiko, A.A. Sobol, and M.I.
Timoshechkin, “Spectroscopic study of stoichiometry deviation in crystals with garnet structure,” Phys. Stat. Sol. (A) 42 (1977) 101-110
[Bernhardt-1978] HJ. Bernhard, “Bound polarons in YAG crystals,” Phys. Stat. Sol. (B)
87 , (1978) 213-219
[Blasse-1967(a)] G. Blasse and A. Bril, “A new phosphor for flying-spot cathode-ray
3+ tubes for color television: yellow-emitting Y 3Al 5O12 -Ce ,” Appl. Phys. Lett. 11 (1967)
53-55
[Geller-1967] S. Geller, “Crystal chemistry of the garnets,” Zeitschrift für
Kristallographie 125 (1967) 1-47
[Geller-1972] S. Geller, J.P. Espinosa, L.D. Fullmer, and P.B. Grandall, “Thermal
expansion of some garnets,” Materials Research Bulletin 7 (1972) 1219
[Geusic-1964] J.E. Geusic, H.M. Marcos, and L.G. Van Uitert, “Laser oscillations in Nd-
doped yttrium aluminum, yttrium gallium and gadolinium garnets,” Appl. Phys. Lett. 4
(1964) 182-184
[Gibbons-1973] E.F. Gibbons, T.Y. Tien, R.G. DeLash, P.J. Zacmanidis, and H.L.
3+ Stadler, “Some factors influencing the luminous decay characteristics of Y 3Al 5O12 :Ce ,”
J. Electrochem. Soc. 120 (1973) 835-837
8 [Guerassimova-2001] N. Guerassimova, N. Garnier, C. Dujardin, A.G. Petrosyan, and C.
Pédrini, “X-ray excited charge transfer luminescence in YAG:Yb and YbAG,” J. Lumin.
94-95 (2001) 11-14
[Guerassimova-2002] N. Guerassimova, C. Dujardin, N. Garnier, C. Pédrini, A.G.
Petrosyan, I.A. Kamenskikh, V.V. Mikhailin, I.N. Shipinkov, D.A. Spassky, K.L.
Ovanesyan, G.O. Shirinyan, R. Chipaux, M. Cribier, J. Mallet, and J.-P. Meyer, “Charge- transfer luminescence and spectroscopic properties of Yb 3+ in aluminum and gallium
garnets,” Nuclear Instruments and Methods in Physics Research A 486 (2002) 278-282
[Hayes-1980] W. Hayes, M. Yamaga, D.J. Robbins, and B. Cockayne, “Optical detection of EPR of recombination centers in YAG,” Journal of Physics C: Solid State Physics 13
(1980) L1085-L1089
[Hewitt-2011] J.D. Hewitt, T.M. Spinka, A.A. Senin, and J.G. Eden, “Internal
3+ 2 4 photopumping of Nd ( H9/2 , F5/2 ) states in yttrium aluminum garnet by excitation
transfer from oxygen deficiency centers and Fe 3+ continuum emission,” Appl. Phys. Lett.
99 (2011) 031101
[Jung-2007] K.Y. Jung and H.W. Lee, “Enhanced luminescent properties of
3+ 3+ Y3Al 5O12 :Tb ,Ce phosphor prepared by spray pyrolysis,” Journal of Luminescence 126
(2007) 469-474
[Kirm-2000] M. Kirm, A. Lushchik, Ch. Lushchik, and G. Zimmerer, “Investigation of
3+ luminescence properties of pure and Ce doped Y 3Al 5O12 crystals using VUV radiation,”
ECS Proceedings 99-40 (2000) 113-122
[Kubo-1992] T.S. Kubo and T.J. Kane, “Diode-pumped lasers at five eye-safe
wavelengths,” J. Quant. Elect. 28 (1992) 1033-1040
9 [Kuklja-1999] M.M. Kuklja and R. Pandey, “Atomistic modeling of point defects in yttrium aluminum garnet crystals,” Journal of the American Ceramic Society 82 (1999)
2881-2886
[Kuklja-2000] M.M. Kuklja, “Defects in yttrium aluminum perovskite and garnet
crystals: atomistic study,” J. Phys.: Condens. Matter 12 (2000) 2953-2967
[Landron-1996] C. Landron, S. Lefloch, M. Gervais, J.P. Coutures, and D. Bazin,
“Yttrium coordination shell in YAG related structures,” Phys. Stat. Sol. (B) 196 (1996)
25-31
[Liu-2009] B. Liu, M. Gu, Z. Liu, S. Huang, and C. Ni, “Formation energies of antisite
defects in Y 3Al 5O12 : a first-principles study,” Appl. Phys. Lett. 94 (2009) 121910
[Mackie-2000] R.A. Mackie, S. Singh, J. Laverock, S.B. Dugdale, and D.J. Keeble,
“Vacancy defect positron lifetimes in strontium titanate,” Physical Review B 79 (2009)
014102
[Milanese-2004] C. Milanese, V. Buscaglia, F. Maglia, and U. Anselmi-Tamburini,
“Disorder and nonstoichiometry in synthetic garnets A 3B5O12 (A=Y, Lu-La, B=Al, Fe,
Ga). A simulation study,” Chem. Mater. 16 (2004) 1232-1239
[Moszynski-1994] M. Moszy ński, T. Ludziejewski, D. Wolski, W. Klamra, and L.O.
Norlin, “Properties of the YAG:Ce scintillator,” Nucl. Instr. and Meth. Phys. Res. A 345
(1994) 461-467
[Murk-1995] V. Mürk and N Yaroshevich, “Exciton and recombination processes in
YAG crystals,” J. Phys.: Condens. Matter 7 (1995) 5857-5864
[Murk-1997] V. Mürk, “Exciton relaxation in wide-gap complex oxides,” Mat. Sci.
Forum 239-241 (1997) 537-542
10 [Nikl-2005] M. Nikl, “Energy transfer phenomena in the luminescence of wide band-gap scintillators,” Phys. Stat. Sol. (A) 202 (2005) 201-206
[Pankove-1971] J.I. Pankove, Optical Processes in Semiconductors , Dover Publications,
Inc., New York, NY (1971)
[Patel-2008] A.P. Patel, M.R. Levy, R.W. Grimes, R.M. Gaume, R.S. Feigelson, K.J.
McClellan, and C.R. Stanek “Mechanisms of nonstoichiometry in Y 3Al 5O12 ,” Appl. Phys.
Lett 93 (2008) 191902
[Pujats-2001] A. Pujats and M. Springis, “The F-type centers in YAG crystals,” Radiat.
Eff. & Def. Solids 155 (2001) 65-69
[Robbins-1979(a)] D.J. Robbins, B. Cockayne, B. Lent, C.N. Duckworth, and J.L.
Glasper, “Investigation of competitive recombination processes in rare-earch activated
garnet phosphors,” Phys. Rev. B 19 (1979) 1254-1269
[Robbins-1979(b)] D.J. Robbins, B. Cockayne, J.L. Glasper, and B. Lent, “The
temperature dependence of rare-earth activated garnet phosphors I. Intensity and lifetime
measurements on undoped and Ce-doped Y 3Al 5O12 ,” J. Electrochem. Soc. 126 (1979)
1213-1220
[Robbins-1979(c)] D.J. Robbins, B. Cockayne, J.L. Glasper, and B. Lent, “The temperature dependence of rare-earth activated garnet phosphors II. A comparative study
3+ 3+ 3+ 3+ of Ce , Eu , Tb , and Gd in Y 3Al 5O12 ,” J. Electrochem. Soc. 126 (1979) 1221-1228
[Robbins-1979(d)] D.J. Robbins, B. Cockayne, B. Lent, and J.L. Glasper, “The relationship between concentration and efficiency in rare earth activated phosphors,” J.
Electrochem. Soc. 126 (1979) 1556-1563
11 [Rotman-1985] S.R. Rotman and C. Warde, “Defect luminescence in cerium-doped yttrium aluminum garnet,” J. Appl. Phys. 58 (1985) 522-525
[Rotman-1989] S.R. Rotman, C. Warde, H.L. Tuller, and J. Haggerty, “Defect-poperty correlations in garnet crystals. V. Energy transfer in luminescent yttrium aluminum- yttrium iron garnet solid solutions,” J. Appl. Phys. 66 (1989) 3207-3210
[Rotman-1990] S.R. Rotman, “Comment on ‘Optical and electron paramagnetic
resonance studies of Fe impurities in yttrium aluminum garnet crystals,’” Phys. Rev. B 41
(1990) 791-792
[Sirdeshmukh-2001] D.R. Sirdeshmukh, L. Sirdeshmukh, K.G. Subhadra, K. Kishan Rao,
and S. Bal Laxman, “Systematic hardness measurements on some rare earth garnet
crystals,” Bull. Mater. Sci. 24 (2001) 469-473
[Springis-1991] M. Springis, A. Pujats, and J. Valbis, “Polarization of luminescence of
colour centres in YAG crystals,” J. Phys.: Condens. Matter 3 (1991) 5457-5461
[Stanek-2008] C.R. Stanek, M.R. Levy, K.J. McClellan, B.P. Uberuaga, and R.W.
Grimes, “Defect identification and compensation in rare earth oxide scintillators,” Nucl.
Instr. and Meth. Phys. Res. B 266 (2008) 2657-2664
[Varney-2012(c)] C.R. Varney, M.A. Khamehchi, J. Ji, and F.A. Selim, “X-ray luminescence based spectrometer for investigation of scintillation properties,” Review of
Scientific Instruments 83 (2012) 103112
[Vedda-2004] A. Vedda, D. Di Martino, M. Martini, J. Mares, E. Mihokova, M. Nikl, N.
Solovieva, K. Blazek, and K. Nejezchleb, “Trap levels in Y-aluminum garnet scintillating crystals,” Radiat. Meas. 38 (2004) 673-676
12 [Weber-2004] M.J. Weber, “Scintillation: mechanisms and new crystals,” Nucl. Instr. and
Meth. Phys. Res. A 527 (2004) 9-14
[Williams-1990] R.T. Williams and K.S. Song “The self-trapped exciton,” J. Phys.
Chem. Solids 51 (1990) 679-716
[Wong-1984] C.M. Wong, S.R. Rotman, and C. Warde, “Optical studies of cerium doped yttrium aluminum garnet single crystals,” Appl. Phys. Lett. 44 (1984) 1038-1040
[Zorenko-2004(b)] Y. Zorenko, A. Voloshinovskii, I. Konstankevych, V. Kolobanov, V.
Mikhailin, and D. Spassky, “Luminescence of excitons and antisite defects in the phosphors based on garnet compounds,” Radiat. Meas. 38 (2004) 677-680
[Zorenko-2007] Yu. Zorenko, A. Voloshinovskii, V. Savchyn, T. Voznyak, M Nikl, K
Nejezchleb, V. Mikhailin, V. Kolobanov, and D. Spassky, “Exciton and antisite defect-
related luminescence in Lu 3Al 5O12 and Y 3Al 5O12 garnets,” Phys. Stat. Sol. (B) 244 (2007)
2180-2189
[Zorenko-2010(a)] Y. Zorenko, T. Zorenko, T. Voznyak, A. Mandowski, Q. Xia, M.
+ Batentschuk, and J. Friedrich, “Luminescence of F and F centers in Al 2O3-Y2O3 oxide compounds,” IOP Conf. Series: Mat. Sci. Eng. 15 (2010) 012060
[Zorenko-2011] Yu. Zorenko, T. Voznyak, V. Gorbenko, E. Zych, S. Nizankovski, A.
Dan’ko, and V. Puzikov, “Luminescence properties of Y 3Al 5O12 :Ce nanoceramics,” J.
Lumin. 131 (2011) 17-21
[Zorenko-2012] Y. Zorenko, T. Zorenko, V.V. Gorbenko, T. Voznyak, V. Savchyn, P.
Bilski, and A. Twardak, “Peularities of luminescent and scintillation properties of
YAG:Ce phosphor prepared in different crystalline forms,” Opt. Mat. 34 (2012) 1314-
1319
13 [Zych-2000] E. Zych, C. Brecher, and J. Glodo, “Kinetics of cerium emission in a
YAG:Ce single crystal: the role of traps,” J. Phys.: Condens. Matter 12 (2000) 1947-1958
14 CHAPTER TWO
EXPERIMENTAL METHODS
This chapter is included for reference throughout the dissertation. Section 2.1 describes the samples, their growth conditions and post-growth treatments, and discusses some impurities found in the samples.
Experimental techniques are typically discussed at the beginning of relevant chapters or sections. Included in this chapter is the description of a new spectrometer we developed to study photo- and x-ray-stimulated luminescence of samples.
2.1 SAMPLES
Table 2.1.1 presents descriptions of each sample. Samples were acquired from the manufacturers United Crystals, Marketech International, and Crytur, Ltd.. Samples were grown by the Czochralski process in the <1 1 1> direction. Some samples were delivered as cut straight from the boule, some were ground to a rough but uniform surface, and others were optically polished. Samples as they were received from the manufacturers had dimensions of 5x5x1 mm, 10x10x1 mm, 10 mm diameter x 1 mm thick (though one sample Yb:YAG 5%, was determined to be a few tenths of a millimeter thicker than 1 mm), and 10 mm diameter x 5 mm thick. A number of samples were cut using a diamond saw either along a diagonal or parallel to the surfaces, the latter of which produced very rough edges that were polished by hand. These cut samples were slightly thicker than 1 mm thick. Yb:YAG 5%, originally 10 mm diameter x 1 mm thick, was cut into quarters.
15 Some unpolished samples were polished by hand using B 4C and Al 2O3 fine powders. 35 µm B 4C powder was used for the samples cut using a diamond saw, which had a very rough surface. Samples were polished by 15, 9, 5, 3, 1, 0.3, and 0.05 µm
Al 2O3 powders, gradually stepping down to finer powder to create a smooth optical
finish.
Samples were annealed in one of two tube furnaces, a Lindberg Model 59744-A
three-zone furnace or a Lindberg Model 54233-V-1.25B single zone furnace, either in air
or under constant flow of a gas. YAG has a high melting point of 1940°C [Caslavsky-
1980] but the maximum temperature of either furnace is 1500°C, so samples were
typically annealed at 1200°C for long durations of several days at a time. Annealing
atmospheres included air, argon, nitrogen, and a mixed atmosphere of argon and
hydrogen. Pure oxygen was attempted as an annealing atmosphere, but it appeared to
exothermically react with hydrogen impurities in the samples to damage some rubber
tubes used for gas flow. Air was used in place of pure oxygen flow for oxidizing anneal.
The presence of Al vacancies is of some concern in YAG. A procedure has been
suggested by Selim et al. [Selim-2007] and Solodovnikov et al. [Solidovnikov-2008] to
attempt to implant Al into the sample to fill Al vacancies, and these procedures are
emulated with some samples in this thesis. The process begins with annealing or growth
in an atmosphere containing oxygen to try to fill oxygen vacancies to isolate Al
vacancies. Next, Al is sputtered onto the samples for 30 seconds at the Cleanroom user
facility at the Center for Materials Research at Washington State University and then
annealed in an atmosphere of Ar for 24 hours at 600°C.
16 The table below describes: dopant concentration given as percent of yttrium atoms replaced by the dopant, dimensions of the crystal, the manufacturer the sample was acquired from, the growth atmosphere of the sample, the condition of the surfaces of the crystals, and any treatments such as annealing performed on the crystal. Samples grown by United Crystals were grown in an atmosphere of pure Ar, samples grown by
Marketech were grown in an undisclosed mixture of 5-30% H 2 in Ar, samples from by
Crytur were grown in 100 ppm O 2 in Ar (undoped YAG) or 40% H 2 in Ar (undoped
YAG and Cr:YAG 0.14%). These growth conditions were specifically chosen to study their effects on crystal growth.
Table 2.1.1 Samples studied in this thesis. Some samples were slightly larger than 1 mm thick, represented by (>1). Some samples came in different shapes and sizes, had different surface conditions, and/or underwent multiple annealing procedures. These varied parameters are combined in the table but labeled, but samples with different growth atmospheres are kept separate.
Sample Size (mm) Manuf. Growth Surface Treatments
Atmos. Condition
YAG* 1) 5x5x1 United Ar 1) fine ground 1) 96 hour air anneal
(undoped) 2) 10x10x1 Crystals 2) polished at 1200°C (fine
3) polished by ground 5x5x1 only)
hand (10x10x1 2) 48 hour air anneal
only) at 1200°C (polished
5x5x1 only)
17 Sample Size (mm) Manuf. Growth Surface Treatments
Atmos. Condition
YAG 1) 10x5 Crytur 40% H 2 1) as cut 48 hour air anneal at
(undoped) 2) 10x(>1) in Ar 2) fine ground 1200°C, Al sputtered
3) polished by and Ar annealed 24
hand hours at 600°C
YAG 1) 10x5 Crytur 100 ppm 1) as cut 1) 50% H 2 annealed
(undoped) 2) 10x(>1) O2 in Ar 2) fine ground 24 hours at 1200°C,
3) polished by further Ar+H 2
hand annealed another
51.5 hours at
1200°C, further Ar
annealed 19 hours at
1200°C, further Ar
annealed 48 hours at
1200°C
2) Al sputtered and
Ar annealed 24
hours at 600°C
Ce:YAG 10x1 Marketech 5-30% polished 36 hour Ar anneal at
0.1% H2 in Ar 600°C
18 Sample Size (mm) Manuf. Growth Surface Treatments
Atmos. Condition
Ce:YAG 1) 10x5 Crytur 40% H 2 1) as cut 1) 48 hour air anneal
0.14% 2) 10x(>1) in Ar 2) fine ground at 1200°C
3) polished by 2) 24 hour anneal in
hand vacuum at 800°C
Ce:YAG 10x1 Marketech 5-30% polished 48 hour air anneal at
0.15% H2 in Ar 1200°C
Ce:YAG 10x1 Marketech 5-30% polished 96 hour air anneal at
0.2% H2 in Ar 1200°C, further Al
sputtered and Ar
annealed 24 hours at
600°C, further 36
hour Ar anneal at
600°C
Ce:YAG 5x5x1 United Ar 1) fine ground 1) 96 hour air anneal
0.3% Crystals 2) polished at 1200°C, further
48 hour N 2 anneal at
1200°C (fine ground
only)
2) 48 hour air anneal
at 1200°C (polished
only)
19 Sample Size (mm) Manuf. Growth Surface Treatments
Atmos. Condition
Nd:YAG 5x5x1 United Ar 1) fine ground 48 hour air anneal at
1% Crystals 2) polished by 1200°C
hand
Tm:YAG 5x5x1 United Ar 1) fine ground 48 hour air anneal at
0.8% Crystals 2) polished by 1200°C
hand
Yb:YAG 5x5x0.8 Marketech 5-30% 1) fine ground
1% H2 in Ar 2) polished by
hand
Yb:YAG 5x5x0.5 Marketech 5-30% 1) fine ground
3% H2 in Ar 2) polished by
hand
Yb:YAG 10x1 United Ar polished 96 hour air anneal at
5%** Crystals 1200°C
Yb:YAG 10x(>1) United Ar polished 96 hour air anneal at
5% Crystals 1200°C
Yb:YAG 5x5x1 United Ar polished 48 hour air anneal at
10% Crystals 1200°C
* Throughout the chapters of this dissertation, the 5x5x1 mm sample will be be referred to as “Ar- grown YAG” and the 10x10x1 mm sample will specifically be mentioned by its size.
** This sample was marked as 5% Yb-doped, but x-ray photoelectron spectroscopy (XPS),
20 Wavelength dispersive x-ray spectroscopy (WDS), and energy dispersive x-ray (EDS) measurements as well as our own optical measurements revealed no presence of Yb in the sample.
Wavelength dispersive x-ray spectroscopy (WDS) measurements revealed similar concentrations of Al and Y in undoped YAG samples, regardless of growth condition.
As will be seen in later chapters, this is a significant result. These measurements also showed that there is no Yb present in one of the samples labeled Yb:YAG 5%.
Glow discharge mass spectroscopy (GDMS) revealed numerous impurities in
YAG samples. Notable impurities detected at measurable concentrations in O 2-grown
YAG include: Mg (0.37 ppm at.), Ca (0.46 ppm at.), Cr (0.31 ppm at.), and Fe (0.48 ppm
at.). Concentration of Tb impurities, suggested later to be responsible for emission in
photoluminescence (Section 3.2) and thermoluminescence (Chapter 4) measurements, is
below the limit for its detection (<0.2 ppm at).
2.2 NEWLY DEVELOPED EXPERIMENTAL TECHNIQUE TO STUDY
LUMINESCENCE
Photoluminescence (Section 3.2) and X-ray stimulated luminescence (Section 3.4)
measurements were conducted using our own newly-developed spectrometer [Varney-
2012(c)], depicted in Fig. 2.2.1.
21
Figure 2.2.1 Schematic diagram for XSL and some PL measurements. The dashed box represents the
boundary of the radiation-safe chamber that shields x rays and stops outside light from getting in.
X rays were generated using a Cu x-ray tube which were passed through a monochromator and collimated to form a focused monoenergetic x-ray beam. The sample was held by a clip with x, y, and z translational and 180° rotational freedom, which can be replaced for measurements on powders by a cup to hold powders with the same possible translation and rotation. The light from the sample was collected by a collimated collection lens and transmitted to the spectrometer via a solarized optical fiber.
The spectrometer used for this setup was an Ocean Optics 2000+ USB spectrometer with a Sony charge-coupled device (CCD) detector with a 200 µm slit. The spectral range of this detector is 200-850 nm with 1 nm resolution. The spectrometer is controlled by the computer program Spectrasuite. Collimating lenses and solarized optical fibers (the fibers transmit UV efficiently) transmitted the excitation light to and emitted light from the sample. It is unclear whether the software SPECTRASUITE corrects data taken using this detector for spectrometer response. The grating 600 lines/mm and is blazed at 400 nm.
22 This spectrometer uses excitation from a Sandhouse Design light-emitting diode
(LED) array or Ocean Optics PX-2 Pulsed Xenon Light Source with a broad emission range. The xenon lamp emits in the range 220-700 nm and was operated at 200 Hz flash rate with a pulse duration full width at half maximum (FWHM) of 5 µs at4.5 µJ per pulse
maximum. The xenon lamp is also used for alignment of the collection lens. The LED
array provided excitations at 240, 270, 310, and 455 nm in continuous wave (CW) mode
with powers of 2 µW, 15 µW, 15 µW, and 1 mW respectively. The LED array can be
controlled manually or by the computer program Sandhouse LED controller. The LED
array is installed such that it can be used as an additional excitation alongside x rays or as
the sole source of excitation. This flexibility gives this setup use in photoluminescence
and x ray luminescence.
2.3 REFERENCES
[Selim-2007] F.A. Selim, D. Solodovnikov, M.H. Weber, and K.G. Lynn, “Identification
of defects in Y 3Al 5O12 crystals by positron annihilation spectroscopy,” Applied Physics
Letters 91 (2007) 104105
[Solodovnikov-2008] D. Solodovnikov, M.H. Weber, and K.G. Lynn, “Improvement in
scintillation performance of Ce, Er codoped yttrium aluminum garnet crystals by means
of a postgrowth treatment,” Applied Physics Letters 93 (2008) 104102
[Varney-2012(c)] C.R. Varney, M.A. Khamehchi, J. Ji, and F.A. Selim, “X-ray
luminescence based spectrometer for investigation of scintillation properties,” Review of
Scientific Instruments 83 (2012) 103112
23 CHAPTER THREE
OPTICAL PROPERTIES OF YAG
YAG is an insulator material with a wide band gap of 6.5 eV, corresponding to an absorption edge at about 190 nm [Dong-2006, Robbins-1979(a), Robbins-1979(d),
Rotman-1989, Xu-1999, Zych-2000]. Adding a dopant creates new acceptor-like and/or donor-like levels within the forbidden band gap just above the valence band or below the conduction band, respectively, giving the crystal new possible energy levels and thus new optical properties based on the electron transitions between these energy levels.
In Ce:YAG, a Ce 3+ ion replaces Y 3+ in the dodecahedral position. The energy
levels of Ce 3+ include the ground state 4f level, which exists within the band gap of YAG,
and the 5d excited state, which exists partially within the band gap and partially within
2 2 the conduction band. The 4f level is split into two components, F5/2 and F7/2 , by spin- orbit interaction and the 5d state is split into 5 states, each doubly degenerate due to
Kramer’s degeneracy, which states an atom containing an odd number of spin-½ particles is at least doubly degenerate in the absence of a magnetic field. There is much disagreement over where exactly the 5d states lie arising from identification of absorption peaks as defect-related or excited states of Ce 3+ . For example, Zych et al. [Zych-2000]
state that three levels can be observed in absorption spectra at wavelengths 230 nm, 340
nm, and 450 nm, with the 230 nm band located within the conduction band and the 340
nm band residing at the very bottom of it. In contrast, Dong et al. [Dong-2006] used
vacuum UV synchrotron radiation to observe a higher Ce 3+ level at 188 nm and attributed
a peak seem in some Ce:YAG samples at 265 nm to Ce 3+ , thus claiming to have observed
24 all five 5d levels. The 265 nm absorption peak is a point of contention because it resides so close to the Fe 3+ transition in YAG at 255 nm, and iron impurities have been argued to
be the main impurities in YAG crystals [Mori-1977]. Since transitions to the highest
states occur at energies at or above that of the conduction band, it is not possible to
observe them by absorption measurements.
Nd:YAG has a much better understood energy level structure than Ce:YAG.
Nd 3+ has a multitude of energy levels within the band gap of YAG. The ground state is
4 4 4 4 split into I9/2 , I11/2 , I13/2 , and I15/2 , and even those components are split further into
4 4 4 multiplets. The lowest excited state is F3/2 , and it is the F3/2 → I11/2 transition that produces the strong 1064 nm emission that is used in Nd:YAG lasers. A comprehensive list of excited levels of Nd 3+ ions can be found in literature on Nd:YAG [e.g. Batygov-
1991, Niklas-1983]. Work involved in this thesis identified transitions in Ne:YAG at
room temperature using beta irradiation [Reda-2012].
The abundance of energy levels lends itself to Nd:YAG as a four-level laser
system. In a four-level laser system, an excited electron undergoes a fast, non-radiative
transition to a special lower state. The electron then emits a photon as it undergoes a
slower transition to a lower state well above the ground state. This step is where laser
emission occurs. Another fast non-radiative transition relaxes the electron to the ground
state. Ideally, the lower laser level is approximately empty and does not appreciably
populate during laser operation. This avoids reabsorption of the laser radiation.
The Tm 3+ ion is similar to Nd 3+ in that it has a large number of electronic energy
3 states that can be observed in YAG. The ground state is H6 and the observable (by
3 3 1 1 3 absorption measurements) excited states are H4, F3, G4, D2, and P2, in order of
25 increasing energy, and each state is further split into multiplets, similar to Nd:YAG.
Tm:YAG emits a strong luminescence at 2020 nm used in laser applications, which is the
3 3 result of the H5→ H6 (ground state) transition [Kubo-1992, Song-2004]. Tiseanu et al.
produced a schematic diagram showing the energy levels in Tm:YAG [Tiseanu-1995].
As opposed to Nd:YAG and Tm:YAG, Yb:YAG is an example of a laser crystal
with few energy levels. Yb 3+ in YAG emits 1030 nm light at its lasing transition of
2 2 F5/2 → F7/2 . The next highest energy level would be the 5d level, which lies far above the
conduction band of YAG [Lacovara-1991].
Tm:YAG and Yb:YAG are quasi-three-level systems, which means that the lowest excited level is very close to the ground state and thus is populated with an appreciable amount of electrons. The result is reabsorption at the laser wavelength.
2.1 ABSORPTION
Perhaps the simplest and most direct approach to probing the band structure of a
crystal is the absorption measurement. If a certain wavelength corresponds to an allowed
transition between energy states, an incident photon of that energy can be absorbed,
otherwise the sample is transparent at that wavelength. Following that simple model, one
can shine a known intensity of light through a crystal and record how much makes it
through to a detector on the other side. Of course, this approach relies heavily on a well-
polished sample, as a rough surface can scatter the light, directing it away from the
director, which gives the false appearance of a higher coefficient of absorption. To avoid
this, absorption spectra for polished samples are reported in this thesis whenever possible
and unpolished samples are only reported when data from a polished sample is
26 unavailable. Absorption spectra show peaks at individual transitions, such as the transitions between the ground state and excited states of the dopant or defect centers, and show high absorption above the energy of the band gap, viz., the sample is nearly opaque to shorter wavelengths.
The absorption measurements were performed at room temperature using a Perkin
Elmer Lambda 35 UV/VIS spectrometer. This instrument uses double-beam operation
for measurement stability, a concave grating with 1053 lines per mm, and deuterium and
tungsten lamps with an automatic switchover at 326 nm, which occasionally appears in
absorption spectra as a sudden jump at this wavelength. This spectrometer can scan over
its wavelength range of 190-1100 nm with a wavelength resolution of 1 nm and accuracy
of ±0.1nm. For our measurements, we used a scan speed of 120 nm/minute and 1 nm slit
width. The unit of absorption used by this spectrometer is absorbance, which is given by
the Beer-Lambert law [Bauman-1962, Pankove-1971]:
I ln 0 = A (2.1.1) I
where I 0 is the intensity of the beam of light before it reaches the sample, I is the intensity of light that remains after passing through the sample, and A is the absorbance.
The machine was not designed to accommodate the small sizes of the samples we worked with, so the holder was modified to include special inserts in order to be able to measure our samples.
There is an abundance of literature on absorption measurements on YAG crystals
[e.g. Chen-1988, Chugunova-2011, Dong-2005, Dong-2006, Fagundes-Peters-2007,
Feldman-2003, Fredrich-Thornton-2010, Hamilton-1989, Korzhik-1989, Kovaleva-1981,
Lacovara-1991, Landron-1996, Malikowski-1995, Masumoto-1985, Miniscalo-1978
27 Mori-1977, Pedrini-1985, Pieterson-2000, Robbins-1979(a), Robbins-1979(d), Rotman-
1985, Rotman-1989, Rotman-1990, Rotman-1992, Selim-2007, Solodovnikov-2008,
Song-2004, Springis-1991, Tiseau-1995, Tomiki-1991, Varney-2012(a), Vedda-2004,
Wong-1984, Xu-2003, Yang-2002, Zorenko-2011, Zorenko-2012]. Since absorption measurements are perhaps the most fundamental approach to probing the band structure, it remains a crucial first step to studying the optical properties of our crystals. A common feature in many YAG absorption measurements is a peak around 256 nm that has been suggested to result from iron impurities. Due to its size, an iron impurity replaces an Al 3+
ion and enters into either the octahedral or tetrahedral sites [Rimal-1966]. Several
studies, both experimental and theoretical, have been conducted to investigate the effects
of iron in YAG to confirm the relation between the 256 nm absorption peak and the
presence of iron [Chen-1988, Dong-2005, Dong-2006, Hewitt-2011, Korzhik-1989,
Masumoto-1985, Milanese-2004, Mori-1977, Ricci-2009, Rimai-1966, Rotman-1989,
Rotman-1990, Varney-2011, Varney-2012(a)]. There are multiple theories regarding the
specific origins of this peak. Mori [Mori-1977] proposed that the 254 nm absorption
peak seen in his samples was due to O 2-→Fe 3+ charge transfer and assigned a peak at 313 nm to Fe 2+ ions. Masumoto and Kuwamo [Masumoto-1985] expanded on Mori’s
hypothesis to suggest that the 255 nm absorption band consists of more than one
absorption center. Meil’man [Meilman-1984] et al. suggested that the 255 nm absorption
band is actually overlapping charge-transfer bands from Fe 3+ the octahedral and tetrahedral sites. Iron impurities are very common defects in YAG and, as will be seen, play a significant role in the samples involved in this thesis.
28 3.1.1 RESULTS
Undoped YAG
Measurement of the absorption spectra of undoped YAG crystals gives a baseline
of sorts to which measurements conducted on other samples may be compared. These
measurements can help determine which structures of the absorption spectra are intrinsic
to YAG itself, which makes it easier to see how the dopants add their own features and
affect those intrinsic to the host.
Figure 3.1.1 shows the absorption spectra for the as grown polished undoped
YAG crystals. This figure clearly displays the band edge at about 190 nm. The
difference in growth conditions can be seen in the low wavelength peaks. The O 2-grown
YAG shows one peak at 256 nm. The Ar-grown samples do actually show a very weak peak at 256 nm but it is greatly subdued compared to O2-grown YAG. Annealing
processes (Figs. 3.1.2-3.1.3) and color center studies (Section 3.3) can further explore the
effects of oxidation on the 256 nm absorption peak. It has been shown computationally
that Fe prefers to exist in the YAG crystal lattice in the divalent state, rather than the
trivalent state, and the charge is compensated by an oxygen vacancy [Milanese-2004].
Annealing the sample in the presence of oxygen oxidizes Fe 2+ to Fe 3+ and fills oxygen
vacancies, providing oxygen ligands near Fe 3+ ions to enhance Fe-O charge transfer.
Also distinguishable is an increase in absorption in wavelengths below the 256 nm peak, possibly due to a weak band around 210 nm. Literature shows a peak at 210 nm due to the charge transfer of Yb in YAG [Friedrich-Thornton-2010, Guerassimova-
2002, Kamesnkikh-2005, Pieterson-2000], so one possible cause is the presence of Yb impurities in our crystals. As will be seen, the intensity of this peak corresponds well
29 with concentration of O 2-, so it seems likely that it is related to some form of charge transfer with an impurity.
Energy (eV) 6.2 3.1 2.1 1.6 0.3 Ar-grown Ar-grown 10x10x1
O2-grown
0.2 Absorbance
0.1
200 400 600 800 Wavelength (nm)
Figure 3.1.1 Absorption spectra for undoped YAG single crystals. The jump at 326 nm is due to the lamp
change of the spectrometer and is not indicative of crystal absorption.
Ar-grown undoped samples were annealed for 48 hours in air at 1200°C. The post-annealing absorption spectrum of Ar-grown YAG is compared to its as grown spectra in Figs. 3.1.2. A pair of O 2-grown YAG crystals were annealed in a mixed
atmosphere of Ar+H 2 with a nearly 1:1 ratio of gasses at 1200°C for 24 hours, measured,
and then annealed further in the same conditions for another 51.5 hours. The absorption
spectra for these measurements are compared to each other and the measurements
conducted on the as grown samples in Fig. 3.1.3. H2-grown and O 2-grown samples were sputtered with a small amount of aluminum onto one surface and then annealed at 600°C in an Ar atmosphere for 24 hours with the Al-sputtered side up in an attempt to alleviate the Al deficiency inherent in our samples. The absorption spectrum of the O 2-grown
30 sample after Al sputtering and Ar annealing is included in Fig. 3.1.3 to be compared to the as grown and Ar+H 2 annealed samples.
Energy (eV) 6.2 3.1 2.1 1.6 0.3
As grown Annealed
0.2 Absorbance
0.1
200 400 600 800 Wavelength (nm)
Figure 3.1.2 Absorption spectra of Ar-grown YAG before and after annealing 48 hours in air at 1200°C.
Annealing the Ar-grown sample in air oxidized it and there is a large increase in the 256 nm band. Also visible is an increase in the wavelengths below this peak.
31 Energy (eV) 6.2 3.1 2.1 1.6 0.5
As grown H2+Ar annealed 0.4 H2+Ar annealed further Al sputtered + Ar annealed
0.3 Absorbance
0.2
0.1
200 400 600 800 Wavelength (nm)
Figure 3.1.3 Absorption spectra of O 2-grown YAG as grown, after annealing 24 hours in Ar+H 2 at 1200°C,
and after Al sputtering and Ar annealing 24 hours at 600°C. These samples annealed in Ar+H 2 are not as
well polished as the other samples. The jump at 326 nm is due to the lamp change of the spectrometer and
is not indicative of crystal absorption.
From Fig. 3.1.3, it can be seen that Al sputtering had no appreciable effect on the absorption of the O 2-grown sample. Annealing in Ar+H 2, however, bleached the 256 nm
peak. The reduction of Fe 3+ to Fe 2+ combined with the potential removal of oxygen and
insertion of hydrogen ions decreased O 2-→Fe 3+ charge transfer absorption. The change
in the overall absorption between samples is due to surface conditions.
Undoped YAG conclusions
The band edge of YAG was observed at 190 nm corresponding to a band gap of
6.5 eV. A band at 256 nm, associated with trivalent iron, increased in intensity when
samples were oxidized, whether by growth atmosphere or by annealing. Addition of
oxygen can fill oxygen vacancies, providing oxygen ligands near Fe 3+ ions and thus
32 enhancing the charge transfer between iron and oxygen ions, and can oxidize Fe 2+ to
Fe 3+ . Iron typically incorporates into the crystal structure as divalent unless grown in an
oxidizing atmosphere.
Ce:YAG
Ce:YAG has a number of well-defined transitions observable in absorption
measurements. Absorption measurements on as grown Ce:YAG samples are shown in
Fig. 3.1.4. The unpolished Ce:YAG 0.3% Ar-grown samples were annealed for 96 hours
at 1200°C in air, measured, and then further annealed in nitrogen for 48 hours at 1200°C.
Absorption of this sample before annealing, between anneals, and after the final anneal
are displayed in Fig. 3.1.5. The Ce:YAG 0.2% samples were annealed for 96 hours in air
at 1200°C and later sputtered with aluminum and annealed in an Ar atmosphere for 24
hours at 600°C. Results of the absorption measurements on the Ce:YAG 0.2% crystals in
each of its states can be found in Fig. 3.1.6.
Energy (eV) 6.2 3.1 2.1 3.0 Ce:YAG 0.1% Ce:YAG 0.15% 2.5 Ce:YAG 0.2% Ce:YAG 0.3% 2.0
1.5 Absorbance
1.0
0.5
0.0
200 400 600 Wavelength (nm)
Figure 3.1.4 Absorption spectra of Ce:YAG single crystals.
33 Comparison of Figs. 3.1.4 and 3.1.1 shows that the peaks at 225, 340, and 458 nm clearly arise from the Ce dopant. It has been argued that the weak peak at 265 nm is another transition level of Ce 3+ and that it appears so weak due to the proximity of the
Fe 3+ absorption peak [Dong-2006], or it could just be the 256 nm peak associated with
iron but shifted slightly.
Figure 3.1.4 also demonstrates some of the differences between Ce:YAG samples
grown in different conditions and of different dopants. The higher Ce concentration
crystal grown in a pure argon atmosphere displays significantly decreased absorption
intensity when compared to the samples grown in a mixture of argon and hydrogen,
suggesting that the presence of hydrogen as a reducing atmosphere is crucial to
incorporate Ce ions in the trivalent state into the crystal matrix as opposed to the
tetravalent state [Varney-2012(a)]. The spectra of the 0.2% and 0.15% atomic
concentration samples are nearly identical while the 0.1% sample shows a significant
increase in intensity over all wavelengths with a noticeably pronounced increase below
the 340 nm peak. The increased absorbance of the Ce:YAG 0.1% sample at lower
wavelengths seems to suggest a more oxidized sample, by analogy to the undoped
samples. This could be a consequence of the size of cerium ions, which are notably
larger than the yttrium ions they replace [Emsley-1989, Varney-2012(a)]. It is possible
that incorporating this large of an ion into the crystal matrix dislocates nearby oxygen
ions, producing oxygen vacancies. It is tempting to suggest that the sample with lower
cerium concentration does not contain as many oxygen vacancies as the other samples,
and thus the higher concentration of oxygen can lead to increased iron luminescence as
discussed above.
34 Energy (eV) 6.2 3.1 2.1 3.0 As grown, unpolished Air annealed
Further N 2 annealed 2.5 Absorbance 2.0
1.5
200 400 600 Wavelength (nm)
Figure 3.1.5 Absorption spectra of Ce:YAG 0.3% 5x5x1 unpolished crystal grown in pure argon
atmosphere after two sequential anneals, first in air at 1200°C for 96 hours and second in nitrogen at
1200°C for 48 hours.
It is unclear why in Fig. 3.1.5 the as grown sample shows so much more background absorption than its annealed counterparts, but this does suggest that the condition of the surface was altered. It is still clear in the spectra that the 256 nm peak increases after oxidation, as it did in the undoped samples, and decreases slightly after annealing in a reducing atmosphere, demonstrating similar behavior as it does in undoped
YAG. The stability of this peak to nitrogen annealing suggests the explanation that the
256 nm absorption peak relies most heavily on the presence of oxygen and, having filled oxygen vacancies that initially stabilized Fe 2+ , it is difficult for trivalent iron in the air annealed sample to reduce to divalent.
35 Energy (eV) 6.2 3.1 2.1
As grown 3 Air annealed Al sputtered + Ar annealed
2 Absorbance
1
0 200 400 600 Wavelength (nm)
Figure 3.1.6 Absorption spectra of Ce:YAG 0.2% as grown, after annealing 96 hours in air at 1200°C, and
after Al sputtering and Ar annealing 24 hours at 600°C.
Ce:YAG 0.2% shows a decrease in absorbance in the two largest Ce 3+ transition peaks after oxidative annealing, demonstrating that this anneal has oxidized a number of
Ce 3+ ions to the Ce 4+ state. Air annealing also increased the intensity of the 256 nm peak.
Figure 3.1.6 shows that sputtering with aluminum and annealing in argon slightly
decreased these peaks but did not noticeably affect the cerium peaks. It is possible that
this decrease is caused by Al 3+ ions displacing Fe 3+ ions in the octahedral and tetrahedral
positions of the lattice.
Figure 3.1.7 shows the effect of vacuum annealing on as cut Ce:YAG 0.14%. The
surfaces of both samples are very rough, but the increased intensity seems to be real, as
the sample turned darker green after this treatment (as opposed to yellow before
annealing).
36 Energy (eV) 6.2 3.1 2.1 4.0 As grown Vacuum annealed 3.5
3.0 Absorbance 2.5
2.0
1.5 200 400 600 Wavelength (nm)
Figure 3.1.7 Unpolished (as cut) Ce:YAG 0.14% before and after annealing in vacuum 24 hours at 800°C.
Ce:YAG conclusions
Doping YAG with Ce adds a series of absorption bands corresponding to allowed electronic transitions in Ce 3+ . The intensity of these peaks is affected by growth
atmosphere, as a reducing atmosphere helps put Ce in the trivalent state that has these
transitions. Growth in an inert atmosphere allows a large concentration of Ce to
incorporate as tetravalent.
Very low concentration of Ce doping possibly allowed for increased oxygen
concentration over Ce:YAG samples with higher dopant levels. This is likely due to the
physical size of Ce ions, which are larger than the Y ions they replace, and their size may
be compensated by an O vacancy.
Nd:YAG
Figure 3.1.8 shows Nd:YAG 1% before and after annealing in air. The figure
clearly displays the abundance of transitions between states in the Nd 3+ ion and a shifted
37 band edge as compared to undoped YAG. In this Nd:YAG sample, the band edge appears to reside at 235 nm, differing significantly from the 190 nm of undoped YAG, due to the strong band at 225 nm attributed to Nd 3+ [Kovaleva-1981]. Several of the
4 most prominent transitions with the I9/2 ground state are marked in Fig. 3.1.10, using the
notation and calculations of Batygov et al. [Batygov-1991].
Energy (eV) 6.2 3.1 2.1 1.6 1.2 1.0 As grown Air annealed
2K ,4G 2 4 13/2 9/2 H2 9/2 , F5/2
4S ,4F Absorbance 0.5 3/2 7/2 2 4G ,2G P3/2 5/2 7/2 4 4 F3/2 F9/2
200 400 600 800 1000 Wavelength (nm)
Figure 3.1.8 Absorption spectra of Nd:YAG 1% before and after oxidative air anneal at 1200°C for 48
hours.
Annealing in air significantly increased the intensity of the 256 nm absorption peak but did not affect any of the Nd 3+ peaks. Neodymium ions are found most often in
the trivalent state and have no higher oxidative state [Emsley-1989]. The oxidative
anneal demonstrated that the vast majority, if not all, neodymium ions in the as grown
Nd:YAG sample were trivalent.
38 Tm:YAG
The absorption spectra of the Tm:YAG 0.8% sample before and after annealing
3 are shown in Fig. 3.1.9. All transitions to the ground state H6 over the measured range of wavelengths are marked in the plot using the calculations and notation of Tiseanu et al.
[Tiseanu-1995].
Energy (eV) 6.2 3.1 2.1 1.6 1.2 1.0 As grown Annealed 48 hours in air 0.8 3 3 3 F3 P2, P1, 0.6 3P ,1I 0 6 Absorbance 3H 0.4 1D 4 2 3 1 F G4 2 0.2
200 400 600 800 1000 Wavelength (nm)
Figure 3.1.9 Absorption spectra of Tm:YAG 0.8% before and after oxidative air anneal at 1200°C for 48
hours.
Again the absorption spectra show an increase in the 210 and 256 nm peaks, the latter slightly hidden under a Tm 3+ peak. There is no change in Tm 3+ absorption despite
Tm possessing a divalent state. A Tm ion can be divalent or trivalent [Emsley-1989]. It
is possible that the majority of the thulium present in the as grown sample was already
trivalent, so oxidative annealing did not increase the concentration of Tm 3+ ions this
sample.
39 Yb:YAG
The absorption spectra of various Yb:YAG samples are displayed in Fig 3.1.10.
Figure 3.1.11 shows absorption spectra of Yb:YAG with 10% atomic doping
concentration before and after annealing in air.
Energy (eV) 6.2 3.1 2.1 1.6 1.2 1.0 Yb:YAG 1% (M) Yb:YAG 3% (M) 0.8 Yb:YAG 5% air annealed 96 hr (UC) Yb:YAG 10% (UC)
0.6 Absorbance 0.4
0.2
0.0 200 400 600 800 1000 Wavelength (nm)
Figure 3.1.10 Absorption spectra of as grown Yb:YAG single crystals.
Similar to Nd doping, doping with Yb gives the appearance of a shifted band edge, this time to almost 240 nm, due to the strong charge-transfer band with a maximum at about 210 nm [Friedrich-Thornton-2010, Guerassimova-2002, Kamesnkikh-2005,
2 2 Pieterson-2000]. The peaks between 900 and 1000 nm are a result of the F5/2 → F7/2 transition, both are which are multiplets, leading to the peak structure observed. These peaks increase with increasing dopant concentration. Whereas it was seen in Ce:YAG that high dopant concentration decreases absorbance, no such limit appears for Yb:YAG.
This suggests that higher concentrations of Yb can positively affect the optical properties of Yb:YAG. In fact, fully doped Yb:YAG, or YbAG, where all yttrium ions have been
40 replaced with ytterbium, have been synthesized [Fagundes-Peters-2007, Guerassimova-
2001, Guerassimova-2002, Wang-2002].
Energy (eV) 6.2 3.1 2.1 1.6 1.2 1.0 As grown Air annealed 48 hours 0.8
0.6 Absorbance 0.4
0.2
0.0 200 400 600 800 1000 Wavelength (nm)
Figure 3.1.11 Absorption spectra of Yb:YAG 10% before and after annealing in air at 1200°C for 48 hours
Figure 3.1.11 shows that this sample shows remarkable robustness to oxidative annealing, aside from the slight increase to the 256 nm peak. Ytterbium has two oxidation states: divalent and trivalent [Emsley-1989]. Annealing in air might have been expected to have oxidized some Yb 2+ to Yb 3+ and, based on the conclusion of Fig. 3.1.10
that increased Yb 3+ concentration increases absorbance, it seems reasonable to expect to
observe an increase in intensity of the absorption peaks. However, no such increase is
observed in either sample, suggesting that ytterbium is either stable in the divalent state
against oxidation or Yb ions in the crystal are almost entirely in the trivalent state in the
as grown sample. The size of Yb 2+ could rule out the former based on the same argument used above for Ce:YAG. Yb 2+ is significantly larger than the Y 3+ ion it replaces in the lattice and as a result its presence in the crystal may displace adjacent oxygen ions,
41 creating oxygen vacancies. Oxidative annealing should fix this and add oxygen ions around the Yb 2+ ion, which would change the charge of ions surrounding the Yb 2+ ion, making the conditions favorable for Yb 2+ to donate an electron, and thus it should not be
stable against such annealing. Yb 3+ , however, is actually slightly smaller than Y 3+ , so it
should have no such straining effect on the surrounding crystal lattice, thus it is believed
that the vast majority of ytterbium ions incorporated into the YAG lattice are trivalent.
