AN ABSTRACT OF THE DISSERTATION OF

Transcription

1 AN ABSTRACT OF THE DISSERTATION OF Joayoung Jeong for the degree of Doctor of Philosophy in Chemistry presented on December 6, Title: High Quantum-Yield Phosphors via Quantum Splitting and Upconversion Abstract approved: Douglas A. Keszler The Gd 3+ ion has been used to induce quantum splitting in luminescent materials by using cross-relaxation energy transfer (CRET). In Nd:LiGdF 4, quantum splitting results from a two-step CRET between Gd 3+ and Nd 3+, first involving a transition 6 G 6 I on Gd 3+ and an excitation within the 4f 3 configuration of Nd 3+ followed by a second CRET that brings Gd 3+ to 6 P 7/2. The excited Nd 3+ ion rapidly relaxes nonradiatively to the emitting 4 F 3/2. The excited Gd 3+ ion then transfers its energy back to Nd 3+, which gives rise to the second photon. The result is a quantum yield of 1.05 ± 0.35 with emission in the NIR following excitation at 175 nm. GdF 3 :Pr 3+, Eu 3+ also exhibits quantum splitting, but only at very low concentration of Pr 3+ (0.3%) and Eu 3+ (0.2%), resulting in a quantum yield of approximately 20% under 160-nm excitation. Host intrinsic emission via a self-trapped exciton (STE) was also examined as a means to sensitize Gd 3+ emission. The material ScPO 4 :Gd 3+ exhibits a high absolute quantum yield of 0.9 ± 0.2 under 170-nm excitation, demonstrating a potentially new and efficient pathway for exciting quantum splitting phosphors. Single crystals of the material GdZrF 7 were grown, and its structure was established via single-crystal X-ray diffraction methods. Doped samples of GdZrF 7 :Yb 3+, Er 3+ exhibit bright up-conversion luminescence with light output that is up to twice that of a commercial material based on the host Gd 2 O 2 S. When doped with Eu 3+, the fluoride also emits a nearly white color under vacuum ultraviolet excitation with an absolute quantum yield near 0.9. The new compound Gd 4.67 (SiO 4 ) 3 S

2 was synthesized and studied. The structure was established via single-crystal X-ray methods, and the luminescence of Tb 3+ samples was investigated.

3 Copyright by Joayoung Jeong December 6, 2007 All Rights Reserved

4 High Quantum-Yield Phosphors via Quantum Splitting and Upconversion by Joayoung Jeong A DISSERTATION submitted to Oregon State University in partial fulfillment of the requirements for the degree of Doctor of Philosophy Presented December 6, 2007 Commencement June 2008

5 Doctor of Philosophy dissertation of Joayoung Jeong presented on December 6, 2007 APPROVED: Major Professor, representing Chemistry Chair of the Department of Chemistry Dean of the Graduate School I understand that my dissertation will become part of the permanent collection of Oregon State University libraries. My signature below authorizes release of my dissertation to any reader upon request. Joayoung Jeong, Author

6 ACKNOWLEDGMENTS First of all I want to say thank you to Douglas Keszler for his endless support, his patience, his guidance, and his acceptance of me as a doctoral student working under his guidance. I want to thank my committee members Drs. Janet Tate, Philip Watson, Michael M. Lerner, William W. Warren, Jr., and John E. Baham for their encouragement in helping me to successfully complete my