Yb:YAG conclusions
Yb appears to incorporate primarily in the trivalent state in YAG. Increased Yb
concentration corresponded to increased absorption, but no such change in intensity was
observed after annealing the samples.
YAG Absorption conclusions
Absorption measurements revealed the presence of iron in almost every sample
studied, which supports the hypothesis by Mori et al. that iron impurities are the main
impurities in YAG [Mori-1977]. Oxidative annealing increased the absorbance at low
wavelengths, increasing the 256 nm band and a weak broad band at lower wavelengths,
and did not have a significant effect on the absorption of most dopants other than cerium.
For Ce-doped YAG, a reducing atmosphere, whether by growth or anneal, was crucial to
putting cerium in the trivalent state and increasing optical response of the crystal.
Different dopants created new allowed electronic transitions, which added new
absorption bands.
42 3.2 PHOTOLUMINESCENCE
Introduction
In the process of photoluminescence (PL), an incident photon is absorbed by an electron, which is raised to an excited energy state. The electron returns to the ground state with the emission of a photon. A luminescence center is defined as a point-like lattice imperfection that alters the band structure of the crystal such that excited electrons can deexcite at these sites with the emission of a photon. Luminescent centers can result from positively- or negatively-charged vacancies, interstitial ions, or from activators such as dopants or antisite defects.
PL measurements serve as an effective and reliable way to probe the
luminescence centers of a crystal provided the measurement can excite the crystal with
the energy necessary to excite electrons at luminescence centers. PL measurements
utilizing the broad emission of the xenon lamp provided a rough first approximation to
the luminescence spectra since it excited the crystal with UV and a broad range of
wavelengths. The emission found using the xenon lamp could then be further explored
using more specialized measurements. Excitation spectra were scanned for emission
peaks found by using the broad UV excitation.
Literature in photoluminescence of YAG crystals is numerous [Babin-2005,
Babin-2011, Bagdasarov-1977, Blasse-1967(a), Blasse-1967(b), Blazek-2004,
Chakrabarti-1988, Dong-2006, Fredrich-Thornton-2010, Gluchowski-2009, Grigorjeva-
2010, Guerassimova-2002, Hewitt-2011, Kamenskikh-2005, Kirm-2000, Kovaleva-1981,
Malinowski-1995, Meng-2008, Murk-1995, Murk-1997, Pieterson-2000, Pujats-2001,
Ricci-2009, Rotman-1985, Rotman-1989, Springis-1991, Suzuki-1997, Tisseanu-1995,
43 Tomiki-1991, Varney-2011, Varney-2012(a), Wong-1984, Zorenko-2010(a), Zorenko-
2011, Zorenko-2012, Zych-2000]. As discussed above in Section 3.1, iron impurities are common in YAG crystals. Iron impurities in YAG luminesce in the range 700-1000
[Hewitt-2011, Ricci-2009, Rotman-1989, Varney-2011] nm and its intensity and peak position depend on concentration.
Experimental setup
Photoluminescence was recorded using two different spectrometers, both methods conducted at room temperature. One setup was previously described in Section 2.2, and used an Ocean Optics 2000+ USB spectrometer. A Sandhouse Design light-emitting diode (LED) array or Ocean Optics PX-2 Pulsed Xenon Light Source was used for excitation. It is unclear whether the software SPECTRASUITE corrects data recorded using this detector for spectrometer response.
Measurements using the USB2000+ spectrometer were conducted with settings deemed appropriate for the measured sample and were consistent across all samples of that type, e.g. the same settings were used for all undoped samples, but they were different from measurements conducted on Ce:YAG. The spectrometer allows for integration time in the range of 1 ms-65 s, which must be set to an appropriate value to collect enough photons for the PL spectra while not collecting too many to saturate the detector. Noise can be reduced in two ways: averaging scans and boxcar. Averaging scans averages intensities at specific wavelengths across multiple measurements to reduce variable noise across measurements, and this was typically set to 3. Boxcar averages intensity values across multiple wavelengths in an individual measurement, and
44 this was typically set to 3. The electronic noise of the spectrometer was subtracted before each measurement.
The other spectrometer allowed for any excitation over the range 190-600 nm and could also scan excitation wavelengths while holding fixed the emission wavelength detected. This used a computer-controlled JY-Horiba Fluorolog-3 spectrofluorometer, which is equipped with double grating excitation and emission monochromators with variable slit sizes and a R9288 photomultiplier tube (PMT). The spectral range of this detector is 300-850 nm. Two light sources, a 450W xenon CW lamp and a deuterium lamp, with different emission wavelengths were used to provide excitation. The measured spectra were corrected for detector response. Whereas the Ocean Optics spectrometer produces spectra at all detectable wavelengths simultaneously, the JY-
Horiba spectrofluorometer scanned spectra such that it observed only one wavelength at a time. This setup is more reproducible than the other but it is still not perfect, so care must still be taken when comparing intensity across measurements.
Both methods of PL measurement have their shortcomings. The emission spectra of the LEDs used in the first spectrometer show weak emission at wavelengths above peak emission that can be misleading in measurements. Varying the intensity of each
LED was determined to affect the peak shape. When operating the JY-Horiba spectrofluorometer, care had to be taken to avoid higher orders of diffraction of the excitation. Thus, with these measurements, there are gaps in the PL spectra where the shutter was closed or the measurement was simply discontinued to avoid saturating the detector.
45 3.2.1 RESULTS
Undoped YAG
As with absorption measurements, undoped YAG PL measurements can
determine a baseline to which measurements collected on doped samples can be
compared. By measuring a number of undoped samples we can gain an understanding of
what impurities might be present in the doped samples and their effect on the dopant’s
luminescence.
Figure 3.2.1 shows undoped YAG PL with 270 nm excitation, along with the
excitation spectrum for the iron emission peak at 775 nm. The iron emission peak can
clearly be observed in the Ar-grown YAG. Observable in all 3 samples, most notably in
the O 2-grown sample, is the series of peaks starting with the largest at 384 nm and
5 7 continuing to about 480 nm. These peaks are attributed to the D3 → Fj transitions in terbium impurities [Choe-2001, Hirata-1996, Jung-2007, Lo-1996, Milliken-2012,
Potdevin-2005, Weg-1985, Zhang-2003]. It has been shown [Jung-2007, Lo-1996,
Potdevin-2005, Weg-1985, Zhang-2003] that in Tb-doped YAG crystals, these peaks are largest for trace concentrations of Tb while these peaks quench at higher concentrations
5 7 and higher wavelength peaks arising from D4 → Fj transitions at 490, 544, 585, 592, and 625 [Kang-1999] become more intense. Though the peaks in Fig. 3.2.1 are unmistakably characteristic of Tb 3+ impurities in YAG, glow discharge mass
spectrometry (GDMS) showed that the concentration of Tb in the O 2-grown sample is
below the detection limit (<0.2 ppm at.). It is also noted that Tb in YAG has a strong
excitation peak near 270 nm [Choe-2001, Hakuta-2003, Jung-2007, Lo-1996, Zhang-
2003], which explains why it shows up so strong in this figure. Figure3.2.2 shows the
46 excitation spectra for various undoped YAG samples for 385 nm luminescence, the strongest Tb 3+ peak observed in our samples.
2000000 Ar-grown
H2-grown O -grown 1500000 2 775 nm excitation
1000000
500000 PL Intensity (counts) PL Intensity 0
200 400 600 800 Wavelength (nm)
Figure 3.2.1 PL of undoped YAG samples with 270 nm excitation (Horiba spectrometer).
5000 Ar-grown 4000 Ar-grown annealed O2-grown YAG
H2-grown YAG 3000
2000
1000 PL Intensity (counts) PL Intensity 0
250 300 350 Wavelength (nm)
Figure 3.2.2 Excitation spectra for undoped YAG samples for maximum Tb trace dopant luminescence at
385 nm (Horiba spectrometer).
47
In Fig. 3.2.2, there is concern that the large emission peak near 800 nm could be a higher order diffraction of the excitation wavelength. Figure 3.2.3 plots several undoped
YAG samples excited with the 270 nm LED. This figure confirms that there does exist a peak around 800 nm in the PL spectra of these crystals. The location and intensity of this peak varies between samples, but typically it is located at about 775 nm. Investigation of
Fig. 3.2.1 shows that the structure up to about 700 nm arises from the LED itself and is not characteristic of the sample, and thus that data is omitted. Annealing in oxygen allows oxygen to diffuse into the sample, and this appears to increase the intensity of the iron emission peak. Air annealed samples display a more intense peak than as grown samples. YAG grown in an oxidizing atmosphere shows the most intense peak of as grown samples. These effects agree with the effects of oxidation on the 256 nm absorption peak discussed earlier in Section 3.1, which also is believed to originate from iron.
48 2000
O2-grown
H2-grown 1500 Ar-grown Ar-grown air annealed Ar-grown 10x10 1000
500 PL Intensity (counts) Intensity PL 0
700 750 800 850 900 Wavelength (nm)
Figure 3.2.3 Photoluminescence spectra of undoped YAG excited with 270 nm LED using 65 second
integration (Ocean Optics spectrometer).
We excited as grown and air annealed Ar-grown YAG using the maximum excitation energy available, 6.5 eV (190 nm), which lies just within the vacuum ultraviolet, to explore the luminescence from the maximum pumping energy of the system. This is also where the band edge is located in YAG. Results are shown in Fig.
3.2.4. An emission peak was found in the UV. Excitation scans for this UV peak revealed peak excitation at 195 nm (Fig. 3.2.5). Figure 3.2.6 displays the PL spectra of undoped YAG samples with 195 nm excitation.
49 40000 Ar-grown 30000 Ar-grown air annealed
20000
10000
0 PL Intensity (counts) Intensity PL
-10000 300 400 500 600 700 800 Wavelength (nm)
Figure 3.2.4 PL spectra of Ar-grown YAG with 190 nm excitation (the maximum pumping energy of the
Horiba spectrofluorometer).
H 2-grown 20000 Ar-grown Ar-grown air annealed
O 2-grown
10000 PL Intensity (counts) PL Intensity
0 200 250 300 Wavelength (nm)
Figure 3.2.5 Excitation spectra for 325 nm emission in undoped YAG samples (Horiba spectrometer).
50
50000
O2-grown
40000 H2-grown
30000
20000
10000 PL Intensity (counts) PL Intensity 0
400 600 800 Wavelength (nm)
Figure 3.2.6 PL spectra of undoped YAG with 195 nm excitation (Horiba spectrometer).
Figures 3.2.4-3.2.6 most notably show the UV emission in YAG, the origin of
3+ which has long been a point of controversy. Some argue that this peak arises from Y Al antisite defects (ADs) [Babin-2005, Murk-1995, Zorenko-2004(b), Zorenko-2007,
Zorenko-2010(b), Zorenko-2011, Zorenko-2012]. Others suggest it to arise from self- trapped excitons (STEs) [Kirm-2000] that may arise from Al vacancies. Still others attribute it to luminescent oxygen vacancies [Rotman-1985, Rozenfeld-1993]. Robbins et al. suggested it to be a combination of defects on the basis that it shifted in different temperature regimes [Robbins-1979(a,b,c)], later suggested to arise from cation vacancies at all three sites in the crystal structure [Hayes-1980]. Recent works seem to agree that luminescence of the AD is at about 320 nm and the STE luminesces near 260 nm [Babin-
2005, Murk-1995, Zorenko-2004(b), Zorenko-2007, Zorenko-2010(b), Zorenko-2011,
Zorenko-2012].
51 Also evident in Figs. 3.2.4 and 3.2.6 is emission from chromium impurities around 700 nm [Bagdasarov-1977, Gluchowski-2009, Hewitt-2011]. This is most easily seen in O2-grown YAG in Fig. 3.2.6, but can also be distinguished in the as grown and air
annealed Ar-grown samples in 3.2.4. No discernible luminescence signal can be seen in
the H 2-grown sample around 700 nm. GDMS measurements confirmed the presence of both Fe and Cr in the O 2-grown sample, present at 1 ppm wt. (0.48 ppm at.) and 0.59 ppm wt. (0.31 ppm at.), respectively. It is worth noting that iron can have emission around 700 nm as well, but its peaks have a significantly different appearance [Rotman-
1989, Varney-2012(c)]. There appear to be trace Ce impurities in H 2-grown YAG
Further measurements exploring the emission of trace Cr, Fe, and Ce impurities in our crystals are shown in Figs. 3.2.7-3.2.9, focusing on the O 2-grown and H 2-grown samples since they showed a significant difference in 700 nm emission in Fig. 3.2.6. Note that in
Fig. 3.2.7, the slit widths used for the two measurements were different, so direct intensity comparisons cannot be made.
52 50000
O2-grown H -grown 40000 2
30000
20000
10000 PL Intensity (counts) Intensity PL 0
300 400 500 600 Wavelength (nm)
Figure 3.2.7 PL excitation spectra for 700 nm emission in O 2-grown and H 2-grown undoped YAG, performed at different slit widths of 3 nm (O 2-grown) and 5 nm (H 2-grown) (Horiba spectrometer).
8000000
H2-grown O -grown 6000000 2
4000000
2000000 PL Intensity (counts) Intensity PL 0
500 550 600 650 700 750 800 850 Wavelength (nm)
Figure 3.2.8 PL spectra for undoped YAG with 450 nm excitation (Horiba spectrometer).
53
H2-grown Ar-grown air annealed 4000000
2000000 PL Intensity (counts) PL Intensity
0
450 500 550 600 650 700 750 800 Wavelength (nm)
Figure 3.2.9 PL spectra for undoped YAG with 425 nm excitation (Horiba spectrometer).
Figure 3.2.7 displays the excitation spectra for the 700 nm emission of the undoped O 2-grown and H 2-grown YAG crystals. The excitation spectrum clearly differs
between the two samples. The excitation spectrum of the O 2-grown YAG is
characteristic of Cr while the excitation spectrum of H 2-grown YAG is actually similar to that of Ce:YAG (see below), suggesting that this sample has trace Ce impurities. The luminescence of this sample also seems to show Ce emission, seen in Figs. 3.2.8-10.
According to Głuchowski et al. [Gluchowski-2009], the narrow peak seen at 688 nm
2 4 arises from the Eg → A2g transition, the diffused bands centered at 675 (anti-stokes vibronic sidebands), 707, and 726 nm correspond to vibronic sidebands (meaning that these are phonon-assisted transitions) assisting fluorescence transition, and the broad
4 band peaked at 695 nm emission underlying these narrower peaks is ascribed to T2g →
4 A2g transitions. The excitation bands at about 430 and 600 nm are assigned to the spin-
54 4 4 4 4 allowed A2 → T2 and A2 → T1 transitions, respectively. Figure 3.2.8 uses the peak excitation for Ce in YAG and actually shows that both samples display Ce emission, the
2 H2-grown YAG signal being much more intense. The O -grown sample in Fig. 3.2.8 does not appear to have luminescent Ce impurities.
In Fig. 3.2.8, the Cr emission can clearly be seen in O 2-grown YAG while it cannot be seen at the displayed scale in the spectrum of the H2-grown sample, in agreement with Fig. 3.2.6. Upon closer inspection, there appear to be a very weak series of peaks around 700 nm, indicating that there may be a very small amount of Cr present in the H 2-grown sample. In Figure 3.2.9, when excited at the characteristic excitation wavelength for 700 nm emission of Cr at 425 nm, emission at 700 nm can clearly be seen in both the H2-grown sample and the air annealed Ar-grown sample.
Also worth discussing as undoped YAG is the sample labeled as Yb:YAG 5% that was determined to contain no Yb. This sample displays very weak Nd 3+ emission,
shown in Fig. 3.2.10, with maximum excitation near 230 nm, a prominent feature in
Nd:YAG absorption (Section 3.1). The small peaks in the PL spectrum of this sample
correspond exactly with Nd 3+ emission peaks in YAG (see Section 3.4). The gaps in data are where the m=2 and m=3 order diffractions of the spectrometer grating are located.
55 80000 Undoped YAG ("Yb:YAG 5%")
60000
40000
20000 PL Intensity (counts) Intensity PL 0
300 400 500 600 700 800 Wavelength (nm)
Figure 3.2.10 PL of undoped YAG (labeled as Yb:YAG 5%) excited by 230 nm (Horiba spectrometer).
Figure 3.2.11 also shows the emission of Fe 3+ in this undoped sample, excited by a 270 nm LED.
56 8000 Undoped YAG ("Yb:YAG 5%") Undoped YAG ("Yb:YAG 5%") annealed 6000
4000
2000 PL Intensity (counts) Intensity PL 0
700 750 800 850 900 Wavelength (nm)
Figure 3.2.11 PL spectra of undoped YAG (labeled as Yb:YAG 5%) excited using 270 nm LED using 65
second integration (Ocean Optics spectrometer).
This sample also contains Ce and Cr impurities, as shown in Fig. 3.2.12.
Ironically, even though this sample labeled is labeled as Yb-doped, it appears to contain almost every dopant studied in this thesis short of Yb itself.
57 Excitation scan using 590 nm emission Emission scan using 460 nm excitation 1.0
0.5
Normalized PL Intensity (a.u.) Intensity PL Normalized 0.0 200 300 400 500 600 700 800 900 Wavelength (nm)
Figure 3.2.12 Excitation spectrum for 590 nm, possibly including trace Ce 3+ emission (line), and PL
emission (points) spectrum for 460 nm excitation of undoped YAG (labeled as Yb:YAG 5%). The
measurements used significantly different settings so they are normalized for ease of viewing (Horiba
spectrometer).
Undoped YAG conclusions
PL measurements on undoped YAG revealed the effects of numerous impurities.
Ce luminescence was detected in H 2-grown YAG while emission from Tb, Cr, and Fe
impurities was detected in O 2- and Ar-grown YAG. Strong UV luminescence was also
observed, corresponding to defects.
Ce:YAG
Figures 3.2.7, 3.2.8, and 3.2.9 clearly show PL of Ce:YAG for very low
concentrations of Ce. The excitation peaks are the same as peaks seen in the absorption
spectra, so the transitions have already been discussed. The emission spectra of Ce in
YAG is due to 5d →4f transitions, typically from the lowest 5d state (5d 1) to the 4f state,
58 2 2 which is split into the doublet f5/2 and f7/2 and is peaked around 525-550 nm [Babin-
2005, Baciero-1999, Blasse-1967(a), Blasse-1967(b), Blazek-2004, Dong-2006,
Fagundes-Peters-2007, Gibbons-1973, Hamilton-1989, Kirm-2000, Liu-2009, Meng-
2008, Miniscalo-1978, Moszynski-1991, Pedrini-1986, Pedrini-2005, Robbins-1979(a),
Robbins-1979(b), Robbins-1979(c), Rotman-1992, Suzuki-1997, Tomiki-1991, Vedda-
2004, Wong-1985, Varney-2012(a), Zorenko-2007, Zorenko-2010(b), Zorenko-2011,
Zorenko-2012, Zych-2000]. Ce 3+ emits in the visible range in YAG due to crystal-field
splitting in YAG which induces very large splitting of the 5d states, making the lowest
state very low [Blasse-1967(b)]. It has been suggested that there exists a much weaker
higher energy emission peak around 300 nm resulting from 5d 2→4f transitions, but it has also been suggested that this is just AD emission. As will be seen later in Chapter 6, the intensity and fast decay time of the 525 nm peak in Ce:YAG make it very promising as a scintillator crystal.
Figures 3.2.13 shows the excitation spectra of select Ce:YAG samples for 550 nm emission. Two large peaks are found in this range at 340 nm and near 450 nm,
5 5 4 corresponding to d1 and d2 → f, as seen in Figs. 3.1.4 – 3.1.6.
59 600000 Ce:YAG 0.15% Ce:YAG 0.15% annealed Ce:YAG 0.3% 400000
200000 PL Intensity (counts) Intensity PL 0
300 400 500 Wavelength (nm)
Figure 3.2.13 Excitation for 550 nm luminescence of Ce:YAG samples (Horiba spectrometer).
Figures 3.2.14 and 3.2.15 show various Ce:YAG samples excited with the 455 nm
LED.
Ce:YAG 0.1% (M) 40000 Ce:YAG 0.15% (M) Ce:YAG 0.14% (C) Ce:YAG 0.3% (UC) 30000
20000
10000 PL Intensity (counts) Intensity PL
0
100 200 300 400 500 600 700 800 900 Wavelength (nm)
Figure 3.2.14 As grown Ce:YAG samples with 455 nm LED excitation using 15 millisecond integration and 10 averaged scans. (M) signifies the sample was supplied by Marketech International, (C) signifies the
60 sample was supplied by Crytur, (UC) signifies that the sample was supplied by United Crystals (Ocean
Optics spectrometer).
0.15% (M) 40000 0.15% annealed (M) 0.2% annealed (M) 0.3% (UC) 30000 0.3% annealed (UC)
20000
10000 PL Intensity (counts) Intensity PL
0
100 200 300 400 500 600 700 800 900 Wavelength (nm)
Figure 3.2.15 Comparison to annealed Ce:YAG with 455 nm LED excitation using 15 millisecond
integration and 10 averaged scans. (M) signifies the sample was supplied by Marketech International, (UC)
signifies that the sample was supplied by United Crystals (Ocean Optics spectrometer).
Oxidative annealing of Ce:YAG can potentially be detrimental to Ce emission.
The intense emission of the 525 nm peak arises from Ce 3+ ions. As discussed above, annealing in the presence of oxygen might oxidize Ce 3+ ions to Ce 4+ , which does not have the same luminesce as its trivalent counterpart.. Thus the tradeoff for filling anion vacancies in Ce:YAG is decreased concentration of luminescent Ce 3+ ions. Figure 3.2.15
actually shows no change to Ce 3+ emission after air annealing, as the annealed spectra are overlaid on top of their as grown counterpart.
61 As is seen in Fig. 3.2.14, increased concentration of Ce does not necessarily translate to increased emission. The optimal doping level seems to be in the range 0.1-
0.15%. The growth conditions play a much larger role than Ce concentration. The reducing atmosphere of argon and hydrogen helped to make the cerium ions trivalent.
Ce 3+ are significantly larger than Y 3+ ions, making them difficult to incorporate into the
crystal structure, but Ce 4+ is a smaller ion that can easily be incorporated into the YAG
crystal structure. Too high of a concentration oversized Ce 3+ ions can put too much strain on the YAG crystal structure and the crystal will fall apart, thus Ce may prefer the tetravalent state for high dopant concentrations. Crystal growth in an atmosphere containing hydrogen reduces cerium to Ce 3+ but growth in pure argon does nothing to
influence the oxidation state, thus a high quantity of Ce ions in the Ar-grown samples
will be tetravalent.
Iron emission can be seen in Ce:YAG samples, as shown in Figs. 3.2.16 and
3.2.17. Figure 3.2.17 uses the 270 nm LED for excitation and thus features below the
intense Ce peak cannot be distinguished.
62 Ce:YAG 0.15% 3000000 CeYAG 0.15% annealed CeYAG 0.3% CeYAG 0.3% annealed 2000000
1000000 PL Intensity (counts) IntensityPL
0
300 400 700 800 900 Wavelength (nm)
Figure 3.2.16 PL spectra of Ce:YAG with 270 nm excitation (Horiba spectrometer).
30000 Ce:YAG 0.15% Ce:YAG 0.15% annealed Ce:YAG 0.2% annealed 20000
10000 PL Intensity (counts) Intensity PL 0
500 600 700 800 900 Wavelength (nm)
Figure 3.2.17 PL spectra of Ce:YAG with 270 nm LED excitation using 10 second integration (Ocean
Optics spectrometer).
63 In Fig. 3.2.16, the tail of the Ce emission dominates the spectra. The majority of the peak is located near the m=2 order diffraction and is thus omitted. Air annealing increases the iron emission at 800 nm, as seen with the undoped YAG. Also visible is the
Tb emission around 400 nm, as identified in the undoped YAG PL measurements. The iron peaks are too weak to distinguish under excitation by the LED.
There are several factors that come into play when annealing Ce:YAG samples in the presence of oxygen. As mentioned above, the Ce3+ ion is oversized for the YAG lattice and is expected to expel nearby oxygen ions to accommodate its size. These vacancies can trap charge carriers and delay Ce lumiscence. Annealing in the presence of oxygen should allow for oxygen to enter the sample at these sites to fill the oxygen vacancies and improve luminescence efficiency of Ce3+ . However, as stated above,
oxidative annealing can oxidize Ce 3+ to Ce 4+ . Figure 3.2.16 shows an inverse relation
between 550 nm and ~800 nm emission, demonstrating that the luminescence of the Fe 3+
ions competes with the Ce 3+ luminescence [Varney 2011]. Oxidative anneal activates
Fe 3+ luminescence, potentially competing with Ce 3+ . As stated above, Fe 3+ impurities are believed to be the main impurities in YAG and thus oxidative annealing in almost any
YAG sample has the potential to activate Fe 3+ luminescence centers to compete with Ce 3+ emission. Thus oxidative annealing of Ce:YAG crystals becomes a trade-off of luminescence efficiency for luminescence speed, both of which are aspects of scintillation efficiency.
64 Ce:YAG conclusions
The optimal Ce-doping level for maximum emission appears to be 0.1-0.15%.
Oxidation can decrease Ce luminescence due to oxidation to the tetravalent state. Fe
impurities appear to compete with Ce emission.
Yb:YAG
Figure 3.2.20 shows the PL of as grown and air annealed Yb:YAG 5% using 273
nm excitation. The narrow series of peaks in Fig. 3.2.18 starting around 400 nm shows
the apparent presence of Tb 3+ ions, which appears stronger after annealing. Note the similarity of emission at higher wavelengths to the spectrum of Ce:YAG 0.3% in Fig.
3.2.16.
10000000 Yb:YAG 5% slit=5 t_int=.3 Yb:YAG 5% annealed slit=5 t_int=.3 Yb:YAG 5% annealed slit=1 t_int=.3 8000000
6000000
4000000
PL Intensity (counts) Intensity PL 2000000
0 400 600 800 Wavelength (nm)
Figure 3.2.18 PL spectrum of Yb:YAG 5% excited with 273 nm (Horiba spectrometer).
65 3.3 COLOR CENTERS
Introduction
Color centers were already briefly introduced in Section 3.2. In this section, we will study them in detail. Color centers can be optically active oxygen vacancies (V O)
containing one, two, or three trapped electrons, denoted as F +-, F-, and F --centers,
respectively. More generally, a color center is defined as a defect that absorbs light in a
spectral region that the crystal does not normally absorb light. Color center activation
refers to the process in which a defect such as an oxygen vacancy traps an electron that in
turn can exist in different energy levels within the trap, thus giving the trap special
absorption and luminescence properties. This section of this will focus on color centers
in YAG activated by light excitation at room temperature.
Color center studies in YAG are important due to the applications of YAG
crystals in high energy optics. Rare-earth-doped YAG is used in such fields as lasers and
scintillation detectors, and in those types of applications they are exposed to high energy
photons or particles that will induce color centers. Thus, these crystals will usually
operate in a color center activated state, therefore it is important to understand how they
behave under such conditions.
Although literature on color centers in YAG is extensive, explanations of color
centers are still mostly speculative and even sometimes appear to be contradictory.
UV induced stable absorption bands at room temperature in YAG were first
reported by Bass and Paladino [Bass-1967], but no explanation to their origin was
suggested.
66 Mori [Mori-1977] found a number of color centers in YAG activated by electron trapping. In this work, new absorption peaks in additively colored YAG at 353, 495, and
833 nm were resolved and appeared to be connected, identified by electron spin resonance and thermoluminescence measurements as oxygen vacancy trapping. Mori noted that UV excitation decreased the absorption peak at 255 nm and increased absorption peaks at 313, 403, and 543 nm, the first of which Mori attributed to Fe 2+ ions but the latter two were left unidentified. All identified absorption peaks were observed at room temperature. Masumoto and Kuwano [Masumoto-1985] discussed the Mori’s findings and how they were affected by varying degrees of oxidation. They attributed the series of peaks at 353, 495, and 833 to F +-centers.
Bernhardt [Bernhardt-1978] discussed sites containing defect electrons at traps
consisting of Mn 2+ or Mn 4+ ions at Al 3+ sites with band maxima at 350 nm, 400 nm, or
460 nm, all apparent at room temperature. Another center discussed appeared under 360
nm excitation and had absorption peaks at 800 and 650 nm at liquid neon temperature (27
K). This was ascribed to an electron trapped between two oxygen ions.
Kovaleva et al. [Kovaleva-1981] studied color centers in Nd:YAG crystals and
noted the effects of co-doping with Cr, Fe, and Nd. The presence of Cr was used to
explain a new absorption peak near 500 nm after gamma excitation. A band at 260 nm
was attributed to Fe 3+ impurities. Co-doping with Ce appeared to prevent the formation
of color centers typically associated with Nd:YAG. Since Kovaleva et al. characterized
Ce 3+ to likely act as hole traps, this finding seemed to suggest that the typical color centers of Nd:YAG resulted from hole trapping. An absorption peak found at 375 nm with luminescence at 410 nm was found in vacuum-grown Nd:YAG samples and was
67 attributed to O - hole centers alongside interstitial cations, which were deemed more likely to have lower valence. One interesting note from this work regarding their studies on growth conditions is that they found that crystal growth in a vacuum removed transition elements from the melt because of their high volatility.
Kaczmarek [Kaczmarek-1999] studied the effects of several dopants on color centers of YAG and other cyrstals. Their studies on YAG included doping with Ce, Fe,
Mg, V, Cr, and Pr. Kaczmarek primarily chose to simply display the data graphically instead of discussing them quantitatively. One of the most applicable absorption spectra to this thesis Kaczmarek displayed is the Cr:YAG spectrum, which shows the addition of a peak just below 300 nm and a broad peak just below 500 nm after coloring. This is deemed relevant to this thesis in light of the findings in the PL section above that displayed the presence of Cr in many samples.
Bagdasarov et al. [Bagdasarov-1977] studied color centers in Cr-doped YAG crystals. They found two absorption bands they attributed to chromium at 380 nm and
480 nm and a third absorption band at 300 nm unassociated with Cr that was attributed to ionization defects in the YAG lattice, all three of which increased after excitation. The
Cr bands at 380 and 480 nm only were observed in samples of Cr concentrations in the range 0.05-0.1 wt. % and were quenched by heating to ~600K. They noted that the increase in 380 and 480 nm peaks was due to a decrease in Cr 3+ concentration since they coincided with the decrease of an absorption band at 600 nm attributed to Cr 3+ in YAG.
Several other UV peaks at 220, 270, 290, and 340 nm were mentioned but did not show consistently in samples and were not explored further.
68 Ashurov el al. [Ashurov-2001] observed a luminescence peak near 500 nm at room temperature that did not appear to come from recombination of opposite sign charge carriers. The peak disappeared with heating but would quickly reappear with high
γ-irradiation, suggesting that the peak arose from trapped electrons at an oxygen vacancy.
Comparison of the luminescence decay time of this peak to the known F +-center decay
+ time of Al 2O3 suggested that this 500 nm peak arose from F -centers by arguments of analogy.
Springis et al. [Springis-1991] found an emission peak at 400 nm at room
temperature with excitation peaks at 235 and 370 nm. They first mention the possibility
of oxygen divacancies, but suggest that the high concentration of these centers makes that
unlikely. They ultimately ascribe it to F +-centers. They expanded upon this idea in a
later paper [Pujats-2001] in which they discuss F +-, F-, and F --centers at 80K. They relate 460 nm luminescence with 240 nm excitation previously proposed to be F-center absorption by Bunch [Bunch-1977]. Measurements showed a higher ratio of 460 nm emission to 400 nm emission after x-irradiating the samples, supporting their assignments to F- and F +-centers, respectively. They discuss a third type of center with absorption bands at 360 nm, 480 nm, and 830 nm, the same center observed by Mori [Mori-1977], and suggest based on similar measurements that they arise from F --centers.
Zorenko et al. [Zorenko-2010(a)] arrived at the same conclusions as Pujats and
Springis [Pujats-2001] about F+- and F-centers and studied their absorption and
luminescence in greater detail, identifying multiple peaks and shoulders in measurements
conducted at 10K. From these measurements, they were able to map out the energy
levels of the F +- and F-centers. They found F +-centers to have three excited energy states
69 with absorption energies at 5.54, 6.0, and 6.51 eV relative to the ground state, denoted as
1A→2A, 1A→2B, and 1A→3P. Two F-center absorption peaks were found with energies of 5.19 and 5.62 eV, labeled as 1S→3P and 1S→1P transitions.
It is relevant to mention that Chakrabarti [Chakrabarti-1988] only observed 240 and 370 nm bands after neutron irradiation and did not observe these peaks after UV excitation, the band at 240 nm ascribed to F-centers and the one at 370 nm ascribed to F +-
centers. Chakrabarti did, however, observe a peak at 310 nm after UV excitation. After
neutron irradiation, UV irradiation was found to decrease the 240 nm peak and increase
370 nm, suggesting that this direct F-center excitation freed trapped electrons and thus
increased F +-center concentration. Irradiation by γ rays had the opposite effect,
suggesting that F +-centers present in the lattice trapped electrons to form F-centers.
Chakrabarti observed two luminescence peaks at 425 and 500 nm after coloring: the former was attributed to F-centers and the latter was suggested to arise from an unknown impurity.
To better understand the mechanisms occurring during color center activation, it is important to examine the impurities present and which ones trap which type of charge carrier. Batygov et al. [Batygov-1972] studied YAG crystals that were undoped or activated with Ce, Pr, Nd, Sm, Eu, Tb, Dy, Yb, Lu, Er, or Ho separately. Their findings showed that only Eu 3+ and Yb 3+ , both of which readily transition to the divalent state, can
act as electron traps at room temperature. Sm 3+ has the third largest electron affinity of the studied activators, but it was found that Sm 2+ is unstable at room temperature. In the
same way, Pr 3+ and Tb 3+ were found to act as hole traps due to their ease of transition to
70 the tetravalent state. They argued that the same effect must be produced by Ce 3+ , but the
strong absorption characteristic of Ce 3+ in YAG made this difficult to observe.
Babin et al. [Babin-2011], Springis et al. [Springis-1991], and Zorenko et al.
[Zorenko-2010(a)] make mention that F +-AD and F-AD dimers can be observed, an interesting observation in light of the computations of Stanek et al. [Stanek-2008] and
Milanese et al. [Milanese-2004] showing that anion vacancy–isovalent cation AD clusters are energetically unfavorable to ADs alone and thus highly unlikely. However,
Grigorjeva et al. [Grigorjeva-2010] showed with electron paramagnetic resonance studies
3+ that Y Al ions can associate with V O defects in yttrium aluminum perovskite (YAP,
2+ + YAlO 3) and can even form Y Al -F -center pairs under certain conditions with UV
excitation. It is also worth mentioning the volume occupied by Y 3+ ions as opposed to the Al 3+ ions they replace. The ionic radius of Y 3+ is 106 pm [Emsley-1989] while the
ionic radius of Al 3+ is merely 57 pm [Emsley-1989], thus the antisite Y3+ ion is oversized
for the Al 3+ site. This great difference in size at ADs strains the YAG crystal structure and deviates it from cubic symmetry and theoretical lattice constants [Dong-1991, Geller-
1967, Kuklja-2000, Stanek-2008]. It seems reasonable to suggest that a comparably oversized antisite Y 3+ ion on an Al 3+ site could displace neighboring oxygen, forming a
3+ YAl -VO dimer, as discussed.
Oxygen vacancies have also been discussed as present in the YAG crystal structure as charge compensation for divalent impurities, and vise versa [Milanese-2004,
Xu-2003, Yang-2002].
It is worth mentioning that Rozenfeld and Rotman [Rozenfeld-1993] speculated on possible energy levels for oxygen vacancies, but their intent was to explain UV
71 luminescence in YAG, thus the transitions discussed do not apply to the emission related to VOs postulated above.
Experimental setup
Our measurements consisted of UV excitation of the samples, typically for 30
minutes, followed by absorption measurements. This process activates color centers by
exciting electrons with enough energy to reach the conduction band, allowing them to
roam freely throughout the crystal until they either relax back into orbit around an atom
or become trapped at a defect. Some defects, such as the discussed oxygen vacancy, gain
optical properties upon trapping an electron. A large enough concentration of these
activated color centers can change the color of the crystal. Even if the change in color
cannot be noted with the naked eye, absorption measurements usually can distinguish
subtle changes in color. UV excitation of samples was performed with the Ocean Optics
PX-2 Pulsed Xenon Light Source discussed in Section 2.2. Absorption measurements
were conducted using the Perkin Elmer UV/VIS Spectrometer discussed in Section 3.1.
To free the trapped electrons from defects, allowing them to return to the valence band,
thermal energy was applied by heating the samples to 400°C. Sample heating was
performed on an Instec, Inc. HCS302-01 hot/cold stage.
Presentation of results
Absorption spectra are presented for ambient and UV irradiated samples. To
display how the absorption changes after the color center activation processes, absorption
shift plots are created by subtracting the ambient absorption spectrum from the absorption
spectrum of the crystal after UV excitation or heating. Following the example of Mori
72 [Mori-1977], some of these subtracted spectra were fit to Gaussians to attempt to resolve individual peaks and understand exactly which centers are present.
3.3.1 RESUTS
Undoped YAG
Excitation of undoped YAG varies strongly with growth conditions. Figure 3.3.1
shows the change in Ar-grown YAG absorption immediately after 30 minute UV
excitation and after heating to 400°C. Figure 3.3.2 shows these measurements for air
annealed Ar-grown YAG and Figs. 3.3.3 shows the spectra of UV-excited O 2-grown and
H2-grown YAG.
Energy (eV) Energy (eV) 6.2 3.1 2.1 1.6 1.2 1.0 6.2 3.1 2.1 1.6 1.2 1.0 0.20 0.010 Ambient UV excited 30 minutes UV excited 30 minutes Heated to 400C Heated to 400C 0.15 0.005 Absorbance Absorbance
0.10 0.000
0.05 -0.005 200 400 600 800 1000 1200 200 400 600 800 1000 1200 Wavelength (nm) Wavelength (nm)
Figure 3.3.1 Ar-grown YAG absorption spectra (left) and absorption spectra shift (right) after UV
excitation and heating.
73
Energy (eV) Energy (eV) 6.2 3.1 2.1 1.6 1.2 1.0 6.2 3.1 2.1 1.6 1.2 1.0 0.20 0.02 Ambient UV excited 30 minutes 0.00 Heated to 400C 0.15 -0.02 Absorbance Absorbance
0.10 -0.04 UV excited 30 minutes Heated to 400C
-0.06 0.05 200 400 600 800 1000 1200 200 400 600 800 1000 1200 Wavelength (nm) Wavelength (nm) Figure 3.3.2 Air annealed Ar-grown YAG absorption spectra (left) and absorption spectra shift (right) after
UV excitation and heating.
Energy (eV) Energy (eV) 6.2 3.1 2.1 1.6 6.2 3.1 2.1 1.6 1.2 1.0 0.2
H2-grown YAG H 2-grown YAG 3 UV exicited 30 minutes O 2-grown YAG
O2-grown YAG UV exicited 30 minutes 0.0
Absorbance 2 Absorbance
1 -0.2 200 400 600 800 200 400 600 800 1000 1200 Wavelength (nm) Wavelength (nm)
Figure 3.3.3 O2-grown and H 2-grown YAG absorption (left) and absorption spectra shift (right) after 30
minute UV excitation.
Air annealed Ar-grown YAG turned an orangish color after UV excitation and
O2-grown YAG turned grayish.
From Figs. 3.3.1-3.3.3, it can be seen that 256 nm O2-→Fe 3+ charge transfer peak decreases after UV excitation in samples containing a low concentration of oxygen vacancies and increases in the other samples. The decrease of the 256 nm peak has been
74 suggested to arise from electron capture by iron [Mori-1977], though the for the decrease to occur a change in the charge state of either Fe 3+ or O 2- must occur such that the O2-
→Fe 3+ charge transfer can no longer occur. In all samples, there appears to be a peak
around 310 or 320 nm. This might be the 313 nm peak presented by Mori [Mori-1977]
and proposed to arise from Fe 2+ . This appears to correspond well in these samples with
the decrease of the 256 nm peak in O 2-grown and air annealed Ar-grown YAG, though
H2-grown YAG does not appear to show these features. However, later measurements in
Section 3.4 appear to provide evidence that the decrease in 256 nm absorption does not
arise from electron capture by Fe 3+ but rather by hole capture by oxygen.
Fitting the shift in H 2-grown YAG absorption to Gaussians proved inconclusive
regarding the presence of iron-related color centers. It is possible that the 256 nm peak
increases for this sample after UV excitation, which would likely be due to the release of
charge carriers from oxygen to form O 2-, which could be localized around O 2- as charge compensation for other impurities such as hydrogen impurities or V Os. It appears that there may be a peak near 300 nm, near the 313 nm peak attributed to Fe 2+ [Mori-1977].
The large decrease in the absorption spectrum of H 2-grown YAG at 230 nm could be from a decrease in F-center concentration. F-center absorption has one thin peak near
240 nm [Burch-1977, Pujats-2001, Zorenko-2010(a)]. F +-center absorption has peaks at
both 235 and 370 nm [Pujats-2001, Springis-1991, Zorenko-2010(a)], and since no such
decrease is noted at 370 nm, this is ruled out. In fact, there appears to be a slight increase
peaked around 370 nm, which, if arising from F +-centers, reaffirms the suspicion that the decrease at 230 nm is due to the decrease in F-center concentration, since it is possible that some F-centers would only lose one electron to form F +-centers. Another possibility
75 is a change in Nd 3+ impurities, which have a peak at 225 nm [Kovaleva-1981], but no other measurements conducted on this sample have shown the presence of Nd in this sample.
Figures 3.3.1-3.3.3 display broad peaks similar to the absorption shift in samples studied by Mori [Mori-1977]. Following Mori’s technique, fitting to Gaussians resolved a weak peak near 500 nm in all samples except the H2-grown YAG. PL measurements
3+ confirmed the presence of Cr ions in each of these samples except H 2-grown YAG, thus it seems appropriate to attribute this additional peak that arises after coloring to chromium, as suggested by Bagdasarov et al. [Bagdasarov-1977], Kovaleva et al.
[Kovaleva-1981], and Kaczmarek [Kaczmarek-1999].
The as grown Ar-grown YAG may show F --centers. Although the shift in absorption for this sample is very weak, it seems to show peaks around 350, 500, and 800 nm, which is similar to the peaks at 360, 480, and 830 reported to be characteristic of F -- centers [Pujats-2001]. These peaks appear to be robust against heating to 400°C.
After heating to 400°C, the color centers in the O 2-grown sample persisted.
However, the sample did become clear again after heating to 450°C.
Figure 3.3.4 displays the change in absorption spectra of the labeled as Yb:YAG
5% that was found to contain no Yb. The figure shows the change in absorption after
UV-excitation for the as grown and air annealed samples.
76 Energy (eV) Energy (eV) 6.2 3.1 2.1 1.6 6.2 3.1 2.1 1.6 1.2 1.0 1.0 0.10 As grown As grown 0.8 UV excited 30 minutes Air annealed 96 hrs Annealed in air 96 hours 0.05 UV excited 30 minutes 0.6 Absorbance Absorbance 0.4 0.00
0.2 -0.05
0.0 200 400 600 800 200 400 600 800 1000 1200 Wavelength (nm) Wavelength (nm)
Figure 3.3.4 Undoped YAG (labeled as Yb:YAG 5%) as grown and air annealed absorption spectra (left)
and absorption spectra shift (right) after 30 minute UV excitation.
Figure 3.3.4 shows that the increase over the range of visible wavelengths with a maximum near 310 nm increases independent of the increase of decrease of the 256 nm peak. Thus, it appears that this is not actually related to Fe 2+ and is instead of some
unknown origin.
Nd:YAG, Tm:YAG, and Yb:YAG
Color centers were examined in some of the other samples as well. The change in
absorption spectra after UV-excitation of Nd:YAG 1%, air annealed Nd:YAG 1%, and
air annealed Tm:YAG 0.8% are shown in Fig 3.3.5. Figure 3.3.6 presents the change in
absorption spectra for UV-excited Yb:YAG 1%, 3%, and 10% while Fig. 3.3.7 shows the
effect of air annealing on Yb:YAG 10% and Yb:YAG 5%. In all of these samples,
heating to 400°C bleached the visible color centers.
77 Energy (eV) Energy (eV) 6.2 3.1 2.1 1.6 1.2 1.0 6.2 3.1 2.1 1.6 1.2 1.0 0.04 Tm:YAG 0.8% annealed Nd:YAG 1% 0.03 0.4 UV excited 30 minutes Nd:YAG 1% 0.02 Nd:YAG 1% annealed
0.01 Absorbance Absorbance 0.2 0.00 -0.01
-0.02
0.0 -0.03 200 400 600 800 1000 1200 200 400 600 800 1000 1200 Wavelength (nm) Wavelength (nm)
Figure 3.3.5 As grown Nd:YAG absorption spectra before and after excitation (left) and as grown
Nd:YAG and air annealed Tm:YAG 0.8% and Nd:YAG 1% absorption spectra shift (right) after 30 minute
UV excitation.
Energy (eV) Energy (eV) 6.2 3.1 2.1 1.6 1.2 1.0 6.2 3.1 2.1 1.6 1.2 1.0 0.8 Yb:YAG 10% 0.04 Yb:YAG 1% (fg)(M) UV excited 30 minutes Yb:YAG 3% (fg)(M) 0.6 Yb:YAG 10% (UC) 0.02
0.4 0.00 Absorbance Absorbance
-0.02 0.2
-0.04 0.0 200 400 600 800 1000 1200 200 400 600 800 1000 1200 Wavelength (nm) Wavelength (nm)
Figure 3.3.6 Yb:YAG 10% absorption spectra before and after UV excitation (left) and Yb:YAG 1%, 3%, and 10% absorption spectra shift (right) after 30 minute UV excitation. (M) signifies that the sample was procured from Marketech International, (UC) signifies that the sample was procured from United Crystals, and (fg) means the sample was unpolished (fine ground). The feature around 400 nm in the spectrum for the Yb:YAG 1% sample is a feature of the spectrometer and does not represent an actual absorption peak.
78
Energy (eV) Energy (eV) 6.2 3.1 2.1 1.6 1.2 1.0 6.2 3.1 2.1 1.6 1.2 1.0 0.6 0.02 Yb:YAG 5% annealed 96 hrs Yb:YAG 10% as grown UV excited 60 minutes 0.01 Yb:YAG 10% air annealed Yb:YAG 5% air annealed 0.4 0.00 Absorbance Absorbance -0.01 0.2 -0.02
0.0 -0.03 200 400 600 800 1000 1200 200 400 600 800 1000 1200 Wavelength (nm) Wavelength (nm) Figure 3.3.7 Air annealed Yb:YAG 5% absorption spectra before and after UV excitation (left) and as
grown Yb:YAG 10% and air annealed Yb:YAG 5% and Yb:YAG 10% absorption spectra shift (right) after
UV excitation. The as grown sample underwent 30 minute excitation, the annealed sample was excited for
40 minutes.
Figure 3.3.5 shows a significant change in the 256 nm peak in all samples shown, increased in as grown Nd:YAG 1% but decreased for both air annealed samples. It was shown in Section 3.1 that annealing in air oxidizes Fe 2+ to Fe 3+ , so this result is to be
expected. The majority of iron in the as grown sample is divalent, while the majority of
iron ions in the air annealed samples are trivalent. On top of that, there exist more
oxygen vacancies in the as grown sample. As a result, UV excitation of the as grown
sample frees electrons in Fe 2+ ions that become trapped at oxygen vacancies, leaving behind Fe 3+ ions and thus increasing 256 nm absorption. UV excitation of air annealed
samples works differently because the Fe 3+ ions do not have the extra electron to excite and on top of that there are fewer V Os. With such a low concentration of V O, electrons can trap at other locations, including Fe 3+ ions, which forms Fe 2+ and decreases the absorbance at 256 nm. Figure 3.3.6 also shows a very strong increase at 256 for lower dopant concentrations of Yb.
79 Figure 3.3.5 shows that the increase over the range of visible wavelengths with a maximum near 310 nm increases independent of the increase of decrease of the 256 nm peak. Thus, it appears that this is not actually related to Fe 2+ and is instead of some
unknown origin.
There appears to be a decrease in UV absorption in as grown Nd- and Yb-doped
samples but an increase in the UV of the air annealed Nd:YAG and Yb:YAG. This likely
arises from a similar mechanism to the 256 nm peak discussed above. Yb 3+ ions are
known electron traps in YAG, but Nd 3+ ions were determined to trap neither holes nor electrons [Batygov-1972], thus this effect is most likely dependent on oxygen concentration. This is explored more in detail below with Yb:YAG.
In Figs. 3.3.6 and 3.3.7, the absorption spectrum of Yb:YAG 10% actually decreases and appears to shift in similar fashion but opposite direction to some undoped samples. Yb 3+ ions act as electron traps, and their concentration in this sample is very high. The growth atmosphere of argon allows for an appreciable concentration of V Os,
and one method of charge compensation is the inclusion of Yb in the divalent state
instead of the trivalent state. It is known that the bands leading to the increase in
absorption spectra in undoped YAG arises from electron traps [Mori-1977]. Thus it is
expected that, for undoped YAG, whatever these traps are become filled after excitation
frees electrons from their bound states at other sites in the lattice. In the case of the
Yb:YAG 10% sample, the Yb 3+ ions act as the primary electron traps after excitation,
forming Yb 2+ ions. This is further evidenced by the large decrease in O2-→Yb 3+ charge-
transfer absorption at low wavelengths as well as the decrease in the Yb 3+ absorption
band around 1000 nm. The overall decrease in absorption comes from excitation freeing
80 electrons from the electron traps responsible for those particular bands and subsequent trapping at ytterbium. An oxidizing anneal almost entirely removed these coloring effects, as evidenced by Fig. 3.3.7, where the only changes seem to be an increase in O2-
→Yb 3+ charge-transfer absorption and very slight decreases in O2-→Fe 3+ charge-transfer
3+ absorption and Yb absorption near 1000 nm. The oxidizing anneal filled V Os and
oxidized impurities and dopants, changing some divalent ions to trivalent or even
tetravalent. It is possible that electrons at oxygen sites that formed O 3- were freed to trap at Fe 3+ and Yb 3+ or Yb 4+ , thus increasing the availability of O 2- ions for charge transfer.
Overall, the decrease in concentration of Fe 3+ ions led to the decrease in O2-→Fe 3+
charge-transfer absorption. It is difficult to tell how much the concentration of Yb 3+ changed since some Yb 3+ was ionized to Yb 2+ while some Yb 4+ was ionized to Yb 3+ , but
the increase in O2-→Yb 3+ charge-transfer absorption seems to suggest that the increase in
O2- concentration was much higher than any potential decrease in Yb 3+ concentration.
It is also possible that antisite Y can act as an electron trap, as ytterbium can be
divalent, although electron trapping at Y 3+ , which has an electronic configuration of [Kr],
3+ would be unlikely. On top of that, Zorenko et al. [Zorenko-2007] argued that Y Al could form a shallow electron trap, so stability of this trap at room temperature for long enough time to measure the absorption spectrum of the sample is even less likely.
Color centers conclusions
Based on the measurements presented above, it is possible to form conclusions regarding color center activation processes in YAG. In samples deficient in oxygen that contain a high concentration of oxygen vacancies, iron primarily exists in the divalent state, and UV excitation frees electrons from O 2- and Fe 2+ , as both of those are capable of
81 losing electrons during UV excitation. As a result, excitation greatly increases Fe 3+
concentration, increasing O 2-→Fe 3+ charge transfer, increasing the 256 nm peak. Other
dopants that do not easily transition to other valence states see a decrease in charge
transfer absorption due to the decreased O 2- concentration. After annealing in air, the
oxygen concentration in the sample is much higher, and thus there are few oxygen
vacancies. Thus, UV excitation can not have the same effect as it did on the as grown
sample since there are less electron traps available. In the annealed samples, there is also
a high concentration of trivalent iron. To charge compensate some lattice defects such as
aluminum vacancies, oxygen can incorporate as O - instead of O 2-. UV excitation allows electrons to trap at O - to form O 2-. Fe 3+ in these conditions also acts as an electron trap,
hence the 256 nm decreases after UV excitation while charge transfer bands with other
dopants increases.
Ytterbium presents a special case where the dopant can act as an electron trap.
This is most clearly observed in Figs. 3.3.6 and 3.3.7, where the Yb:YAG 10% sample
decreases over visible wavelengths in the same way that it increases in other samples.
The traps that cause this absorption shift in other samples are more stable under ambient
conditions in this sample. UV excitation frees electrons from these traps and they trap at
ytterbium ions, hence the observed decrease. Annealing this sample filled oxygen
vacancies and stabilized Yb 3+ , making it so that Yb 3+ could no longer act as an electron trap. Thus, UV excitation did not cause electrons to trap at ytterbium ions and there was no change over the majority of the absorption spectrum aside from the charge transfer band.
82 3.4 RADIOLUMINESCENCE
Introduction
Radioluminescence (RL) is the optical phenomenon in which luminescence in a material is stimulated by ionizing radiation such as beta particles or high-energy photons.
The ionizing radiation simultaneously activates all optically active centers, allowing for quick investigation of the luminescent properties of a sample. As will be discussed later in Chapter 6, this characteristic of RL measurements is very useful in scintillation testing measurements.
X-ray excitation, as used in part of these studies, can have different effects at high enough energy. Low energy x rays can only induce the Compton effect. Higher energy x rays with energy around 10-100 keV can induce the photoelectric effect as well as the
Compton effect. Very high energy x-rays can also result in pair production. The x rays used in this work were produced by K α transitions in Cu of energy 8.0478 keV (0.15406 nm) and are expected to induce the Compton and photoelectric effects [Li-2000]. The excitation creates excitons, color centers, and other defect aggregates. These can quickly produce luminescence by recombination, form metastable color centers, or non- radiatively decay in a process that produces only heat [Rajan-2005].
Experimental setup
Samples were excited for RL measurements using beta particles or x rays. The
samples were exposed to beta particles from a 100 µCi 60 Co source obtained from Eckert
& Ziegler. A schematic diagram showing the setup for beta-stimulated luminescence is shown in Fig. 3.4.1. The source sample itself emitted a weak peak around 800 nm, so a thin piece of black paper was placed between the 60 Co source and the sample to eliminate
83 this emission. Literature pertaining to beta-particle-stimulated luminescence of YAG crystals is limited [Zych-2000].
X-ray stimulated luminescence was conducted using the setup described in
Section 2.2, redisplayed in Fig. 3.4.2. XSL has been used before to study the UV and optical emission of YAG crystals [Batygov-1991, Guerassimova-2001, Hayes-1980,
Nikl-2005, Niklas-1983, Pujats-2001, Smol’skaya-1987, Varney-2012(c), Wisniewska-
2001, Zorenko-2004(a), Zorenko-2004(b), Zorenko-2007].
Both types of RL measurements were conducted at room temperature and used a
Ocean Optics USB2000+ spectrometer, described previously in Section 2.2. All measurements were conducted by averaging 3 scans and using a boxcar setting of 3 to reduce noise.
Figure 3.4.1 Schematic diagram for RL measurements conducted with the 60 Co source.
84
Figure 3.4.2 Schematic diagram for XSL and some PL measurements. The dashed box represents the
boundary of the radiation-safe chamber that shields x rays and stops outside light from getting in.
Both x rays and beta particles are forms of ionizing radiation. The x rays are the much stronger form of excitation due to the much higher quantity of x rays incident upon the sample than was achievable than beta particles, but the beta particles are much higher energy. Some plots using the 60 Co source are included in the section for comparison.
One major challenge faced when trying to use the 60 Co source was infrared emission from the source itself. Two different methods were employed to stop this luminescence signal from interfering with measurements: 1) the dark reference spectrum was taken with the spectrometer collecting this infrared emission from the 60 Co source with no sample in place and was subtracted from the spectrum with the sample in place or 2) a thin sheet of black paper was placed between the 60 Co source and the sample, which
significantly reduced signal intensity, as shown in Fig. 3.4.1. Neither method
demonstrated a distinct advantage over the other and data recorded by the first method
was reported due to its stronger luminescence signal [Reda-2012]. Though
measurements taken using the XSL system were highly repeatable, measurements
conducted using beta excitation were not.
85 60 Co emits a beta particle and two gamma rays of energies 1.17 and 1.33 MeV during decay to 60 Ni [Deutsch-1945]. Though no definitely conclusive evidence could be found, evidence, namely that the luminescence is observed only when the sample is very close to or in contact with the source, suggests that they are beta particles. Further evidence that the exciting particles are betas is the fact that betas are much more efficient at excitation than gammas. Along those lines, the beta particles emitted from 60 Co are electrons, and electron injection has long been used to study luminescence in YAG crystals by a process called cathodoluminescence [Ashurov-2001, Gibbons-1973,
Robbins-1979(a), Robbins-1979(b), Robbins-1979(c), Robbins-1979(d), Zorenko-2005,
Zorenko-2007, Zorenko-2010(b)].
Beta particle radiation by 60 Co involves emission of a high energy electron. Upon entering a sample, this electron undergoes a series of elastic and inelastic collisions with the lattice atoms. The inelastic collisions produce a variety of useful signals such as generation of electron-hole pairs, secondary electron emission, and characteristic x rays
[Yacobi-1986]. The injected electrons lose energy in these collisions by energy transfer until they finally come to a rest. The process measured in our RL measurements involves excitation of valence or dopant electrons to the conduction band and subsequent recombination with holes with the emission of a photon. Excitation by high energy electrons allows for simultaneous excitation of all luminescence centers present in the crystal, differing from PL measurements, where the luminescence was often heavily dependent on the excitation used.
86 3.4.1 RESULTS
Undoped YAG
In photoluminescence measurements (Section 3.2), undoped YAG did not display
any very intense luminescence. The most intense peak came from trace Fe 3+ impurities, and most other peaks were identified as originating from other trace impurities. UV
Luminescence showed up very weak at the highest energy excitation. As seen in Fig.
3.4.3, XSL presents very different luminescence spectra of YAG and portrays the ADs as the dominant luminescence center in undoped YAG.
O2-grown O -grown Ar+H2 annealed 20000 2
H2-grown Ar-grown Ar-grown annealed Ar-grown 10x10x1 10000
0 Luminescence Intensity (counts) Intensity Luminescence 100 200 300 400 500 600 700 800 900 Wavelength (nm)
Figure 3.4.3 XSL spectra of undoped YAG using 65 second integration.
The large peak at 340 nm is attributed to ADs, as discussed in Section 3.2. Also discussed in Section 3.2 is the STE, which appears as a slight shoulder to the AD peak near 260 nm and is clearly much less intense than the AD peak. The Fe 3+ peak at 800 nm is also barely visible in these spectra. An array of peaks around 700 nm is easily distinguishable. In Section 3.2, a collection of peaks in this same range was attributed to
87 Cr 3+ ions. However, the shapes and positions of these peaks are different, and in fact it is believed that these also arise from Fe 3+ ions [Rotman-1989, Varney-2012(c)].
There appears to be a weak peak centered just below 500 nm in most samples.
Yb 3+ -O2- charge transfer luminescence emits a peak near that wavelength [Fredrich-
Thornton-2010, Guerassimova-2001, Kamesnkikh-2005, Pieterson-2000], so trace Yb 3+ impurities as suggested in Section 3.1 are a possibility, though this emission is typically almost completely quenched at room temperature. Ashurov et al. [Ashurov-2001] attributed a peak near 500 nm to F +-centers, as discussed in Section 3.3. Also discussed in Section 3.3 were the works of Pujats and Springis [Pujats-2001] and Zorenko et al.
[Zorenko-2010(a)], who attributed 460 nm luminescence to F-centers.
The counterargument to either of the ideas involving V Os is that the intensity of
this peak seems to increase with increasing oxygen concentration. Let us investigate the
matter under the assumption that the emission arises from F-centers. This requires that
oxygen vacancies contain two trapped electrons, which would be necessary for charge
neutrality in the oversimplified case where every cation is trivalent and there only exist
anion vacancies. It may be of some benefit to examine another luminescent phenomenon
observed in Fig. 3.4.3 in order to gain a clearer picture of what it going on.
Let’s step back from that for a moment to discuss the ADs. Stanek et al. [Stanek-
2008] and Milanese et al. [Milanese-2004] separately calculated that anion vacancy–
isovalent cation AD clusters are energetically unfavorable to ADs alone and thus highly
unlikely. However, these arguments were founded on calculations based on electrostatic
forces and crystal structure formation and ignore the size of Y 3+ ions compared to the
Al 3+ ion sites they replace [Emsley-1989, Geller-1967, Milanese-2004, Xu-1999]. As
88 discussed in Section 3.3, the much larger Y 3+ ion sitting on the Al 3+ site will likely eject adjacent oxygen ions. An aluminum ion adjacent to the ejected oxygen may also be ejected as charge compensation [Selim-2007].
The mechanism of AD luminescence in YAG is not completely understood. It is
believed to involve trapping and detrapping of charge carriers at antisite defects. The
emission itself arises from recombination at cation or anion sites perturbed by antisite
3+ YAl [Zorenko-2007]. Thus, the presence of oxygen is integral to AD emission.
Due to the nature of the AD, it seems prudent to primarily compare AD
luminescence between samples from the same manufacturer, otherwise we would simply
be discussing which company produces the more stoichiometrically correct YAG crystal
(which, to that end, it appears that Crytur grows more stoichiometric YAG than United
Crystals). From Fig. 3.4.3, it is clear that reducing conditions seem to decrease AD
luminescence while oxidative conditions seem to increase it. However, the samples
should still contain the same ratio of Y 3+ :Al 3+ , as WDS measurements show (Section
2.1), thus the concentration of ADs must be constant. Instead, as discussed above, the
3+ concentration of oxygen plays a large role in AD luminescence. Y Al effectively
catalyzes the recombination in oxygen ions, thus it is clear why samples possessing a
higher concentration of oxygen display greater AD emission in Fig. 3.4.3.
It seems that it would take a large difference in VO concentration to have this noticeable of a difference on AD emission. Assuming this to be true, H 2-grown YAG
appears to have a much larger concentration of V O than O 2-grown YAG. During x-ray excitation of the sample, a number of electrons become trapped at oxygen vacancies. In
Section 3.3, it was determined that these traps appear to be unstable at room temperature;
89 however under constant irradiation, these traps will constantly be filled. For very high concentrations of V O, the number of anion vacancies will be large compared to the number of excited electrons that can be trapped, thus the majority of V Os will appear as
F+-centers, appearing to be maintained as such due to the constant detrapping of electrons
due to instability of these traps at room temperature. Following the same arguments for a
lower concentration of V O, more V Os will trap two electrons simply due to the lack of presence of V Os; forming F-centers with the emission observed in Fig. 3.4.3 peaked just
below 500 nm. Annealing or growth in oxygen does not fill all oxygen vacancies,
3+ especially localized around oversized impurity ions such as Y Al , where there is insufficient room to include the anion, or near cation vacancies where charge neutrality is maintained by a V O. Thus, it is understood how the samples containing a greater
concentration of oxygen display more intense F-center emission.
In subsequent chapters, thermoluminescence (Chapter 4) and positron annihilation
lifetime spectroscopy (Chapter 5) measurements more closely examine the charge and
concentration of defects present in the crystal.
Undoped YAG conclusions
UV emission in undoped YAG arises from defects related to the crystal’s inherent
nonstoichiometry. Growth or annealing in an atmosphere containing oxygen improves
this luminescence while growth or annealing in an atmosphere containing hydrogen
decreases this emission. Luminescence from oxygen vacancies and other impurities,
mainly Fe, are also identified in spectra.
90 Ce:YAG
Figures 3.4.4 and 3.4.5 show XSL for Ce:YAG samples at 65 and 20 second integration, respectively. The two different integration times were used to examine the
UV and Ce 3+ emission peaks, respectively.
0.1% (M) 60000 0.15% (M) 0.15% annealed (M) 0.2% annealed (M) 0.3% (UC) 40000 0.3% annealed (UC)
20000
0 Luminescence Intensity (counts) Intensity Luminescence 100 200 300 400 500 600 700 800 900 Wavelength (nm)
Figure 3.4.4 XSL spectra of Ce:YAG of different dopant concentrations using 65 second integration. (M)
signifies the sample was supplied by Marketech International, (UC) signifies that that sample was supplied
by United Crystals Ltd.
91
40000 0.1% (M) 0.15% (M) 0.15% annealed (M) 0.2% annealed (M) 0.3% (UC) 0.3% annealed (UC)
20000
0 Luminescence Intensity (counts) Intensity Luminescence 100 200 300 400 500 600 700 800 900 Wavelength (nm)
Figure 3.4.5 XSL spectra of Ce:YAG of different dopant concentrations using 20 second integration. (M)
signifies the sample was supplied by Marketech International, (UC) signifies that that sample was supplied
by United Crystals Ltd.
As with above, it is appropriate to only compare samples from the same company during AD discussions. Another point to recall is that Ce 3+ ions are significantly larger than Y 3+ and will possibly eject adjacent oxygen ions due to their size [Emsley-1989,
Varney-2012(a)]. Oxidative anneal on Ce:YAG can both fill V Os and oxidize cerium ions from trivalent to tetravalent, so it is uncertain if it improves Ce 3+ emission. Figure
3.4.5 displays mixed results on this subject. Comparing Figs. 3.4.4 to 3.4.3 shows significantly decreased AD emission for Ce-doped samples, potentially due to the Ce 3+
ions absorbing a large portion of the incident excitation energy before it can reach ADs.
One significant feature of the spectra of Figs. 3.4.4 and 3.4.5 is the dip that
appears in the AD peak centered at 340 nm. Recalling Figs. 3.1.4-3.1.7, this is a strong
92 absorption peak of Ce 3+ . This implies that some AD emission is absorbed and reemitted
by Ce 3+ ions. This adds a long time component to luminescence decay, which decreases
Ce:YAG scintillation performance [Ludziejewski-1997, Moszynski-1994, Weber-2004,
Zych-2000]. This is the major obstacle for Ce:YAG as a scintillator crystal, which is
further discussed later in Chapter 6.
Some Ce:YAG samples were used in beta-excitation measurements, shown in Fig.
3.4.6.
1400 Ce:YAG 0.1% Ce:YAG 0.15% 1200 Ce:YAG 0.2% annealed Ce:YAG 0.3% (UC) 1000 800
600 400 200
Luminescence(arb. units) Luminescence(arb. 0 -200 200 300 400 500 600 700 Wavelength (nm)
Figure 3.4.6 RL spectra of Ce:YAG samples excited using 60 Co beta particles using 65 second integration.
(UC) signifies that the Ce:YAG 0.3% sample was supplied by United Crystals. All other samples were
supplied by Marketech International.
Above 700 nm is not shown in Fig 3.4.6 because it is distorted by source luminescence. In the figure, the AD peak can barely be seen in the samples and is most noticeable in the Ce:YAG 0.3% sample, consistent with Fig. 3.4.4. It is known that the
Ce:YAG 0.15% sample used in this measurement is the same one that was later annealed
93 and labeled as the annealed sample in Figs. 3.4.4 and 3.4.5, so it is clear that there is a significant difference between the two Ce:YAG 0.15% samples. That appears to be the only major difference between the two figures. Using the annealed Ce:YAG 0.2% sample as a benchmark, annealing in air increases Ce:YAG 0.15% luminescence.
Ce:YAG conclusions
XSL of Ce:YAG shows very strong emission from Ce 3+ ions, previously
discussed in Section 3.2. The defect-related UV emission seen in undoped YAG is also
present in Ce:YAG, though at a reduced intensity. However, this UV emission overlaps
an absorption peak of Ce 3+ , and thus there is observable reabsorption, leading to reemission and an increased scintillation decay time, as will be further examined in
Chapter 6.
Effect of various treatments on undoped and Ce-doped YAG XSL
Two each of H2-grown YAG, O 2-grown YAG, and Ce:YAG 0.2% were sputtered
with Al for 30 seconds at the Cleanroom user facility at the Center for Materials Research
at Washington State University and then annealed in an atmosphere of Ar for 24 hours at
600°C. The H 2-grown and Ce:YAG 0.2% samples were first annealed in air to fill oxygen vacancies. This process was intended to follow the process detailed by Selim et al. [Selim-2007] and Solodovnikov et al. [Solodovnikov-2008] to inject aluminum into the samples after first filling oxygen vacancies. Figures 3.4.7-9 show the XSL emission of each sample. One of the H 2-grown YAG samples has a large chip along the edge and were thus the two H 2-grown YAG samples are distinguishable from each other, but the
O2-grown YAG and Ce:YAG 0.2% samples were both indistinguishable.
94
As grown 1 15000 As grown 2 Al sputtered and Ar annealed 1 Al sputtered and Ar annealed 2
10000
5000
0 Luminescence Intensity (counts) Intensity Luminescence 200 300 400 500 600 700 800 900 Wavelength (nm)
Figure 3.4.7 XSL spectra of H2-grown YAG with attempted Al insertion using 65 second integration.
As grown 15000 As grown Al sputtered and Ar annealed Al sputtered and Ar annealed
10000
5000
0 Luminescence Intensity (counts) Intensity Luminescence 200 300 400 500 600 700 800 900 Wavelength (nm)
Figure 3.4.8 XSL spectra of O2-grown YAG with attempted Al insertion using 65 second integration.
95
60000 1500 Air annealed Air annealed 50000 Air annealed Air annealed Al sputtered and Ar annealed Al sputtered and Ar annealed 40000 Al sputtered and Ar annealed 1000 Al sputtered and Ar annealed 30000
20000 500 10000
0 0 100 200 300 400 500 600 700 800 900 200 300 400 Luminescence Intensity(counts) (counts) IntensityLuminescence Wavelength (nm) Wavelength (nm)
Figure 3.4.9 XSL spectra of Ce:YAG 0.2% with attempted Al insertion using 20 second integration.
Figures 3.4.7-9 do not appear to show a significant change in luminescence after
Al sputtering and Ar annealing. Experiments in later chapters further explore these samples after Al sputtering and Ar annealing.
A Ce:YAG sample from Crytur was annealed in a vacuum for 24 hours at 800°C and measured. Results are shown as Fig. 3.4.10.
96
50000 As grown vacuum annealed 40000
30000
20000
10000
0 Luminescence Intensity (counts) Intensity Luminescence 100 200 300 400 500 600 700 800 900 Wavelength (nm)
Figure 3.4.10 XSL spectra of Ce:YAG 0.14% before and after vacuum anneal using 20 second integration.
The vacuum anneal decreased Ce 3+ emission. Vacuum anneal can remove anions
from the crystal lattice and as a result might reduce the cations, which would transition
Ce 4+ ions to Ce 3+ . However, Fig. 3.4.10 displays a decrease in emission, suggesting the removal of anions to be more impactful. It is important to note as well that the sample may not have been properly cleaned before annealing, so it is possible that annealing may have allowed new impurities to diffuse into the sample. Ce 3+ emission has been shown
above to be sensitive to defects, and it is possible that the decrease in emission is due to
the introduction of these new impurities.
Nd:YAG, Tm:YAG, and Yb:YAG
Figures. 3.4.11 and 3.4.12 show annealed Nd:YAG and Tm:YAG together and
multiple Yb:YAG samples, respectively.
97
50000 Nd:YAG 1% annealed Tm:YAG 0.8% annealed
40000
30000
20000
10000
0 Luminescence Intensity (counts) Intensity Luminescence 100 200 300 400 500 600 700 800 900 Wavelength (nm)
Figure 3.4.11 XSL spectra of annealed Nd:YAG 1% and Tm:YAG 0.8% using 65 second integration.
Yb:YAG 1% (M) Yb:YAG 3% (M) Yb:YAG 5% annealed (UC) Yb:YAG 10% annealed (UC)
500
0 Luminescence Intensity (counts) Intensity Luminescence 200 300 400 500 600 700 800 900 Wavelength (nm)
Figure 3.4.12 XSL spectra of Yb:YAG before and after anneal in air using 65 second integration. (M)
denotes that the sample was supplied by Marketech International, (UC) denotes that the sample was
supplied by United Crystals.
98 Figure 3.4.11 shows all emission peaks for Nd:YAG and Tm:YAG over the displayed range 200-850 nm. The transitions responsible for each individual peak were discussed previously in Sections 3.1 and 3.2. These transitions have been observed previously at low temperatures but have not been observed before at room temperature
[Reda-2012].
The XSL spectra of Fig. 3.4.12 are almost entirely O2-→Yb 3+ charge-transfer luminescence, aside from the Fe 3+ emission clearly visible in the Yb:YAG 1% sample.
Though this emission is typically observed at low temperatures [Guerassimova-2001,
Kamesnkikh-2005, Pieterson-2000], it has been seen weakly at room temperature
[Fredrich-Thornton-2010]. Emission appears to decrease quickly with concentration, but it appears that annealing does increase charge-transfer luminescence. This concentration quenching has been noticed before [Pieterson-2000, Yang-2002]. It is possible that the small size of Yb 3+ ions of 86 pm, as opposed to 106 pm for Y 3+ [Emsley-1989], allows for easier insertion of oxygen ions during annealing, so higher concentrations of Yb-doping allows for the addition of more oxygen to fill V Os and thus annealing higher concentration Yb-doped YAG produces more intense charge transfer luminescence than annealed Yb-doped YAG of lower concentration.
Beta excitation by 60 Co of the Nd:YAG sample clearly shows each of the
transition states in the visible spectrum, as demonstrated by Fig. 3.4.13.
99 1000 Nd:YAG 1% 800
600
400
200
0
Luminescence(arb. units) Luminescence(arb. -200
200 300 400 500 600 700 wavelength(nm)
Figure 3.4.13 RL spectra of Nd:YAG 1% samples excited using 60 Co beta particles using 65 second
integration.
Comparison of Fig. 3.4.13 to Fig. 3.4.11 shows similarity in the range up to 700 nm, above which is omitted due to misleading source emission.
A very interesting feature in Figs. 3.4.11-13 is the lack of AD luminescence in
Nd:YAG, Tm:YAG, and Yb:YAG. This has to do with charge transfer between oxygen and the cations. AD luminescence is based on a charge transfer-like mechanism for excitation. However, O 2-→Nd 3+ , O 2-→Tm 3+ , and O 2-→Yb 3+ are much stronger charge
transfer mechanisms and thus these dopants dominate the charge transfer physics of the
lattice. As a result, the AD emission is not observed in these samples.
Effect of color center activation on undoped YAG XSL
A color center measurement was performed on O 2-grown YAG using XSL. A sample was measured, colored by UV excitation for 30 minutes without moving the sample or lens, and then measured again, shown in Fig. 3.4.14.
100
30000 As-grown UV excited 30 minutes 25000
20000
15000
10000
5000
0 Luminescence Intensity (counts) Intensity Luminescence 100 200 300 400 500 600 700 800 900 Wavelength (nm)
Figure 3.4.14 XSL spectra of O2-grown YAG before and after 30 minute UV excitation using 65 second
integration.
The decrease in the AD peak likely arises from hole capture of oxygen ions.
Combining Fig. 3.4.14 with results from Section 3.3, it appears that UV excitation released electrons from oxygen ions that trap, albeit rarely, at oxygen vacancies and that
Fe 3+ concentration actually increases. It was argued by Rotman et al. [Rotman-1989] and
Hewitt et al. [Hewitt-2011] that excited states in Fe 3+ can be populated by F +-, F-, or F --
center →Fe 3+ excitation transfer, which would support this measurement where the more
intense F-center emission corresponds with the larger iron emission around 700 and 800
nm.
101 3.5 REFERENCES
[Ashurov-1977] M.Kh. Ashurov, Yu.K. Voronko, V.V. Osiko, A.A. Sobol, and M.I.
Timoshechkin, “Spectroscopic study of stoichiometry deviation in crystals with garnet structure,” Physica Status Solidi (A) 42 (1977) 101-110
[Ashurov-2001] M.Kh. Ashurov, A.F. Rakov, and R.A. Erzin, “”Luminescence of defect centers in yttrium-aluminum garnet crystals,” Solid State Communications 120 (1002)
491-494
[Babin-2005] V.Babin, K. Blazek, A. Krasnikov, K. Nejezchleb, M. Nikl, T. Savikhina,
and S. Zazubovich, “Luminescence of undoped LuAG and YAG crystals,” Physica Status
Solidi (C) 2 (2005) 97-100
[Babin-2011] V. Babin, V.V. Laguta, A. Maaroos, A. Makhov, M. Nikl, and S.
+ Zazubovich, “Luminescence of F -type centers in undoped Lu 3Al 5O12 single crystals,”
PPhysica Status Solidi B 248 (2011) 239-242
[Baciero-1999] A. Baciero, L. Placentino, K.J. McCarthy, L.R. Barquero, A. Ibarra, and
B. Zurro, “Vacuum ultraviolet and x-ray luminescence efficiencies of Y 3Al 5O12 :Ce phosphor screens,” Journal of Applied Physics 85 (1999) 6790-6796
[Bagdasarov-1977] Kh.S. Bagdasarov, L.B. Pasternak, and B.K. Sevast’yanov,
3+ “Radiation color centers in Y 3Al 5O12 :Cr crystals,” Soviet Journal of Quantum
Elecronics 7 (1977) 965-968
[Bass-1967] M. Bass and A.E. Paladino, “Color centers in Yttrium Gallium Garnet and
Yttrium Aluminum Garnet,” Journal of Applied Physics 38 (1967) 2706-2707
102 [Batygov-1972] S.Kh. Batygov, Yu.K. Voron’ko, B.I. Denker, A.A. Maier, V.V. Osiko,
V.S. Radyukhin, and M.I. Timoshechkin, “Color centers in Y 3Al 5O12 crystals,” Soviet
Physics – Solid State 14 (1972) 839-841
[Batygov-1991] S.Kh. Batygov, Yu.K. Voron’ko, N.G. Margiani, and N.N. Ryskin,
“Luminescence of Nd 3+ ions in garnets under high-energy excitation,” OSA Proceedings on Advanced Solid-State Lasers, G. Dubé and L. Bhase (eds.), 10 (1991) 368-370
[Bernhardt-1978] HJ. Bernhard, “Bound polarons in YAG crystals,” Physica Status
Solidi (B) 87 , (1978) 213-219
[Blasse-1967(a)] G. Blasse and A. Bril, “A new phosphor for flying-spot cathode-ray
3+ tubes for color television: yellow-emitting Y 3Al 5O12 -Ce ,” Applied Physics Letters 11
(1967) 53-55
[Blasse-1967(b)] G. Blasse and A. Bril, “Investigation of some Ce 3+ -activated
phosphors,” Journal of Chemical Physics 47 (1967) 5139-5145
[Blazek-2004] K. Blazek, A. Krasnikov, K. Nejezchleb, M. Nikl, T. Savikhina, and S.
3+ Zazubovich, “Luminescence and defects creation in Ce -doped Lu 3Al 5O12 crystals,”
Physica Status Solidi (B) 241 (2004) 1134-1140
[Bauman-1962] R.P. Bauman, Absorption Spectroscopy , John Wiley & Sons, Inc., New
York, NY, (1962)
[Bunch-1977] J.M. Bunch, “Mullwo-Ivey relation between peak color-center absorption energy and average oxygen ion spacing in several oxides of group-II and –III metals,”
Physical Review B 16 (1977) 724-725
103 [Caslavsky-1980] J.L. Caslavsky and D.J. Viechnicki, “Melting behaviour and metastability of yttrium aluminium garnet (YAG) and YAlO 3 determined by optical
differential thermal analysis,” Journal of Materials Science 15 (1980) 1709-1718
[Chakrabarti-1988] K. Chakrabarti, “Photobleaching and photoluminescence in neutron-
irradiated YAG,” Journal of Physics and Chemistry in Solids 49 (1988) 1009-1011
[Chen-1988] C.Y. Chen, G.J. Pogatchnik, and Y. Chen, “Optical and electron
paramagnetic resonance studies of Fe impurities in yttrium aluminum garnet crystals,”
Physical Review B 38 (1988) 8555-8561
[Choe-2001] J.Y. Choe, D. Ravichandran, S.M. Bloomquist, D.C. Morton, K.W.
Kirchner, M.H. Ervin, and U. Lee, “Alkoxy sol-gel derived Y 3-xAl 5O12 :Tb x thin films as efficient cathodoluminescent phosphors,” Applied Physics Letters 78 (2001) 2300-3802
[Chugunova-2011] M.M. Chugunova, I.A. Kamenskikh, V.V. Mikhailin, and S.A.
Usenko, “Charge Transfer Luminescence Kinetics in a YAG:Yb Monocrystal and
Ceramics,” Moskow University Physics Bulletin 66 (2011) 249-253
[Deutsch-1945] M. Deutsch, L.G. Elliot, and A. Roberts, “Disintegration schemes of
radioactive substances VIII. Co 60 ,” Physical Review 68 (1945) 193-197
[Dong-1991] J. Dong and K. Lu, “Noncubic symmetry in garnet structures studied using extended x-ray-absorption fine-structure spectra,” Physical Review B 43 (1991) 8808-
8821
[Dong-2005] D. Dong, K. Xiao-Xu, G. Jian-Jun, W. Hui, and Z. Kang-Wei, “Optical absorption and EPR study of the octahedral Fe 3+ center in yttrium aluminum garnet,”
Physical Review B 72 (2005) 073101
104 [Dong-2006] Y. Dong, G. Zhou, J. Xu, G. Zhao, F. Su, L. Su, G. Zhang, D. Zhang, H. Li, and J. Si, “Luminescence studies of Ce:YAG using vacuum ultraviolet synchrotron radiation,” Materials Research Bulletin 41 (2006) 1959-1963
[Emsley-1989] J. Emsley, The Elements , Oxford University Press, New York, NY (1989)
[Fagundes-Peters-2007] D. Fagundes-Peters, N. Martynyuk, K. Lünstedt, V. Peters, K.
Petermann, G. Huber, S. Basun, B. Laguta, and A. Hofstaetter, “High quantum efficiency
YbAG-crystals,” Journal of Luminescence 125 (2007) 238-247
[Feldman-2003] R. Feldman, Y. Shimony, and Z. Burshtein, “Dynamics of chromium ion valence transformations in Cr,Ca:YAG crystals used as laser gain and passive Q- switching media,” Optical Materials 24 (2003) 333-344
[Fredrich-Thornton-2010] S.T. Fredrich-Thornton, “Nonlinear losses in single crystalline and ceramic Yb:YAG thin-disk lasers,” thesis, University of Hamburg (2010)
[Geller-1967] S. Geller, “Crystal chemistry of the garnets,” Zeitschrift für
Kristallographie 125 (1967) 1-47
[Gibbons-1973] E.F. Gibbons, T.Y. Tien, R.G. DeLash, P.J. Zacmanidis, and H.L.
3+ Stadler, “Some factors influencing the luminous decay characteristics of Y 3Al 5O12 :Ce ,”
Journal of the Electrochemical. Society 120 (1973) 835-837
[Gluchowski-2009] P.Gluchowski, R. P ązik, D. Hreniak, and W. Str ęk, “Luminescence
3+ properties of Cr :Y 3Al 5O12 nanocrystals.” Journal of Luminescence 129 (2009) 548-553
[Grigorjeva-2010] L. Grigorjeva, A. Krasnikov, V.V. Laguta, M. Nikl, and S.
Zazubovich, “Luminescence and creation of electron centers in UV-irradiated single crystals,” Journal of Applied Physics 108 (2010) 053509
105 [Guerassimova-2001] N. Guerassimova, N. Garnier, C. Dujardin, A.G. Petrosyan, and C.
Pédrini, “X-ray excited charge transfer luminescence in YAG:Yb and YbAG,” Journal of
Luminescence 94-95 (2001) 11-14
[Guerassimova-2002] N. Guerassimova, C. Dujardin, N. Garnier, C. Pédrini, A.G.
Petrosyan, I.A. Kamenskikh, V.V. Mikhailin, I.N. Shipinkov, D.A. Spassky, K.L.
Ovanesyan, G.O. Shirinyan, R. Chipaux, M. Cribier, J. Mallet, and J.-P. Meyer, “Charge- transfer luminescence and spectroscopic properties of Yb 3+ in aluminum and gallium
garnets,” Nuclear Instruments and Methods in Physics Research A 486 (2002) 278-282
[Hakuta-2003] Y. Hakuta, T-Haganuma, K. Sue, T. Adschiri, and K. Arai, “Continuous production of phosphor YAG:Tb nanoparticles by hydrothermal synthesis in supercritical water,” Materials Research Bulletin 38 (2003) 1257-1265
[Hamilton-1989] D.S. Hamilton, S.K. Gayen, G.J. Pogatshnik, and R.D. Ghen, “Optical-
3+ absorption and photoionization measurements from the excited states of Ce :Y 3Al 5O12 ,”
Physical Review B 39 8807-8815
[Hayes-1980] W. Hayes, M. Yamaga, D.J. Robbins, and B. Cockayne, “Optical detection of EPR of recombination centers in YAG,” Journal of Physics C: Solid State Physics 13
(1980) L1085-L1089
[Hewitt-2011] J.D. Hewitt, T.M. Spinka, A.A. Senin, and J.G. Eden, “Internal
3+ 2 4 photopumping of Nd ( H9/2 , F5/2 ) states in yttrium aluminum garnet by excitation
transfer from oxygen deficiency centers and Fe 3+ continuum emission,” Applied Physics
Letters 99 (2011) 031101
[Hirata-1996] G.A. Hirata, O.A. Lopez, L.E. Shea, J.Y. Yi, T. Cheeks, J. McKittrick, J.
Siqueiros, M. Avalos-Borja, A. Esparza, and C. Falcony, “Pulsed laser deposition of
106 Y3Al 5O12 :Tb photoluminescent thin films,” Journal of Vacuum Science and Technology
A 14 (1996) 1694-4696
[Jung-2007] K.Y. Jung and H.W. Lee, “Enhanced luminescent properties of
3+ 3+ Y3Al 5O12 :Tb ,Ce phosphor prepared by spray pyrolysis,” Journal of Luminescence 126
(2007) 469-474
[Kaczmarek-1999] S.M. Kaczmarek, “Role of the type of impurity in radiation influence
on oxide compounds,” Crystal Research and Technology 34 (1999) 737-743
[Kamenskikh-2005] I. Kamenskikh, C. Dujardin, N. Garnier, N. Guerassimova, G.
Ledoux, V. Mikhaillin, C. Pedrini, A. Petrosyan, and A. Vasil’ev, “Temperature dependence of charge transfer and f-f luminescence of Yb 3+ in garnets and YAP,” Journal
of Physics: Condensed Matter 17 (2005) 5587-5594
[Kang-1999] Y.C. Kang, I.W. Lenggoro, S.B. Park, and K. Okuyama,
“Photoluminescence characteristics of YAG:Tb phosphor particles with spherical
morphology and non-aggregation,” Journal of Physics and Chemistry of Solids 60 (1999)
1855-1858
[Kirm-2000] M. Kirm, A. Lushchik, Ch. Lushchik, and G. Zimmerer, “Investigation of
3+ luminescence properties of pure and Ce doped Y 3Al 5O12 crystals using VUV radiation,”
ECS Proceedings 99-40 (2000) 113-122
[Korzhik-1989] M.V. Korzhik, M.G. Livshits, Kh.S. Bagdasarov, A.M. Kevorkov, T.A.
Melkonyan, and M.L. Meil’man, “Efficient pumping of Nd 0 ions via charge transfer bands of Fe 3+ ions in YAG,” Sov. J. Quantum Electron. 19 (1989) 344-346
107 [Kovaleva-1981] N.S. Kovaleva, A.O. Ivanov, and É.P. Dubrovina, “Relationship between formation of radiation color centers and growth defects in YAG:Nd crystals,”
Soviet Journal of Quantum Electronics 11 (1981) 1485-1488
[Kubo-1992] T.S. Kubo and T.J. Kane, “Diode-pumped lasers at five eye-safe wavelengths,” Journal of Quantum Electronics 28 (1992) 1033-1040
[Kuklja-2000] M.M. Kuklja, “Defects in yttrium aluminum perovskite and garnet crystals: atomistic study,” Journal of Physics: Condensed Matter 12 (2000) 2953-2967
[Lacovara-1991] P. Lacovara, H.K. Choi, C.A. Wang, R.L. Aggarwal, and T.Y. Fan,
“Room-temperature diode-pumped Yb:YAG laser,” Optics Letters 16 (1991) 1089-1091
[Landron-1996] C. Landron, S. Lefloch, M. Gervais, J.P. Coutures, and D. Bazin,
“Yttrium coordination shell in YAG related structures,” Physica Status Solidi (B) 196
(1996) 25-31
[Li-2000] B. Li, Z. Gu, J. Lin, M. Su, “X-ray luminescence properties of rare-earth doped orthotantalate,” Materials Research Bulletin 35 (2000) 1921-1931
[Liu-2009] B. Liu, M. Gu, Z. Liu, S. Huang, and C. Ni, “Formation energies of antisite defects in Y 3Al 5O12 : a first-principles study,” Applied Physics Letters 94 (2009) 121910
[Lo-1996] J. Lo, T. Tseng, J. Tyan, and J.C.M. Huang, “Luminescent properties of sol-
th gel prepared by Y 3Al 5O12 :Tb thin films,” 9 International Vacuum Microelectronics
Conference, St. Petersburg (1996) 197-201
[Ludziejewski-1997] T. Ludziejewski, M. Moszy ński, M. Kapusta, D. Wolski, W.
Klamra, and K Moszy ńska, “Investigation of some scintillation properties of YAG:Ce crystals,” Nuclear Instrumentation and Methods in Physics Research A 398 (1997) 287-
294
108 [Malinowski-1995] M. Malinowski, R. Wolski, Z. Frukacz, T. Lukasiewicz, and Z.
Luczy ński, “Spectroscopic studies of YAG:Sm 3+ crystals,” Journal of Applied
Spectroscopy 62 (1995) 840-843
[Masumoto-1985] T. Masumoto and Y. Kuwano, “Effects of oxygen pressure on optical absorption of YAG,” Jap. Journal of Applied Physics 24 (1985) 546-551
[Meilman-1984] M.L. Meil’man, M.V. Korzhik, V.V. Kuz’min, M.G. Livshit, Kh.S.
Bagdasaov, and A.M. Kevorkov, “Luminescence and energy-level structure of impurity
3+ centers in Y 3Al 5O12 :Fe single crystals,” Soviet Physics – Doklady 29 (1984) 61-63
[Meng-2008] J.X. Meng, J.Q. Li, Z.P. Shi, and K.W. Cheah, “Efficient energy transfer for Ce to Nd in Nd/Ce codoped yttrium aluminum garnet,” Applied Physics Letters 93
(2008) 221908
[Milanese-2004] C. Milanese, V. Buscaglia, F. Maglia, and U. Anselmi-Tamburini,
“Disorder and nonstoichiometry in synthetic garnets A 3B5O12 (A=Y, Lu-La, B=Al, Fe,
Ga). A simulation study,” Chemistry of Materials 16 (2004) 1232-1239
[Milliken-2012] E.D. Milliken, L.C. Oliveira, G. Denis, and E.G. Yukihara, “Testing a
model-guided approach to the development of new thermoluminescent materials using
YAG:Ln produced by solution combustion synthesis,” Journal of Luminescence 132
(2012) 2495-2504
[Miniscalco-1978] W.J. Miniscalco, J.M. Pellegrino, and W.M. Yen, “Measurements of
excited-state absorption in Ce 3+ :YAG,”Journal of Applied Physics 49 (1978) 6109-6111
[Mori-1977] K. Mori, “Transient colour centres caused by UV light irradiation in yttrium aluminium garnet crystals,” Physica Status Solidi A 42 (1977) 375-384
109 [Moszynski-1994] M. Moszy ński, T. Ludziejewski, D. Wolski, W. Klamra, and L.O.
Norlin, “Properties of the YAG:Ce scintillator,” Nuclear Instrumentation and Methods in
Physics Research A 345 (1994) 461-467
[Murk-1995] V. Mürk and N Yaroshevich, “Exciton and recombination processes in
YAG crystals,” Journal of Physics: Condensed Matter 7 (1995) 5857-5864
[Murk-1997] V. Mürk, “Exciton relaxation in wide-gap complex oxides,” Mat. Sci.
Forum 239-241 (1997) 537-542
[Nikl-2005] M. Nikl, “Energy transfer phenomena in the luminescence of wide band-gap
scintillators,” Physica Status Solidi (A) 202 (2005) 201-206
[Niklas-1983] A. Niklas and W. Jele ński, “X-ray luminescence of YAG:Nd 3+ ,” Physica
Status Solidi (A) 77 (1983) 393-398
[Pankove-1971] J.I. Pankove, Optical Processes in Semiconductors , Dover Publications,
Inc., New York, NY (1971)
[Pedrini-1986] C. Pedrini, F. Rogemond, and D.S. McClure, “Photoionization thresholds
2+ 3+ 2+ of rare-earth impurity ions. Eu :CaF 2, Ce :YAG, and Sm :CaF 2,” Journal of Applied
Physics 59 (1986) 1196-1201
[Pedrini-2005] C. Pédrini, “Elecronic processes in rare earth activated wide gap materials,” Physica Status Solidi (A) 202 (2005) 185-194
[Pieterson-2000] L. van Pieterson, M. Heeroma, E. de Heer, and A. Meijerink, “Charge transfer luminescence of Yb 3+ ,” Journal of Luminescence 91 (2000) 177-193
[Potdevin-2005] S. Potdevin, G. Chadeyron, D. Boyer, B. Caillier, and R. Mahiou, “Sol- gel based YAG:Tb 3+ or Eu 3+ phosphors for application in lighting sources,” Journal of
Physics D: Applied Physics 38 (2005) 3251-3260
110 [Pujats-2001] A. Pujats and M. Springis, “The F-type centers in YAG crystals,”
Radiation Effects and Defects in Solids 155 (2001) 65-69
[Rajan-2005] K.G. Rajan and A.J. Lenus, “X-ray excited optical luminescence studies on the system BaXY (X, Y=F, Cl, Br, I),” Pramana 65 (2005) 323-338
[Reda-2012] S.M. Reda, C.R. Varney, and F.A. Selim, “Radio-luminescence and absence of trapping defects in Nd-doped YAG single crystals,” Results in Physics 2 (2012) 123-
126
[Ricci-2009] P.C. Ricci, A. Casu, and A. Anedda, “Effect of the electron-phonon interaction on the luminescence properties of iron ions in YAG crystal,” J of Physical
Chemistry A 113 (2009) 13901-13906
[Rimal-1966] L. Rimal and T. Kushida, “Paramagnetic resonance of Fe 3+ in yttrium
aluminum, lutetium aluminum, and lutetium gallium garnets,” Physical Review 143
(1966) 160-164
[Robbins-1979(a)] D.J. Robbins, B. Cockayne, B. Lent, C.N. Duckworth, and J.L.
Glasper, “Investigation of competitive recombination processes in rare-earch activated
garnet phosphors,” Physical Review B 19 (1979) 1254-1269
[Robbins-1979(b)] D.J. Robbins, B. Cockayne, J.L. Glasper, and B. Lent, “The
temperature dependence of rare-earth activated garnet phosphors I. Intensity and lifetime
measurements on undoped and Ce-doped Y 3Al 5O12 ,” Journal of the Electrochemical
Society 126 (1979) 1213-1220
[Robbins-1979(c)] D.J. Robbins, B. Cockayne, J.L. Glasper, and B. Lent, “The
temperature dependence of rare-earth activated garnet phosphors II. A comparative study
111 3+ 3+ 3+ 3+ of Ce , Eu , Tb , and Gd in Y 3Al 5O12 ,” Journal of the Electrochemical Society 126
(1979) 1221-1228
[Robbins-1979(d)] D.J. Robbins, B. Cockayne, B. Lent, and J.L. Glasper, “The
relationship between concentration and efficiency in rare earth activated phosphors,”
Journal of the Electrochemical Society 126 (1979) 1556-1563
[Rotman-1985] S.R. Rotman and C. Warde, “Defect luminescence in cerium-doped
yttrium aluminum garnet,” Journal of Applied Physics 58 (1985) 522-525
[Rotman-1989] S.R. Rotman, C. Warde, H.L. Tuller, and J. Haggerty, “Defect-poperty correlations in garnet crystals. V. Energy transfer in luminescent yttrium aluminum- yttrium iron garnet solid solutions,” Journal of Applied Physics 66 (1989) 3207-3210
[Rotman-1990] S.R. Rotman, “Comment on ‘Optical and electron paramagnetic
resonance studies of Fe impurities in yttrium aluminum garnet crystals,’” Physical
Review B 41 (1990) 791-792
[Rotman-1992] S.R. Rotman, H.L. Tuller, and C. Warde, “Defect-poperty correlations in
garnet crystals. VI. The electrical conductivity, defect structure, and optical properties of
luminescent calcium and cerium-doped yttrium aluminum garnet,” Journal of Applied
Physics 71 (1992) 1209-1214
[Rozenfeld-1993] Yu.B. Rozenfeld and S.R. Rotman, “The luminescence of defects in
yttrium aluminum garnet,” Physica Status Solidi (A) 139 (1993) 249-262
[Selim-2007] F.A. Selim, D. Solodovnikov, M.H. Weber, and K.G. Lynn, “Identification
of defects in Y 3Al 5O12 crystals by positron annihilation spectroscopy,” Applied Physics
Letters 91 (2007) 104105
112 [Smol’skaya-1987] L.P. Smol’skaya, E.F. Martynovich, A.G. Davydchenko, and S.A.
Smirnova, “X-ray and thermally stimulated luminescence in YAG,” Journal of Applied
Spectroscopy 46 (1987) 44-46
[Solodovnikov-2008] D. Solodovnikov, M.H. Weber, and K.G. Lynn, “Improvement in scintillation performance of Ce, Er codoped yttrium aluminum garnet crystals by means of a postgrowth treatment,” Applied Physics Letters 93 (2008) 104102
[Song-2004] P. Song, Z. Zhao, X. Xu, B. Jang, P. Deng, and J. Xu, “Growth and
properties of Tm:YAG crystals,” Journal of Crystal Growth 270 (2004) 433-437
[Springis-1991] M. Springis, A. Pujats, and J. Valbis, “Polarization of luminescence of colour centres in YAG crystals,” Journal of Physics: Condensed Matter 3 (1991) 5457-
5461
[Stanek-2008] C.R. Stanek, M.R. Levy, K.J. McClellan, B.P. Uberuaga, and R.W.
Grimes, “Defect identification and compensation in rare earth oxide scintillators,”
Nuclear Instrumentation and Methods in Physics Research B 266 (2008) 2657-2664
[Suzuki-1997] Y. Suzuki, T. Sakuma, and M. Hirai, “UV emission from the second
lowest 5d state in Ce 3+ :YAG,” Material Science Forum 239-241 (1997) 219-222
[Tiseanu-1995] C. Tiseanu, A Lupei, and V. Lupei, :Energy levels of Tm 3+ in yttrium aluminum garnet,” Journal of Physics: Condensed Matter 7 (1995) 8477-8486
[Tomiki-1991] T. Tomiki, H. Akamine, M. Gushiken, Y. Kihjoh, M. Miyazato, T.
Miyazato, N. Toyokawa, M. Hiraoka, N. Hirata, Y. Hanaha, and T. Futemma, “Ce 3+
centers in Y 3Al 5O12 (YAG) single crystals,” J. Physics Soc. Japan 60 (1991) 2437-2445
113 [Varney-2011] C.R. Varney, S.M. Reda, D.T. Mackay, M.C. Rowe, and F.A. Selim,
“Strong visible and near infrared luminescence in undoped YAG single crystals,” AIP
Advances 1 (2011) 042170
[Varney-2012(a)] C.R. Varney, D.T. Mackay, S.M. Reda, and F.A. Selim, “On the optical properties of undoped and rare-earth-doped yttrium aluminum garnet single crystals,” Journal of Physics D: Applied Physics 45 (2012) 015103
[Varney-2012(c)] C.R. Varney, M.A. Khamehchi, J. Ji, and F.A. Selim, “X-ray luminescence based spectrometer for investigation of scintillation properties,” Review of
Scientific Instruments 83 (2012) 103112
[Vedda-2004] A. Vedda, D. Di Martino, M. Martini, J. Mares, E. Mihokova, M. Nikl, N.
Solovieva, K. Blazek, and K. Nejezchleb, “Trap levels in Y-aluminum garnet scintillating crystals,” Radiation Measurements 38 (2004) 673-676
[Wang-2002] H.M. Wang, M.C. Simmonds, and J.M. Rodenburg, “Manufacturing of
YbAG coatings and crystallisation of the pure and Li 2O-doped Yb 2O3-Al 2O3 system by a modified sol-gel method,” Materials Chemistry and Physics 77 (2002) 802-807
[Weber-2004] M.J. Weber, “Scintillation: mechanisms and new crystals,” Nuclear
Instrumentation and Methods in Physics Research A 527 (2004) 9-14
[Weg-1985] W.F. van der Weg, Th.J.A. Popma, and A.T. Vink, “Concentration
3+ dependence of UV and electron-excited Tb luminescence in Y 3Al 5O12 ,” Journal of
Applied Physics 57 (1985) 5450-5456
[Wisniewska-2001] M. Wi śniewska, A.J. Wojtowicz, T. Łukasiewicz, Z. Frukacz, Z.
Gał ązka, and M. Malinowski, “Radio- and VUV-excited luminescence of YAG:Ce,
YAP:Pr, and YAG:Pr,” Proceedings of SPIE 4412 (2001) 351-356
114 [Wong-1984] C.M. Wong, S.R. Rotman, and C. Warde, “Optical studies of cerium doped yttrium aluminum garnet single crystals,” Applied Physics Letters 44 (1984) 1038-1040
[Xu-1999] Y.N. Xu and W.Y. Ching, “Electronic structure of yttrium aluminum garnet
(Y 3Al 5O12 ),” Physical Review B 59 (1999) 10530-10535
[Xu-2003] X. Xu, Z. Zhao, G. Zhao, P.X. Song, J. Xu, and P. Deng, “Comparison of
Yb:YAG crystals grown by CZ and TGT method,” Journal of Crystal Growth 257 (2003)
297-300
[Yacobi-1986] B.G. Yacobi and D.B. Holt, “Cathodoluminescence scanning electron microscopy of semiconductors,” Journal of Applied Physics 59 (1986) R1-R24
[Yang-2002] P. Yang, P. Deng, and Z. Yin, “Concentration quenching in Yb:YAG,”
Journal of Luminescence 97 (2002) 51-54
[Zhang-2003] J. Zhang, J. Ning, X. Liu, Y. Pan, and L. Huang, “A novel synthesis of phase-pure ultrafine YAG:Tb phosphor with different Tb concentration,” Materials
Letters 57 (2003) 3077-3081
[Zorenko-2004(a)] Yu.V. Zorenko, I.V. Konstankevych, V.V. Mikha ĭlin, V.N.
Kolobanov, and D.A. Spasski ĭ, “Luminescence of excitons in single-crystal garnets,”
Optics and Spectroscopy 96 (2004) 390-397
[Zorenko-2004(b)] Y. Zorenko, A. Voloshinovskii, I. Konstankevych, V. Kolobanov, V.
Mikhailin, and D. Spassky, “Luminescence of excitons and antisite defects in the
phosphors based on garnet compounds,” Radiation Measurements 38 (2004) 677-680
[Zorenko-2007] Yu. Zorenko, A. Voloshinovskii, V. Savchyn, T. Voznyak, M Nikl, K
Nejezchleb, V. Mikhailin, V. Kolobanov, and D. Spassky, “Exciton and antisite defect-
115 related luminescence in Lu 3Al 5O12 and Y 3Al 5O12 garnets,” Physica Status Solidi (B) 244
(2007) 2180-2189
[Zorenko-2010(a)] Y. Zorenko, T. Zorenko, T. Voznyak, A. Mandowski, Q. Xia, M.
+ Batentschuk, and J. Friedrich, “Luminescence of F and F centers in Al 2O3-Y2O3 oxide compounds,” IOP Conference Series: Material Science Engineering 15 (2010) 012060
[Zorenko-2010(b)] Yu. Zorenko, J.A. Mares, P. Prusa, M. Nikl, V. Gorbenko, V.
Savchyn, R. Kucerkova, and K. Nejezchleb, “Luminescence and scintillation
characteristics of YAG:Ce single crystalline films and single crystals,” Radiation
Measurements 45 (2010) 389-391
[Zorenko-2011] Yu. Zorenko, T. Voznyak, V. Gorbenko, E. Zych, S. Nizankovski, A.
Dan’ko, and V. Puzikov, “Luminescence properties of Y 3Al 5O12 :Ce nanoceramics,”
Journal of Luminescence 131 (2011) 17-21
[Zorenko-2012] Y. Zorenko, T. Zorenko, V.V. Gorbenko, T. Voznyak, V. Savchyn, P.
Bilski, and A. Twardak, “Peularities of luminescent and scintillation properties of
YAG:Ce phosphor prepared in different crystalline forms,” Optical Materials 34 (2012)
1314-1319
[Zych-2000] E. Zych, C. Brecher, and J. Glodo, “Kinetics of cerium emission in a
YAG:Ce single crystal: the role of traps,” Journal of Physics: Condensed Matter 12
(2000) 1947-1958
116 CHAPTER FOUR
DEFECTS CHARACTERIZATION BY TRAP LEVEL MEASUREMENTS
Introduction
In Chapter 3, defects were identified via optical measurements. In Section 3.3, color center measurements showed the presence of charge carrier traps that were stable at room temperature but may be bleached by heating the sample above a certain temperature. This is an example of heat supplying a trapped charge carrier with enough thermal energy to escape the potential barrier of a trap. After escape, it can recombine with an oppositely-charged carrier at a luminescence center. This recombination is usually accompanied by emission of a photon characteristic of the recombination center.
Measuring this thermally-stimulated luminescence characterizing the potential well depth of traps is referred to as thermoluminescence (TL). Charge carriers released from their bound state orbitals by excitation can become trapped at defects, which they can escape when supplied with sufficient thermal energy. TL is the measurement of the emission of the crystal while increasing the temperature to release trapped charge carriers. TL data is often studied as a plot of integrated TL intensity versus temperature, which is referred to as a “glow curve.”
Figure 4.0.1 presents a schematic diagram of the process of thermal stimulation and recombination of trapped charge carriers. The traps can be vacancy, interstitial, substitutional, or antisite defects. Typically, charge carriers trap at defects of the opposite charge, such as vacancies. However, the behavior of some impurity traps depends on the electronic structure of the impurity. For instance, Eu 3+ and Yb 3+ impurities have been
117 found to act as stable electron traps at room temperature due to their large electron affinity. Meanwhile, Pr 3+ , Tb 3+ , and Ce 3+ were found to act as hole traps, also stable at
room temperature, due to how readily these ions transition to the tetravalent state
[Batygov-1972, Zorenko-2012]. At low temperatures, other traps stabilize, such as Sm 3+ electron traps.
Figure 4.0.1 Schematic diagram of the TL process after thermal activation of trapped charge carriers.
TL emission intrinsically occurs at one or few luminescence centers. Thus, TL measurements provide a reliable method by which to record small luminescence intensities. This makes TL ideal for applications to radiation dosimetry and dating of archaeological or paleontological artifacts. TL is also widely used to study trap levels in crystals such as YAG [Bagdasarov-1977, Batygov-1972, Guerassimova-2002, Mackay-
2011, Mackay-2012, Milliken-2012, Mori-1977, Murk-1995, Robbins-1979(b), Selim-
2007, Smol’skaya-1987, Solodovnikov-2008, Varney-2012(b), Vedda-2004, Zorenko-
2012, Zych-2000].
Analysis of glow curves can provide quantitative information about trap levels.
The calculated depth of a trap is referred to as its activation energy. There are several different ways to calculate activation energy from glow curves based on TL kinetics. The
118 initial rise method calculates the activation energy directly by fitting the beginning of TL peak. The initial rise method is applicable to any order kinetics but can only be applied for non-overlapping glow curves, where the peak being studied initially rises from the baseline [Braunlich-1967, Chen-1981, Christodoulides-1985, Garlick-1948, Mahesh-
1989, Varney-2012(b)]. Other methods of activation energy calculation rely on conducting the experiment multiple times with different heating rates. At a faster heating rate, a glow curve peak broadens, decreases in intensity, and shifts to higher temperature
[Bos-2001, Chen-1981, Mahesh-1989]. This effect is shown in Fig. 4.0.2 for glow curves of Ce:YAG 0.2% at heating rates of 5, 10, 20, 30, 40, 50, and 60°C/minute. The method of two heating rates uses two separate constant heating rates to cancel unknown terms in kinetics calculations, thus rendering activation energy calculation trivial [Bos-2001,
Chen-1981, Kitis-1998(a), Mackay-2011, Mahesh-1989, Varney-2012(b)]. This method is limited by peak fitting capabilities and, since peak temperature is usually not very sensitive to the heating rate, can give large uncertainty. This method is typically applied to first order kinetics, but it is possible to apply this method to any order kinetics.
Following the same principles as the method of two heating rates is the variable heating rate method, where several glow curves using different heating rates are constructed
[Mahesh-1989, Varney-2012(b)]. As with the method of two heating rates, unknown terms cancel in kinetics equations, and in this case the activation energy can be determined graphically. The benefit of this method over the method of two heating rates is that the greater number of data points decreases uncertainty. This method is typically applied to first order kinetics, but it can be generalized to any order kinetics. Other
119 methods, such as the half-width method or direct computational glow curve fitting with the activation energy as a fitting parameter, are not explored in this thesis.
5500000 5 C/min 5000000 10 C/min 4500000 20 C/min 30 C/min 4000000 40 C/min 3500000 50 C/min 3000000 60 C/min
2500000 2000000 1500000 1000000 500000 0 Integrated TL Intensity (counts) Intensity TL Integrated -500000 0 50 100 150 200 250 300 350 400 450 Temperature (C)
Figure 4.0.2 Glow curves of Ce:YAG 0.2% at multiple heating rates, showing the effects of heating rate on
glow curve peaks.
Experimental setup
To conduct TL measurements, samples were cooled or heated on an Instec, Inc.
HCS302-01 hot/cold stage, with a temperature range of -190-400°C, controlled by an mK1000 temperature controller operated by a PC. The hot/cold stage consists of a heated/cooled plate that is thermally shielded from the surrounding atmosphere by a quartz window that is transparent over the entire wavelength range of the detector. The hot/cold stage is contained in a case that is temperature moderated by pumped water flow. Cooling the hot/cold state was done by an LN2-P liquid nitrogen cooling system.
The LN2-P pulls liquid nitrogen through the hot/cold stage via suction, heats it to room temperature, and then expells the nitrogen gas, which is directed at the hot/cold stage to
120 defrost the window. The detector was an Ocean Optics USB2000+ spectrometer. The sample was excited with UV bombardment in position on the hot/cold stage using a PX-2
Pulsed Xenon Light Source. Both the spectrometer and flash lamp were described in detail in Section 2.2. Experiments were conducted within a specially made dark box using custom built lensholder units. Figure 4.0.3 shows a schematic diagram of the setup used for TL experiments.
Figure 4.0.3 Schematic diagram of the TL setup. The sample is heated on a heating stage at a constant
heating rate controlled by the mK1000 temperature controller and LN2-P liquid nitrogen pump and the
frame temperature is regulated by a water pump. A collimated lens collects emitted light from the sample
and directs it down an optical fiber to a spectrometer.
Samples were excited for a consistent time by UV irradiation at the lowest temperature of the measurement. Then the sample was linearly heated to higher temperatures while the spectrometer recorded a series of spectra until the sample reached maximum temperature. For almost all measurements, the sample was excited for 30
121 minutes with the xenon lamp at 5 ms/pulse and the spectrometer recorded a spectrum after integrating over 5°C, however that was calculated from the heating rate. All samples used the same settings for high-temperature TL, but some Ce:YAG samples saturated the detector during low-temperature TL at those settings. Instead, low temperature Ce:YAG measurements were conducted after only 15 minutes excitation and spectra were integrated every 1.5 seconds, regardless of heating rate. For all measurements, the settings of the spectrometer control software SPECTRASUITE
(Ocean Optics) were scans to average: 1 and boxcar: 0. The background noise was subtracted before the first recording with the cap on the spectrometer.
Low-temperature measurements were conducted from liquid nitrogen temperature
(-190°C) to room temperature. High-temperature measurements were conducted by heating the samples from room temperature to 400°C. The sample was excited at the lowest temperature prior to heating.
It was found that the timing of the temperature control and spectrometer software did not match. Using a stopwatch and trial-and-error techniques, it was determined that the heating rate must be multiplied by 1.01 to reach the desired heating rate to match the spectrometer software (e.g., setting the heating rate of the system to 60.6°C/minute gave a reliable heating rate of 60°C/minute). This matched the timing between the two systems.
Glow curves were constructed by integrating each spectrum over a range of wavelengths found to contain a unique peak to obtain the TL intensity for the glow curve.
The temperature profile was matched to the spectra to complete the glow curve. Peaks were then fit using an asymmetric Gaussian, shown later in Eq. 4.1.34, and results were
122 analyzed using the appropriate activation energy calculation. To calculate activation energy by the method of two heating rates, samples were heated at constant rates of
60°C/minute and 20°C/minute.
Some shallow trap measurements on Ce:YAG were already reported in the thesis by D. Mackay [Mackay-2011]. Some glow curves and activation energies presented in that work are referenced where applicable.
Measurements can be conducted at low temperatures or high temperature. Low- temperature measurements are considered shallow trap measurements and high- temperature measurements are considered deep trap measurements due to the relative amount of thermal energy required to free trapped charge carriers in those regimes.
The experimental setup for these measurements was highly repeatable. There was a slot drilled into the lens holder that held the lens at the same spot and the experimental parameters are all constant and it is believed that the intensity of each measurement is comparable to other measurements. There were some possible sources of error that may have varied the intensity slightly across measurements. For instance, between measurements, there were instances where the focal length of the lens was changed slightly due to a loose screw. On top of that, the Xe lamp may have degraded after years of use. However, these factors did not appear to have a significant effect on TL intensity and it is still believed that measurements can be compared directly to each other.
4.1 THEORY
To explain TL phenomena, Randall and Wilkins presented a model based on a single defect level within the band gap and assumed negligible re-trapping, in which case
123 the charge carrier can go straight to a luminescence center [Randall-1945(a), Randall-
1945(b), Randall-1945(c)]. This is the basis of the first order kinetics of TL.
Derivation of initial rise method
The probability p of an electron escaping from a trap of depth E at temperature T
can be described by the Arrhenius equation [Bos-2001 Pankove-1971, Randall-1945(b)]
E p = s exp − (4.1.1) k BT
where k B is the Boltzmann constant and s, known as the frequency factor, is given by
[Pankove-1971]
s = N B vσ t (4.1.2) where N B is the density of states in the band in which the carriers escape, v is the carrier’s thermal velocity, and σt is the trap’s capture cross-section. The frequency factor s has units of time -1 and may vary weakly with temperature. The rate of de-trapping –dn/dt at
temperature T can be expressed as
dn E − = s n exp − (4.1.3) dt k BT
where n is the concentration of trapped charge carriers. Determining n(T) follows from
Eq. 4.1.3 by
dn E = −s exp − dt , n k BT
s E = − exp − dT (4.1.4) q k BT
where q = dT/dt is the heating rate. Throughout these derivations, we assume a linear
124 heating rate. Integration gives
s E ln( n) = −∫ exp − dT '+constant . (4.1.5) q k BT '
The prime in T’ is introduced to distinguish between the variable and the upper limit of integration. When t = 0, let n = n 0 and T = T i = some initial temperature, then the
constant is equal to n 0 and we have
n T s E ln = − exp − dT ' , n ∫ q k T ' 0 Ti B or
T s E n = n exp − exp − dT ' . (4.1.6) 0 ∫ q k T ' Ti B
The thermoluminescence intensity I is proportional to the rate that charge carriers are supplied to luminescence centers [Bos-2001, Chen-1981, DeVault-1983, Garlick-
1948, Kitis-2007, Mahesh-1989, Randall-1945(b)]
dn I(T ) = −c , dt
E = c s n exp − . (4.1.7) k BT
Combining this with Eq. 4.1.6,
E T s E I(T ) = c s n exp − exp − exp − dT ' (4.1.8) 0 k T ∫ q k T B Ti B which is known as the Randall-Wilkins first-order expression of a single glow peak
[Randall-1948(b)]. For low temperatures (T ≈ T i) we may estimate n ≈ n 0 and thus write
125 E I ≈ c s n0 exp − . (4.1.9) k BT
The initial rise method, introduced by Garlick and Gibson [Garlick-1948], is based on Eq.
4.1.9. To calculate the activation energy by the initial rise method, we may rewrite Eq.
4.1.9 in a more useful form
E ln(I) = ln ()c s n0 − . (4.1.10) k BT
Plotting ln(I) versus 1/kBT gives the slope = E.
The initial rise method is applicable to first, second, and even general order cases but is limited to the region where Ii ≤ I « Im.
Bräunlich [Braunlich-1967] showed that the values found for E using the initial rise method will be smaller than the actual values when the re-trapping factor
R[h(T i)/f(T i)] » 1, where R is given by
probabilit y of re − trapping R = (4.1.11) probabilit y of recombinat ion
and h(T i) and f(T i) are traps with electrons and the initial concentration of unoccupied recombination centers, respectively (at the temperature T i). It is believed that this does
not affect the majority of our measurements, as it basically implies high order kinetics.
Christodoulides [Christodoulides-1985] estimated a correction to the initial rise
method that may be used if the intensity of the range used for calculation is larger than a
small fraction of the peak height. The activation energy calculated by the initial rise
method E IR can be adjusted to the corrected value of the activation energy E C by
T E = 1( + 74.0 a + .0 082 a )E − 2( a + 22.0 a ) m . (4.1.12) C 1 2 IR 1 2 11605
126 The factors a 1 and a 2 are the fraction of the peak value of the TL intensity observed at
temperatures T 1 and T2, respectively, marked as the lower and upper bounds of the fit, as demonstrated in Fig. 3.1.1.
15
14 I2=a 2Im
13
12 I1=a 1Im ln(Integrated TL emission) TL ln(Integrated
0.0018 0.0020 0.0022 1/T (C-1 )
Figure 4.1.1 TL glow curve of Ce:YAG 0.2% plotted as ln(I) versus 1/T for calculation of the activation
energy by the initial rise method, demonstrating the definitions of a 1 and a 2.
Derivation of methods of two and variable heating rates
The initial rise method is applicable to general order kinetics. However, if peaks are overlapped it is not possible apply this method since the beginning of a peak is not available for analysis. The methods of two and variable heating rates only require that the peak temperature can be determined for calculation of E. These methods can be used with general order kinetics. The methods of two and variable heating rates can be derived from first order kinetics or simplified from general order kinetics. Rather than show the derivation twice, these methods will be derived for general order kinetics and then applied to the case of first order kinetics. It will be necessary to assume that s is
127 independent of temperature to proceed with this derivation. To general order of kinetics b, Eq. 4.1.7 becomes [Bos-2001, Chen-1981, Kitis-1998(b), Kitis-2007, Mahesh-1989,
May-1964]
dn b E I(T ) = −c = c s'n exp − (4.1.13) dt k BT where s’ is the general order pre-exponential factor and s’ = s for b = 1. Equation 4.1.13 can be integrated
n E (n )' b = − s'exp − dt . (4.1.14) ∫ ∫ k T n0 B
This has the solution
s (' b − )1 nb−1 T E 1−b 1−b 0 n = n0 1+ ∫ exp − dT ' , q Ti k BT' or
1 s (' b − )1 nb−1 T E 1−b n = n 1+ 0 exp − dT ' . (4.1.15) 0 q ∫ k T' Ti B
The intensity of Eq. 3.1.13 then becomes
b E s (' b − )1 nb−1 T E b−1 I(T ) = c s'exp − nb 1+ 0 exp − dT ' . (4.1.16) k T 0 q ∫ k T ' B Ti B
The maximum intensity of a peak occurs at temperature T m. At this point, the
derivative of the intensity is zero, as is the derivative of the logarithm of the intensity.
Thus, we obtain at T = T m the relationship
−1 d{ln[ I(T)]} E b s (' b − )1 nb−1 T E 0 = 0 = 2 − 1+ exp − dT ' dT k T b −1 q ∫ k T' T =Tm B m Ti B T =Tm
128 s (' b − )1 nb−1 E 0 × exp − . (4.1.17) q kBTm
This yields
s (' b − )1 nb−1 Tm E b s' nb−1k T 2 E 1+ 0 exp − dT = 0 B m exp − . (4.1.18) q ∫ k T q E k T Ti B B m
The following approximation can be made
T n E E k BT n−1 ∫ exp − dT ' ≅ T exp − ∑ (− )1 n! (4.1.19) k T ' k T n=1 E Ti B B
which, according to Kitis and Pagonis [Kitis-2007], is usually “a very good numerical
approximation.” The series converges quickly, so only the first two terms need to be
taken. Inserting Eq. 4.1.19 into 4.1.18 yields
s (' b − )1 nb−1 E kT 2k 2T 2 b s' nb−1k T 2 E 0 T m 0 B m , 1+ exp − − 2 = exp − q k BTm E E qE k BTm
which becomes
q E E 2k T b−1 B m 2 = s' n0 exp − 1+ (b − )1 . (4.1.20) k BTm k BTm E
At this point, we can calculate the first order solutions for the method of two heating rates and the variable heating rate method by setting b = 1 and s’ = s
q E E 2 = s exp − . (4.1.21) k BTm k BTm
Conducting a pair of TL measurements and recording the maximum temperature of a
peak in both glow curves gives two such equations with different q and T m, but with the
same E and s. Division of one equation by the other gives
129 q T 2 E E 1 m2 2 = exp − , (4.1.22) q2Tm1 k BTm2 k BTm1 which can be solved for E
k T T q T 2 E B m1 m2 1 m2 = ln 2 , (4.1.23) Tm1 − Tm2 q2Tm1 which represents the activation energy to first order kinetics using the method of two heating rates [Bos-2001, Chen-2001, Kitis-1998(a), Kitis-1998(b), Mahesh-1989]. The frequency factor can also be found by Eq. 4.1.21 once we have obtained E.
The variable heating rate method adapts the method of two heating rates to the case of many heating rates, to be solved graphically. From Eq. 4.1.12, for first order kinetics,
2 Tm 1 s k B ln = E − ln . (4.1.24) q k BTm E
Data can easily be plotted in this way to give the activation energy as its slope. Once E is calculated, s may be calculated from the y-intercept.
The general order variable heating rate method and method of multiple heating rates are not well-documented in literature [Mahesh-1989] and, to my knowledge, have not been solved in the following way. The solution of the general order variable heating rate method begins by taking the logarithm of Eq. 4.1.20
b−1 2 E s' n0 k BTm 2k BTm = ln 1+ ()b −1 , (4.1.25) k BTm q E E which can be separated into the parts
2 b−1 E Tm 2k BTm s' n0 k B = ln + ln 1+ (b − )1 + ln . (4.1.26) k BTm q E E
130 The activation energy is typically at least an order of magnitude larger than k BTm
[Mahesh-1989], so we can apply the series expansion for ln(1+x) for x ≤ 1 to get
T 2 E E k T k 2T 2 m b B m b 2 B m ln ≈ ln b−1 + − (2 − )1 + (2 − )1 2 q s' n0 k B k BTm E E
8 k 3T 3 − (b − )1 3 B m +K. (4.1.27) 3 E 3
2 If we let y = ln (Tm q) and x = k BTm and designate y 0 as the logarithmic term independent of both q and T m, Eq. 4.1.27 becomes
E x x 2 8 x 3 y = y + − (2 b − )1 + (b − )1 2 − (b − )1 3 +K, (4.1.28) 0 x E E 2 3 E 3
which can be plotted and fit using a simple fitting program. Mahesh et al. [Mahesh-
1989] solve this in a very different way:
2 b b−1 b−1 b−1 T (c s' n ) E ln I m = bln E + ln 0 + . (4.1.29) m q b k T ()s' b k B B m
Equation 4.1.29 shows dependence on intensity, which may add a further source of
uncertaincy. Detector zero drift, overlapping glow curve peaks, nonlinear intensity
response by the detector, or other misleading intensity measurements will add error to
calculations. Thus, Eq. 4.1.27 is preferred over Eq. 4.1.29 since it contains less factors
that may create further sources of error.
The general order method of two heating rates can be solved by subtracting Eq.
4.1.27 for two different measurements
E 1 1 q T 2 (2 b − )1 k (2 b − )1 2 k 2 1 m2 B T T B T 2 T 2 0 = − + ln 2 + ()m2 − m )1 − 2 ()m2 − m1 k B Tm1 Tm2 q2Tm1 E E
131 (8 b − )1 3 k 3 + B ()T 3 − T 3 −K. (4.1.30) 3E m2 m1
This is easiest solved graphically. After inputting all the known values for T mi and q i and
the constants and letting x = E, we can write Eq. 4.1.30 as
C D E y = Ax + B + − + −K, (4.1.31) x x 2 x 3 which can be plotted to find the x-intercept = E. The substitutions made are as follows:
1 1 1 A = − , k B Tm1 Tm2 q T 2 B 1 m2 = ln 2 , q2Tm1
C = (2 b − )1 k B ()Tm2 − Tm1 , (4.1.32) 2 2 2 2 D = (2 b − )1 k B ()Tm2 − Tm1 , 8 E = (b − )1 3 k 3 ()T 3 − T 3 . 3 B m2 m1
To check this work, we can set b=1 in Eq. 4.1.30 and see that we return the first order kinetics solution of Eq. 4.1.22.
The solution presented by Mahesh et al. [Mahesh-1989] for the general order method of two heating rates is
2 2 Tm1 1 2k BTm1 E Tm2 1 2k BTm2 E + exp − = + exp − , q1 b −1 E k BTm1 q2 b −1 E k BTm2
(4.1.33)
which is not easily solved for E.
Thus we have formulated the theory for the initial rise method (Eq. 4.1.10), which
can be applied to general order kinetics; the first order method of two heating rates (Eq.
4.1.22) and variable heating rate method (Eq. 4.1.24); and general order method of two
132 heating rates (Eq. 4.1.30) and variable heating rate method (Eq. 4.1.27). For our measurements, the method of two heating rates was used most often due to its simplicity.
Results often showed overlapping peaks at low intensities and thus it was not possible to fit the initial rise of a peak, though T m could easily be determined. Three different
methods were applied in these measurements and occasionally compared to confirm
results.
The general order solutions were applied to several peaks, including peaks that
clearly displayed higher order kinetics. However, results for activation energy
calculations never varied by more than a few percent, so the first order solutions are
presented in this thesis. Effectively this is an approximation where kBTm/E is assumed small enough to be ignored, as appears to almost always be the case.
In order to determine T m for activation energy calculations, peaks present in the
glow curves were fit to an asymmetric Gaussian function (Eq. 4.1.34) due to their
intrinsic asymmetry. The equation for this function is
A (T − c) 2 I(T ) = exp − . (4.1.34) 2 2π []σ −α()T − c 2[]σ − α(T − c)
This asymmetric Gaussian peak fit was systematically determined by Bos et al. [Bos-
1993, Bos-1994] to fit TL peaks as well as or better than rigorous computational methods
used by researchers.
Glow peaks were fit using the progran GNUPLOT. T m was calculated from the
fitting parameters
2α 3c + 2α 2σ + σ − 4α 2σ 2 + σ 2 T = , (4.1.35) m 2α 3 which was found by setting the derivative of Eq. 4.1.34 equal to zero.
133 The uncertainty in T m and I m were calculated by partial derivatives [Lyons-1991], i.e.,
dT 2 dT 2 dT 2 dT 2 = m dα 2 + m dc 2 + m dσ 2 (4.1.36) m dα dc dσ
and similarly fot dI m. This yielded
2 2 2 σ 3σ 2σ 3σ 1+ 4α 2 dT m = + + − dα 2 4 2 2 4 α 2 α 1+ 4α 2α
1 1− 1+ 4α 2 + + dσ 2 + dc 2 (4.1.37) 3 α 2α and
2 dI 2 1 T − c α(T − c) 2 = α − m − m ()dT 2 + dc 2 2 2 m I σ −α(Tm − c) σ −α(Tm − c) []σ −α(Tm − c)
2 2 1 (T − c)2 dA 2 m 2 + 2 −1 dσ + 2 σ −α(Tm − c) []σ −α(Tm − c) A
2 2 1 (T − c)3 + T − c − m dα 2 . (4.1.38) 2 σ −α(Tm − c) []σ −α(Tm − c)
The uncertainties in activation energy and frequency factor were obtained from the graphical fit whenever possible, but the method of two heating rates required calculation by hand. These uncertainties were obtained using partial derivatives
2 dE 2 T q T 2 T T q T 2 T = m2 ln 1 m2 − m1 m2 ln 1 m2 − 2 m2 dT 2 2 2 2 2 m1 k B Tm1 − Tm2 q2Tm1 ()Tm1 − Tm2 q1Tm1 Tm1 − Tm2
2 T q T 2 T T q T 2 T m1 ln 1 m2 m1 m2 ln 1 m2 2 m1 dT 2 + 2 + 2 2 + m2 Tm1 − Tm2 q2Tm1 ()Tm1 − Tm2 q1Tm1 Tm1 − Tm2
134 (4.1.39) and
2 2 q E E E 2 ds 2 exp 1 dE 2 dT 2 . (4.1.40) = 2 + + 2 k BTm k BTm k BTm Tm
For these calculations, dq was determined to be negligible compared to dT m due to the stability of the hot/cold stage and our experimental design.
In some cases, there were multiple runs at one heating rate, each with slightly different estimates for T m and dT m. These values were averaged using a weighted average based on uncertainty [Lyons-1991]
E ∑ i i dE 2 E = i (4.1.41) 1 ∑i 2 dE i
1 dE 2 = . (4.1.42) 1 ∑i 2 dE i
4.2 RESULTS
High temperature TL of undoped YAG (as grown)
Figure 4.2.1 shows the contour plot of Ar-grown YAG 10x10x1 nm, plotting contours of intensity to a map of temperature versus wavelength.
135
400
350
300
250
200
150
100
50 Approximate Temperature (C) Temperature Approximate
200 300 400 500 600 700 800 Wavelength (nm)
Figure 4.2.1 Contour plot of TL of Ar-grown YAG 10x10x1 mm, plotting luminescence intensity contours
to a map of temperature versus wavelength. Each contour represents a step of 200 counts.
As seen in Fig. 4.2.1, undoped YAG has two clearly distinguishable recombination centers associated with different traps. Figure 4.2.1 clearly displays the two different centers, one centered near 700 nm and one centered near 450 nm. We reported previously that the peak 700 nm peak likely arises from iron [Varney-2012(b)], but results shown in Section 3.2 and glow discharge mass spectrometry (GDMS) measurements suggest that it may also arise from Cr3+ . GDMS measurements confirmed the presence of both Fe and Cr in the sample, present at 1 ppm wt. (0.31 ppm at.) and
0.59 ppm wt. (0.48 ppm at.), respectively. In Chapter 3, it was seen that Fe 2+ is a stable oxidation state of iron in YAG at room temperature, so Fe 3+ can act as an electron trap.
Color center measurements (Section 2.3) showed that iron can act as a hole or electron trap, depending on the predominant valence state of iron in the sample. At room temperature, chromium is stable as Cr 4+ in YAG [Choi-2001, Shimony-1995], thus Cr 3+
136 can act as a hole trap. It is possible that the emission at 700 nm could be luminescence from either or both ions since both samples have emission at 700 nm and can form recombination centers for TL.
Figure 3.2.1 above in Section 3.2 showed PL emission of Tb 3+ impurities in the same 400-500 nm wavelength range as the low wavelength peak in Fig. 4.2.1. Tb 3+ is
known to act as a stable hole trap at room temperature [Batygov-1972]. The series of
5 7 peaks in this wavelength range are attributed to the D3 → Fj transitions in terbium impurities [Choe-2001, Hirata-1996, Jung-2007, Lo-1996, Milliken-2012, Potdevin-2005,
Weg-1985, Zhang-2003]. As a result, it seems reasonable to attribute this peak to Tb 3+
impurities. However, GDMS showed that the concentration of Tb impurities in O 2- grown YAG was below the detection limit. It is also possible that this emission may arise from recombination at antisite defects (AD), though the luminescence peak is different from what was observed in Sections 3.2 and 3.4 and the UV excitation may not be high enough energy to release electrons from ADs. The likliest source of the low wavelength emission appears to be Tb 3+ ions.
Near 400°C we observe the onset of infrared emission, a characteristic of all TL
measurements performed. It should be noted that the temperature is only an
approximation in Fig. 4.2.1, due to limitations of the plotting software.
From Fig. 4.2.1 we can determine integration wavelength ranges to construct the
glow curve. To study the low wavelength peak, each spectrum is integrated over the
range 340-570 nm. To study the high wavelength peak, each spectrum is integrated over
the range 570-800 nm, stopping at 800 nm to minimize IR noise at higher temperatures.
These integration ranges apply to high-temperature studies on all undoped YAG samples.
137 Figure 4.2.2 shows the glow curves for the low and high wavelength peaks for as grown YAG samples.
450000 Ar-grown Ar-grown 400000 80000 Ar-grown 10x10 Ar-grown 10x10 350000 H -grown ac O2-grown 2 60000 300000 O -grown H2-grown 2 250000
40000 200000 150000 20000 TLIntensity (counts) 100000 50000 0 0
IntegratedTL (counts) Intensity 0 100 200 300 400 Integrated 0 100 200 300 400 Temperature (C) Temperature (C)
Figure 4.2.2 Glow curves of as grown undoped YAG using an integration range of 340-570 nm (left) and
570-800 nm (right).
For the low wavelength peak, H2-grown YAG shows no TL while Ar- and O2-
grown YAG show a broad peak near 250°C. The peak in the O 2-grown YAG glow curve is slightly shifted to lower wavelength, indicating a slightly shallower trap. Batygov et al. [Batygov-1972] reported a single deexcitation maximum of trace Tb-doped YAG near
200°C in which Tb 3+ was reported to be the luminescence center, generally in agreement
with the TL spectra shown in Fig. 4.2.2.
The high wavelength range again shows no TL signal from the H 2-grown YAG.
The other samples all show peaks at 110°C, 185°C, and 250°C and one more high temperature peak. For the Ar-grown YAG samples, this high temperature peak is at
390°C, but it appears at lower temperature, near 360°C, for O 2-grown YAG. This effect is further explored in Fig. 4.2.4, where the effects of oxidation on the glow curves of Ar- grown YAG are shown.
138 Figure 4.2.3 shows the TL spectrum as a function of wavelength for Ar-grown undoped YAG at 252°C, which is the temperature of peak luminescence for the glow curves for both ranges of integration. In this figure, the intense peak at 700 nm can clearly be seen and the weaker peak at lower wavelengths can barely be seen.
2000
1500
1000
500
TL Intensity (counts) Intensity TL 0
-500 100 200 300 400 500 600 700 800 900 Wavelength (nm)
Figure 4.2.3 TL emission spectrum at 252°C for Ar-grown YAG. The emission at 700 nm is much
stronger than the 400-500 nm emission.
High temperature TL of undoped YAG (annealed)
Figure 4.2.4 shows the effects of annealing on the glow curves of Ar-grown undoped YAG.
139
80000 350000 Ar-grown YAG Ar-grown YAG Ar-grown YAG annealed 300000 Ar-grown YAG annealed 60000 250000
200000 40000 150000
20000 TLIntensity (counts) 100000 50000
0 0
IntegratedTL (counts) Intensity 0 50 100 150 200 250 300 350 400 450 Integrated 0 50 100 150 200 250 300 350 400 450 Temperature (C) Temperature (C)
Figure 4.2.4 Glow curves of Ar-grown YAG as grown and annealed in air, using a 340-570 (left) nm and
570-800 nm (right) integration range.
The TL intensity of Ar-grown YAG mostly decreased in both wavelength ranges after oxidative annealing. This treatment clearly decreases the intensity of most peaks.
This could arise from filling oxygen vacancies (V Os) responsible for these traps or from
oxidation of impurities responsible for charge carrier trapping. For instance, if Cr3+ is the
luminescence center, annealing in air can oxidize some Cr 3+ to Cr 4+ and thus fewer
chromium ions can act as hole traps, hence decreased TL intensity.
The highest temperature peak in Ar-grown YAG shifted from 390°C to 350°C
after annealing but remained the same intensity. The 390°C peak in the as grown sample
appears to arise from a defect aggregate including at least one V O.
High temperature TL of undoped YAG (further treatment)
Let us examine the reason for a lack of TL in the H2-grown YAG. Growth in the reducing atmosphere could have maintained accidental dopants in reduced states with more electrons, causing several ions to be much larger and more positively charged than they would have been under growth in different conditions. Ions such as Ce 3+ that
140 incorporate well as trivalent under reducing conditions would govern the optical properties of the crystal. In low-temperature TL measurements below (Fig. 4.2.10), the glow curve of H 2-grown YAG resembles that of Ce:YAG, with emission occurring over
the same wavelength range. Since Ce 3+ ions are such strong emitters and cause very intense thermoluminescence, it must either be present in very low concentration or there must be a lack of charge carrier traps in H 2-grown YAG compared to Ar- or O 2-grown
YAG. Figure 3.3.3 showed that stable traps do exist in H 2-grown YAG at room temperature, though they are very few compared to other samples and they only exist at
UV wavelengths. Thus it appears that H 2-grown YAG does not have much emission during TL because it lacks the impurity defects and traps present in other undoped samples that lead to intense TL emission and that these effects are stabilized by defect aggregates.
An alternative explanation is that impurities are removed from the melt in the reducing atmosphere. Kovaleva et al. [Kovaleva-1981] grew crystals by the Czochralski method under vacuum and found that growth in vacuum removed iron, chromium, and other transition elements from the melt due to the high volatility of the oxides formed by these elements. Vacuum could be considered a reducing atmosphere in this case because it leads to the removal of oxygen from the crystal melt. The reducing atmosphere of hydrogen may have the same effect, and thus oxygen may escape the melt attached to transition elements, and thus the crystal is formed with a decreased concentration of impurities than was initially present in the melt.
H2-grown YAG was annealed in air and sputtered with aluminum followed by annealing in argon. These samples did not show TL after any processing (not shown).
141 Some O 2-grown YAG samples were annealed in a mixed atmosphere of H 2+Ar and some were sputtered with aluminum and Ar annealed. Figure 4.2.5 shows the glow curves at both wavelengths for as grown and annealed O 2-grown YAG.
120000 700000 O -grown YAG 2 O2-grown YAG O -grown YAG Ar+H an. 100000 600000 2 2 O2-grown YAG Ar+H 2 an. O -grown YAG Al sp.+Ar an. O -grown YAG Al sp+Ar an. 500000 2 80000 2
400000 60000
300000 40000
TLIntensity (counts) 200000 20000 100000 0 0
Integrated 0 50 100 150 200 250 300 350 400 450 IntegratedTL Intensity (counts) 0 100 200 300 400 Temperature (C) Temperature (C)
Figure 4.2.5 Glow curves of O 2-grown YAG as grown, and annealed in H 2+Ar, and Al sputtered and Ar
annealed; with a 340-570 nm (left) and 570-800 nm (right) integration range.
A very significant change is seen in Fig. 4.2.5 after Ar+H 2 annealing. The 250°C
peaks still appears to be present in this glow curve as a shoulder to the large new peak at
140°C, which has another shoulder at 100°C. This wavelength range corresponds to a
series of peaks that was attributed above to Tb 3+ impurities. Milliken et al. presented the glow curve for Tb:YAG, showing peaks at exactly the same temperatures and at the over the same wavelengths as the Ar+H 2-annealed O 2-grown YAG glow curve above [Fig.
4(e) in Milliken-2012]. It seems that reductive anneal causes the crystal lattice to be more accommodating to Tb 2+ , likely transitioning several trivalent ions to divalent in the
process. However, divalent terbium is unstable in YAG at high temperatures, hence the
intense TL emission as the trapped holes are released.
142 Hydrogen impurities were suggested above to be present in H 2-grown YAG to further explain the decreased TL intensity. Infrared absorption measurements confirmed the presence of hydrogen in H2- and O2-grown samples, although they appeared to be
incorporated in different ways in either sample. Hydrogen was found to incorporate in
-1 O2-grown YAG as O-H due to a strong absorption peak at 3370cm , but there was not
-1 absorption peak from the H 2-grown sample at 3370cm and instead there was one at
3416cm -1. Devor et al. [Devor-1984] observed faint peaks at 3400 and 3425 cm -1 in
YAG samples with high hydrogen concentration, so it is possible that this peak is related.
Devor et al. suggested that larger than normal lattice perturbations by a high hydrogen
density caused the absorption peak to shift to lower frequencies. Annealing the O 2-grown
sample in a mixed atmosphere of hydrogen and argon slightly increased the intensity of
-1 the 3370cm peak. The TL intensity of O 2-grown YAG appeared to increase significantly after annealing in an atmosphere containing hydrogen whereas the glow curve of H 2-grown YAG does not show observable TL. It seems that annealing in hydrogen may not inject hydrogen into the sample at all but rather remove oxygen and reduce the charge state of impurities. This would lead to a high population of Tb 2+ ions which, due to its instability in YAG at high temperatures, shows in TL measurements.
Reduction of O 2-grown YAG can reduce the charge state of some impurities, adding electron orbitals and increasing ionic radius. The increased ionic radius and reduced charge state can eject an adjacent oxygen ion. This effect is seen in the high temperature peak in Fig. 4.2.5 where reduction moved the peak in opposite the way oxidation moved it in the Ar-grown sample in Fig. 4.2.4, presented again below in Fig.
143 4.2.6. This adds further evidence that this peak arises from defect aggregate involving an impurity and an oxygen vacancy.
200000 60000
Ar-grown YAG O2-grown YAG Ar-grown YAG annealed O -grown YAG Ar+H an. 150000 2 2 40000
100000
20000
TLIntensity (counts) 50000
0 0
Integrated 300 350 400 IntegratedTLIntensity (counts) 300 350 400 Temperature (C) Temperature (C)
Figure 4.2.6 Glow curves of Ar-grown YAG before and after annealing in air (left) and O 2-grown YAG
before and after annealing in H 2+Ar (right), with a 570-800 nm integration range. This shows the way the
way the high temperature peak shifts due to increased (left) or decreased (right) oxygen concentration in
the sample.
Due to the significance of the glow curve of the O2-grown YAG annealed in argon and hydrogen, its contour plot is shown in Fig. 4.2.7 and its emission spectrum at
138°C is shown in Fig. 4.2.8.
144 400
350
300
250
200
150
100
Approximate Temperature (C) Temperature Approximate 50
200 300 400 500 600 700 800 Wavelength (nm)
Figure 4.2.7 Contour plot of TL of O 2-grown YAG annealed in Ar+H 2, plotting luminescence intensity contours to a map of temperature versus wavelength. The first (outer) contour is 150 counts and
each contour above that represents a step of 300 counts.
1200
1000
800
600
400
200
0 TL Intensity (counts) Intensity TL -200
-400 100 200 300 400 500 600 700 800 900 Wavelength (nm)
Figure 4.2.8 TL emission spectrum at -138°C for O 2-grown YAG annealed in a mixed atmosphere of
Ar+H 2.
145 High temperature TL of undoped YAG conclusions
Different impurities lead to different thermoluminescence in undoped YAG. The thermoluminescence in these samples appears to come from Tb (lower wavelength) and
Cr or Fe (higher wavelength) impurities.
After annealing the Ar-grown sample in air, it was seen that all peaks decreased in intensity or in trap depth. The decreased intensity may be due to filling oxygen vacancies or it can be due to changed charge state of electrons. In O 2-grown YAG, this peak
already exists at the lower temperature and increases to the higher temperature after
annealing in a reducing atmosphere containing hydrogen. The shifted energy level of the
highest temperature peak suggests a very significant change, potentially due to oxygen
diffusing into a vacancy cluster involving Al and O vacancies.
Low temperature TL of undoped YAG
Low-temperature measurements on Ar-grown YAG were already reported by
Mackay [Mackay-2011]. It was found that all Ar-grown YAG, even after annealing, had a single peak at -120°C that was obviously of high order kinetics due to its long exponential decay at higher temperatures. This peak was of different origin than the high-temperature peaks and the glow curve was constructed by integration over the wavelength range 300-740 nm. It can be seen below that this is different from the glow curves of H 2- and O 2-grown YAG. Low-temperature glow curves of H 2- and O 2-grown
YAG are shown in Figs. 4.2.9 and 4.2.10, compared to each other due to their similarity.
O2-grown YAG uses the same wavelength integration ranges of high-temperature YAG,
but H 2-grown YAG, containing significantly different impurities, uses different integration ranges of 191-450 nm and 450-700 nm. The range 450-700 nm was
146 determined very roughly by guess and check methods and matches very closely to the range used for Ce:YAG of 470-720 nm.
30000 H2-grown YAG H -grown YAG annealed 20000 2 H2-grown YAG annealed Al sp+Ar an 10000
0
900000 O2-grown YAG TL Intensity (counts) TL Intensity O2-grown YAG Ar+H 2 annealed
600000 O2-grown YAG Ar+H 2 annealedx2 O -grown YAG Al sp+Ar an 2 300000 Integrated Integrated 0 -200 -150 -100 -50 0 50 Temperature (C)
Figure 4.2.9 Glow curves of H2-grown YAG as grown, air annealed, and Al sputtered and Ar annealed; with and integration range of 191-450 nm (top); and O2-grown YAG as grown, and annealed in H 2+Ar, and
Al sputtered and Ar annealed; using a 340-570 nm integration range (bottom).
147
30000 H2-grown YAG
H2-grown YAG annealed 20000 H2-grown YAG annealed Al sp+Ar an 10000
0 400000
O2-grown YAG
300000 O2-grown YAG Ar+H 2 annealed
O2-grown YAG Ar+H 2 annealedx2 200000 O -grown YAG Al sp+Ar an 2 100000 Integrated TL Intensity (counts) Intensity TL Integrated
0 -200 -150 -100 -50 0 50 Temperature (C)
Figure 4.2.10 Glow curves of H 2-grown YAG as grown, air annealed, and Al sputtered and Ar annealed;
with and integration range of 450-700 nm (top); and O 2-grown YAG as grown, and annealed in H2+Ar, and
Al sputtered and Ar annealed; using a 570-800 nm integration range (bottom)
Figures 4.2.9 and 4.2.10 show similar defect structure with slightly different depth, as seen by the shift in temperature. It seems that both samples have very similar shallow traps despite very different luminescence centers. The only significant difference is that the glow curve of H 2-grown YAG does show a peak at about -75°C not present in the glow curve of O 2-grown YAG. The luminescence centers for H 2-grown YAG are
Ce 3+ (for high wavelength TL emission) and some unknown impurity, undeterminable
due to its weak intensity and the fact that this is the only instance of this impurity.
148 Low temperature of undoped YAG conclusions
Undoped YAG shows various shallow traps, corresponding to different impurities. Ar-grown YAG shows extensive retrapping, indicating high order kinetics.
Results on Ar-grown YAG were discussed by Mackay in [Mackay-2011].
High temperature TL of undoped YAG (labeled as Yb:YAG 5%)
The undoped YAG sample labeled as Yb:YAG 5% shows a couple glow curve
peaks at high temperature, one weak peak at 105°C and one strong high order peak at
300°C, shown in Fig. 4.2.11. This glow curve was constructed using the same integration
range for Ce:YAG, 470-720 nm.
350000 Undoped YAG ("Yb:YAG 5%")
300000
250000
200000
150000
100000
50000
0 Integrated TL Intensity (counts) Intensity TL Integrated
0 50 100 150 200 250 300 350 400 450 Temperature (C)
Figure 4.2.11 Glow curve of undoped YAG (mislabeled as Yb:YAG 5%) using 470-720 nm integration.
It is unclear where this emission comes from. This sample appears to contain an
abundance of impurities so it is very likely that one of those is the cause, though it is
unclear which one(s). It is worth noting that this sample was intended to be doped with
Yb but x-ray photoelectron spectroscopy (XPS) and energy dispersive x-ray (EDS)
149 measurements as well as our own optical measurements revealed no presence of Yb in the sample. The only similarity to other undoped YAG is the weak peak near 100°C that appears in undoped YAG. The peak at 300°C, clearly of higher order kinetics, is not observed in other undoped samples.
Low temperature TL of undoped YAG (labeled as Yb:YAG 5%)
Low temperature TL appears similar to Ar-grown YAG and is graphed together with Ar-grown YAG in Fig. 4.2.12. In this figure, Ar-grown YAG was constructed using the integration range 300-740 nm and the mislabeled Yb:YAG 5% sample used the integration range 250-640 nm. It is possible that both contain the same shallow trap.
Mislabeled Yb:YAG 5% Ar-grown undoped YAG 400000
200000
0 Integrated TL Intensity (counts) Intensity TL Integrated
-200 -150 -100 -50 0 50 Temperature (C)
Figure 4.2.12 Glow curves for undoped YAG (labeled as Yb:YAG 5%) sample using 470-720 nm
integration and Ar-grown YAG using 300-740 nm integration.
150 Undoped YAG defect level calculations
Activation energies were calculated from the glow curves of all samples by the method of two heating rates except high-temperature O 2-grown YAG glow curves, which
were analyzed using the method of multiple heating rates.
Ar-grown YAG defect level and frequency factor calculations
Activation energies and frequency factors calculated for Ar-grown YAG samples are presented in Tables 4.2.1 and 4.2.2 for low wavelength and high wavelength ranges, respectively.
Table 4.2.1 Calculated activation energies of Ar-grown YAG samples: polished as grown 5x5x1 (As
grown) and fine ground unpolished 5x5x1 air annealed (fg air annealed); for the glow curves constructed using 340-570 nm integration.
Peak location Activation energy Frequency factor
(q=60°C/min) As grown fg air annealed As grown
250°C 1.47 ± 0.23 eV 0.95 ± 0.09 eV 2.68 x 10 10 ± 9.19 x 10 6 s-1
Table 4.2.2 Calculated activation energies of Ar-grown YAG samples: polished as grown 5x5x1 (As
grown) and fine ground unpolished 5x5x1 air annealed (fg air annealed); for the glow curves constructed using 570-800 nm integration.
Peak location Activation energy Frequency factor
(q=60°C/min) As grown fg air annealed As grown
110°C 1.65 ± 0.01 eV 1.63 ± 0.09 eV 6.47 x 10 17 ± 5.39 x 10 14 s-1
180°C 3.04 ± 2.27 eV 1.44 ± 1.87 eV (not calculated)
250°C 1.92 ± 0.09 eV 2.07 ± 0.15 eV 4.21 x 10 19 ± 4.22 x 10 16 s-1
151 345°C N/A 2.23 ± 0.55 eV 1.14 x 10 22 ± 1.42 x 10 19 s-1
395°C 3.15 ± 0.25 eV N/A N/A
Obviously, the activation energy increases with temperature, as charge carriers at deeper traps require more thermal energy for release. The most noticeable change is the large decrease in activation energy of the highest energy peak after annealing. The depth of the low-wavelength peak at 250°C appears to decrease slightly after annealing, possibly due change in charge state of Tb and/or change of lattice defects surrounding
Tb 3+ ions. No other activation energy appeared to change after annealing, so it seems that annealing only decreased their concentration and not their depth.
O2-grown YAG defect level calculations
Activation energies calculated for O 2-grown YAG samples are presented in
Tables 4.2.3 and 4.2.4 for low wavelength and high wavelength ranges, respectively.
Some peaks were not included due to difficulty of fitting or lack of measurement at a second heating rate. Uncertainties are large for several of these peaks due to their low intensity.
Table 4.2.3 Calculated activation energies of O 2-grown YAG samples: as grown (As grown), after one
anneal in a mixed atmosphere of Ar+ H2 (Ar+H 2 annealed), and after two separate anneals in Ar+ H2
(Ar+H 2 annealed x2); for the glow curves constructed using 340-570 nm integration.
Peak location Activation energy
(q=60°C/min) As grown Ar+H 2 annealed Ar+H 2 annealed x2
-130°C .293 ± 0.05 eV N/A N/A
152 -95°C 1.15 ± 1.44 eV 0.93 ± 2.32 eV 1.19 ± 1.50 eV
-30°C 0.80 ± 0.11 eV N/A N/A
220°C 1.20 ± 0.07 eV N/A N/A
Table 4.2.4 Calculated activation energies of O 2-grown YAG samples: as grown (As grown), after one
anneal in a mixed atmosphere of H 2+Ar (Ar+H 2 annealed), and after two separate anneals in H 2+Ar (Ar+H 2 annealed x2); for the glow curves constructed using 570-800 nm integration.
Peak location Activation energy
(q=60°C/min) As grown Ar+H 2 annealed Ar+H 2 annealed x2
-130°C 0.31 ± 0.06 eV 0.35 ± 0.04 eV 0.30 ± 0.12 eV
-30°C 0.64 ± 0.07 eV 0.54 ± 0.06 eV 0.66 ± 0.05 eV
120°C 6.14 ± 12.13 eV N/A N/A
250°C 1.49 ± 0.38 eV N/A N/A
375 3.34 ± 0.68 eV N/A N/A
Due to the high uncertainty, it is difficult to compare these results to the Ar-grown
YAG activation energies.
Undoped YAG TL conclusions
Oxygen vacancies and impurities in reduced states play key roles in charge carrier trapping in undoped YAG. Upon oxidative anneal, TL intensity of most peaks was reduced, indicating that fewer charge carrier traps were present. The deepest trap measured decreased in depth from ~3 eV to ~2 eV, a large change that suggests this trap is related to an oxygen and aluminum vacancy cluster that decreased in depth after oxygen diffused into the trap.
153 O2-grown YAG had very low thermoluminescence intensity. The low intensity may be due to the fully oxidized state of the majority of impurities and/or the increased oxygen concentration in this sample. Annealing in reducing conditions appeared to have the reverse effect that oxidative annealing had on other samples.
H2-grown YAG showed very little thermoluminescence. This sample contains different impurities due to its heavily reduced nature.
High temperature TL of Ce:YAG
Ce:YAG TL measurements showed intense emission. The only recombination centers in all TL measurements conducted on Ce:YAG samples were Ce 3+ ions. Ce:YAG is grown in a reducing atmosphere containing hydrogen, so this is likely a result of the growth atmosphere, similar to the effect of growth atmosphere on impurity concentration
3+ on H 2-grown YAG described above, as well as the dominance of Ce ions on charge transfer mechanisms within the crystal. Therefore, the integration range used to construct all glow curves for Ce:YAG samples was 470-720 nm. Figure 4.2.13 shows a sample contour plot for high-temperature Ce:YAG TL for Ce:YAG 0.14% and Fig. 4.2.14 shows a spectrum of TL emission as a function of wavelength for the same sample for the maximum emission at 262°C.
154
400
350
300
250
200
150
100
50 Approximate Temperature (C) Temperature Approximate
200 300 400 500 600 700 800 Wavelength (nm)
Figure 4.2.13 Contour plot of TL of Ce:YAG 0.14%, plotting luminescence intensity contours to a map of
temperature versus wavelength. Each Contour represents 1500 counts.
20000
15000
10000
5000 TL Intensity (counts) Intensity TL
0
100 200 300 400 500 600 700 800 900 Wavelength (nm)
Figure 4.2.14 TL emission spectrum at 262°C for Ce:YAG 0.14%.
155 Figure 4.2.15 shows the glow curves for all as grown Ce:YAG samples. Note that the Ce:YAG 0.14% sample from Crytur is the only unpolished sample displayed and thus its glow curve also reflects surface defects, resulting in significantly more intense TL.
6000000 5500000 Ce:YAG 0.1% Ce:YAG 0.15% 5000000 Ce:YAG 0.2% 4500000 Ce:YAG 0.3% 4000000 Ce:YAG (ac) 3500000 3000000 2500000 2000000 1500000 1000000 500000 0 Integrated TL Intensity (counts) Intensity TL Integrated -500000 0 50 100 150 200 250 300 350 400 450 Temperature (C)
Figure 4.2.15 Glow curves of as grown Ce:YAG samples: Ce:YAG 0.1% grown in 5-30% H 2 in Ar,
Ce:YAG 0.15% grown in 5-30% H 2 in Ar, Ce:YAG 0.2%, grown in 5-30% H 2 in Ar, Ce:YAG 0.14%
grown in 40% H 2 in Ar, and Ce:YAG 0.3% grown in pure Ar.
The glow curve of Ce:YAG 0.1% has very low intensity, which is believed to arise from its high concentration of oxygen compared to other Ce:YAG samples, as hypothesized in Section 3.1. Similar to other Ce:YAG samples, it contains peaks at
70°C, 120°C, 170°C, 270°C, and 390°C. Ce:YAG 0.3% was grown in a pure argon environment while all other Ce:YAG samples presented in this figure were grown in an atmosphere containing a mixture of hydrogen and argon. The figure illustrates the effect of growth conditions on TL.
156 High temperature TL of annealed Ce:YAG
Several samples were annealed and remeasured. Figures 4.2.16 and 4.2.17 show the glow curves for Ce:YAG 0.3% and 0.15%, respectively, before and after annealing in air. Figure 4.2.18 shows the glow curves for Ce:YAG 0.2% as grown, after annealing in air, and after subsequently sputtering with aluminum and annealing in Ar.
1500000 Ce:YAG 0.3% Ce:YAG 0.3% annealed
1000000
500000
0 Integrated TL Intensity (counts) Intensity TL Integrated
0 50 100 150 200 250 300 350 400 450 Temperature (C)
Figure 4.2.16 Glow curves of Ce:YAG 0.3% before and after annealing in air using integration over the
wavelengths 570-720 nm.
In Fig. 4.2.16, the shoulder at 250°C and the peak at 325 nm in as grown Ce:YAG
0.3% appear to arise from oxygen vacancies, as they disappeared after annealing in air.
Other peaks remained relatively unchanged in temperature and intensity.
157
3500000 Ce:YAG 0.15% 3000000 Ce:YAG 0.15% annealed
2500000
2000000
1500000
1000000
500000
0 Integrated TL Intensity (counts) Intensity TL Integrated -500000 0 50 100 150 200 250 300 350 400 450 Temperature (C)
Figure 4.2.17 Glow curves of Ce:YAG 0.15% before and after annealing in air using integration over the
wavelengths 570-720 nm.
Ce:YAG 0.2% Ce:YAG 0.2% air annealed Ce:YAG 0.2% air annealed Al sp+Ar an 4000000
2000000
Integrated TL Intensity (counts) Intensity TL Integrated 0 0 50 100 150 200 250 300 350 400 450 Temperature (C)
Figure 4.2.18 Glow curves of Ce:YAG 0.2% as grown, after annealing in air, and after Al sputtering and
Ar annealing using integration over the wavelengths 570-720 nm.
158 The glow curve of Ce:YAG 0.15% does not appear to change much after annealing in air. The intensity actually increased for a couple of peaks. Interestingly, one of the peaks to increase after annealing in air is at a similar location to the shifted peak that arose in undoped YAG glow curves after the addition of oxygen to the crystal, suggesting a similarity in traps between these samples. However, the same peak disappears in the glow curve of Ce:YAG 0.2% after annealing in air. While it is possible that these peaks in different samples at the same temperature arise from different traps that coincidentally have similar depth, it is more likely that annealing affected the same trap differently in either sample. Oxidizing anneal on Ce:YAG can oxidize Ce3+ to Ce 4+ ,
oxidize other impurities, and/or fill oxygen vacancies.
The glow curve of Ce:YAG 0.2% exhibited a drastic decreased in intensity after
annealing in air. This is expected to arise from reduction of impurities as well as
incorporation of oxygen ions to fill vacancies. Annealing in oxygen fills V Os, and thus
removes electron traps. This can stabilize cerium in the trivalent state due to a balance of
surrounding charge while simultaneously trying to oxidize some Ce 3+ to Ce 4+ .
High temperature TL of Ce:YAG conclusions
Effects of annealing on Ce:YAG TL was inconsistent, possibly because oxygen
could not diffuse as easily into some samples as it did in others, perhaps due to hydrogen
impurities or some other defect. However, the majority of deep traps appear to be related
to oxygen vacancies, as oxidation resulted in a large decrease of most peaks and even
elimination of a peak around 350°C in Ce:YAG 0.2% and 0.3%.
159 Low temperature TL of Ce:YAG
Low-temperature TL measurements on most of our Ce:YAG samples were
covered by Mackay [Mackay-2011], therefore only newer low-temperature
measurements on annealed Ce:YAG 0.14% and the measurement on the as grown sample
previously reported by Mackay will be reported here. Activation energies are also
reported below for the as grown and air annealed samples. For consistence with data
presented in this thesis, I refit peaks reported by Mackay and recalculated the activation
energy for more direct comparison to the new measurements reported below. The low-
temperature glow curves for Ce:YAG as grown, after annealing in air, and after annealing
in vacuum are shown in Fig. 4.2.19.
20000000 Ce:YAG 0.14% Ce:YAG 0.14% annealed 18000000 Ce:YAG 0.14% vacuum annealed 16000000 14000000 12000000 10000000
8000000 6000000 4000000 2000000
Integrated TL Intensity (counts) Intensity TL Integrated 0 -200 -150 -100 -50 0 50 Temperature (C)
Figure 4.2.19 Glow curves of Ce:YAG 0.14% as grown, annealed in air, and annealed in vacuum using
470-720 nm integration.
The vacuum anneal did not appear to eliminate any peaks, it merely decreased the intensity of all of them. It is expected that vacuum anneal could remove anions from the
160 crystal and reduce or possibly even remove some cations, as required by charge compensation. After annealing, the sample turned from yellow to dark green and its luminescence properties reduced slightly. It is possible that charge transfer processes were less available to Ce 3+ ions after vacuum annealing, which might explain the
decreased luminescence properties and TL emission, but it is unclear exactly why the
crystal changes in this way after annealing in vacuum.
Annealing in air decreased the overall TL intensity and eliminated the large peak
at -75°C and the smaller peak at -10°C, suggesting these peak were related to V Os.
Elimination of the -75°C peak allowed for observation of two weak peaks at -100°C and -
90°C.
Ce:YAG defect level and frequency factor calculations
Activation energies for Ce:YAG 0.1%, Ce:YAG 0.15%, and Ce:YAG 0.2% are shown in Table 4.2.5. For comparison, activation energies for some peaks in Ce:YAG
0.2% calculated by the initial rise method and the corrected initial rise method proposed by Christodoulies [Christodoulides-1985] are presented in Table 4.2.6. Activation energies for as grown Ce:YAG 0.3%, air annealed Ce:YAG 0.3%, and unpolished
Ce:YAG 0.3% annealed in air and then in nitrogen are presented in Table 4.2.7. Table
4.2.8 reports calculated activation energies for Ce:YAG 0.14% before and after annealing in air.
161 Table 4.2.5 Calculated activation energies of Ce:YAG samples obtained from Marketech International:
Ce:YAG 0.1%, Ce:YAG 0.15%, and Ce:YAG 0.2%; for the glow curves constructed using 470-720 nm integration range.
Peak location Activation energy
(q=60°C/min) Ce:YAG 0.1% Ce:YAG 0.15% Ce:YAG 0.2%
70°C 0.99 ± 0.23 eV 0.73 ± 0.01 eV 0.97 ± 0.32 eV
120°C 1.31 ± 0.22 eV 1.45 ± 0.01 eV 1.18 ± 0.01 eV
170°C 1.88 ± 10.39 eV 1.88 ± 0.32 eV 0.91 ± 0.05 eV
270°C 2.37 ± 0.18 eV 2.30 ± 0.01 eV 1.89 ± 0.04 eV
340°C N/A 2.57 ± 0.03 eV 2.18 ± 0.04 eV
390°C 3.80 ± 1.73 eV N/A 2.66 ± 0.34 eV
Table 4.2.6 Calculated activation energies of Ce:YAG 0.2% calculated by the initial rise and corrected initial rise methods; for the glow curves constructed using 470-720 nm integration range.
Peak location Activation energy Frequency factor
(q=60°C/min) Initial rise Corrected IR
120°C 0.88 ± 0.06 eV 1.46 ± 0.05 eV 3.73 x 10 15 ± 1.70 x 10 15 s-1
270°C 1.15 ± 0.06 eV 1.83 ± 0.10 eV 2.86 x 10 17 ± 3.02 x 10 17 s-1
340°C 1.49 ± 0.07eV 1.97 ± 0.05 eV 3.12 x 10 17 ± 3.25 x 10 17 s-1
Table 4.2.6 shows that the correction to the initial rise method gave good agreement with results from the variable heating rates method. This agreement gives confidence in the results obtained using the methods of two and variable heating rates.
Only a few peaks could be analyzed using the initial rise method due to overlapping
162 peaks, demonstrating why the methods of two and variable heating rates were preferred for analysis in this chapter.
Table 4.2.7 Calculated activation energies of Ce:YAG 0.3% samples obtained from United Crystals: as grown (As grown), annealed in air (annealed in air), and fine ground unpolished annealed 96 hours in air followed by 48 hour anneal in nitrogen (fg annealed in N 2); for the glow curves constructed using 470-720 nm integration range.
Peak location Activation energy
(q=60°C/min) As grown Annealed in air fg annealed in N 2
120°C 1.36 ± 0.25 eV 1.72 ± 0.40 eV 1.46 ± 0.18 eV
170°C N/A 1.47 ± 0.53 eV 1.59 ± 0.70 eV
270°C 2.24 ± 0.70 eV 2.1 ± 0.02 eV 2.12 ± 0.02 eV
325°C 2.69 ± 0.20 eV N/A 2.42 ± 4.09 eV
390°C 2.71 ± 0.23 eV 2.73 ± 0.17 eV 2.91 ± 0.07 eV
Table 4.2.8 Calculated activation energies of Ce:YAG 0.14% samples obtained from Crytur, Ltd.: as grown and annealed in air; for the glow curves constructed using 470-720 nm integration range.
Peak location Activation energy
(q=60°C/min) As grown Annealed in air
-170°C 0.20 ± 0.01 eV 0.37 ± 0.03 eV
-140°C 0.27 ± 0.01 eV 0.39 ± 0.02 eV
-100°C N/A 0.38 ± 0.15 eV
-90°C N/A 0.67 ± 0.06 eV
-75°C 0.55 ± 0.01 eV N/A
163 -45°C 0.66 ± 0.01 eV 0.81 ± 0.01 eV
-10°C 0.90 ± 1.04 eV N/A
In Table 4.2.5, Ce:YAG 0.2% shows a lower activation energy for almost every peak. The activation energies for Ce:YAG 0.2% were calculated using the variable heating rate method, while all other Ce:YAG activation energies were calculated by the method of two heating rates. It is possible that the data recorded from a broader range of heating rates influenced the results to differ, and given that there is no other simple explanation, this is likely the cause. Trap depths were very similar across samples of different concentrations of cerium.
The peak at 325°C in Ce:YAG 0.3% is completely eliminated by annealing in air.
However, annealing in nitrogen seemed to make it weakly reappear. Calculated activation energies are similar to each other and the Marketech samples of Table 4.2.5.
Comparison of Table 4.2.5 to 4.2.7 shows that the activation energy associated with the peak at 70°C is the only peak that appears to be significantly affected by the growth atmosphere containing hydrogen.
Air annealing Ce:YAG 0.2% appeared to increase the depth of all traps. Given that this is systematic and that this effect is not observed in other samples, this is likely error.
Ce:YAG TL conclusion
Low temperature TL measurements on Ce:YAG showed much more intense luminescence than high temperature TL measurements, indicating that far more charge carrier traps are stable at low temperature than at high temperature. These shallow traps
164 can still trap charge carriers at room temperature, but they are unstable and the trapped charge carrier can escape very quickly.
High temperature TL demonstrated a high dependence on oxygen concentration, as seen with undoped YAG.
TL of other YAG samples
Nd:YAG, Tm:YAG, and Yb:YAG samples did not show any TL signal at low or high temperatures, even after annealing. It is unclear why this is, as Section 3.3 clearly showed that each of these samples can form stable charge carrier traps at room temperature.
4.3 REFERENCES
[Bagdasarov-1977] Kh.S. Bagdasarov, L.B. Pasternak, and B.K. Sevast’yanov,
3+ “Radiation color centers in Y 3Al 5O12 :Cr crystals,” Soviet Journal of Quantum
Elecronics 7 (1977) 965-968
[Batygov-1972] S.Kh. Batygov, Yu.K. Voron’ko, B.I. Denker, A.A. Maier, V.V. Osiko,
V.S. Radyukhin, and M.I. Timoshechkin, “Color centers in Y 3Al 5O12 crystals,” Soviet
Physics – Solid State 14 (1972) 839-841
[Bos-2001] A.J.J. Bos, “High sensitivity thermoluminescence dosimetry,” Nuclear
Instruments and Methods in Physics Research B 184 (2001) 3-28
[Braunlich-1967] P. Bräunlich, “Comment on the initial-rise method for determining trap depths,” Journal of Applied Physics 38 (1967) 2516-2519
165 [Chen-1981] R. Chen and Y. Kirsh, Analysis of thermally stimulated processes ,
International Series on the Science of the Solid State, Volume 15, Pergamon Press,
Oxford, England (1981)
[Choi-2001] Y.-S. Choi, “Saturation characteristics of Cr 4+ :YAG crystals and dye films for passive Q switches,” Applied Optics 40 (2001) 5417-5422
[Christodoulides-1985] C. Christodoulides, “Errors involved in the determination of activation energies in TL and TSDC by the initial rise method,” Journal of Physics D:
Applied Physics 18 (1985) 1665-1671
[DeVault-1983] D. DeVault, Govindjee, and W. Arnold, “Energetics of photosynthetic
glow peaks,” Proceedings of the National Academy of Science USA 80 (1983) 983-987
[Devor-1984] D.P. Devor, R.C. Pastor, and L.G. DeShazer, “Hydroxyl impurity effects in
YAG (Y 3Al 5O12 ),” Journal of Chemical Physics 81 (1984) 4104-4117
[Garlick-1948] G.F.J. Garlick and A.F. Gibson, “The electron trap mechanism of
luminescence in sulphide and silicate phosphors,” Proceedings of the Physical Society 60
(1948) 574-590
[Guerassimova-2002] N. Guerassimova, C. Dujardin, N. Garnier, C. Pédrini, A.G.
Petrosyan, I.A. Kamenskikh, V.V. Mikhailin, I.N. Shipinkov, D.A. Spassky, K.L.
Ovanesyan, G.O. Shirinyan, R. Chipaux, M. Cribier, J. Mallet, and J.-P. Meyer, “Charge- transfer luminescence and spectroscopic properties of Yb 3+ in aluminum and gallium garnets,” Nuclear Instruments and Methods in Physics Research A 486 (2002) 278-282
[Kitis-1998(a)] G. Kitis and J.W.N. Tuyn, “A simple method to correct for the
temperature lag in TL glow-curve measurements,” Journal of Physics D: Applied Physics
31 (1998) 2065-2073
166 [Kitis-1998(b)] G. Kitis, J.M. Gomez-Ros, and J.W.N. Tuyn, “Thermoluminescence glow-curve deconvolution functions for first, second and general order kinetics,” Journal of Physics D: Applied Physics 31 (1998) 2636-2641
[Kitis-2007] G.Kitis and V. Pagonis, “Peak shape methods for general order
thermoluminescence glow-peaks: A reappraisal,” Nuclear Instrumentation and Methods
in Physics Research B 262 (2007) 313-322
[Kovaleva-1981] N.S. Kovaleva, A.O. Ivanov, and É.P. Dubrovina, “Relationship between formation of radiation color centers and growth defects in YAG:Nd crystals,”
Soviet Journal of Quantum Electronics 11 (1981) 1485-1488
[Lyons1991] L. Lyons, A Practical Guide to Dana Analysis for Physical Science
Students , Cambridge University Press, New York, NY (1991)
[Mackay-2011] D. Mackay, “Low temperature thermoluminescence of cerium doped
YAG,” thesis, WSU, 2011
[Mackay-2012] D.T. Mackay, C.R. Varney, J. Buscher, and F.A. Selim, “Study of
exciton dynamics in garnets by low temperature thermo-luminescence,” Journal of
Applied Physics 112 (2012) 023522
[Mahesh-1989] K. Mahesh, P.S. Weng, and C. Furetta, Thermoluminescence in solids and its applications , Nuclear Technology Publishing, Ashford, Kent, England (1989)
[Milliken-2012] E.D. Milliken, L.C. Oliveira, G. Denis, and E.G. Yukihara, “Testing a
model-guided approach to the development of new thermoluminescent materials using
YAG:Ln produced by solution combustion synthesis,” Journal of Luminescence 132
(2012) 2495-2504
167 [Mori-1977] K. Mori, “Transient colour centres caused by UV light irradiation in yttrium aluminium garnet crystals,” Physica Status Solidi A 42 (1977) 375-384
[Murk-1995] V. Mürk and N Yaroshevich, “Exciton and recombination processes in
YAG crystals,” Journal of Physics: Condensed Matter 7 (1995) 5857-5864
[Pankove-1971] J.I. Pankove, Optical Processes in Semiconductors , Dover Publications,
Inc., New York, NY (1971)
[Randall-1945(a)] J.T. Randall and H.P. Wilkins, “The phosphorescence of various
solids,” Proceedings of the Royal Society A: Mathematical, Physical, & Engineering
Sciences 184 (1945) 347-364
[Randall-1945(b)] J.T. Randall and H.P. Wilkins, “Phosphorescence and electron traps I.
The study of trap distributions,” Proceedings of the Royal Society A: Mathematical,
Physical, & Engineering Sciences 184 (1945) 366-389
[Randall-1945(c)] J.T. Randall and H.P. Wilkins, “Phosphorescence and electron traps II.
The interpretation of long-period phosphorescence,” Proceedings of the Royal Society A:
Mathematical, Physical, & Engineering Sciences 184 (1945) 390-407
[Robbins-1979(b)] D.J. Robbins, B. Cockayne, J.L. Glasper, and B. Lent, “The
temperature dependence of rare-earth activated garnet phosphors I. Intensity and lifetime
measurements on undoped and Ce-doped Y 3Al 5O12 ,” Journal of the Electrochemical
Society 126 (1979) 1213-1220
[Selim-2007] F.A. Selim, D. Solodovnikov, M.H. Weber, and K.G. Lynn, “Identification of defects in Y 3Al 5O12 crystals by positron annihilation spectroscopy,” Applied Physics
Letters 91 (2007) 104105
168 [Shimony-1995] Y. Shimony, Y. Kalisky, and B.H.T Chai, “Quantitative studies of
Cr 4+ :YAG as a saturable absorber for Nd:YAG laser,” Optical Materials 4 (1995) 547-
551
[Smol’skaya-1987] L.P. Smol’skaya, E.F. Martynovich, A.G. Davydchenko, and S.A.
Smirnova, “X-ray and thermally stimulated luminescence in YAG,” Journal of Applied
Spectroscopy 46 (1987) 44-46
[Solodovnikov-2008] D. Solodovnikov, M.H. Weber, and K.G. Lynn, “Improvement in scintillation performance of Ce, Er codoped yttrium aluminum garnet crystals by means of a postgrowth treatment,” Applied Physics Letters 93 (2008) 104102
[Varney-2012(b)] C.R. Varney, D.T. Mackay, A. Pratt, S.M. Reda, and F.A. Selim,
“Energy levels of exciton traps in yttrium aluminum garnet single crystals,” Journal of
Applied Physics 111 (2012) 063505
[Vedda-2004] A. Vedda, D. Di Martino, M. Martini, J. Mares, E. Mihokova, M. Nikl, N.
Solovieva, K. Blazek, and K. Nejezchleb, “Trap levels in Y-aluminum garnet scintillating crystals,” Radiation Measurements 38 (2004) 673-676
[Zorenko-2012] Y. Zorenko, T. Zorenko, V.V. Gorbenko, T. Voznyak, V. Savchyn, P.
Bilski, and A. Twardak, “Peularities of luminescent and scintillation properties of
YAG:Ce phosphor prepared in different crystalline forms,” Optical Materials 34 (2012)
1314-1319
[Zych-2000] E. Zych, C. Brecher, and J. Glodo, “Kinetics of cerium emission in a
YAG:Ce single crystal: the role of traps,” Journal of Physics: Condensed Matter 12
(2000) 1947-1958
169 CHAPTER FIVE
DEFECT IDENTIFICATION BY POSITRON LIFETIME MEASUREMENTS
Introduction
Thus far in this thesis, arguments regarding defect structure have provided some insight but have been largely speculative. In Chapter 4, charge carrier traps were identified and investigated in terms of their potential depth and the types of traps were discussed. This chapter identifies defect structures present in YAG crystals measured by positron annihilation. Positron interaction with matter is primarily characterized by
Coulombic interactions and culminates with an electron annihilating with the positron, its antiparticle, which results in the emission of two 511 keV γ rays in opposite directions to conserve momentum.
Experiments on positron annihilation in solids can be conducted in three independent ways: angular correlation, Doppler broadening, and lifetime measurements
[Hautojarvi-1979, Krause-Rehberg-1999, Schultz-1988, Siegel-1980, West-1974], all of which are excellent methods by which to probe defect structure. Whereas angular correlation and Doppler broadening measurements both measure subtle changes in the
511 keV γ rays that arise from subtle differences in momenta of annihilating electrons and positrons that can be used to identify defects, positron annihilation lifetime spectroscopy (PALS), the focus of this chapter, is the measurement of the time a positron exists within a sample before annihilation.
Positron annihilation measurements can unambiguously identify open-volume defects as long as they can trap positrons. Thus, this is a reliable method by which to
170 identify such defects as cation vacancies and vacancy clusters. Positron annihilation measurements are very sensitive; measurements conducted at room temperature typically have a sensitivity range for negatively charged vacancies of 5 x 10 15 to 5 x 10 18 cm -3 concentration [Krause-Rehberg-1999].
The most common positron source in use today and the source used in this work is 22 Na [Hautojarvi-1979, Krause-Rehberg-1999, Saoucha-1999, Schultz-1988, Siegel-
1980, West-1974], which decays to 22 Ne with a half-life of 2.6 years [Melissinos-1966]
22 22 + + according to the reaction Na → Ne + β + νe + γ, where β is a positron, νe is a neutrino, and γ is a γ photon of energy 1.274 MeV corresponding to the difference between the excited and ground states of 22 Ne. The excited state of 22 Ne decays to the ground state after 3.7 ps, thus the 1.274 MeV γ ray is considered coincident with creation of a positron. The long half-life of 22 Na provides consistent level of radioactivity for a
series of experiments. 22 Na decay gives a high positron yield of on average 0.9 per
decay.
Positrons generated from decay of the 22 Na source enter a sample with an energy distribution up to 0.54 MeV [Krause-Rehberg-1999]. They quickly (~10 ps) lose most of their energy due to collisions with electrons and ions, a process referred to as thermalization. Once thermal energy is attained, the positrons start to diffuse through the sample. Due to Coulombic interactions, positrons are repelled by nuclei, causing their largest position probability density to lie in interstitial regions. The positron diffusion length L is a function of the diffusion constant D + and the lifetime of positrons in the
defect-free bulk τB [Krause-Rehberg-1999]
L = τ B D+ (5.0.1)
171 with the diffusion constant given by
k T D = τ B (5.0.2) + t m * where τt is the relaxation time of the dominant scattering mechanism and m* is the
effective positron mass. The lifetime of a positron is dependent on the electron density at
the annihilation site. The reciprocal of the positron lifetime τ, the annihilation rate λ, is
given by the overlap integral of electron and positron densities, given by n -(r) and n +(r), respectively [Krause-Rehberg-1999, Sigel-1980]
1 2 λ = = π r 2 c ψ + ()r n ()r γ dr (5.0.3) τ 0 ∫ − where r 0 is the classical electron radius, c is the speed of light, r is the position vector,
+ 2 and n +(r) = | ψ (r)| relates the positron density and wavefunction. The correlation
function γ = 1 + ∆n-/n - describes the increase in electron density as a result of the
Coulombic attraction to a positron, an effect referred to as enhancement.
After a period of time characteristic of the sample, the positron annihilates with an
electron. The positron’s lifetime can be lengthened by trapping at a defect such as a
vacancy, where local charge distribution differs from the bulk of the crystal in such a way
that the Coulombic repulsion between ion nuclei and the positron is locally reduced. The
redistribution of electrons at certain defects can create a negative electrostatic potential,
thus attracting positrons. Cation vacancies are negatively charged defects that can attract
positrons and anion vacancies are positively charged and tend to repel positrons. As a
result, PALS can provide information about the cation vacancies present in a crystal but
typically does not reveal information regarding anion vacancies. PALS can provide
information about the size and concentration of negatively charged defects.
172 It is also possible for a positron and electron to form a bound state, referred to as positronium [Consolati-2001, Hautojarvi-1979, Krause-Rehberg-1999, Schultz-1988,
Siegel-1980, West-1974]. Positronium can exist in the states parapositronium or orthopositronium, with average free space lifetimes of 125 ps and 140 ns, respectively.
Parapositronium is a singlet state with antiparallel spins that decays into two photons of
511 keV each. Orthopositronium is a triplet state with parallel spins that decays into three photons to conserve spin, since 2 γ-annihilation is forbidden by selection rules, with a sum equal to the total rest mass energy 1.022 MeV [Hautojarvi-1979, Krause-Rehberg-
1999, Schultz-1988, Siegel-1980, West-1974]. The lifetime of orthopositronium in a medium is shortened by “pick-off” annihilation, which is where the positron annihilates with a third electron that has opposite spin. Positronium formation typically occurs in large open spaces and is common in polymers where vacancy-like defects can be very large. In our measurements, if the samples do not completely cover the positron source, it is possible that some positrons will go out into the air and can form positronium there, but more often orthopositronium forms on the surface interface of the samples, which annihilates by pick-off annihilation.
It should be emphasized that previous work directly pertaining to this study is lacking. PALS measurements have been extensively reported in metals and semiconductors [e.g. Hautojarvi-1979, Krause-Rehberg-1999, Schultz-1988, Siegel-1980,
West-1974] but very rarely in complex oxides [Mackie-2009] and never before in YAG.
To my knowledge, this is the first publication on PALS results for YAG and one of very few works in PALS measurements conducted on complex oxides. The findings reported
173 here are important to all complex oxides because they can unambiguously identify defects that may be common in other complex oxides.
5.1 METHODS
Experimental setup
A PALS measurement employs a sandwich configuration, shown in Fig. 5.1.1.
The purpose of the sandwich configuration is to reduce noise and ensure that positrons enter into the sample. The 22 Na source was deposited onto a thin foil, which was folded over to help shield the samples from 22 Na and provide a more uniform source foil. The foil used for these experiments was 8 µm thick and made of Kapton, which has been
commonly used and its contributions to positron lifetime measurements have been
studied extensively both theoretically and experimentally [McGuire-2006(a), McGuire-
2006(b), Monge-1994, Saoucha-1999]. The foil was placed between two of the sample
being studied, which were sandwiched between two BaF 2 scintillators and Hamamatsu
H3177 photomultiplier tubes (PMTs). Due to the geometry, PALS requires two identical samples for measurements, creating a 4π geometry surrounding the positron source such
that every positron emitted that escapes the foil will annihilate within the sample being
studied. To this end, the samples being studied must be thick enough to stop all incident
positrons. For YAG the positron implantation depth is of the order of magnitude of 100
µm [Dryzek-2006, Scultz-1988], an order or magnitude shorter than the thickness of all samples studied in this work. A small piece of tape held the two samples together for ease of placement between the detectors. The BaF 2 scintillator crystals are fast scintillators with high γ-ray detection efficiency [Rajainmaki-1987, Wei-2004].
174
Figure 5.1.1 Schematic diagram of the sandwich configuration setup for PALS measurements. The
positron source is sandwiched between two of the same sample which are sandwiched between detectors.
A schematic diagram of the setup used for our PALS measurements is shown in
Fig. 5.1.2. Both PMTs were powered by a Power Designs Inc. NIM Standard HV Power
Supply set to -2300 V. Each PMT was connected to an Ortec 583 Constant Fraction
Differential Discriminator (CFDD), which only allows the signal of a user-defined range of energies to pass further along the electronics and blocks all other energies. One CFDD was set to a window around 1.274 MeV, the energy of the photon that is very nearly coincident with positron emission in 22 Na [Hautojarvi-1979, Krause-Rehberg-1999,
Saoucha-1999, Schultz-1988, Siegel-1980, Staab-2000, West-1974], and the other CFDD was set to a window around 511 keV [Hautojarvi-1979, Krause-Rehberg-1999, Saoucha-
1999, Schultz-1988, Siegel-1980, Staab-2000, West-1974], the energy characteristic of electron-positron annihilation. The 1.274 MeV γ ray provides the start signal and the
0.511 MeV γ ray provides the stop signal for recording the lifetime of a positron. Each
CFDD used two outputs: one to an Ortec 414A Fast Coincidence and one to an Ortec
Model DB463 Delay Box. In the Delay Box, the signals were delayed by different
amounts of time to give a cleaner readout on the computer. Different settings were used
175 throughout our measurements, but for the most recent measurements the start signal was delayed 63 ns and the stop signal was delayed 95 ns. After delay, both the start and stop signals were sent to the START and STOP inputs on an Ortec 567 Time-Amplitude
Converter (TAC), set to the coincidence setting between signals. The signal from the
CFDD to the Fast Coincidence gated the TAC. The TAC only output data if the start and stop signal were coincident with each other within 50 ns, as per the settings of the Fast
Coincidence. The TAC converts the duration of time between the two measurements to voltage amplitude. The TAC was set to a data range of 100 ns. This voltage amplitude signal was sent to an Ortec 926 ADCAM Multichannel Buffer (MCB), where the data was compiled into a histogram that was then displayed on the computer connected to the
MCB. The computer software used for these experiments was MAESTRO (ORTEC), which was set to display 8192 channels. The time calibration over the range of channels used for current measurements was determined to be 12.3 ps/channel. Time calibration measurements were performed by sending the output of both CFDDs through to the
Delay Box as both the start and stop signal and then setting different delay settings and recording the channel of output. All electronics except for the HV power supply and computer were powered by an Ortec 4001A or 4001C BIN power supply.
176
Figure 5.1.2 Schematic diagram of the setup for PALS measurements. The signal originates with positron
creation in the 22 Na source (sandwiched between two identical sandwiches) and annihilation in the samples
and then travels through the devices as shown.
The setup used for PALS measurements was optimized for the best energy resolution of the 1.274 MeV and 0.511 MeV photopeaks. The power supply was adjusted to minimize scintillator and PMT signal rise time, measured using a Tektronix
TDS 640 oscilloscope. The output of each CFDD was observed using the same TDS 640 oscilloscope and the upper and lower limits of the windows were adjusted to create a small window around 0.511 or 1.274 MeV. The CFDD windows were set to 1.05-1.56 for the start signal and 0.52-0.85 for the stop signal, which do not correspond directly to
MeV. Delay times were adjusted so that the lifetime spectra were centrally located on the
ADC readout. The Fast Coincidence window of 50 ns was chosen because it produced an
177 acceptable signal-to-noise ratio. Table 5.1.1 summarizes the settings of the electronics used in these experiments.
Table 5.1.1 Summary of equipment settings used for PALS experiments.
Equipment Settings
Power Designs Inc. NIM Standard HV Power Supply Voltage: -2300 V
H3177 PMTs BaF 2 scintillators
Ortec 583 CFDD Limits (not MeV):
1.05-1.56 (1)
0.52-0.85 (2)
Ortec 414A Fast Coincidence 50 ns
Ortec Model DB463 Delay Box 63 ns (start signal)
95 ns (stop signal)
Time-Amplitude Converter Data range: 100 ns
Coincident gating
Ortec 926 ADCAM Multichannel Buffer/MAESTRO software 8192 channels
The source foil was prepared by depositing aqueous 22 NaCl onto a flat sheet of
Kapton foil, usually one or two drops at a time to reduce spot size. The foil was then allowed to dry in air for a few days within an enclosed container for protection. After the
22 NaCl solution dries, another drop was added and the process is repeated until the source
reached the desired strength. Then the Kapton foil was carefully folded over with the
source located near the crease. This was important for alignment of the source between
178 samples. Spot sizes for positron sources used varied slightly but were typically approximately 3 mm in diameter.
Fitting
Analysis of the spectra was performed using PATFIT-88, and a couple of runs were also fit using the program MELT for comparison. The resolution function was fit for each run using RESOLUTIONFIT prior to positron lifetime fitting and it was found to vary slightly between runs due to drift in the electronics. The resolution function was found to fit to three Gaussians, which is typical of PALS measurements with BaF 2 detectors [Kirkegaard-1989]. Though it varied slightly between measurements, a typical resolution function used is 0.28 ns FWHM (85% intensity), 0.32 ns FWHM (5% intensity, -0.08 ns shift relative to the largest peak), 0.22 ns FWHM (10% intensity, 0.12 ns shift relative to the largest peak). The resolution function determined using
RESOLUTIONFIT was then input as fixed parameters into the positron lifetime fitting routine POSITRONFIT, which fits the lifetime data to decaying exponentials
[Kirkegaard-1989]. A maximum of two lifetimes characteristic of the sample and one longer ~2 ns lifetime could be resolved in measurements conducted on our samples due to low statistics. The long-lived component is the pick-off lifetime of positronium that occurs at the surface of the samples and will not be considered in discussion [Bernal-
1995, Kirkegaard-1989, Krause-Rehberg-1999, Schultz-1988, Siegel-1980, West-1974].
The model typically used for positron fitting assumes that the lifetime spectrum closely resembles an exponential decay [Chitambo-2002, Hautojarvi-1979, Kirkegaard-
1989, Krause-Rehberg-1999, Ranki-2004, Schultz-1988, Siegel-1980, Tuomisto-2003,
West-1974], which may not always be the case. In samples that are usually studied with
179 PALS (e.g. metals, semiconductors, etc), this assumption works because the crystal structure is simple. In YAG and other complex oxides, the crystal structure is very complex and nonuniform by comparison. As mentioned above, a unit cell of YAG contains 160 atoms with different distances between different bonds [Dong-1991, Geller-
1967, Landron-1996, Stanek-2008, Xu-1999].
The lifetimes of positrons in NaCl crystals and Kapton have been extensively studied and are found to be about 430 and 382 ps [Djourelov-1996, McGuire-2006(a),
McGuire-2006(b), Staab-1996], respectively. Kapton was chosen as the foil material because its long lifetime is easy to distinguish above the much shorter lifetimes characteristic of single crystal samples and can thus be easily subtracted. The intensity of the 382 ps lifetime in Kapton was found experimentally to be 8.5% and the intensity of the positron lifetime in NaCl was determined to be a function of source strength and was typically near 1%. The intensities of positron lifetimes in Kapton and NaCl as a function of Kapton thickness and sample properties have been studied theoretically and experimentally [McGuire-2006(a), McGuire-2006(b), Monge-1994, Saoucha-1999,
Staab-1996]. The above stated intensities match intensities in previous experimental work conducted on complex oxides [Mackie-2009] but do not match theoretical calculated intensities. Once these constant correction terms were determined they were subtracted from lifetime spectra during fitting to allow for easier fitting of the lifetimes characteristic to the samples.
Aluminum annealed in nitrogen was run as a diagnostic test whenever necessary, as its lifetime has been extensively studied and is well known. Al annealed in a nitrogen atmosphere contains no vacancies due to its low annealing temperature and simple
180 structure, giving one lifetime component of around 160 ps [Calloni-2005, Djourelov-
1996, Ferragut-1998, Monge-1994, Saoucha-1999, Staab-1996]. Annealing Al in an atmosphere containing oxygen forms Al 2O3, hence why it is annealed in nitrogen.
Analysis
When two lifetimes are resolved, the second lifetime τ2 and its intensity I 2 are
characteristic of crystal structure defects and bulk lifetime τB is characteristic of the defect-free bulk of the crystal. τ1, known as the reduced lifetime, and its intensity I 1 are
reduced to be in contrast to the defect and bulk lifetimes such that the relation in Eq. 5.1.1
holds. The bulk lifetime can be calculated from measured values τ1, τ2, I 1, and I 2 by
[Hautojarvi-1979, Krause-Rehberg-1999, Mackie-2009, McGuire-2006(b), Schultz-1988,
Siegel-1980, West-1974]
−1 I I 1 2 τ B = + , (5.1.1) τ 1 τ 2
where I i is given as a ratio of the total intensity and I 1 + I 2 = 1. Another useful quantity to
calculate is the average lifetime, which is a simple average calculation weighted by
intensity [Hautojarvi-1979, Krause-1990, Krause-Rehberg-1999, Mackie-2009, McGuire-
2006(b), Polity-1997(a), Schultz-1988, Siegel-1980, Somieski-1996, Staab-2000,
Tuomisto-2003, West-1974]
τ AV = I1τ 1 + I 2τ 2 . (5.1.2)
A general rule is if τAV > τB then vacancy-type defects are present. Uncertainties of τB and τAV calculated by partial derivatives are shown in Eqs. 5.1.3 and 5.1.4 respectively.
181 2/1 2 2 dI dI I dτ I dτ 2 1 2 1 1 2 2 dτ B = τ B + + 2 + 2 , (5.1.3) τ τ τ τ 1 2 1 2
2 2 2 2 2/1 dτ AV = [()()()()I1dτ 1 + I 2 dτ 2 + τ 1dI 1 + τ 2 dI 2 ] . (5.1.4)
It is also possible to calculate the trapping rate at defects κd, proportional to the
defect concentration, from the defect lifetime τ2 [Hautojarvi-1979, Krause-Rehberg-1999,
Mackie-2009, Polity-1997(a), West-1974, Zhang-2004]
1 1 κ d = I 2 − (5.1.5) τ 1 τ 2 with uncertainty, calculated from partial derivatives,
2/1 2 2 2 1 1 I I 2 2 dκ d = − dI 2 + 2 dτ 1 + 2 dτ 2 . (5.1.6) τ τ τ τ 1 2 1 2
The quantity κd allows for estimation of the concentration of defects C, given by
κ d = µC (5.1.7) where µ is known as the positron trapping coefficient, a constant for a given defect determined by independent studies. For this work, we used the typical estimate at room temperature for µ of 10 15 s -1 for negatively charged monovacancies and 10 14 s -1 for neutral monovacancies [Krause-Rehberg-1999, Polity-1997]. Positively charged monovacancies do not trap positrons at room temperature and are not considered in the analysis of these measurements. The approximations for negatively charged and neutral monovacancies were formed based on positron annihilation in semiconductors and are expected to apply to YAG to order of magnitude. According to Krause-Rehberg [Krause-
Rehberg-1999], vacancy clusters are accounted for by µ cluster = i µv, where µ v is the
182 positron trapping coefficient at a monovacancy and i is the number of vacancies.
Calculations in this chapter assume this is applicable to YAG and void trapping rates reflect small adjustments. Normally this applies to vacancies of the same ion, however in
YAG vacancy clusters usually involve both cations and anions. Conversion to more workable units is accomplished by
−1 3 −1 µ[s ] µ[]cm s = (5.1.8) N at where N at is the number of atoms per unit volume. In YAG, using the density of 4.55 g
cm -3 and the molar mass of 593.616, I calculate the number densities of each constituent
22 -3 22 -3 22 -3 to be NY = 1.38 x 10 cm , N Al = 2.31 x 10 cm , and N O = 5.54 x 10 cm , giving a total atomic concentration of N = 9.23 x 10 22 cm -3. In YAG, cation vacancies are
expected to be Al. Studies on stoichiometry have shown that YAG crystals tend to be
nonstoichiometric with excess Y despite stoichiometric amounts of ingredients [Ashurov-
1977, Geller-1967] and formation energy calculations show that this nonstoichiometry is
3+ accounted for by Y Al ADs and Al vacancies [Kuklja-1999]. Also, an Al vacancy has a
higher formation energy compared to a Y vacancy (53 eV to 49 eV according to Kuklja
and Pandey [Kuklja-1999]), which implies that Al deficiency results from difficulty of
incorporation of Al during crystal growth. An Al vacancy is a negatively charged
vacancy due to the missing Al 3+ ion surrounded by negatively charged oxygen ions and
15 7 -3 thus is calculated to have a concentration C = κd/µ = N Al κd/10 = 2 x 10 κd s cm , calculated from the above number density of Al. Vacancy clusters containing an aluminum vacancy and at least one oxygen vacancy, henceforth denoted as Al+ nO
(where n is an integer ≥1), are often formed together as a defect aggregate for charge compensation and are assumed in this thesis to be neutral, balanced by neighboring
183 9 -3 impurities such as hydrogen, with defect concentration of about C = 10 κd s cm . After
oxidation, however, electrons can be added to its surroundings or an oxygen ion may
even diffuse into the vacancy cluster, and thus Al+nO clusters are assumed to inherit a
8 -3 negative charge state after oxidation, with defect concentration of C = 10 κd s cm .
Hydrogen impurities in aluminum vacancies, denoted as Al+H I, are treated the same at Al
vacancies.
PALS results and calculations pertaining to defects, viz. τ2, its intensity, bulk and average lifetimes, κd, and defect concentration estimates based on the defect most likely to lead to the observed trapping, were averaged over all measurements using a weighted average based on uncertainty [Lyons-1991]
x ∑ i i dx 2 x = i (5.1.9) 1 ∑i 2 dx i
1 dx 2 = . (5.1.10) 1 ∑i 2 dx i
All individual PALS results and calculations, viz. τ1, τ2, their intensities, bulk and
average lifetimes, κd, and defect concentration estimates based on the defect most likely to lead to the observed trapping are tabulated in Appendix A. For clarity, figures demonstrating the data fits are included with some data in this section.
Typically, at least two million counts in the peak were desired for good statistics
[Consolati-1998, Hautojarvi-1979, Kirkegaard-1989, Krause-Rehberg-1999, West-1974].
184 5.2 RESULTS
Overview of presentation of results
PALS results are tabulated by sample, showing relevant results averaged over multiple measurements. In Appendix A, results of each individual positron lifetime measurement are presented, usually entailing multiple measurements conducted on the same sample to verify consistent and repeatable results. The results presented in this chapter are the averages of those series of measurements. Shown in the tables included in this section are the defect lifetimes and intensities (left blank if there is no defect lifetime), the bulk and average lifetimes, defect trapping rate, the type of positron- trapping vacancy I hypothesized to exist for calculation of defect concentration, and approximate defect concentration. Lifetimes are given in nanoseconds, intensities are given as a percent of the total amount of observed positron annihilation events within the
-1 crystal (I 1 + I 2 = 100%), defect trapping rate κd is given in ns , and defect concentrations are given in cm -3.
Positron penetration depth in YAG is of the order of magnitude of 100 µm
[Dryzek-2006, Scultz-1988]. Therefore, surface conditions are not expected to significantly affect PALS measurements and lifetimes found for unpolished samples can thus be compared side-by-side with lifetimes found for polished samples. This is different from all measurements in previous chapters, where it was advantageous for the sample to be well polished.
Undoped YAG
Table 5.2.1 presents data for polished Ar-grown undoped YAG 5x5x1 mm. Table
5.2.2 presents data for unpolished (fine ground) Ar-grown undoped YAG 10x10x1 mm,
185 also as grown. Table 5.2.3 shows data for unpolished (fine ground) Ar-grown undoped
YAG 5x5x1, which includes results for both as grown and air annealed samples.
Table 5.2.1 PALS results for polished Ar-grown undoped YAG 5x5x1 mm as grown.
-1 -3 τ2 [ns] (I 2) τB [ns] τAV [ns] κd [ns ] Defect C [cm ]
As grown
0.2931 ± 0.0032 0.1671 ± 0.1742 ± 0.38 ± Al+nO 3.7 x 10 17 ±
(12.89 ± 0.55%) 0.0012 0.0019 0.02 2 x 10 16
Table 5.2.2 PALS results for unpolished (fine ground) Ar-grown undoped YAG 10x10x1 mm as grown.
-1 -3 τ2 [ns] (I 2) τB [ns] τAV [ns] κd [ns ] Defect C [cm ]
As grown
0.3091 ± 0.0066 0.1669 ± 0.1742 ± 0.23 ± Al+nO 2.3 x 10 17 ±
(7.38 ± 0.62%) 0.0013 0.0023 0.02 2 ± 10 16
Table 5.2.3 PALS results for unpolished (fine ground) Ar-grown undoped YAG 5x5x1 mm as grown and annealed in air.
-1 -3 τ2 [ns] (I 2) τB [ns] τAV [ns] κd [ns ] Defect C [cm ]
As grown
0.3031 ± 0.0046 0.1661 ± 0.1731 ± 0.33 ± Al+nO 3.3 x 10 17 ±
(10.68 ± 0.62%) 0.0014 0.0022 0.02 2 x 10 16
Annealed in air 96 hours at 1200°C
186 0.2691 ± 0.0037 0.1699 ± 0.1759 ± 0.38 ± Al+nO 3.8 x 10 17 ±
(14.71 ± 0.93%) 0.0021 0.0030 0.03 3 x 10 16
Tables 5.2.13 show good agreement in lifetime values for the as grown samples, but the 10x10x1 mm sample shows less defect trapping. After annealing, the defect lifetime decreases while its intensity increases, potentially due to decreased defect size and changed charge state. Calculation of C does not reveal an increase in concentration, remaining near 1 part per million (ppm), instead the vacancies present in the crystal changed in size and charge state such that positrons can more easily trap at them and then annihilate quicker with an electron. An explanation is that this defect is a vacancy cluster
3+ containing both anion and cation vacancies. It has been suggested that Y Al ADs can
evoke large vacancy cluster defects by ejecting adjacent oxygen ions due to its size and a
nearby aluminum ion as charge compensation [Babin-2011, Selim-2007, Springis-1991,
Zorenko-2010(a)]. This model fits the results presented in Tables 5.2.1-3, as annealing in
air could potentially fill one or more of the oxygen vacancies, decreasing the size of the
vacancy while giving it a more negative charge, effecting the observed changes.
I chose Al+nO as the defect after annealing in Table 5.2.3 because the lifetime is
still large. The lifetime is shorter, which likely reflects a decrease in defect size. This
may result from, for instance, an oxygen ion diffusing into Al+2O clusters to form Al+O.
3+ Charge compensation and localized Y Al antisite defects (ADs) may make it difficult for a second oxygen ion to diffuse into the vacancy cluster, which stops it from going all the way to a single Al vacancy.
187 For reference, Fig. 5.2.1 shows the fit of one of the runs on the polished Ar-grown sample.
100000
Data Fit 10000
1000
Counts
100
10 0.0 0.5 1.0 1.5 2.0 Lifetime (ns)
Figure 5.2.1 PALS fit for a measurement conducted on polished Ar-grown YAG 5x5x1 fit to two lifetimes.
Tables 5.2.4 and 5.2.5 show results for H 2- and O 2-grown YAG, respectively.
Different treatments were applied to each sample to observe their effects on defects using
PALS. Both samples underwent Al sputtering and Ar annealing treatment, although the
H2-grown sample was annealed in air before this treatment in an attempt to fill oxygen vacancies to make it easier for the deposited aluminum ions to incorporate into the sample to fill Al vacancies. For these measurements, the Al sputtered side of both samples was placed closest to the 22 Na source in the sandwich configuration as an attempt
to maximize the likelihood that any effects of the treatment could be observed.
188 Table 5.2.4 PALS results for H2-grown undoped YAG 10 mm dia. x ~1 mm as grown, annealed in air, and
subsequently sputtered with Al and annealed in argon.
-1 -3 τ2 [ns] (I 2) τB [ns] τAV [ns] κd [ns ] Defect C [cm ]
As grown
0 0.1492 ± 0.1492 ± 0 (none) 0
0.0001 0.0001
Annealed in air 48 hours at 1200°C
0.2356 ± 0.0389 0.1509 ± 0.1529 ± 0.16 ± Al 3.3 x 10 15
(6.38 ± 0.0389%) 0.0103 0.0155 0.15 2.9 x 10 15
0 0.1521 ± 0.1521 ± 0 (none) 0
0.0002 0.0002
After above anneal in air, sputtered with Al and annealed in Ar 48 hours at 1200°C
0 0.1463 ± 0.1463 ± 0 (none)
0.0001 0.0001
Table 5.2.5 PALS results for O2-grown undoped YAG 10 mm dia. x ~1 mm as grown, annealed in a mixed
atmosphere consisting of (~50%) H 2 in Ar, further annealed in a mixed atmosphere consisting of (~50%)
H2 in Ar, further annealed in Ar, further annealed in Ar, and sputtered with Al and annealed in argon.
-1 -3 τ2 [ns] (I 2) τB [ns] τAV [ns] κd [ns ] Defect C [cm ]
As grown
15 0.2097 ± 0.0063 0.1596 ± 0.1633 ± 0.41 ± Al, Al+H I 8.2 x 10 ±
(18.92 ± 3.71%) 0.0078 0.0102 0.09 1.8 x 10 15
Annealed in a mixed atmosphere of (~50%) H2 in Ar 24 hours at 1200°C
189 16 0.1747 ± 0.0054 0.1584 ± 0.1612 ± 0.96 ± Al, Al+H I 1.9 x 10 ±
(59.90 ± 14.52%) 0.0343 0.0328 0.41 8 x 10 15
Further annealed in a mixed atmosphere of (~50%) H2 in Ar 51.5 hours at 1200°C
16 0.1872 ± 0.0042 0.1575 ± 0.1626 ± 1.24 ± Al, Al+H I 2.5 x 10 ±
(55.64 ± 7.83%) 0.0185 0.0182 0.26 5 x 10 15
Further annealed in a Ar 19 hours at 1200°C
16 0.1998 ± 0.0066 0.1581 ± 0.1629 ± 0.68 ± Al, Al+H I 1.4 x 10 ±
(33.00 ± 7.39 %) .0161 0.0186 0.18 4 x 10 15
Further annealed in a Ar 48 hours at 1200°C
16 0.1927 ± 0.0057 0.1561 ± 0.1608 ± 0.91 ± Al, Al+H I 1.8 x 10 ±
(43.92 ± 8.61%) 0.0193 0.0206 0.23 5 x 10 15
Sputtered with Al and annealed in Ar 48 hours at 1200°C
15 0.1984 ± 0.0064 0.1550 ± 0.1581 ± 0.39 ± Al, Al+H I 7.8 x 10 ±
(17.23 ± 4.14%) 0.0084 0.0112 0.10 2.1 x 10 15
Both Tables 5.2.4 and 5.2.5 show shorter defect lifetimes (when present) than Ar- grown YAG showed in Tables 5.2.1-3. It is apparent that growth in an atmosphere containing oxygen helps fill oxygen vacancies. It could likewise be surmised that growth in hydrogen may lead to the inclusion of hydrogen ions in the crystal lattice. As mentioned in Chapter 4, IR absorption measurements confirmed the presence of hydrogen in both samples. As will be discussed below, these factors cause a significant difference in PALS results of measurements conducted on YAG grown in different atmospheres.
190 In Table 5.2.4, it appears that there are no defects capable of trapping positrons
present in the as grown H 2-grown sample. However, Wavelength dispersive x-ray spectroscopy (WDS) measurements a similar concentration of Al in this sample to other undoped YAG samples, meaning Al vacancies, and thus positron-trapping defects, should be just as prevalent in this sample as in others. A second defect lifetime appeared after annealing the sample in air, although it reflects a very low defect concentration of less than 0.1 ppm. IR absorption measurements showed different results for hydrogen absorption in H 2- and O2-grown YAG, resulting from the great difference hydrogen
concentrations between the two samples [Devor-1984]. The very high hydrogen
concentration may lead to different incorporation of hydrogen in the H2-grown as it
incorporates in the O 2-grown sample. It is possible that hydrogen incorporates in H 2-
grown undoped YAG in such a way that it fills negatively-charged and neutral vacancies,
giving them a positive charge, thus these vacancies and vacancy clusters cannot trap
positrons at room temperature. This also explains the lack of color centers (Section 3.3)
or TL (Chapter 4) in the as grown samples and only weak TL from shallow traps after
annealing in air diffused oxygen ions into the lattice, changing the charge state of these
otherwise neutral or positive vacancy clusters. Sputtering with aluminum and annealing
in argon did not appear to significantly change the positron lifetime.
After annealing in air, the second the second lifetime of the annealed H 2-grown
sample initially shows a second lifetime which disappears in subsequent measurements.
The most likely explanation for this arises from the concentration of these defects. As
seen in Table 5.2.4, after annealing there is a very low concentration of ~10 15 cm -3, which
191 is at the lower limit of sensitivity of PALS. It is likely that this defect lifetime simply could not be resolved consistently due to the very low defect concentration.
Table 5.2.5 shows that there exist vacancy structures present in O 2-grown YAG.
The shorter defect lifetime with high intensity shows that these defects are possibly in the form of single Al vacancies, and they appear to arise in concentrations of about 0.1 ppm, rather low compared to the defects found in Ar-grown samples. Based on the data from individual measurements, presented in Appendix A, it appears that there may be two defects present in this sample, but they can not be independently resolved due to the statistics of the measurements and how close they are to each other, hence the middle lifetime gets coupled somehow to the defect or reduced lifetime. It seems that the two separate defect lifetimes may be around 170 and 230 ps. Some individual measurements do show a second lifetime as high as 230 ps, which seems a more reasonable value for Al vacancies than 200 ps. It should be noted that IR absorption measurements confirmed the presence of hydrogen in O 2-grown YAG, primarily on the form of O-H, and annealing in
Ar+H 2 appeared to slightly increase the concentration of O-H. Though it is unclear exactly what defects the two lifetimes correspond to, it is possible to form a hypothesis based on the results of Table 5.2.5 and IR absorption measurements. The long lifetime trapping defects appear to be Al vacancies and the shorter lifetime trapping defects are Al vacancies containing an H impurity. Assuming a single H impurity carries a charge of
+1, this still leaves the vacancy with a strong -2 charge and effectively a smaller size, which would bring about the decreased lifetime. This is evidenced by the decrease in defect lifetime and concentration after annealing in an atmosphere containing hydrogen
192 and the subsequent increase of this lifetime and defect concentration after annealing in argon to remove hydrogen.
Sputtering with aluminum and annealing in Ar appeared to slightly decrease the defect concentration in the O 2-grown samples. Aluminum vacancies in the O 2-grown
samples are mostly isolated due to growth conditions and are not often coupled to oxygen
vacancies, as is the case for H 2-grown YAG, thus incorporation of Al into its vacancy site
is favorable. The H 2-grown YAG was annealed in an atmosphere containing oxygen but
it did not show a significant change after Al sputtering.
Some of the limitations of the fitting software are apparent in Tables 5.2.4 and
5.2.5 that was not apparent in Tables 5.2.1-3. When the two lifetimes are similar in
value, the fitting program has difficulty resolving both, resulting in large uncertainties of
lifetimes and intensities. This can be expected for further results reported below in which
both the defect lifetime is similar to the bulk lifetime.
Figure 5.2.2 compares a spectrum of H 2-grown YAG fit to one lifetime to a spectrum of H 2-grown YAG annealed in air fit to two lifetimes to demonstrate the appearance of a weak second lifetime.
193
100000 As grown As grown fit Annealed 10000 Annealed fit
1000
Counts
100
10 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Lifetime (ns)
Figure 5.2.2 PALS fits for a measurement conducted on as grown and air annealed H2-grown YAG fit to
one lifetime and two lifetimes, respectively.
For graphical comparison of PALS on undoped YAG in various growth
conditions, Fig. 5.2.3 shows PALS spectra for Ar-, H2-, and O2-grown undoped YAG.
The figure clearly shows that Ar-grown YAG has the longest lifetime and H 2-grown
YAG has the shortest, as seen in the tabulated data.
194
100000 Ar-grown YAG
H2-grown YAG O -grown YAG 10000 2
1000
Counts
100
10 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Lifetime (ns)
Figure 5.2.3 Comparison of PALS measurements conducted on Ar-, H 2-, and O 2-grown undoped YAG.
Undoped YAG conclusions
Several defects were identified in undoped YAG samples using positron lifetime measurements. Ar-grown YAG showed a long defect lifetime, indicating a large vacancy cluster, attributed to an Al and multiple O vacancies. After annealing, the defect lifetime decreased but was still significantly larger than other undoped samples. Thus, it appears that annealing diffuses a lone O into this vacancy cluster, reducing the size of the vacancy cluster.
In YAG grown in an atmosphere containing oxygen, a short defect lifetime is observed, corresponding to small vacancies, attributed to isolated Al vacancies.
Hydrogen impurities can also exist within these vacancies, decreasing the characteristic positron lifetime of the trap.
Growth in an atmosphere containing hydrogen caused hydrogen impurities to behave differently than it did in other samples. H impurities filled vacancies so
195 thoroughly in H 2-grown YAG that positrons could not trap, thus no defect lifetime was observed. After annealing, a weak defect lifetime was observed, revealing the existence of vacancy-like defects once some H impurities were diffused out of the traps.
Ce:YAG
A unique series of experiments were conducted on Ce:YAG 0.1% to test the effects of photonic excitation on positron lifetime. The samples were excited using the
Xenon Pulsed Light Source and the 270 nm and 455 nm LEDs, all of which were described in Section 3.2. To excite the samples, a lens was placed about 1 cm above the samples in position sandwiched between the detectors, aligned so that the beam of light was centered between the two samples, and light was shined down onto the samples.
Table 5.2.6 shows this series of PALS measurements. There is an abundance of data for
Ce:YAG 0.1% under ambient conditions because it was occasionally used as a follow-up diagnostic sample to Al for PALS measurements, run whenever the system was changed somehow or did not seem to be working properly.
Table 5.2.6 PALS results for Ce:YAG 0.1% 10 mm dia. x 1 mm as grown, during UV excitation using the
Xenon Pulsed Light Source, after this excitation, after heating to remove any effects of this excitation, during excitation using the 455 nm LED, and during excitation using the 270 nm LED.
-1 -3 τ2 [ns] (I 2) τB [ns] τAV [ns] κd [ns ] Defect C [cm ]
As grown
15 0.2367 ± 0.0145 0.1526 ± 0.1554 ± 0.20 ± Al, Al+H I 4.1 x 10 ±
(8.16 ± 2.57%) 0.0050 0.0074 0.07 1.4 x 10 15
196 0 0.1514 ± 0.1514 ± 0 (none) 0
0.0001 0.0001
During UV excitation using the Xenon Pulsed Light Source
15 0.2041 ± 0.0201 0.1465 ± 0.1488 ± 0.27 ± Al, Al+H I 5.5 x 10 ±
(12.44 ± 7.78%) 0.0147 0.0196 0.18 3.6 x 10 15
0 0.1483 ± 0.1483 ± 0 (none) 0
0.0003 0.0003
After UV excitation using the Xenon Pulsed Light Source
16 0.1596 ± 0.0021 0.1414 ± 0.1453 ± 1.39 ± Al, Al+H I 2.8 x 10 ±
(68.73 ± 5.60) 0.0125 0.0111 0.29 6 x 10 15
0 0.1474 ± 0.1474 ± 0 (none) 0
0.0002 0.0002
After heating following UV excitation using the Xenon Pulsed Light Source
16 0.1601 ± 0.0093 0.1447 ± 0.1469 ± 1.16 ± Al, Al+H I 2.3 x 10 ±
(63.60 ± 32.79%) 0.0708 0.0669 0.95 1.9 x 10 16
0 0.1484 ± 0.1484 ± 0 (none) 0
0.0003 0.0003
During excitation using the 455 nm LED
16 0.1672 ± 0.0174 0.1461 ± 0.1480 ± 0.65 ± Al, Al+H I 1.3 x 10 ±
(42.91 ± 44.24%) 0.0912 0.0953 0.79 1.6 x 10 16
0 0.1488 ± 0.1488 ± 0 (none) 0
0.0002 0.0002
During excitation using the 270 nm LED
197 16 0.1721 ± 0.0130 0.1423 ± 0.1455 ± 0 0.84 ± Al, Al+H I 1.7 x 10
(40.73 ± 23.40%) 0.0470 0507 0.56 1.1 x 10 16
0 0.1486 ± 0.1486 ± 0 (none) 0
0.0002 0.0002
A second lifetime could not consistently be resolved in these measurements.
When a defect lifetime is resolved, the calculated defect concentration is so low that it is near the lower limit of sensitivity of PALS, which may explain the inability to consistently resolve the second lifetime. The runs during excitation were conducted just before the weak source used was retired for a newer, stronger source, so it is possible that the source correction terms may have been applied too large for these measurements with no way to check source strength. It is also possible that the difficulty in resolving a second lifetime comes from the very short defect lifetime, rendering its resolution above the bulk difficult. The very low calculated defect concentration of about 0.01 ppm is potentially too low for PALS measurements to accurately probe [Krause-Rehberg-1999].
It appears that the measurement after UV excitation displays slightly shorter positron lifetime than during UV excitation or after heating, all of which appear to have shorter lifetime than the ambient sample. Upon UV excitation, electrons can become trapped at defects, giving them a negative charge that is attractive to positrons and creating new energy level transitions in the crystal. These new allowed electronic transitions could help expedite positron thermalization, as the last and longest part of thermalization is heavily reliant on energy loss steps equivalent to an allowed electron energy transition. During UV excitation, electrons on average are more energetic, which
198 may result in longer positron lifetime as some electrons have enough energy to escape the
Coulombic potential well of the positron. After excitation has ceased, electrons do not have this extra energy and thus the positron can annihilate quickly easier. After heating, as was shown in Section 3.3, certain color centers persist, and based on how short lifetimes are for these measurements compared to lifetimes in the ambient sample it seems clear that some of the defects that sped up thermalization, and consequently annihilation, persist. However, the lifetime does increase slightly after heating, as is to be expected since heating removes some trapped electrons from their traps and thus less allowed energy transfers are available, increasing thermalization time. The 455 nm and
270 nm LEDs did not appear to significantly change the lifetimes. These LEDs do not emit sufficient energy to free electrons from nuclei, which seems to be the important step to changing positron lifetime by photonic irradiation.
Unfortunately, the defect trapping rate and concentration are too inconsistent to provide any reliable insight, though if anything it seems that the defect concentration increases upon any sort of optical excitation process.
Table 5.2.7 shows data from PALS measurements conducted on as grown
Ce:YAG 0.15%.
Table 5.2.7 PALS results for Ce:YAG 0.15% 10 mm dia. x 1 mm as grown.
-1 -3 τ2 [ns] (I 2) τB [ns] τAV [ns] κd [ns ] Defect C [cm ]
As grown
0.3096 ± 0.0101 0.1537 ± 0.1562 ± 0.10 ± Al+nO 9.7 x 10 16 ±
(2.77 ± 0.31%) 0.0007 0.0012 0.01 1.1 x 10 16
199
The second lifetime of these measurements shows poor agreement but the bulk and average lifetimes are quite consistent. These samples seem to show large vacancy- type defects, most likely caused by the large vacancy clusters evoked by inclusion of oversized ADs or even Ce impurities, with a relatively high concentration of about 1 ppm. Comparison to Table 5.2.6 shows that these vacancy clusters are larger and possibly more common in Ce:YAG 0.15% than in Ce:YAG 0.1%,. Comparison to TL results from Chapter 4 shows strong agreement with this conclusion, as Ce:YAG 0.1% has very few charge carrier traps but Ce:YAG 0.15% has several and shows very strong
TL signal. Thus, Ce:YAG 0.15% contains negatively charged vacancy clusters, possibly an oxygen vacancy and aluminum vacancy, which would have an effective charge of -1.
Table 5.2.8 shows results for PALS measurements conducted on Ce:YAG 0.2%.
These samples were annealed in air and later sputtered with aluminum and annealed in argon. Measurements were taken at each of these steps.
Table 5.2.8 PALS results for Ce:YAG 0.2% 10 mm dia. x 1 mm as grown, annealed in air, and subsequently sputtered with Al and annealed in argon.
-1 -3 τ2 [ns] (I 2) τB [ns] τAV [ns] κd [ns ] Defect C [cm ]
As grown
0.2710 ± 0.0103 0.1554 ± 0.1562 ± 0.06 ± Al+nO 6.4 x 10 16 ±
(2.02 ± 0.40%) 0.0008 0.0012 0.01 1.3 x 10 16
Annealed in air 96 hours at 1200°C
200 0.2617 ± 0.0156 0.1559 ± 0.1579 ± 0.13 ± Al+nO 1.3 x 10 17 ±
(4.67 ± 1.29%) 0.0025 0.0041 0.04 4 x 10 16
0 0.1516 ± 0.1516 ± 0 (none) 0
.0002 0.0002
After above anneal in air, sputtered with Al and annealed in Ar 48 hours at 1200°C
15 0.1700 ± 0.0022 0.1483 ± 0.1496 ± 0.45 ± Al+H I 9.1 x 10 ±
(34.50 ± 0.06%) 0.0012 0.0012 0.04 7 x 10 14
The as grown Ce:YAG 0.2% sample shows similar defect structure to the
Ce:YAG 0.15% sample. Comparison to TL glow curves in Chapter 4 can verify this, as both samples showed intense TL emission. As with Ce:YAG 0.15%, this contrasts lifetime measurements of Ce:YAG 0.1%.
Annealing in air seems to have a significant effect, suggesting an effect similar to
Ar-grown YAG where annealing in air fills an oxygen vacancy that is part of a larger vacancy cluster, resulting in shorter defect lifetimes. It is curious that the second lifetime disappears in later measurements of the annealed sample. This effect is potentially due to the abundance of measurements conducted on these samples, such as XLS and TL, and various treatments, such as heating and high energy photon excitation, that may accidentally be applied to the samples during these measurements. The defect concentration decreases by an order of magnitude after Al sputtering from about 1 ppm to about 0.1 ppm, which seems to show that the Al sputtering procedure had an effect on the crystal, likely filling an appreciable concentration of Al vacancies. Based on the very low defect lifetime, it appears that Al may have diffused into isolated Al vacancies but
201 was unable to diffuse into Al vacancies in which a hydrogen impurity was present, blocking Al.
Ce:YAG 0.14% PALS results are shown in Table 5.2.9 for the as grown and air annealed samples.
Table 5.2.9 PALS results for Ce:YAG 0.14% 10 mm dia. x ~1 mm as grown and annealed in air.
-1 -3 τ2 [ns] (I 2) τB [ns] τAV [ns] κd [ns ] Defect C [cm ]
As grown
15 0.1953 ± 0.0114 0.1480 ± 0.1502 ± 0.31 ± Al, Al+H I 6.2 x 10 ±
(14.98 ± 6.27%) 0.0121 0.0156 0.14 2.8 x 10 15
0 0.1493 ± 0.1493 ± 0 (none) 0
0.0002 0.0002
Annealed in air 48 hours at 1200°C
16 0.1674 ± 0.0074 0.1493 ± 0.1523 ± 0.97 ± Al+H I 1.9 x 10 ±
(58.60 ± 20.73%) 0.0459 0.0444 0.60 1.2 x 10 16
0 0.1541 ± 0.1541 ± 0 (none) 0
0.0003 0.0003
Table 5.2.9 shows the PALS results for Ce:YAG 0.14% is comparable to
Ce:YAG 0.1%. There exists a low concentration of small vacancy-type defects. After annealing, the defect lifetime appears to become even shorter due to oxidation. A possible explanation of the results is that there are Al vacancies and there are Al vacancies that contain hydrogen impurities. After annealing in air, more hydrogen
202 diffuses into the sample, possibly from moisture in the air, thus the concentration of Al vacancies containing hydrogen impurities rises while the concentration of Al vacancies consequently decreases. It is clear that the fitting program had considerable difficulty fitting a second lifetime to these spectra. The very low defect concentration of around
10 16 cm -3 (about 0.1 ppm) may cause of the difficulty in fitting, as the low concentration
is very near the limit of PALS sensitivity. The change in defect trapping rate can also
draw comparisons to O 2-grown YAG in which there were actually three lifetimes present in the measurements on the as grown sample that could not be completely resolved, coupling the second lifetime to the first and third, and annealing seemingly decreased the intensity of or possibly eliminated the longer lifetime.
Tables 5.2.10 and 5.2.11 show the results of PALS measurements conducted on polished and unpolished (fine ground) Ce:YAG 0.3%, respectively. The polished samples were only measured as grown, but the unpolished samples were measured before and after annealing in air and further annealing in nitrogen. As can be seen in the table, there was clearly a problem with the PALS measurement conducted on the as grown unpolished sample. This result is only included to demonstrate the error and can be disregarded. The error likely arises from poor 22 Na source alignment between the crystals, causing a significant portion of the generated positrons to go into air instead of the crystals, resulting in higher intensity source terms than could be accounted for and long-lived positronium. Higher source terms are suggested as a result of poor alignment because positrons that go into air can form positronium and are less likely to annihilate within the 50 ns window set by the Fast Coincidence and thus a lower percentage of annihilation events of positrons that escape the source foil will be recorded by the MCB.
203 Unfortunately, the samples were annealed before this error was noticed, so the error could not be fixed. The PALS results of the annealed unpolished samples can still be compared to the as grown polished sample since the penetration depth of positrons is deeper than few micron surface aberrations characteristic of the unpolished surfaces.
Table 5.2.10 PALS results for polished Ce:YAG 0.3% 5x5x1 mm as grown.
-1 -3 τ2 [ns] (I 2) τB [ns] τAV [ns] κd [ns ] Defect C [cm ]
As grown
0.3254 ± 0.0032 0.1570 ± 0.1632 ± 0.24 ± Al+nO 2.4 x 10 17 ±
(6.69 ± 0.21%) 0.0004 0.0008 0.01 1 x 10 16
Table 5.2.11 PALS results for unpolished (fine ground) Ce:YAG 0.3% 5x5x1 mm annealed in air and subsequently annealed in nitrogen.
-1 -3 τ2 [ns] (I 2) τB [ns] τAV [ns] κd [ns ] Defect C [cm ]
Annealed in air 96 hours at 1200°C
0.2949 ± 0.0138 0.1591 ± 0.1627 ± 0.17 ± Al+nO 1.7 x 10 17 ±
(5.60 ± 1.02%) 0.0021 0.0036 0.03 3 x 10 16
Further annealed in nitrogen 48 hours at 1200°C
0.3139 ± 0.0060 0.1595 ± 0.1639 ± 0.18 ± Al+nO 1.8 x 10 17 ±
(5.63 ± 0.39%) 0.0008 0.0015 0.01 1 x 10 16
It appears that annealing in air may have slightly decreased the defect lifetime.
Annealing in nitrogen reduced the charge states of impurities, which removed electrons from ions adjacent to vacancy-type defects, resulting in fewer electrons available for
204 annihilation for trapped positrons. This returned the lifetime to the original values before annealing in air. It is also worth pointing out that the defect concentration was consistent across all measurements at about 1 ppm, showing that the annealing procedures simply altered the defects present.
Ce:YAG conclusions
PALS measurements conducted on Ce:YAG revealed a wide range of defect types and concentrations. Samples grown in Ar showed large Al and O vacancy clusters while samples grown in an atmosphere containing H occasionally showed smaller defects.
Oxygen concentration was found to play a large role in defect size. Defect concentrations in Ce:YAG 0.1%, 0.14%, and 0.2% were near the lower limit of detection and were difficult to resolve.
A common characteristic of these PALS measurements seems to be that samples provided by United Crystals tend to have larger vacancy clusters than the samples provided by Crytur. X-ray stimulated luminescence measurements (Section 3.4) showed significantly stronger AD emission in United Crystals, suggesting less stoichiometry, which may also indicate a larger concentration of Al vacancies which are charge compensated by oxygen vacancies, which could explain this trend in PALS results.
Nd:YAG
Results for PALS measurements conducted on Nd:YAG 1% are shown in Table
5.2.12.
205 Table 5.2.12 PALS results for Nd:YAG 1% 5x5x1 mm as grown.
-1 -3 τ2 [ns] (I 2) τB [ns] τAV [ns] κd [ns ] Defect C [cm ]
As grown
0.3053 ± 0.0047 0.1704 ± 0.1758 ± 0.22 ± Al+nO 2.2 x 10 17 ±
(6.95 ± 0.43%) 0.0010 0.0017 0.01 1 x 10 16
Nd:YAG 1% has the largest bulk lifetime of any YAG sample. Its defect concentration, about 1 ppm, is similar to other samples acquired from United Crystals. It is clear that the large vacancy cluster apparent in all other samples procured from United
Crystals also exist in the Nd:YAG 1%.
Tm:YAG
Results for PALS measurements conducted on Tm:YAG 0.8% are shown in Table
5.2.13.
Table 5.2.13 PALS results for Tm:YAG 0.8% 5x5x1 mm as grown.
-1 -3 τ2 [ns] (I 2) τB [ns] τAV [ns] κd [ns ] Defect C [cm ]
As grown
0.3018 ± 0.0072 0.1562 ± 0.1589 ± 0.09 ± Al+nO 8.8 x 10 16 ±
(2.24 ± 0.24%) 0.0005 0.0011 0.01 9 x 10 15
As with Nd:YAG 1%, it is clear that the large vacancy cluster apparent in all other samples procured from United Crystals also exist in the Tm:YAG 0.8% samples at the similar 1 ppm concentration.
206 Comparison of PALS for undoped and Ce-, Nd-, and Tm-doped YAG
Comparing to measurements conducted on other samples clearly displays the significantly increased bulk lifetime, as shown in Fig. 5.2.4.
100000 Ar-grown undoped Ce:YAG 0.1% Nd:YAG 1% Tm:YAG 0.8%
10000
Counts 1000
100 0 1 2 Lifetime (ns)
Figure 5.2.4 Comparison of raw data from PALS measurements conducted on Ar-grown undoped YAG,
Ce:YAG 0.1%, Nd:YAG 1%, and Tm:YAG 0.8%.
Yb:YAG
Few reliable measurements were conducted on Yb:YAG samples. Yb:YAG 5% was cut into quarters, allowing for PALS measurements. However, cutting these samples into quarters made them similar in size to the 22 Na spot on the Kapton foil used for measurement. This does present potential for error, but results below are consistent with
Yb:YAG 10%, suggesting that sample size did not affect the measurements. Results of
PALS measurements conducted on Yb:YAG 5% and Yb:YAG 10% are displayed in
Tables 5.2.14 and 5.2.15, respectively.
207 Table 5.2.14 PALS results for Yb:YAG 5% 10 mm dia. x ~1 mm cut into quarters annealed in air.
-1 -3 τ2 [ns] (I 2) τB [ns] τAV [ns] κd [ns ] Defect C [cm ]
Annealed in air 96 hours at 1200°C
.3639 ± 0.0058 0.1557 ± 0.1610 ± 0.17 ± Al+nO 1.7 x 10 17 ±
(4.36 ± 0.20%) 0.0005 0.0009 0.01 1 x 10 16
Table 5.2.15 PALS results for Yb:YAG 10% 5x5x1 mm as grown and annealed in air.
-1 -3 τ2 [ns] (I 2) τB [ns] τAV [ns] κd [ns ] Defect C [cm ]
As grown
0.3499 ± 0.0077 0.1582 ± 0.1632 ± 0.17 ± Al+nO 1.7 x 10 17 ±
(4.61 ± 0.31%) 0.0007 0.0013 0.01 1 x 10 16
Annealed in air 48 hours at 1200°C
0.3624 ± 0.0058 0.1622 ± 0.1695 ± 0.23 ± Al+nO 2.3 x 10 17 ±
(6.40 ± 0.30%) 0.0007 0.0013 0.01 1 x 10 16
The effect annealing in air appears has on defect lifetime in Yb:YAG is unclear, but it does appear to slightly increase defect concentration, which was similarly observed in most other samples after annealing. These samples show a long defect lifetime that seems to be characteristic of samples grown by United Crystals.
Undoped YAG (labeled as Yb:YAG 5%)
Table 5.2.16 shows results from PALS measurements conducted on the undoped
YAG sample that was intended to contain 5% Yb doping.
208 4
Table 5.2.16 PALS results for undoped YAG (labeled as Yb:YAG 5%) 10 mm dia. x 1 mm cut into quarters as grown and UV excited 30 minutes such that both samples showed visible coloring in the region probed by positrons.
-1 -3 Τ2 [ns] (I 2) τB [ns] τAV [ns] κd [ns ] Defect C [cm ]
As grown
0.3184 ± 0.0064 0.1670 ± 0.1737 ± 0.27 ± Al+nO 2.8 x 10 17 ±
(8.75 ± 0.64%) 0.0014 0.0024 0.02 2 x 10 16
UV excited 30 minutes such that both samples are colored
0.3884 ± 0.0036 0.1779 ± 0.1919 ± 0.40 ± Al+nO 4.0 x 10 17 ±
(11.50 ± 0.30%) 0.0008 0.0014 0.01 1 x 10 16
Table 5.2.15 compares well to Table 5.2.12. This sample was determined to contain an observable concentration of Nd, thus it is reasonable to expect both to have similar PALS behavior. It seems that Nd impurities create complex defect structures. No other samples have bulk or average lifetimes as large as these samples. Color center measurements (Section 3.3) showed this sample to possess strong charge carrier traps.
PALS measurements, however, show these traps to differ significantly from those in other samples.
PALS conclusion
Various defects were identified in YAG using PALS. Isolated vacancies were identified, particularly in samples containing high concentration of oxygen. Al vacancies containing H impurities were also observed in these samples. Larger vacancy clusters involving Al and O vacancies were observed most commonly in samples grown in an
209 inert atmosphere. No trapping was observed in H 2-grown YAG due to H impurities
localizing at vacancy-like defects, preventing positrons from trapping at these defects.
Samples doped with Nd or Yb displayed a very long defect lifetime, indicating
vacancy clusters containing significantly decreased electron concentration, likely due to
larger physical size.
5.3 REFERENCES
[Babin-2011] V. Babin, V.V. Laguta, A. Maaroos, A. Makhov, M. Nikl, and S.
+ Zazubovich, “Luminescence of F -type centers in undoped Lu 3Al 5O12 single crystals,”
PPhysica Status Solidi B 248 (2011) 239-242
[Bernal-1995] M.J. Bernal, R.M. de la Cruz, R. Pareja, and J.M. Riveiro, “Mechanical amorphization in the Ni-Zr system investigated by positron lifetime spectroscopy,”
Journal of Non-Crystalline Solids 180 (1995) 164-170
[Calloni-2005] A. Calloni, A. Dupasquier, R. Ferragut, P. Folegati, and M.M. Iglesias,
“Positron localization effects on the Doppler broadening of the annihilation line:
aluminum as a case study,” Physical Review B 72 (2005) 054112
[Chithambo-2002] M.L. Chithambo, E.J. Sendezera, and A.T. Davidson, “A preliminary thermoluminescence and positron annihilation study of α-Al 2O3:C,” Radiation Protection
Dosimetry 100 (2002) 269-272
[Consolati-2001] G. Consolati, M. Levi and S. Turri, “Positron annihilation and phase separation phenomena in polyurethane crosslinked coatings with different polyether segments,” Polymer 42 (2001) 9723-9728
210 [Devor-1984] D.P. Devor, R.C. Pastor, and L.G. DeShazer, “Hydroxyl impurity effects in
YAG (Y 3Al 5O12 ),” Journal of Chemical Physics 81 (1984) 4104-4117
[Djourelov-1996] N. Djourelov and M. Misheva, “Source correction in positron
annihilation lifetime spectroscopy,” Journal of Physics: Condensed Matter 8 (1996)
2081-2087
[Dong-1991] J. Dong and K. Lu, “Noncubic symmetry in garnet structures studied using extended x-ray-absorption fine-structure spectra,” Physical Review B 43 (1991) 8808-
8821
[Dryzek-2006] J. Dryzek and Doug Singleton, “Implantation profile and linear absorption
coefficients for positrons injected in solids from radioactive sources 22 Na and 68 Ge\68 Ga,”
Nuclear Instrumentation and Methods in Physics Research B 252 (2006) 197-204
[Ferragut-1998] R. Ferragut, A. Somoza, and A Dupasquier, “Positron lifetime
spectroscopy and decomposition processes in commercial Al-Zn-Mg-based alloys,”
Journal of Physics: Condensed Matter 10 (1998) 3903-3918
[Geller-1967] S. Geller, “Crystal chemistry of the garnets,” Zeitschrift für
Kristallographie 125 (1967) 1-47
[Hautojarvi-1979] P. Hautojärvi, A. Vehanen, P.E. Mijnarends, R.N. West, R.M.
Nieminen, M.J. Manninen, and A. Dupasquier, Positrons in Solids , ed. P. Hautojärvi,
Topics in Current Physics vol. 7, Springer-Verlag, Berlin, Germany (1979)
[Kirkegaard-1989] P. Kirkegaard, N.J. Pedersen, and M. Eldrup, PATFIT-88: A Data-
Processing System for Positron Annihilation Spectra on Mainfraime and Personal
Computers , Risø National Laboratory, Grafisk Service, Risø, Denmark, (1989)
211 [Krause-Rehberg-1999] R. Krause-Rehberg and H.S. Leipner, Positron annihilation in
semiconductors: defect studies , Springer series in solid-state sciences 127, Springer, New
York, NY (1999)
[Landron-1996] C. Landron, S. Lefloch, M. Gervais, J.P. Coutures, and D. Bazin,
“Yttrium coordination shell in YAG related structures,” Physica Status Solidi (B) 196
(1996) 25-31
[Mackie-2000] R.A. Mackie, S. Singh, J. Laverock, S.B. Dugdale, and D.J. Keeble,
“Vacancy defect positron lifetimes in strontium titanate,” Physical Review B 79 (2009)
014102
[McGuire-2006(a)] S. McGuire and D.J. Keeble, “Positron lifetime and implantation in
kapton,” Journal of Physics D: Applied Physics 39 (2006) 3388-3393
[McGuire-2006(b)] S. McGuire and D.J. Keeble, “Positron lifetimes of polycrystalline metals: A positron source correction study,” Journal of Applied Physics 100 (2006)
103504
[Melissinos-1966] A.C. Melissinos, Experiments in Modern Physics , Academic Press,
Inc., San Diego, CA, 92101
[Monge-1994] M.A. Monge and J. del Rio, “Positron annihilation in Kapton source- supporting foils,” Journal of Physics: Condensed Matter 6 (1994) 2643-2646
[Polity-1997(a)] A. Polity, F. Rudolg, C. Nagel, S. Eichler, and R. Krause-Rehberg,
“Defects in electron-irradiated GaAs studied by positron lifetime spectroscopy,” Physical
Review B 55 (1997) 10480-10486
[Polity-1997(b)] A. Polity and T. Engelbrecht, “Defects in electron-irradiated InP studied by positron lifetime spectroscopy,” Physical Review B 55 (1997) 10467-10479
212 [Rajainmaki-1987] H. Rajainmäki, “High-resolution positron lifetime spectrometer with
BaF 2 scintillators,” Applied Physics A 42 (1987) 205-208
[Ranki-2004] V. Ranki and K. Saarinen, “Formation of thermal vacancies in highly As and P doped Si,” Physical Review Letters 93 (2004) 255502
[Saoucha-1999] A. Saoucha “Positron transmission into kapton: the effective mass absorption coefficient of relevance to positron lifetime experiments,” Journal of Applied
Physics 85 (1999) 1802-1810
[Schultz-1988] P.J. Schultz and K.G. Lynn, “Interaction of positron beams with surfaces, thin films, and interfaces,” Reviews of Modern Physics 60 (1988) 701-779
[Selim-2007] F.A. Selim, D. Solodovnikov, M.H. Weber, and K.G. Lynn, “Identification
of defects in Y 3Al 5O12 crystals by positron annihilation spectroscopy,” Applied Physics
Letters 91 (2007) 104105
[Siegel-1980] R.W. Siegel, “Positron annihilation spectroscopy,” Annual Reviews of
Materials Science 10 (1980) 393-425
[Somieski-1996] B. Somieski, T.E.M. Staab, and R. Krause-Rehberg, “The data treatment influence on the spectra decomposition in positron lifetime spectroscopy Part 1:
On the interpretation of multi-component analysis studied by Monte Carlo simulated model spectra,” Nuclear Instrumentation and Methods in Physics Research A 381 (1996)
128-140
[Springis-1991] M. Springis, A. Pujats, and J. Valbis, “Polarization of luminescence of colour centres in YAG crystals,” Journal of Physics: Condensed Matter 3 (1991) 5457-
5461
213 [Staab-1996] T.E.M. Staab, B. Somieski, and R. Krause-Rehberg, “The data treatment influence on the spectra decomposition in positron lifetime spectroscopy Part 2: The effect of source corrections,” Nuclear Instrumentation and Methods in Physics Research
A 381 (1996) 141-151
[Staab-2000] T.E.M. Staab, E. Zschech, and R. Krause-Rehberg, “Positron lifetime measurements for characterization of nano-structural changes in the age hardenable
AlCuMg 2024 alloy,” Journal of Material Science 35 (2005) 4667-4672
[Stanek-2008] C.R. Stanek, M.R. Levy, K.J. McClellan, B.P. Uberuaga, and R.W.
Grimes, “Defect identification and compensation in rare earth oxide scintillators,”
Nuclear Instrumentation and Methods in Physics Research B 266 (2008) 2657-2664
[Tuomisto-2003] F. Tuomisto, V. Ranki, and K. Saarinen, “Evidence of the Zn vacancy
acting as the dominant acceptor in n-type ZnO,” Physical Review Letters 91 (2003)
205502
[Wei-2004] L. Wei, Z. Zhang, C. Ma, C. Zhou, B. Wang, T. Chang, and S. Ren, “A high- resolution BaF 2 positron lifetime spectrometer with high counting rate,” Material Science
Forum 445-446 (2004) 516-518
[West-1974] R.N. West, Positron Studies of Condensed Matter , Taylor & Francis
monographs on physics, Taylor & Francis LTD, London, England (1974)
[Xu-1999] Y.N. Xu and W.Y. Ching, “Electronic structure of yttrium aluminum garnet
(Y 3Al 5O12 ),” Physical Review B 59 (1999) 10530-10535
[Zhang-2004] X.Y. Zhang, W. Sprengel, T.E.M. Staab, H. Inui, and H.-E. Schaefer,
“Formation of thermal vacancies on the Si sublattice of the intermecallic compound
MoSi 2,” Physical Review Letters 92 (2004) 155502
214 [Zorenko-2010(a)] Y. Zorenko, T. Zorenko, T. Voznyak, A. Mandowski, Q. Xia, M.
+ Batentschuk, and J. Friedrich, “Luminescence of F and F centers in Al 2O3-Y2O3 oxide compounds,” IOP Conference Series: Material Science Engineering 15 (2010) 012060
215 CHAPTER SIX
SCINTILLATION PROPERTIES
Introduction
Scintillation detectors are one of the most widely used types of particle and high energy photon detectors. They rely on similar phenomena as was discussed in Section
3.4, then pertaining to radioluminescence measurements. Certain materials can emit luminescence when excited with ionizing radiation. A crystal that responds to ionizing radiation with intense emission is a scintillator. The scintillator is mounted on a photomultiplier tube (PMT). The luminescence from the scintillator crystal is incident upon the photocathode, a thin covering on the entry window of the PMT, where the incident photons eject electrons from the photocathode by the photoelectric effect. The electrons are multiplied by secondary emission at a series of dynodes, enhancing the signal.
In order for a crystal to be considered a candidate for scintillation, it should possess a series of characteristics, such as: high stopping power for ionizing radiation, high sensitivity to energy, high efficiency of conversion of incident ionizing radiation to luminescence, and fast time response [Leo-1994, Zych-2000]. High stopping power generally requires a host crystal with a high density. Sensitivity to energy means a linear response with increasing energy and is a common characteristic of most scintillators.
Energy conversion efficiency requires generation of a large number of charge carriers upon irradiation and their efficient transport to and recombination at luminescence
216 centers. Fast time response involves short rise and decay times of luminescence, allowing for high count rates since the recovery time of the scintillator is low.
Ce:YAG has been discussed as a scintillator crystal [Babin-2005, Blazek-2004,
Ludziejewski-1997, Moszynski-1994, Nikl-2005, Rodnyi-2001, Selim-2007, Solodnikov-
2008, Stanek-2008, Vedda-2004, Zorenko-2007, Zorenko-2010(b), Zorenko-2012, Zych-
2000]. As seen in previous chapters, it emits a strong luminescence peak around 525 nm, which is ideal for several PMTs available today. The host crystal YAG has a high density of 4.55 g cm -3, though it has been argued that YAG consequently has relatively
low stopping power for γ rays compared to competing scintillation materials [Blazek-
2004, Ludziejewski-1997, Nikl-2005]. It has a high light output of about 20000
photons/MeV of ionizing radiation, which compares favorably at twice bismuth
germinate (BGO, Bi 4Ge 3O12 ), the current standard scintillator [Ludziejewski-1997,
Moszynski-1994, Nikl-2005, Zych-2000]. Ce:YAG under gamma excitation has a very fast rise time of a few nanoseconds and a decay time less than 100 ns [Ludziejewski-
1997, Moszynski-1994, Zych-2000], faster than most competing scintillation materials.
The decay lifetime has two components: one short lifetime of about 60-70 ns characteristic of charge carrier migration directly to Ce 3+ ions upon excitation and one
3+ long lifetime of a few hundred nanoseconds potentially arising from Y Al ADs
[Moszynski-1994, Robbins-1979(b), Vedda-2004, Zych-2000]. In Section 3.4, it was
3+ 3+ seen that Y Al emission was partly absorbed by Ce ions, likely causing this apparent
3+ 3+ delay in Ce emission. It has been suggested that charge carrier migration from Y Al antisite defects (ADs) to Ce 3+ prolongs emission and thus slows scintillation decay [Nikl-
2005, Zorenko-2007], as our X-ray stimulated luminescence (XSL) measurements show
217 [Varney-2012(c)]. Overall, Ce:YAG appears to demonstrate the above stated characteristics of scintillator crystals. However, very little is discussed about undoped
YAG. In fact, to our knowledge, these are the first measurements conducted on undoped
YAG as a scintillator crystal. Its emission arises entirely from defects so it is difficult to gauge. In this chapter, we further investigate the scintillation properties of Ce:YAG as well as surprising new findings about the scintillation properties of undoped YAG.
6.1 METHODS
Experimental setup
A series of measurements were conducted to explore some scintillation characteristics and efficiency of undoped YAG and Ce:YAG and then compared to typical scintillators such as Tl:NaI and BaF 2. XSL measurements, described in Section
3.4, were used to gauge the light response of Ce:YAG to ionizing radiation by varying
the intensity of the incident irradiation. A scintillation detector was built following the
instructions of Leo [Leo-1994] by mounting a polished YAG sample to a Hamamatsu
R329 PMT, which was in turn connected to an Ortec 473 Spectroscopy Amplifier and
Gated Integrator and the Ortec 926 ADCAM Multichannel Buffer (MCB) discussed in
Chapter 5, as shown in Fig. 6.1.1. The anode output of the PMT provides faster signal
than the dynode signal but occasionally leads to energy saturation, thus the anode output
is used for decay time measurements and usually the dynode is used for energy resolution
measurements. Amplifier settings used for scintillation measurements varied. While
using the anode output of the PMT for energy resolution measurements, the settings are
shaping time: 2 µs and coarse gain: 500 (YAG and BaF 2) or 50 (Tl:NaI), the difference in
218 coarse gain necessitated by the size and efficiency difference of reference crystals. While using the dynode output of the PMT for energy resolution measurements, the settings are: shaping time (0.5 µs) and coarse gain (500 for YAG or 50 for Tl:NaI and BaF 2). The amplifier serves to amplify signal from the PMT and shape it into a more usable shape and maximum voltage for the ADCAM to record. The PMT is powered by the HV
Power Supply discussed in Chapter 5 and is set to -1200 V for measurements using the anode output, which includes all decay time measurements and some energy resolution measurements, and -1500 V for measurements using the much weaker dynode output signal. The sample is coupled to the PMT by a thin layer of silicone optical grease.
137 Cs, 133 Ba, and 22 Na radioactive sources were placed near the scintillator to provide γ rays, allowing for use of multiple energies of ionizing radiation. The PMT was connected directly to an oscilloscope to observe decay lifetimes. Results are recorded as a histogram by the computer program Maestro mentioned in Chapter 5, compiled into what is referred to as a pulse height spectrum. For clarity, pulse height spectra created using YAG scintillator crystals are smoothed using ORIGIN (OriginLab Corporation) by the Savitzky-Golay method using a ten point window.
Figure 6.1.1 Schematic diagram of scintillation measurements conducted using γ rays ( 137 Cs, 133 Ba, and
22 Na sources) as ionizing radiation. The PMT is powered by a high voltage power supply set to -1200 V
(not shown).
219 The Hamamatsu R329 PMT is sensitive to wavelengths in the range 300-650 nm with peak sensitivity at 420 nm. It has high sensitivity and efficiency in the range of the
UV peak in YAG, near 320 nm, and shows adequate sensitivity around 525 nm for use with the Ce 3+ emission peak in Ce:YAG. Key specifications for the R329 PMT are
tabulated in Table 6.1.1 and compared to the H3177 PMT used in PALS measurements
(Chapter 5), as recorded from the Hamamatsu website for the R329 PMT
(http://sales.hamamatsu.com/en/products/electron-tube-
division/detectors/photomultiplier-tubes/part-r329.php and
http://sales.hamamatsu.com/en/products/electron-tube-division/detectors/photomultiplier-
modules/part-h3177-50.php, respectively). A very thin layer of silicone optical grease
coupled the scintillator crystal to the PMT window to matching the index of refraction.
The PMT window is about 5.1 cm in diameter, much larger than the 1 cm diameter of the
majority of YAG crystals used in this study. The size of a sample can affect scintillation
if it is not large enough to stop all γ rays. Aluminum foil (10 µm thick) was loosely fitted over the crystal and the PMT window. The aluminum foil acts as a mirror-like reflector, increasing the efficiency of light collection [Leo-1994]. It was fitted loosely because it is useful to leave a layer of air between the scintillator and reflector to maximize internal reflection within the crystal. The end of the PMT was wrapped in black electrical tape, aluminum foil and all, to prevent outside light from interfering with measurements.
220 Table 6.1.1 Key specifications for R329 PMT
Photomultiplier tube H3177 R329
Size (diameter) 51 mm 51 mm
Minimum wavelength 160 nm 300 nm
Maximum wavelength 650 nm 650 nm
Peak sensitivity 420 nm 420 nm
Window Silica glass Borosilicate
Cathode type Bialkali Bialkali
Gain 2 x 10 7 1.1 x 10 6
Rise time 1.3 ns 2.6 ns
Transit time (not listed) 48 ns
Transit time spread 0.55 ns 1.1 ns
Number of dynodes 12 12
In Table 6.1.1, it is clear that the H3177 PMT is much faster than the R329 PMT used in these measurements. Unfortunately, the H3177 PMT was unavailable for these measurements, but the R329 PMT provided an adequate representation of the scintillation properties of the samples studied that were able to lead to some conclusions. However, it is obvious that better energy and time resolution could be obtained with the H3177 PMT.
The different sources of excitation provided different energies of irradiation. The decay of 22 Na was discussed in Chapter 5 to coincide with emission of a γ ray or 1.274
MeV. Since 22 Na decay also results in creation of a positron, 0.511 MeV γ rays can also
be expected from annihilation. 137 Cs has a half-life of 30.17 years and decays to 137 Ba
221 with the emission of a γ ray of 0.662 MeV [Audi-2003, Melissinos-1966]. 133 Ba decays into 133 Cs with a half-life of 10.51 years. 133 Ba decay has several possible photon
emission energies, the most intense being a γ ray with 0.356 MeV energy [Audi-2003,
Hennecke-1967]. 60 Co decays to 60 Ni with a half-life of 5.2 years and decays with two nearly coincident gammas of 1.17 and 1.33 MeV [Melissinos-1966]. The known energies of the large peaks of 133 Ba and 137 Cs are used for energy calibration.
The setup was maintained with as much repeatability across similar samples as
possible. Measurements on 137 Cs were conducted for an hour using each sample as scintillator at identical electronics settings. The only parameters that may have varied with different scintillation crystals were silicone optical grease thickness and distance to the source of γ rays. The distance from the source to the detector was kept close to 1 cm.
Silicone optical grease thickness was minimized as much as possible. Its thickness was reflected in Fabry-Perot interference patterns in the spectrometer decay time measurements, as seen weakly in Fig. 6.1.2, and the sample was removed and a new layer of grease was applied if a noticeably different interference pattern appeared. Due to high repeatability, measurements conducted on different samples were directly comparable, even for samples of similar size and shape, as the ratio of intensity at the Compton edge to the intensity of the photopeak can demonstrate the dependence of efficiency on crystal size.
Analysis
The energy resolution of each scintillation crystal used can be calculated by
∆E FWHM = (6.1.1) E E
222 where E is the energy of a peak and FWHM is the full width at half maximum of that peak. The energy resolution is commonly expressed as a percentage.
Decay times were measured from maximum intensity to the time when the intensity falls to 1/e of its maximum value, adjusted for background noise. The decay time was measured on the oscilloscope and not by a more sophisticated technique involving fitting software. On the oscilloscope readout, the decay time was determined as the time between one point and a second point with 1/e the intensity of the first. This means that multiple overlapping decay times can skew measurements. Figure 6.1.2 shows a typical oscilloscope display from which the decay time can be deduced. This figure shows the emission intensity decay of Al sputtered and Ar annealed Ce:YAG
0.2%, with a decay time of 112 ns. Small Fabry-Perot interference is visible in the figure and does not affect the measurement.
Figure 6.1.2 Representative decay measurement of Ce:YAG 0.2% Al sputtered and Ar annealed using
137 Cs source. Vertical divisions represent 2 mV, horizontal divisions represent 100 ns. The decay time is
measured to be 112 nm.
223 Reference scintillators
Tl:NaI and BaF 2 scintillator crystals were used to compare to the samples used in this study. Tl:NaI is used as the model high efficiency scintillator [Leo-1994] while BaF 2
is the fastest high-Z scintillation detector in use today [Leo-1994, Rajainmaki-1987,
Rodnyi-2001, Saito-2002, Wei-2004]. The Tl:NaI scintillator was used with the same
PMT as was used with the YAG crystals in this study. The BaF 2 scintillator crystal used in these experiments is not one of the two used in positron annihilation lifetime spectroscopy measurements above in Chapter 5.
6.2 RESULTS
Reference scintillators Tl:NaI and BaF 2
Measurements using Tl:NaI and BaF 2 as scintillator crystals were taken for reference. Decay times measured were 330 ns for Tl:NaI and 4.7 ns for BaF 2. Table
6.1.1 shows that the R329 PMT has a rise time and transit time spread comparable to the
decay time of BaF 2, so it is likely that the decay time is much smaller than 4.7 ns and its duration is lengthened by the PMT. The transition responsible for fast BaF 2 emission is
self-trapped exciton (STE) emission [Weber-2004]. Measurements were conducted using
137 Cs as a source of γ rays for the Tl:NaI and BaF 2 scintillator crystals. Measured using the anode signal, the energy resolution of Tl:NaI and BaF 2 are calculated to be 10% and
17%, respectively, calculated from the 0.662 MeV peak. From the dynode signal, the energy resolution of Tl:NaI and BaF 2 are calculated to be 8% and 17%, respectively.
Tl:NaI has slow scintillation while BaF 2 is a very fast scintillator, and it is clear that
Tl:NaI has improved resolution using the slower dynode signal while BaF 2 has similar
224 energy resolution while using either signal. These measurements demonstrate that Tl:NaI is the more efficient scintillator and has higher energy resolution. However, decay times show that BaF 2 is the better scintillator for fast timing, hence why BaF 2 is used in such applications as positron annihilation lifetime spectroscopy where very high time resolution is important.
Characteristic of these scintillation measurements is a strong peak at 0.662 MeV, which is the photopeak of 137 Cs gamma emission. An intense and broad feature at lower energy results from Compton scattering, where the gamma ray scatters off the scintillator crystal but escapes and is an unavoidable characteristic of all pulse height spectra. A rapid decrease in signal intensity after the highest energy Compton scattering occurs at what is known as the Compton edge. The ratio of the intensity of the Compton edge to the photopeak can provide insight into scintillator efficiency as it can be used to estimate the relative ratio of γ rays that Compton scatter and escape to those that provide good useful emission. The Tl:NaI crystal is very large, about 5 mm diameter x 5 mm thick, thus the sample can stop more γ rays and Compton scattering is relatively low intensity.
On the other hand, the BaF 2 sample is smaller, about 2.5 mm diameter x 2.5 mm thick, making it easier for γ rays to escape and leading to higher Compton scattering relative to
the photopeak. The YAG samples studied in this work are significantly smaller than
even the BaF 2 scintillator and thus all have very intense Compton scattering compared to the photopeak(s).
For reference, their XSL spectra are shown in Fig. 6.2.1. BaF 2 shows similar emission to undoped YAG, which will be seen later to be an important result, and Tl:NaI
225 peaks near 420, which matches perfectly with the PMT, which has peak sensitivity at 420 nm.
8000 Tl:NaI 6000 BaF 2
6000 4000 4000
2000 2000
0 0
100 200 300 400 500 600 700 800 900 100 200 300 400 500 600 700 800 900 Luminescence IntensityLuminescence (counts) IntensityLuminescence (counts) Wavelength (nm) Wavelength (nm)
Figure 6.2.1 XSL of BaF 2 (left) and Tl:NaI (right).
Ce:YAG
XSL measurements on Ce:YAG were previously discussed in Section 3.4. It was observed that the UV emission coincides with an absorption band of Ce 3+ . This emission and reabsorption delays the luminescence decay time, decreasing performance in one key aspect of scintillation. Of the samples measured, Ce:YAG 0.1% had the lowest measured intensity of the UV peak, Ce:YAG 0.15% and 0.2% were similar in intensity, and
Ce:YAG 0.3% had very large UV emission intensity. As will be seen below, this corresponds well with measured decay times.
Ce:YAG has already been shown to have great potential as a scintillator. The strong emission and relatively fast decay of the Ce3+ peak are very attractive properties
for scintillation. Figure 6.2.2 shows the pulse height spectra for some Ce-doped YAG
samples of different sizes. In this figure, Ce:YAG 0.14% is 10 mm diameter x 5 mm
226 thick, Ce:YAG 0.15% is 10 mm diameter x 1 mm thick, and Ce:YAG 0.3% is 5 x 5 x 1 mm. This figure is included to emphasize the importance of scintillator size.
Ce:YAG 0.14% (10x5) Ce:YAG 0.15% (10x1) 400 Ce:YAG 0.3% (5x5x1)
200 Counts
0 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Energy (MeV)
Figure 6.2.2 Pulse height spectra of 137 Cs using Ce:YAG 0.14% (10 mm dia. x 5 mm), Ce:YAG 0.15% (10
mm dia. x 1 mm), and Ce:YAG 0.3% (5 x 5 x 1 mm) as scintillator crystals.
The rest of the pulse height spectra measured using Ce:YAG samples are shown in Figs. 6.2.3 and 6.2.4, where Fig. 6.2.4 expands the region containing the photopeak.
Included in these spectra are Ce:YAG 0.1%, Ce:YAG 0.1% annealed 36 hour in an atmosphere of Ar at 600°C, Ce:YAG 0.15% (same as shown in Fig. 6.2.2), Ce:YAG
0.2% after annealing in air 96 hours at 1200°C and further Al sputtering and Ar annealing
24 hours at 600°C, and Ce:YAG after annealing in air, Al sputtering and Ar annealing, and further annealing in Ar 36 hours at 600°C. All samples included in Fig. 6.2.3 are 10 mm diameter x 1 mm thick.
227
Ce:YAG 0.1% 600 Ce:YAG 0.1% Ar an. Ce:YAG 0.15% Ce:YAG 0.2% Al sp.+Ar an. Ce:YAG 0.2% Al sp.+Ar an. (x2) 400
Counts 200
0 0.2 0.3 0.4 0.5 0.6 0.7 Energy (MeV)
Figure 6.2.3 Pulse height spectra of 137 Cs using various 10 mm dia. x 1 mm Ce:YAG samples as
scintillator crystals.
80 Ce:YAG 0.1% 70 Ce:YAG 0.1% Ar an. Ce:YAG 0.15% 60 Ce:YAG 0.2% Al sp.+Ar an Ce:YAG 0.2% Al sp.+Ar an. (x2) 50
40
30 Counts 20
10
0 0.5 0.6 0.7 0.8 Energy (MeV)
Figure 6.2.4 Pulse height spectra of 137Cs using various 10 mm dia. x 1 mm Ce:YAG samples as
scintillator crystals, with the photopeak expanded for examination.
228 As is seen in Fig. 6.2.2, larger samples result in more emission and higher efficiency. The larger the sample, the more likely that the sample stops γ rays. The ratios
of Compton edge to photopeak intensity are 6.0 for the largest sample (Ce:YAG 0.14%),
6.4 for the thinner sample (Ce:YAG 0.15%), and 10.2 for the smallest sample (Ce:YAG
0.3%). It should be noted that the other 10 mm diameter x 1 mm thick samples have
ratios of 13.1 (Ce:YAG 0.1%), 9.4 (Ce:YAG 0.1% Ar annealed), 9.9 (Ce:YAG 0.2% air
annealed then Al sputtered and Ar annealed), and 4.0 (Ce:YAG 0.2% air annealed then
Al sputtered and Ar annealed and further Ar annealed). On average, thickness plays the
largest role in the intensity of Compton scattering.
The shapes of the photopeaks in Figs. 6.2.2-4 demonstrate varied energy
resolution across samples. Table 6.2.1 shows the energy resolutions calculated for each
Ce:YAG sample along with the decay time of emission to 1/e of max intensity. Not yet
discussed but included in the table is a 1 mm thick irregularly shaped shard of Ce:YAG
0.14% otherwise identical to the 10 mm diameter x 5 mm thick sample for direct
comparison to a larger sample of the same composition. Energy resolution measurements
were conducted using both the dynode and anode output signals of the PMT for
comparison. As seen above with the reference crystals Tl:NaI and BaF 2, which signal
provides higher resolution is sample-dependent. For most samples, the anode signal
saturated the detector at an energy below 0.662 MeV and thus their results cannot be
reported. Measurements conducted on Ce:YAG 0.14% using the anode signal saturated
the detector above 0.662 MeV, and results are presented in Table 6.2.1. Anode signal
measurement was not attempted on Ce:YAG 0.3%, nor was dynode signal measurement
229 attempted on the 1 mm thick Ce:YAG 0.14% shard, since the anode signal showed it to have identical resolution its thicker counterpart.
Table 6.2.1 Energy resolution and decay time for 0.662 MeV photopeak of 137 Cs decay using Ce:YAG
samples as scintillator crystals. Energy resolution measurements were attempted using both the anode and
dynode signals of the PMT. Decay time measurements used the anode signal only.
Sample Energy resolution Decay time
Dynode Anode
Ce:YAG 0.14% (10 mm dia. x 5 mm) 17% 8% 86 ns
Ce:YAG 0.14% (1 mm thick irregular shard) N/A 8% 82 ns
Ce:YAG 0.1% 13% N/A 88 ns
Ce:YAG 0.1% Ar an. 15% N/A 77 ns
Ce:YAG 0.15% 11% N/A 110 ns
Ce:YAG 0.2% air an., Al sp.+Ar an. 10% N/A 112 ns
Ce:YAG 0.2% air an., Al sp.+Ar an., Ar an 11% N/A 116 ns
Ce:YAG 0.3% (grown in Ar only) 20% N/A 112 ns
The best energy resolution was measured using Ce:YAG 0.14%. Ce:YAG 0.2% after Al sputtering had the best resolution of the measured samples by the dynode signal.
3+ However, its lifetime was still long. The long lifetime effectively results from Y Al
ADs, which are believed to persist after this treatment. However, Al sputtering may fill
Al vacancies, which in turn may decrease STE emission.
The thickness of the Ce:YAG 0.14% sample does not appear to affect decay time or energy resolution, nor does it appear to have any significant effect on either parameter
230 in other samples. The only effect the size appeared to have on scintillation measurements was that ratio of photopeak to Compton scattering intensity was better for the larger sample, which results from the larger crystal having more stopping power due to its increased volume. The smallest sample, Ce:YAG 0.3%, shows the worst resolution at
20%, but it appears that this is a result of growth conditions. All other samples besides
Ce:YAG 0.3% are grown in a reducing environment involving hydrogen, but this sample was grown in pure argon. Other samples show consistently worse resolution after annealing in argon. Thus, it is possible that hydrogen improves efficiency of Ce 3+ ion
emission in Ce:YAG, which may be expected as it helps to protect Ce 3+ from oxidation to
Ce 4+ . From Table 6.2.1, it is possible to conclude that sample size (in the range of sizes
studied) does not affect energy resolution or decay time of the scintillator.
In previous chapters, Ce:YAG 0.1% and 0.14% consistently showed the weakest
trapping phenomena of all Ce-doped YAG samples studied. In Section 3.4, Ce:YAG
0.1% had the lowest UV emission in XSL measurements. In Chapter 4, both samples
were shown to have the weakest thermoluminescence (TL) emission of Ce:YAG samples
for temperatures above room temperature, which are the most pertinent measurements to
scintillation measurements conducted at room temperatures since defects identified by
low temperature TL measurements are shallow and cannot trap charge carriers at room
temperature. In Chapter 4, both samples were shown to have the lowest concentration of
positron-trapping defects of Ce:YAG samples. In Table 6.2.1, these samples have the
shortest decay time by a large margin, and Ar annealing improves this noticeably. It is
obvious that these samples have a significantly lower concentration of charge carrier
trapping defects that cause slow scintillation than is present in the other samples. It is
231 3+ believed that Ce:YAG scintillation signal is slowed by charge carrier transfer from Y Al
ADs to Ce 3+ luminescence centers [Nikl-2005, Zorenko-2007], which is evidenced by comparing decay time to the intensity of UV emission in XSL measurements (Fig. 6.2.5 below), which can be reabsorbed by the 340 nm absorption band of Ce 3+ in YAG.
Ce:YAG 0.1% has the weakest UV emission and thus the least reabsorption and fastest decay time. Ce:YAG 0.15% has more intense UV emission and thus higher reabsorption and longer decay time. Ce:YAG 0.3% has by far the most intense UV emission, but it also has a fast decay component that is much stronger than it is in undoped samples. As discussed below for undoped YAG, the short decay time likely arises from STE emission
[Babin-2005, Kirm-2000, Zorenko-2004(a), Zorenko-2004(b)], which has peak emission near 300 nm, similar to AD emission. Fig 6.2.6 shows the decay time of Ce:YAG 0.3%.
Also included in Fig. 6.2.6 is the much shorter decay of Ce:YAG 0.1% for comparison.
A similar phenomenon’s observed in Fig. 6.2.5 with Ce:YAG 0.1% after annealing in Ar.
The UV emission increases but the decay time decreases, which may result from STEs.
12000 0.1% (M) 0.1% as grown 0.15% (M) 2000 0.1% Ar annealed 0.3% (UC) 8000
1000
4000
0 0 200 250 300 350 400 450 200 250 300 350 400 450 Luminescence IntensityLuminescence (counts) Luminescence Intensity(counts) Wavelength (nm) Wavelength (nm)
Figure 6.2.5 XSL spectra of Ce:YAG samples, included to demonstrate UV emission and reabsorption by
Ce 3+ ions. Measurements are compared for select as grown samples (left) and Ce:YAG 0.1% before and
after annealing in Ar (right).
232
Figure 6.2.6 Decay measurements of Ce:YAG 0.3%, measured to be 112 ns (top), and Ce:YAG 0.1%,
measured to be 88 ns (botton), using 137 Cs source. For both plots, the scale is 100 ns per horizontal division
and 2 mV per vertical division.
Comparison of Fig. 6.2.6 to Fig. 6.1.2 shows that Ce:YAG 0.3% emission has a very short decay component that does not appear to be present in the Al sputtered
Ce:YAG 0.2% sample, at least not at such high intensity.
Ce:YAG emission as a function of x-ray power
X ray emission can also provide information about the linearity and proportionality of the material response to radiation, which are important scintillation characteristics that can define its energy resolution. This was studied for Ce:YAG 0.15% by changing the input power of the x-ray machine and recording the emission. Figure
6.2.7 shows the emission spectra at different x-ray tube power and Fig. 6.2.8
233 demonstrates the linearity of the integrated intensity versus x-ray tube power. Data presented in Fig 6.2.7 are integrated over the Ce 3+ emission peak to produce Fig. 6.2.8.
20000 1400 W 16000 1200 W 1000 W 800 W 12000 600 W 400 W 200 W
8000
4000 Luminescence (counts) Luminescence 0
100 200 300 400 500 600 700 800 900 Wavelength (nm)
Figure 6.2.7 Luminescence spectra of Ce:YAG 0.15% at various x-ray powers.
1800000 Integrated data 1600000 fit
1400000
1200000
1000000
800000 600000
400000
200000
Integrated intensity (arb. units) (arb. intensity Integrated 0 0 200 400 600 800 1000 1200 1400 X-ray machine power (W)
Figure 6.2.8 Light yield calculated by plotting the integrated 530 nm peak versus x-ray tube power for
Ce:YAG 0.15%
234
Direct photoluminescence decay measurement of Ce:YAG
An experiment was conducted using the JY-Horiba spectrometer discussed in
Section 3.2 to directly examine the decay of Ce 3+ luminescence at 530 nm. A 270 nm flash LED with <1 ns pulse duration was used for excitation. The sample used in this experiment was Ce:YAG 0.14% 10 mm diameter x 5 mm thick. The measurement is included below as Fig 6.2.9. The lifetime of decay was measured as 66.0 ± 1.4 ns.
Ce:YAG 0.14% 530 nm luminescence 10000 270 nm LED prompt
1000
100 Intensity (Counts) Intensity
10
0 50 100 Time (ns)
Figure 6.2.9 Photoluminescence decay measurement of Ce:YAG 0.14% of luminescence at 530 nm using
270 nm flash LED. The lifetime of the luminescence decay was measured to be 66.0 ± 1.4 ns.
Ce:YAG conclusions
Ce:YAG scintillation was measured to have very high energy resolution and good decay time that is much faster than Tl:NaI. It also showed excellent linear response to varied intensity of incident ionizing radiation, which indirectly indicates that that
Ce:YAG also has such linear response to change in energy.
235 Ce:YAG 0.14% has the best energy resolution (8%) and nearly the fastest scintillation decay (82 ns) of Ce:YAG samples measured. Ce:YAG 0.3%, the only sample grown in pure Ar atmosphere, had the worst energy resolution (20%) and nearly the slowest scintillation decay (112 ns). Annealing in Ar decreased energy resolution but quickened scintillation decay time in Ce:YAG 0.15%, giving the fastest decay of any
Ce:YAG sample measured (77 ns). Though direct comparison is not available, it appears that Al sputtering and Ar annealing improved energy resolution, though it is unclear how this treatment affected decay time, if at all. Ce:YAG scintillation decay arises from two components: one shorts component of ~60-70 ns arising from Ce 3+ ions and another longer decay of ~1 µs arising from defects.
Undoped YAG
Ce:YAG has very strong emission at 525 nm, but the PMT has relatively weak sensitivity at and around this wavelength. Undoped YAG has very strong emission in the
UV, peaked at 320 nm, where the PMT is much more sensitive. Scintillation measurements were conducted using undoped YAG and yielded surprisingly promising results. Undoped YAG samples investigated are: Ar-grown, H 2-grown, O 2-grown, and
O2-grown Al sputtered and annealed in Ar. This section to our knowledge presents the
first pulse height spectra measurements conducted on undoped YAG.
Undoped YAG showed two easily distinguishable decay times of a few ns and ~1
µs. It has been suggested that the short ~ns decay time arises from STE emission and the long ~ µs decay time results from AD emission [Babin-2005, Kirm-2000, Zorenko-
2004(a), Zorenko-2004(b)], both of which have peak emission near 300 nm. Figure
6.2.10 shows the decay time measurement of Ar-grown YAG 10x10x1 mm. Two
236 separate measurements are presented due to the difference in scale, as the short-decay component is much more intense than the long-decay component. A very rough estimate of a ratio of intensities is 10:1 STE:AD.
Figure 6.2.10 Decay measurement of Ar-grown undoped YAG using 137 Cs source. Vertical divisions
represent 10 mV (left) and 1 mV (right), horizontal divisions represent 5 ns (left) and 250 ns (right). The
decay time of the short component is measured to be 3 ns (left) and the decay time of the long component is
measured to be 1 µs (right).
It is interesting to compare the decay time of undoped YAG to that of BaF 2, since
fast decay emission of both samples arises from STEs [Weber-2004]. Figure 6.2.11
shows the XSL spectra of Ar-, H 2-, and O 2-grown undoped YAG to demonstrate the
strong emission used for scintillation, which are very similar to BaF 2 luminescence.
237 20000
O2-grown H -grown 15000 2 Ar-grown 10x10x1
10000
5000
0
200 300 400 500 600 Luminescence (counts) Intensity Wavelength (nm)
Figure 6.2.11 XSL spectra of undoped YAG, included to demonstrate the UV emission by undoped YAG
used for scintillation to compare to BaF 2. The PMT is sensitive to emission in the wavelength range 300-
650 nm.
Figures 6.2.12 and 6.2.13 show the pulse height spectra for 137 Cs using the larger
10 mm diameter x 5 mm thick H 2- and O 2-grown YAG samples as scintillator crystals,
compared to the Ce:YAG 0.14% sample of the same physical dimensions for reference.
Figures 6.2.14 and 6.2.15 show the pulse height spectra for 137 Cs using the 10 mm
diameter x 1 mm thick Ar-grown YAG and O 2-grown YAG sputtered with aluminum and
annealed in argon 24 hours at 600°C, compared to Ce:YAG 0.15% for reference. Figures
6.2.12 and 6.2.14 include the Compton edge while Figs. 6.2.13 and 6.2.15 focus on the
region containing the 0.662 MeV photopeak.
238
2200 2000 Ce:YAG 0.14% O -grown YAG 1800 2 H -grown YAG 1600 2 1400 1200
1000
Counts 800 600 400 200 0 0.4 0.5 0.6 0.7 0.8 Energy (MeV)
137 Figure 6.2.12 Pulse height spectra of Cs using 10 mm diameter x 5 mm thick H2- and O 2-grown undoped
YAG and Ce:YAG 0.14% samples as scintillator crystals.
600 Ce:YAG 0.14%
O2-grown YAG
H2-grown YAG 400
Counts 200
0 0.5 0.6 0.7 0.8 Energy (MeV)
137 Figure 6.2.13 Pulse height spectra of Cs using 10 mm diameter x 5 mm thick H2- and O 2-grown undoped
YAG and Ce:YAG 0.14% samples as scintillator crystals, with the photopeak expanded for examination.
239
800 Ce:YAG 0.15% 700 Ar-grown undoped YAG O -grown undoped YAG Al sp.+Ar an. 600 2
500
400
300 Counts
200
100
0 0.3 0.4 0.5 0.6 0.7 0.8 Energy (MeV)
Figure 6.2.14 Pulse height spectra of 137 Cs using 10 mm diameter x 1 mm thick Ar-grown undoped YAG,
O2-grown undoped YAG Al sputtered and Ar annealed, and Ce:YAG 0.15% samples as scintillator
crystals.
240
100 Ce:YAG 0.15% Ar-grown undoped YAG 80 O2-grown undoped YAG Al sp.+Ar an.
60
40 Counts
20
0 0.55 0.60 0.65 0.70 0.75 0.80 Energy (MeV)
Figure 6.2.15 Pulse height spectra of 137 Cs using 10 mm diameter x 1 mm thick Ar-grown undoped YAG,
O2-grown undoped YAG Al sputtered and Ar annealed, and Ce:YAG 0.15% samples as scintillator
crystals, with the photopeak expanded for examination.
From Figs. 6.2.12-15 it is apparent that undoped YAG has good energy resolution and efficiency comparable with Ce:YAG. It must be emphasized that the PMT used is significantly more sensitive to the wavelengths of undoped YAG emission than it is to the emission range of Ce:YAG, so it is possible that the apparent increased efficiency for undoped YAG over Ce:YAG is due to the fact that PMT sensitivity better matches undoped YAG luminescence than Ce:YAG luminescence. The ratios of intensity of the
Compton edge to the intensity of the photopeak are 6.2 for H 2-grown YAG, 5.4 for O 2-
grown YAG, 8.6 for Ar-grown YAG, and 9.0 for Al sputtered and Ar annealed O 2-grown
YAG, which shows efficiency comparable to Ce:YAG. Table 6.2.2 shows the energy
resolution and decay time for each undoped YAG sample measured. Energy resolution
241 measurements were conducted using both the dynode and anode output signals of the
PMT for comparison. For these undoped samples, weaker emission actually presented an advantage over Ce:YAG because the anode signal did not saturate the detector and thus higher resolution measurements were recorded. The short decay time rendered precise measurement difficult, so some short decay times are given as a range of values.
Table 6.2.2 Energy resolution and decay time for 0.662 MeV photopeak of 137Cs decay using undoped
YAG samples as scintillator crystals. Energy resolution measurements were attempted using both the anode and dynode signals of the PMT. Decay time measurements used the anode signal only.
Sample Energy resolution Decay time
Dynode Anode Short Long
Ar-grown YAG 13% 11% 3 ns 1000 ns
H2-grown YAG 18% 16% 3-4 ns 750 ns
O2-grown YAG 14% 13% 3-4 ns 750 ns
O2-grown YAG Al sp. + Ar an. 15% 15% 4 ns 850 ns
H2-grown YAG has the worst energy resolution for the undoped samples. It seems that hydrogen decreases scintillation efficiency and energy resolution. Color center (Section 3.3), XSL (Section 3.4), TL (Chapter 4), and PALS (Chapter 5) measurements appeared to show a lack of trapping defects in H2-grown YAG, which may
acts as a disadvantage for scintillation. An alternate possibility is that H 2-grown YAG
lacks luminescent centers that are present in undoped YAG grown in inert or oxidizing
atmospheres. According to Babin et al., Kirm et al., and Zorenko et al. [Babin-2005,
Kirm-2000, Zorenko-2004(a), Zorenko-2004(b)], the fast component of UV emission in
242 YAG arises from STEs. STEs are formed when a charge carrier escapes an ion and traps at a nearby ion or defect but remains bound to the oppositely-charged carrier left behind and lacks the energy for escape and recombination [Williams-1990]. It is believed that an Al vacancy can act as a trap for the STE formation [Hayes-1980, Kirm-2000]. PALS measurements present the ideal method by which to probe Al vacancy concentration, but no such feature was found in H 2-grown YAG until after annealing in air, thus it was proposed that Al vacancies were present but filled by hydrogen. XSL measurements as well showed decreased UV emission, which may at least partially arise from a lower concentration of STEs. Thus, it is tempting to conclude that STEs cannot form as easily in this sample as in other samples. On top of that, XSL measurements showed that H 2-
grown YAG has the weakest UV emission of undoped samples, further decreasing
efficiency. Growth in an atmosphere containing hydrogen leads to decreased scintillation
efficiency in undoped YAG. This interesting conclusion is opposite to the conclusion
formed for Ce-doped YAG, and it arises from the differences in luminescence centers
used for scintillation.
The decay time measurements of undoped YAG are limited by the PMT. A very
fast PMT would be required to fully appreciate the speed of undoped YAG scintillation
decay.
Direct photoluminescence decay measurement of Ar-grown undoped YAG
An experiment was conducted using the JY-Horiba spectrometer discussed in
Section 3.2 to directly examine the decay of luminescence at 300 nm in undoped YAG.
A 270 nm flash LED with <1 ns pulse duration was used for excitation. The sample used
in this experiment was Ar-grown YAG 10 x 10 x 1 mm, shown in Table 6.2.2 to have the
243 fastest decay. The measurement is included below as Fig 6.2.16. The lifetime of the decay is clearly less than the resolution of the system. The LED pulse duration is <1 ns, but no more specific details are given. There was also very weak emission. As a result, it is very difficult to accurately measure the decay time of this measurement. BaF 2 has a fast decay time of ~0.8 ns [Sobolev-1994]. An estimate of Ar-grown YAG decay time using fitting software gave 0.2 ± 0.1 ns, which if correct makes Ar-grown YAG emission almost an order of magnitude faster than BaF 2!
Ar-grown YAG 300 nm luminescence 10000 270 nm LED prompt
1000
100 Intensity (Counts) Intensity
10 -5 0 5 10 15 Time (ns)
Figure 6.2.16 Photoluminescence decay measurement of Ar-grown undoped YAG luminescence at 300 nm
using 270 nm flash LED. The lifetime of the luminescence decay was estimated to be 0.2 ± 0.1 ns. It
should be emphasized that this is below the resolution of the system.
It appears that only the fast signal is excited by 270 nm excitation, as no long lifetime component was observed. This measurement was also attempted at 320 nm but only the prompt was observed. The measurement was also attempted at 385 nm, where a peak attributed emission from to Tb 3+ impurities (Section 3.2) is observed, and a long
244 lifetime component is observed. For reference, these measurements are shown in Fig.
6.2.17
Ar-grown YAG 320 nm 10000 Ar-grown YAG 385 nm 10000 270 nm LED prompt 270 nm LED prompt
1000
1000
100 Intensity(Counts) Intensity(Counts)
100 10 -5 0 5 10 15 20 25 0 50 Time (ns) Time (ns)
Figure 6.2.17 Photoluminescence decay measurements of Ar-grown undoped of luminescence at 320 nm
(left) and 385 nm (right) YAG using 270 nm flash LED. No significant luminescence was recorded at 320
nm and the measurement at 385 nm displays a lifetime of ~30 ns.
Undoped YAG discussion and comparison to BaF 2
O2-grown YAG and Ar-grown YAG both show better resolution than BaF 2 and
similar decay time. Of the two, Ar-grown YAG shows the slightly better resolution. Ar-
grown YAG has more intense UV emission in XSL measurements (Section 3.4) than O 2-
grown YAG. PALS measurements revealed that Ar-grown YAG contains large vacancy
clusters while O 2-grown YAG likely contains single Al vacancies, with a higher density of defects present in Ar-grown YAG, possibly suggesting more efficient STE creation and recombination in Ar-grown YAG. Color center studies showed Ar-grown YAG to have very weak response to UV excitation while O 2-grown YAG showed a large change
in optical characteristics, suggesting that charge carrier trapping defects may not be as
prevalent in Ar-grown YAG as it is in O 2-grown YAG. TL glow curves seem at odds
245 with this, but while heating to 400°C bleached color in Ar-grown YAG, the O 2-grown
YAG sample remained colored after heating to 400°C, so it is likely that more intense TL
emission exists for O 2-grown YAG at temperatures above the capabilities of our
measurements. Thus it appears that growth in an inert pure argon atmosphere forms the
best undoped YAG crystal for scintillation.
The O 2-grown YAG sample sputtered with Al and annealed in Ar shows subtle
decrease to resolution and increase to decay time. Assuming the above statement that
STEs rely on Al vacancies, the cause of this is trivial. XSL (Section 3.4) showed
decreased UV emission and PALS (Chapter 5) showed decreased defect concentration,
both of which point to a decreased concentration of Al vacancies, suggesting some of the
Al successfully incorporated into the crystal. This degrades scintillation efficiency of the
crystal. For efficient, high resolution, and fast scintillation, an undoped YAG crystal
must maximize Al vacancies to increase STE formation.
It has been shown in previous work that scintillation performance of Ce:YAG is
improved by diffusing Al into the sample [Solodovnikov-2008]. This is because
Ce:YAG as a scintillator relies on the Ce 3+ emission, the efficiency of which can be decreased by any competing decay centers such as those responsible for UV luminescence. However, the Al sputtering and annealing treatment worsens undoped
YAG scintillation properties because its emission comes from Al-deficiency defects. It is important to recognize the distinction here that Ce:YAG scintillation benefits from Al diffusion treatment while undoped YAG scintillation is worsened by the same process.
To investigate the applicability of undoped YAG to PALS, a measurement was conducted on 60 Co. The two γ rays emitted by the decay of 60 Co are nearly coincident
246 and close in energy. The purpose of this measurement is to investigate if a Constant
Fraction Differential Discriminator, such as was discussed in Chapter 5, was set to a tight window around an individual photopeak so that specific incidents, such as creation of a positron in 22 Na indicated by a coincident 1.274 MeV γ ray or positron annihilation
indicated by a 0.511 MeV γ ray, was recorded. Figure 6.2.18 shows the pulse height
60 spectrum of Co using O 2-grown undoped YAG as a scintillator. Both peaks are observed, the 1.17 MeV showing as a shoulder to the Compton edge. For comparison, the same measurement was conducted using BaF 2 as the scintillator, also shown in Fig.
6.2.18. It appears that undoped YAG has better resolution, confirming above results.
Based on Fig. 6.2.18 it appears that O 2-grown YAG could find use as a scintillator crystal for PALS.
60 Co using O -grown YAG 250000 2
100 200000
150000
100000 50 Counts Counts
50000 60 Co using BaF 2 0 1.0 1.1 1.2 1.3 1.4 1.5 1.0 1.1 1.2 1.3 1.4 1.5 Energy (MeV) Energy (MeV)
60 Figure 6.2.18 Pulse height spectra of Co using 10 mm diameter x 5 mm thick O2-grown undoped YAG
(left) and BaF 2 (right) as the scintillator crystal.
Undoped YAG is hence presented as a potential ultrafast scintillator.
Measurements show that it has faster decay time to and higher energy resolution than
BaF 2, the current standard for fast scintillators [Leo-1994, Rajainmaki-1987, Rodnyi-
247 2001, Saito-2002, Wei-2004]. To our knowledge, undoped YAG has not been studied as a scintillator. Based on our PL measurements, we showed that it has potential in scintillation based on optical luminescence measurements [Varney-2011]. It should be mentioned that the results presented in this chapter use a relatively slow PMT with a sensitivity range that poorly matches undoped YAG, so a more high quality PMT that is more sensitive to wavelengths around 300 nm will undoubtedly show better scintillation performance of undoped YAG. These results demand further exploration into its scintillation properties and potential as a fast scintillator.
Undoped YAG conclusions
Undoped has high energy resolution and very fast decay. The best energy resolution measured was 11%, easily surpassing the energy resolution of BaF 2, measured on our system as 17%. Decay time of this same sample was found to be significantly faster than BaF 2. The high energy resolution and ultrafast scintillation suggests that YAG may be useful in experiments such as PALS, where short decay time is very important for good time resolution.
Investigation into the effects of growth atmosphere and post-growth treatments
revealed the importance of defects on undoped YAG scintillation. Ar-grown YAG,
shown in previous chapters to have the highest concentration of defects, demonstrates the
best scintillation properties of the samples tested. H 2-grown YAG has the worst energy
resolution (16%), potentially an effect of abundant H impurities. Al sputtering decreased
energy resolution and increased decay time of O 2-grown YAG, the opposite effect it had
on Ce:YAG scintillation.
248 6.3 REFERENCES
[Audi-2003] G. Audi, O. Bersillon, J. Blachot, A.H. Wapstra, “The NUBASE evaluation of nuclear and decay properties,” Nuclear Physics A 729 (2003) 3-128
[Babin-2005] V.Babin, K. Blazek, A. Krasnikov, K. Nejezchleb, M. Nikl, T. Savikhina,
and S. Zazubovich, “Luminescence of undoped LuAG and YAG crystals,” Physica Status
Solidi (C) 2 (2005) 97-100
[Blazek-2004] K. Blazek, A. Krasnikov, K. Nejezchleb, M. Nikl, T. Savikhina, and S.
3+ Zazubovich, “Luminescence and defects creation in Ce -doped Lu 3Al 5O12 crystals,”
Physica Status Solidi (B) 241 (2004) 1134-1140
[Hayes-1980] W. Hayes, M. Yamaga, D.J. Robbins, and B. Cockayne, “Optical detection
of EPR of recombination centers in YAG,” Journal of Physics C: Solid State Physics 13
(1980) L1085-L1089
[Hennecke-1967] H.J. Hennecke, J.C. Manthuruthil, and O. Bergman, “Internal- conversion-electron study of the decay of Ba 133 to Cs 133 ,” Physical Review 159 (1967)
955-968
[Kirm-2000] M. Kirm, A. Lushchik, Ch. Lushchik, and G. Zimmerer, “Investigation of
3+ luminescence properties of pure and Ce doped Y 3Al 5O12 crystals using VUV radiation,”
ECS Proceedings 99-40 (2000) 113-122
[Leo-1994] W.R. Leo, Techniques for Nuclear and Particle Physics Experiments: and
How-to Approach ,” Springer-Verlag, Berlin, Germany (1994)
[Ludziejewski-1997] T. Ludziejewski, M. Moszy ński, M. Kapusta, D. Wolski, W.
Klamra, and K Moszy ńska, “Investigation of some scintillation properties of YAG:Ce
249 crystals,” Nuclear Instrumentation and Methods in Physics Research A 398 (1997) 287-
294
[Melissinos-1966] A.C. Melissinos, Experiments in Modern Physics , Academic Press,
Inc., San Diego, CA, 92101
[Moszynski-1994] M. Moszy ński, T. Ludziejewski, D. Wolski, W. Klamra, and L.O.
Norlin, “Properties of the YAG:Ce scintillator,” Nuclear Instrumentation and Methods in
Physics Research A 345 (1994) 461-467
[Nikl-2005] M. Nikl, “Energy transfer phenomena in the luminescence of wide band-gap scintillators,” Physica Status Solidi (A) 202 (2005) 201-206
[Rajainmaki-1987] H. Rajainmäki, “High-resolution positron lifetime spectrometer with
BaF2 scintillators,” Applied Physics A 42 (1987) 205-208
[Robbins-1979(b)] D.J. Robbins, B. Cockayne, J.L. Glasper, and B. Lent, “The temperature dependence of rare-earth activated garnet phosphors I. Intensity and lifetime measurements on undoped and Ce-doped Y 3Al 5O12 ,” Journal of the Electrochemical
Society 126 (1979) 1213-1220
[Rodnyi-2001] P.A. Rodnyi, “Progress in fast scintillators,” Radiation Measurements 33
(2001) 605-614
[Saito-2002] H. Saito, Y. Nagashima, T. Kurihara, and T. Hyodo, “A new positron
lifetime spectrometer using a fast digital oscilloscope and BaF 2 scintillators,” Nuclear
Instrumentation and Methods in Physics Research A 487 (2002) 612-617
[Selim-2007] F.A. Selim, D. Solodovnikov, M.H. Weber, and K.G. Lynn, “Identification of defects in Y 3Al 5O12 crystals by positron annihilation spectroscopy,” Applied Physics
Letters 91 (2007) 104105
250 [Sobolev-1994] B.P. Sobolev, E.A. Krivandina, S.E. Derenzo, W.W. Moses, A.C. West,
“Suppression of BaF 2 slow component of x-ray luminescence in non-stoichiometric
Ba 0.9 R0.1 F2.1 crystals (R = rare earth element),” Proceedings of The Material Research
Society: Scintillator and Phosphor Materials, M.J. Weber, ed., San Francisco, CA (1994)
277–283
[Solodovnikov-2008] D. Solodovnikov, M.H. Weber, and K.G. Lynn, “Improvement in scintillation performance of Ce, Er codoped yttrium aluminum garnet crystals by means of a postgrowth treatment,” Applied Physics Letters 93 (2008) 104102
[Stanek-2008] C.R. Stanek, M.R. Levy, K.J. McClellan, B.P. Uberuaga, and R.W.
Grimes, “Defect identification and compensation in rare earth oxide scintillators,”
Nuclear Instrumentation and Methods in Physics Research B 266 (2008) 2657-2664
[Varney-2011] C.R. Varney, S.M. Reda, D.T. Mackay, M.C. Rowe, and F.A. Selim,
“Strong visible and near infrared luminescence in undoped YAG single crystals,” AIP
Advances 1 (2011) 042170
[Vedda-2004] A. Vedda, D. Di Martino, M. Martini, J. Mares, E. Mihokova, M. Nikl, N.
Solovieva, K. Blazek, and K. Nejezchleb, “Trap levels in Y-aluminum garnet scintillating crystals,” Radiation Measurements 38 (2004) 673-676
[Weber-2004] M.J. Weber, “Scintillation: mechanisms and new crystals,” Nuclear
Instrumentation and Methods in Physics Research A 527 (2004) 9-14
[Wei-2004] L. Wei, Z. Zhang, C. Ma, C. Zhou, B. Wang, T. Chang, and S. Ren, “A high- resolution BaF 2 positron lifetime spectrometer with high counting rate,” Material Science
Forum 445-446 (2004) 516-518
251 [Williams-1990] R.T. Williams and K.S. Song “The self-trapped exciton,” Journal of
Physics and Chemistry in Solids 51 (1990) 679-716
[Zorenko-2004(a)] Yu.V. Zorenko, I.V. Konstankevych, V.V. Mikha ĭlin, V.N.
Kolobanov, and D.A. Spasski ĭ, “Luminescence of excitons in single-crystal garnets,”
Optics and Spectroscopy 96 (2004) 390-397
[Zorenko-2004(b)] Y. Zorenko, A. Voloshinovskii, I. Konstankevych, V. Kolobanov, V.
Mikhailin, and D. Spassky, “Luminescence of excitons and antisite defects in the phosphors based on garnet compounds,” Radiation Measurements 38 (2004) 677-680
[Zorenko-2007] Yu. Zorenko, A. Voloshinovskii, V. Savchyn, T. Voznyak, M Nikl, K
Nejezchleb, V. Mikhailin, V. Kolobanov, and D. Spassky, “Exciton and antisite defect- related luminescence in Lu 3Al 5O12 and Y 3Al 5O12 garnets,” Physica Status Solidi (B) 244
(2007) 2180-2189
[Zorenko-2010(b)] Yu. Zorenko, J.A. Mares, P. Prusa, M. Nikl, V. Gorbenko, V.
Savchyn, R. Kucerkova, and K. Nejezchleb, “Luminescence and scintillation
characteristics of YAG:Ce single crystalline films and single crystals,” Radiation
Measurements 45 (2010) 389-391
[Zorenko-2012] Y. Zorenko, T. Zorenko, V.V. Gorbenko, T. Voznyak, V. Savchyn, P.
Bilski, and A. Twardak, “Peularities of luminescent and scintillation properties of
YAG:Ce phosphor prepared in different crystalline forms,” Optical Materials 34 (2012)
1314-1319
[Zych-2000] E. Zych, C. Brecher, and J. Glodo, “Kinetics of cerium emission in a
YAG:Ce single crystal: the role of traps,” Journal of Physics: Condensed Matter 12
(2000) 1947-1958
252 CHAPTER SEVEN
CONCLUSION
Summary of work
The subject of this research was to study trapping and luminescence phenomena in YAG crystals. Optical measurements involving absorption, luminescence, and color center activation provided information pertaining to electronic transitions and processes in YAG and served as the foundation for further investigations in later chapters. A new spectrometer was developed for luminescence investigation. Thermoluminescence (TL) measurements revealed information regarding trap depth. TL measurements revealed very high shallow trap concentration for Ce:YAG, showing high TL intensity for peaks below room temperature, which shows that the majority of charge carrier trapping defects in Ce:YAG are unstable at room temperature and hence Ce:YAG has more efficient optical and scintillation properties at room temperature than at lower temperatures. There are also a high concentration of deep traps in Ce:YAG stable at high temperatures, but they do not to be in such abundance due to decreased TL intensity. The first positron annihilation lifetime spectroscopy (PALS) measurements ever reported on YAG are included in this work. PALS measurements indicate the presence of vacancy clusters including both anion (oxygen) and cation (aluminum) vacancies. Result of PALS measurements are important to complex oxides in general as they unambiguously show the types of defects present in this sort of crystal. PALS and TL measurements alike showed that oxidation and reduction can alter defect size and structure in YAG.
Annealing in an atmosphere containing oxygen can add an oxygen ion at the site of a
253 vacancy, shrinking the physical size, potential depth, and charge state of the defect.
Scintillation measurements lead to the discovery of a new material for fast scintillation.
Undoped YAG showed better decay time to BaF 2, the current standard for fast scintillation, with much improved energy resolution.
Positron measurements offered new insight into the defect structure of YAG. The very long lifetimes observed in these measurements confirmed the hypotheses of several separate researchers who previously postulated that defect aggregates including both anion and cation vacancies exist in YAG. The PALS measurements that show a decreased defect size but similar defect concentration after annealing in oxygen further support this, as does the observation of small cation vacancy defects in oxygen grown samples.
Scintillation measurements confirmed the high energy resolution and efficiency of
Ce:YAG previously studied and revealed very fast and high resolution scintillation for undoped YAG. Scintillation of Ce:YAG is based on intense Ce 3+ emission peaked in the
visible range. For these samples, minimizing defect concentration is crucial to obtaining
fast decay and good energy resolution. Hydrogen and growth in reducing conditions are
important for Ce:YAG scintillation and diffusion of Al into the sample improved
scintillation, reflecting increased energy resolution. This is almost the opposite of what is
required for good undoped YAG scintillation. Undoped YAG scintillation relies on UV
3+ emission that arises from lattice defects, postulated to be Y Al antisite defects (ADs) and self-trapped excitons (STEs) connected with aluminum vacancies. ADs are not desired since their emission has a very long decay, but Al vacancies are very important and a higher concentration results in faster decay and higher energy resolution. The octahedral
254 Al site can form an AD far more favorably than the tetrahedral Al site can, so it may be beneficial to attempt to synthesize undoped YAG crystals with a high concentration of tetrahedral Al vacancies and full octahedral Al sites such that AD concentration is minimized and Al vacancy concentration is maximized, maximizing the fast decay signal while minimizing the long decay component of scintillation. One possible method for this is to include lightly dope a sample with an impurity such as iron that has a slight preference to the octahedral site over the tetrahedral. Iron is not a good example, though, as it can easily enter either Al site. Perhaps larger nonluminescent dopants could be explored in future work to this end. Al vacancies must remain vacant and impurities such as hydrogen located in an Al vacancy can decrease scintillation performance. Often with optical applications the goal is to minimize defects, but for undoped YAG as a scintillator it is important to control certain defects and imperative that they are included.
The attempt to diffuse Al into the samples by sputtering and annealing in argon produced mixed results. When the aluminum vacancy was already occupied by hydrogen or there simply was not enough oxygen around the vacancy to make incorporation of an
Al 3+ ion favorable and Al was unable to enter into the sample, as was the case when this
treatment was attempted on the H 2-grown YAG sample. When the primary defects
present in the sample were isolated aluminum vacancies, as was likely the case for O 2- grown YAG, the aluminum was able to incorporate into the crystal structure. This process can eliminate Al vacancies, which is beneficial for most applications but is not advised for undoped YAG fast scintillator crystals.
255 Future work
The studies presented in this work suggest that further investigation into undoped
YAG as a fast scintillator may be a wise next step. Ce:YAG is a well-known scintillator with high efficiency, but undoped YAG is a new proposal. PALS measurements could be attempted using undoped YAG samples as scintillators and the resolution and lifetimes can be compared to measurements conducted using BaF2 detectors.
Another direction for future work is to expand upon the library of PALS
measurements on YAG begun in this work. A systematic approach would be to study
samples containing other dopants. Alternatively, studies could be designed to more
directly compare PALS results to scintillation efficiency. In the work presented in this
thesis, both PALS and undoped YAG scintillation were new and largely unknown, so
connections were made after the measurements were taken. Now that the foundation has
been set, it may be possible to use one measurement to predict results of the other. It may
also be necessary to compare results of x-ray stimulated luminescence (XSL), PALS, and
scintillation measurements to fully understand trapping and defect structure in order to
work toward designing a more efficient scintillation crystal, especially since its
scintillation is based on defects.
Low temperature film and powder synthesis has been shown to eliminate ADs
since there is insufficient thermal energy available to create ADs. This begs for the
synthesis of nonstoichiometric YAG in an attempt to increase Al vacancies and STE
signal. This could theoretically result in samples that have very low AD emission and
intense STE emission, which would provide very fast and reliable scintillation. Of
course, films and powders by themselves are much too small to be useful as scintillators,
256 but powders can be pressed to form large ceramic samples and can undergo treatments to become transparent, an essential characteristic of scintillator crystals.
257 APPENDIX A
PALS RESULTS TABLES
Results of each individual positron lifetime measurement are presented, usually entailing multiple measurements conducted on the same sample to verify consistence. Shown in the tables are the lifetimes and intensities of the fit lifetimes, the variance of the fit (Var.), and, for cases where the spectrum is fit to two lifetimes, the bulk and average lifetimes, defect trapping rate, and approximate defect concentration and the type of positron- trapping vacancy defect hypothesized to exist. It should be noted that a variance of unity is ideal and a variance of less than 1.000 can indicate overfitting of either the resolution function of the lifetimes. Lifetimes are given in nanoseconds, intensities are given as a percent of the total amount of observed positron annihilation events within the crystal (I 1
-1 + I 2 = 100%), defect trapping rate κd is given in ns , and defect concentrations are given
in cm -3. Results for each sample are tabulated in the chronological order in which they were recorded.
Table A.1 PALS results for polished Ar-grown undoped YAG 5x5x1 mm as grown.
-1 -3 τ1 [ns] (I 1) τ2 [ns] (I 2) Var. τB [ns] τAV [ns] κd [ns ] C [cm ]
As grown
0.1559 ± 0.0007 0.2947 ± 0.0039 1.145 0.1667 ± 0.1750 ± 0.42 ± 4 x 10 17 ±
(86.24 ± 0.71) (13.76 ± 0.71) 0.0016 0.0025 0.02 2 x 10 16
Al+nO
258 0.1595 ± 0.0008 0.2936 ± 0.0063 1.066 0.1681 ± 0.1745 ± 0.32 ± 3 x 10 17 ±
(88.79 ± 0.98) (11.21 ± 0.98) 0.0021 0.0034 0.03 3 x 10 16
Al+nO
0.1572 ± 0.0017 0.2790 ± 0.0110 1.143 0.1666 ± 0.1730 ± 0.36 ± 4 x 10 17 ±
(87.05 ± 2.19) (12.95 ± 2.19) 0.0048 0.0073 0.06 6 x 10 16
Al+nO
Table A.2 PALS results for unpolished (fine ground) Ar-grown undoped YAG 10x10x1 mm as grown.
-1 -3 τ1 [ns] (I 1) τ2 [ns] (I 2) Var. τB [ns] τAV [ns] κd [ns ] C [cm ]
As grown
0.1597 ± 0.0006 0.3166 ± 0.0075 1.221 0.1656 ± 0.1710 ± 0.22 ± 2 x 10 17 ±
(92.82 ± 0.66) (7.18 ± 0.66) 0.0014 0.0025 0.02 2 x 10 16
Al+nO
0.1616 ± 0.0015 0.2840 ± 0.0137 1.028 0.1682 ± 0.1728 ± 0.24 ± 2 x 10 17 ±
(90.87 ± 1.93) (9.13 ± 1.93) 0.0042 0.0066 0.05 5 x 10 16
Al+nO
259 Table A.3 PALS results for unpolished (fine ground) Ar-grown undoped YAG 5x5x1 mm as grown and annealed in air.
-1 -3 τ1 [ns] (I 1) τ2 [ns] (I 2) Var. τB [ns] τAV [ns] κd [ns ] C [cm ]
As grown
0.1576 ± 0.0006 0.3031 ± 0.0046 1.443 0.1661 ± 0.1731 ± 0.33 ± 3 x 10 17 ±
(89.32 ± 0.62) (10.68 ± 0.62) 0.0014 0.0022 0.02 2 x 10 16
Al+nO
Annealed in air 96 hours at 1200°C
0.1607 ± 0.0009 0.2757 ± 0.0055 1.068 0.1705 ± 0.1765 ± 0.36 ± 4 x 10 17 ±
(86.27 ± 1.25) (13.74 ± 1.25) 0.0028 0.0041 0.03 3 x 10 16
Al+nO
0.1593 ± 0.0012 0.2689 ± 0.0062 1.187 0.1698 ± 0.1759 ± 0.39 ± 4 x 10 17 ±
(84.87 ± 1.62) (15.13 ± 1.62) 0.0036 0.0052 0.04 4 x 10 16
Al+nO
0.1525 ± 0.0024 0.2556 ± 0.0086 1.010 0.1666 ± 0.1741 ± 0.55 ± 6 x 10 17 ±
(79.08 ± 3.29) (20.92 ± 3.29) 0.0074 0.0101 0.09 9 x 10 16
Al+nO
0.1615 ± 0.0031 0.2510 ± 0.0024 1.045 0.1681 ± 0.1714 ± 0.24 ± 2 x 10 17 ±
(88.96 ± 5.63) (11.04 ± 5.63) 0.0121 0.0172 0.13 1 x 10 17
Al+nO
260 Table A.4 PALS results for H2-grown undoped YAG 10 mm dia. x ~1 mm as grown, annealed in air, and
subsequently sputtered with Al and annealed in argon.
-1 -3 τ1 [ns] (I 1) τ2 [ns] (I 2) Var. τB [ns] τAV [ns] κd [ns ] C [cm ]
As grown
0.1504 ± 0.0001 1.414 0.1504 ± 0.1504 ± 0 0
(100) 0.0001 0.0001
0.1502 ± 0.0003 1.121 0.1502 ± 0.1502 ± 0 0
(100) 0.0003 0.0003
0.1466 ± 0.0002 1.192 0.1466 ± 0.1466 ± 0 0
(100) 0.0002 0.0002
0.1468 ± 0.0002 1.260 0.1468 ± 0.1468 ± 0 0
(100) 0.0002 0.0002
Annealed in air 48 hours at 1200°C
0.1473 ± 0.0028 0.2356 ± 0.0389 1.034 0.1509 ± 0.1529 ± 0.16 ± 3 x 10 15 ±
(93.62 ± 5.42) (6.38 ± 5.41) 0.0103 0.0155 0.15 3 x 10 15
Al
0.1371 ± 0.0089 0.1788 ± 0.0175 1.208 0.1485 ± 0.1509 ± 0.56 ± 1 x 10 16 ±
(66.95 ± 27.58) (33.05 ± 27.58) 0.0565 0.0827 0.53 1 x 10 16
Al, Al+H I
0.1530 ± 0.0002 1.107 0.1530 ± 0.1530 ± 0 0
(100) 0.0002 0.0002
0.1512 ± 0.0002 1.131 0.1512 ± 0.1512 ± 0 0
(100) 0.0002 0.0002
261 After above anneal in air, sputtered with Al and annealed in Ar 48 hours at 1200°C
0.1467 ± 0.0002 1.124 0.1467 ± 0.1467 ± 0 0
(100) 0.0002 0.0002
0.1467 ± 0.0002 1.113 0.1467 ± 0.1467 ± 0 0
(100) 0.0002 0.0002
0.1456 ± 0.0002 0.965 0.1456 ± 0.1456 ± 0 0
(100) 0.0002 0.0002
Table A.5 PALS results for O2-grown undoped YAG 10 mm dia. x ~1 mm as grown, annealed in a mixed
atmosphere consisting of (~50%) H 2 in Ar, further annealed in a mixed atmosphere consisting of (~50%)
H2 in Ar, further annealed in Ar, further annealed in Ar, and sputtered with Al and annealed in argon.
-1 -3 τ1 [ns] (I 1) τ2 [ns] (I 2) Var. τB [ns] τAV [ns] κd [ns ] C [cm ]
As grown
0.1510 ± 0.0023 0.2285 ± 0.0109 1.262 0.1600 ± 0.1639 ± 0.37 ± 8 x 10 15 ±
(83.39 ± 4.53) (16.61 ± 4.53) 0.0095 0.0127 0.11 2 x 10 15
Al, Al+H I
0.1411 ± 0.0068 0.2048 ± 0.0137 1.133 0.1569 ± 0.1617 ± 0.71 ± 1 x 10 16 ±
(67.69 ± 13.63) (32.31 ± 13.63) 0.0295 0.0345 0.34 7 x 10 15
Al, Al+H I
0.1521 ± 0.0042 0.2241 ± 0.0209 0.935 0.1605 ± 0.1638 ± 0.34 ± 7 x 10 15 ±
(83.78 ± 9.19) (16.22 ± 9.19) 0.0193 0.0254 0.21 4 x 10 15
Al, Al+H I
262 0.1384 ± 0.0157 0.1820 ± 0.0138 0.973 0.1582 ± 0.1612 ± 0.90 ± 2 x 10 16 ±
(47.76 ± 32.90) (52.24 ± 32.90) 0.0756 0.0759 0.75 1 x 10 16
Al, Al+H I
0.1446 ± 0.0057 0.2049 ± 0.0155 1.129 0.1566 ± 0.1604 ± 0.53 ± 1 x 10 16 ±
(73.87 ± 13.19) (26.13 ± 13.19) 0.0279 0.0336 0.29 6 x 10 15
Al, Al+H I
Annealed in a mixed atmosphere of (~50%) H2 in Ar 24 hours at 1200°C
0.1223 ± 0.0361 0.1660 ± 0.0071 1.078 0.1568 ± 0.1589 ± 1.80 ± 4 x 10 16 ±
(16.97 ± 26.34) (83.03 ± 26.34) 0.0667 0.0550 2.09 4 x 10 16
Al, Al+H I
0.1376 ± 0.0125 0.1860 ± 0.0119 1.052 0.1583 ± 0.1619 ± 0.95 ± 2 x 10 16 ±
(49.82 ± 24.42) (50.18 ± 24.42) 0.0561 0.0571 0.59 1 x 10 16
Al, Al+H I
0.1396 ± 0.0122 0.1870 ± 0.0121 0.997 0.1597 ± 0.1632 ± 0.90 ± 2 x 10 16 ±
(50.58± 24.81) (49.42 ± 24.81) 0.0573 0.0586 0.57 1 x 10 16
Al, Al+H I
Further annealed in a mixed atmosphere of (~50%) H2 in Ar 51.5 hours at 1200°C
0.1302 ± 0.0107 0.1858 ± 0.0074 0.924 0.1577 ± 0.1626 ± 1.34 ± 3 x 10 16 ±
(41.81 ± 15.12) (58.19 ± 15.12) 0.0360 0.0349 0.52 1 x 10 16
Al, Al+H I
0.1253 ± 0.0104 0.1841 ± 0.0061 1.045 0.1562 ± 0.1617 ± 1.58 ± 3 x 10 16 ±
(38.02 ± 12.44) (61.98 ± 12.44) 0.0301 0.0282 0.53 1 x 10 16
Al, Al+H I
263 0.1366 ± 0.0079 0.1958 ± 0.0088 1.223 0.1587 ± 0.1639 ± 1.02 ± 2 x 10 16 ±
(53.86 ± 13.50) (46.14 ± 13.50) 0.0310 0.0328 0.37 7 x 1015
Al, Al+H I
Further annealed in a Ar 19 hours at 1200°C
0.1445 ± 0.0053 0.2083 ± 0.0129 0.903 0.1582 ± 0.1625 ± 0.60 ± 1 x 10 16 ±
(71.81 ± 11.15) (28.19 ± 11.15) 0.0240 0.0288 0.26 5 x 10 15
Al, Al+H I
0.1365 ± 0.0082 0.1921 ± 0.0097 1.098 0.1569 ± 0.1615 ± 0.95 ± 2 x 10 16 ±
(55.09 ± 15.41) (44.91 ± 15.41) 0.0347 0.0368 0.40 8 x 10 15
Al, Al+H I
0.1442 ± 0.0060 0.2049 ± 0.0128 1.058 0.1587 ± 0.1645 ± 0.64 ± 1 x 10 16 ±
(69.91 ± 12.86) (31.09 ± 12.86) 0.0280 0.0327 0.29 6 x 10 15
Al, Al+H I
Further annealed in a Ar 48 hours at 1200°C
0.1436 ± 0.0087 0.1951 ± 0.0141 1.022 0.1593 ± 0.1628 ± 0.69 ± 1 x 10 16 ±
(62.68 ± 20.22) (37.32 ± 20.22) 0.0450 0.0496 0.43 9 x 10 15
Al, Al+H I
0.1391 ± 0.0079 0.1942 ± 0.0110 1.031 0.1574 ± 0.1617 ± 0.84 ± 2 x 10 16 ±
(59.00 ± 16.08) (41.00 ± 16.08) 0.0359 0.0390 0.39 8 x 10 15
Al, Al+H I
0.1310 ± 0.0073 0.1912 ± 0.0080 1.022 0.1542 ± 0.1597 ± 1.15 ± 2 x 10 16 ±
(52.25 ± 11.81) (47.75 ± 11.81) 0.0266 0.0279 0.36 7 x 10 15
Al, Al+H I
264 Sputtered with Al and annealed in Ar 48 hours at 1200°C
0.1309 ± 0.0070 0.1888 ± 0.0090 1.108 0.1506 ± 0.1556 ± 1.00 ± 2 x 10 16 ±
(57.27 ± 12.99) (42.73 ± 12.99) 0.0280 0.0303 0.37 7 x 10 15
Al, Al+H I
0.1420 ± 0.0048 0.2060 ± 0.0149 1.040 0.1532 ± 0.1571 ± 0.52 ± 1 x 10 16 ±
(76.46 ± 10.73) (23.54 ± 10.73) 0.0221 0.0273 0.26 5 x 10 15
Al, Al+H I
0.1508 ± 0.0027 0.2365 ± 0.0245 1.132 0.1561 ± 0.1589 ± 0.23 ± 5 x 10 15 ±
(90.60 ± 5.20) (9.40 ± 5.20) 0.0104 0.0150 0.13 3 x 10 15
Al, Al+H I
0.1408 ± 0.0056 0.2020 ± 0.0128 0.973 0.1547 ± 0.1589 ± 0.64 ± 1 x 10 16 ±
(70.39 ± 12.15) (29.61 ± 12.15) 0.0257 0.0304 0.29 6 x 10 15
Al, Al+H I
Table A.6 PALS results for Ce:YAG 0.1% 10 mm dia. x 1 mm as grown, during UV excitation using the
Xenon Pulsed Light Source, after this excitation, after heating to remove any effects of this excitation, during excitation using the 455 nm LED, and during excitation using the 270 nm LED.
-1 -3 τ1 [ns] (I 1) τ2 [ns] (I 2) Var. τB [ns] τAV [ns] κd [ns ] C [cm ]
As grown
0.1475 ± 0.0016 0.2431 ± 0.0160 1.062 0.1525 ± 0.1554 ± 0.22 ± 4 x 10 15 ±
(91.70 ± 2.69) (8.30 ± 2.69) 0.0052 0.0079 0.08 2 x 10 15
Al, Al+H I
265 0.1526 ± 0.0001 1.083 0.1526 ± 0.1526 ± 0 0
(100) 0.0001 0.0001
0.1513 ± 0.0026 0.2065 ± 0.0346 1.063 0.1541 ± 0.1550 ± 0.12 ± 2 x 10 15 ±
(93.21 ± 8.42) (6.79 ± 8.42) 0.0166 0.0218 0.16 3 x 10 15
Al, Al+H I
0.1511 ± 0.0002 1.090 0.1511 ± 0.1511 ± 0 0
(100) 0.0002 0.0002
0.1507 ± 0.0002 0.895 0.1507 ± 0.1507 ± 0 0
(100) 0.0002 0.0002
0.1497 ± 0.0002 1.092 0.1497 ± 0.1497 ± 0 0
(100) 0.0002 0.0002
0.1493 ± 0.0002 0.930 0.1493 ± 0.1493 ± 0 0
(100) 0.0002 0.0002
During UV excitation using the Xenon Pulsed Light Source
0.1409 ± 0.0030 0.2041 ± 0.0201 1.278 0.1465 ± 0.1488 ± 0.27 ± 5 x 10 15 ±
(87.56 ± 7.78) (12.44 ± 7.78) 0.0147 0.0196 0.18 4 x 10 15
Al, Al+H I
0.1483 ± 0.0003 1.280 0.1483 ± 0.1483 ± 0 0
(100) 0.0003 0.0003
After UV excitation using the Xenon Pulsed Light Source
0.1093 ± 0.0055 0.1594 ± 0.0021 1.233 0.1410 ± 0.1451 ± 2.06 ± 4 x 10 16 ±
(28.46 ± 5.78) (71.54 ± 5.78) 0.0131 0.0114 0.37 7 x 10 15
Al, Al+H I
266 0.1474 ± 0.0002 1.341 0.1474 ± 0.1474 ± 0 0
(100) 0.0002 0.0002
0.1366 ± 0.0070 0.1804 ± 0.0195 1.054 0.1456 ± 0.1478 ± 0.45 ± 9 x 10 15 ±
(74.44 ± 22.67) (25.56 ± 22.67) 0.0447 0.0518 0.44 8 x 10 15
Al, Al+H I
After heating following UV excitation using the Xenon Pulsed Light Source
0.1484 ± 0.0003 1.267 0.1484 ± 0.1484 ± 0 0
(100) 0.0003 0.0003
0.1239 ± 0.0170 0.1601 ± 0.0092 1.248 0.1447 ± 0.1469 ± 1.16 ± 2 x 10 16 ±
(36.40 ± 32.79) (63.60 ± 32.79) 0.0708 0.0669 0.95 2 x 10 16
Al, Al+H I
During excitation using the 455 nm LED
0.1488 ± 0.0002 1.040 0.1488 ± 0.1488 ± 0 0
(100) 0.0002 0.0002
0.1335 ± 0.0135 0.1672 ± 0.0174 1.097 0.1461 ± 0.1480 ± 0.65 ± 1 x 10 16 ±
(57.09 ± 44.24) (42.91 ± 44.24) 0.0912 0.0953 0.79 2 x 10 16
Al, Al+H I
During excitation using the 270 nm LED
0.1272 ± 0.0094 0.1721 ± 0.0130 1.081 0.1423 ± 0.1455 ± 0.84 ± 2 x 10 16 ±
(59.27 ± 23.40) (40.73 ± 23.40) 0.0470 0.0507 0.56 1 x 10 16
Al, Al+H I
0.1486 ± 0.0002 1.535 0.1486 ± 0.1486 ± 0 0
(100) 0.0002 0.0002
267
Table A.7 PALS results for Ce:YAG 0.15% 10 mm dia. x 1 mm as grown.
-1 -3 τ1 [ns] (I 1) τ2 [ns] (I 2) Var. τB [ns] τAV [ns] κd [ns ] C [cm ]
As grown
0.1512 ± 0.0004 0.3343 ± 0.0121 0.957 0.1535 ± 0.1563 ± 0.10 ± 1 x 10 17 ±
(97.24 ± 0.33) (2.76 ± 0.33) 0.0007 0.0013 0.01 1 x 10 16
Al+nO
0.1498 ± 0.0023 0.2212 ± 0.0278 1.104 0.1534 ± 0.1550 ± 0.16 ± 2 x 10 17 ±
(92.65 ± 5.59) (7.35 ± 5.59) 0.0109 0.0152 0.13 1 x 10 17
Al+nO
0.1528 ± 0.0007 0.2763 ± 0.0247 1.116 0.1546 ± 0.1561 ± 0.08 ± 8 x 10 16 ±
(97.36 ± 0.99) (2.64 ± 0.99) 0.0191 0.0033 0.03 3 x 10 16
Al+nO
Table A.8 PALS results for Ce:YAG 0.2% 10 mm dia. x 1 mm as grown, annealed in air, and subsequently sputtered with Al and annealed in argon.
-1 -3 τ1 [ns] (I 1) τ2 [ns] (I 2) Var. τB [ns] τAV [ns] κd [ns ] C [cm ]
As grown
0.1491 ± 0.0007 0.2629 ± 0.0116 0.977 0.1527 ± 0.1553 ± 0.16 ± 2 x 10 17 ±
(94.55 ± 1.06) (5.45 ± 1.06) 0.0021 0.0033 0.03 3 x 10 16
Al+nO
268 0.1546 ± 0.0005 0.2992 ± 0.0238 1.341 0.1560 ± 0.1574 ± 0.06 ± 6 x 10 16 ±
(98.09 ± 0.58) (1.91 ± 0.58) 0.0012 0.0021 0.02 2 x 10 16
Al+nO
0.1550 ± 0.0006 0.3032 ± 0.0559 1.258 0.1557 ± 0.1564 ± 0.03 ± 3 x 10 16 ±
(99.09 ± 0.64) (0.91 ± 0.64) 0.0013 0.0023 0.02 2 x 10 16
Al+nO
Annealed in air 96 hours at 1200°C
0.1547 ± 0.0017 0.2721 ± 0.0420 1.139 0.1571 ± 0.1589 ± 0.10 ± 1 x 10 17 ±
(96.43 ± 2.43) (3.57 ± 2.43) 0.0048 0.0079 0.07 7 x 10 16
Al+nO
0.1521 ± 0.0010 0.2600 ± 0.0168 1.200 0.1554 ± 0.1576 ± 0.14 ± 1 x 10 17 ±
(94.90 ± 1.52) (5.10 ± 1.52) 0.0030 0.0047 0.04 4 x 10 16
Al+nO
0.1263 ± 0.0172 0.1674 ± 0.0112 1.168 0.1478 ± 0.1507 ± 1.15 ± 2 x 10 16 ±
(40.64 ± 32.33) (59.36 ± 32.33) 0.0709 0.0685 0.93 2 x 10 16
Al, Al+H I
0.1526 ± 0.0003 1.206 0.1526 ± 0.1526 ± 0 0
(100) 0.0003 0.0003
0.1511 ± 0.0002 1.141 0.1511 ± 0.1511 ± 0 0
(100) 0.0002 0.0002
After above anneal in air, sputtered with Al and annealed in Ar 48 hours at 1200°C
269 0.1389 ± 0.0014 0.1700 ± 0.0022 1.039 0.1483 ± 0.1496 ± 0.45 ± 9 x 10 15 ±
(65.50 ± 0.06) (34.50 ± 0.06) 0.0012 0.0012 0.04 7 x 10 14
Al+H I
0.1251 ± 0.0162 0.1656 ± 0.0121 1.023 0.1450 ± 0.1479 ± 1.10 ± 2 x 10 16 ±
(43.79 ± 33.39) (56.21 ± 33.39) 0.0712 0.0700 0.91 2 x 10 16
Al+H I
0.1418 ± 0.0073 0.1895 ± 0.0309 1.215 0.1488 ± 0.1507 ± 0.33 ± 7 x 10 15 ±
(81.40 ± 23.92) (18.60 ± 23.92) 0.0472 0.0572 0.46 9 x 10 15
Al+H I
Table A.9 PALS results for Ce:YAG 0.14% 10 mm dia. x ~1 mm as grown and annealed in air.
-1 -3 τ1 [ns] (I 1) τ2 [ns] (I 2) Var. τB [ns] τAV [ns] κd [ns ] C [cm ]
As grown
0.1419 ± 0.0025 0.2023 ± 0.0151 1.005 0.1480 ± 0.1502 ± 0.29 ± 6 x 10 15 ±
(86.20 ± 6.77) (13.80 ± 6.77) 0.0130 0.0170 0.15 3 x 10 15
Al, Al+H I
0.1401 ± 0.0051 0.1859 ± 0.0174 1.079 0.1482 ± 0.1502 ± 0.39 ± 8 x 10 15 ±
(77.89 ± 16.66) (22.11± 16.66) 0.0331 0.0392 0.32 6 x 10 15
Al, Al+H I
0.1493 ± 0.0002 1.016 0.1493 ± 0.1493 ± 0 0
(100) 0.0002 0.0002
Annealed in air 48 hours at 1200°C
270 0.1226 ± 0.0225 0.1649 ± 0.0084 1.240 0.1503 ± 0.1530 ± 1.50 ± 3 x 10 16 ±
(28.18 ± 28.33) (71.82 ± 28.33) 0.0659 0.0589 1.25 3 x 10 16
Al+H I
0.1541 ± 0.0003 1.142 0.1541 ± 0.1541 ± 0 0
(100) 0.0003 0.0003
0.1324 ± 0.0123 0.1760 ± 0.0154 1.148 0.1483 ± 0.1513 ± 0.81 ± 2 x 10 16 ±
(56.66 ± 30.43) (43.34 ± 30.43) 0.0641 0.0677 0.68 1 x 10 16
Al+H I
Table A.10 PALS results for polished Ce:YAG 0.3% 5x5x1 mm as grown.
-1 -3 τ1 [ns] (I 1) τ2 [ns] (I 2) Var. τB [ns] τAV [ns] κd [ns ] C [cm ]
As grown
0.1482 ± 0.0003 0.3080 ± 0.0041 1.238 0.1535 ± 0.1587 ± 0.23 ± 2 x 10 17 ±
(93.40 ± 0.31) (6.60 ± 0.31) 0.0006 0.0011 0.01 1 x 10 16
Al+nO
0.1550 ± 0.0004 0.3702 ± 0.0059 1.256 0.1611 ± 0.1690 ± 0.24 ± 2 x 10 17 ±
(93.49 ± 0.30) (6.51 ± 0.30) 0.0007 0.0013 0.01 1 x 10 16
Al+nO
0.1497 ± 0.0010 0.3049 ± 0.0090 1.056 0.1577 ± 0.1652 ± 0.34 ± 3 x 10 17 ±
(89.98 ± 1.09) (10.02 ± 1.03) 0.0022 0.0037 0.04 4 x 10 16
Al+nO
271 Table A.11 PALS results for unpolished (fine ground) Ce:YAG 0.3% 5x5x1 mm annealed in air and subsequently annealed in nitrogen.
-1 -3 τ1 [ns] (I 1) τ2 [ns] (I 2) Var. τB [ns] τAV [ns] κd [ns ] C [cm ]
Annealed in air 96 hours at 1200°C
0.1549 ± 0.0009 0.2949 ± 0.0138 1.112 0.1591 ± 0.1627 ± 0.17 ± 2 x 10 17 ±
(94.40 ± 1.02) (5.60 ± 1.02) 0.0021 0.0036 0.03 3 x 10 16
Al+nO
Further annealed in nitrogen 48 hours at 1200°C
0.1543 ± 0.0010 0.3121 ± 0.0080 1.101 0.1597 ± 0.1647 ± 0.21 ± 2 x 10 17 ±
(93.45 ± 0.60) (6.55 ± 0.60) 0.0013 0.0022 0.02 2 x 10 16
Al+nO
0.1557 ± 0.0010 0.3170 ± 0.0111 1.261 0.1595 ± 0.1633 ± 0.15 ± 2 x 10 17 ±
(95.28 ± 0.58) (4.72 ± 0.58) 0.0012 0.0022 0.02 2 x 10 16
Al+nO
0.1543 ± 0.0010 0.3145 ± 0.0164 0.969 0.1589 ± 0.1635 ± 0.19 ± 2 x 10 17 ±
(94.29 ± 1.05) (5.71 ± 1.05) 0.0022 0.0039 0.04 4 x 10 16
Al+nO
272 Table A.12 PALS results for Nd:YAG 1% 5x5x1 mm as grown.
-1 -3 τ1 [ns] (I 1) τ2 [ns] (I 2) Var. τB [ns] τAV [ns] κd [ns ] C [cm ]
As grown
0.1662 ± 0.0006 0.3506 ± 0.0090 1.395 0.1714 ± 0.1768 ± 0.18 ± 2 x 10 17 ±
(94.25 ± 0.50) (5.75 ± 0.50) 0.0012 0.0021 0.02 2 x 10 16
Al+nO
0.1607 ± 0.0008 0.3006 ± 0.0069 1.202 0.1681 ± 0.1739 ± 0.27 ± 3 x 10 17 ±
(90.55 ± 0.87) (9.45 ± 0.87) 0.0019 0.0031 0.03 3 x 10 16
Al+nO
0.1579 ± 0.0017 0.2671 ± 0.0091 1.168 0.1684 ± 0.1745 ± 0.39 ± 4 x 10 17 ±
(84.78 ± 2.40) (15.22 ± 2.40) 0.0061 0.0108 0.07 7 x 10 16
Al+nO
Table A.13 PALS results for Tm:YAG 0.8% 5x5x1 mm as grown.
-1 -3 τ1 [ns] (I 1) τ2 [ns] (I 2) Var. τB [ns] τAV [ns] κd [ns ] C [cm ]
As grown
0.1547 ± 0.0004 0.3855 ± 0.0216 1.276 0.1562 ± 0.1584 ± 0.06 ± 6 x 10 16 ±
(98.41 ± 0.26) (1.59 ± 0.26) 0.0006 0.0012 0.01 1 x 10 16
Al+nO
0.1517 ± 0.0007 0.2852 ± 0.0095 1.140 0.1568 ± 0.1609 ± 0.21 ± 2 x 10 17 ±
(93.10 ± 0.90) (6.90 ± 0.90) 0.0018 0.0031 0.03 3 x 10 16
Al+nO
273 0.1501 ± 0.0010 0.3027 ± 0.0130 1.013 0.1556 ± 0.1608 ± 0.24 ± 2 x 10 17 ±
(92.97 ± 1.09) (7.03 ± 1.09) 0.0022 0.0039 0.04 4 x 10 16
Al+nO
Table A.14 PALS results for Yb:YAG 5% 10 mm dia. x ~1 mm cut into quarters annealed in air.
-1 -3 τ1 [ns] (I 1) τ2 [ns] (I 2) Var. τB [ns] τAV [ns] κd [ns ] C [cm ]
Annealed in air 96 hours at 1200°C
0.1527 ± 0.0007 0.3635 ± 0.0168 1.161 0.1560 ± 0.1604 ± 0.14 ± 1 x 10 17 ±
(96.32 ± 0.50) (3.68 ± 0.50) 0.0011 0.0022 0.02 1 x 10 16
Al+nO
0.1518 ± 0.0005 0.3628 ± 0.0115 1.039 0.1551 ± 0.1595 ± 0.14 ± 1 x 10 17 ±
(96.37 ± 0.33) (3.63 ± 0.33) 0.0008 0.0015 0.01 2 x 10 16
Al+nO
0.1516 ± 0.0007 0.3738 ± 0.0158 1.013 0.1555 ± 0.1609 ± 0.16 ± 2 x 10 17 ±
(95.82 ± 0.50) (4.18 ± 0.50) 0.0011 0.0022 0.02 2 x 10 16
Al+nO
0.1511 ± 0.0005 0.3619 ± 0.0082 1.114 0.1564 ± 0.1633 ± 0.22 ± 2 x 10 17 ±
(94.22 ± 0.38) (5.78 ± 0.38) 0.0008 0.0016 0.02 2 x 10 16
Al+nO
274 Table A.15 PALS results for Yb:YAG 10% 5x5x1 mm as grown and annealed in air.
-1 -3 τ1 [ns] (I 1) τ2 [ns] (I 2) Var. τB [ns] τAV [ns] κd [ns ] C [cm ]
As grown
0.1541 ± 0.0005 0.3514 ± 0.0092 1.152 0.1582 ± 0.1631 ± 0.17 ± 2 x 10 17 ±
(95.42 ± 0.37) (4.58 ± 0.37) 0.0008 0.0016 0.01 1 x 10 16
Al+nO
0.1543 ± 0.0007 0.3465 ± 0.0138 1.198 0.1584 ± 0.1633 ± 0.17 ± 2 x 10 17 ±
(95.32 ± 0.59) (4.68 ± 0.59) 0.0013 0.0024 0.02 2 x 10 16
Al+nO
Annealed in air 48 hours at 1200°C
0.1572 ± 0.0009 0.3565 ± 0.0112 1.065 0.1637 ± 0.1713 ± 0.25 ± 3 x 10 17 ±
(92.91 ± 0.68) (7.09 ± 0.68) 0.0016 0.0029 0.03 3 x 10 16
Al+nO
0.1571 ± 0.0005 0.3794 ± 0.0078 1.195 0.1631 ± 0.1710 ± 0.23 ± 2 x 10 17 ±
(93.72 ± 0.37) (6.26 ± 0.37) 0.0008 0.0016 0.01 1 x 10 16
Al+nO
0.1507 ± 0.0010 0.3147 ± 0.0143 0.961 0.1555 ± 0.1605 ± 0.21 ± 2 x 10 17 ±
(94.05 ± 0.92) (5.95 ± 0.92) 0.0019 0.0035 0.03 3 x 10 16
Al+nO
275 Table A.16 PALS results for undoped YAG (labeled as Yb:YAG 5%) 10 mm dia. x 1 mm cut into quarters as grown and UV excited 30 minutes such that both samples showed visible coloring in the region probed by positrons.
-1 -3 τ1 [ns] (I 1) τ2 [ns] (I 2) Var. τB [ns] τAV [ns] κd [ns ] C [cm ]
As grown
0.1594 ± 0.0008 0.3141 ± 0.0076 1.236 0.1671 ± 0.1739 ± 0.29 ± 3 x 10 17 ±
(90.62 ± 0.83) (9.38 ± 0.83) 0.0018 0.0031 0.03 3 x 10 16
Al+nO
0.1600 ± 0.0011 0.3294 ± 0.0121 0.963 0.1667 ± 0.1732 ± 0.25 ± 3 x 10 17 ±
(92.19 ± 1.00) (7.81 ± 1.00) 0.0022 0.0039 0.03 3 x 10 16
Al+nO
UV excited 30 minutes such that both samples are colored
0.1664 ± 0.0005 0.3941 ± 0.0043 1.289 0.1778 ± 0.1917 ± 0.39 ± 4 x 10 17 ±
(88.89 ± 0.34) (11.11 ± 0.34) 0.0009 0.0016 0.01 1 x 10 16
Al+nO
0.1650 ± 0.0009 0.3743 ± 0.0068 1.060 0.1781 ± 0.1925 ± 0.45 ± 4 x 10 17 ±
(86.84 ± 0.70) (13.16 ± 0.70) 0.0017 0.0031 0.02 2 x 10 16
Al+nO
276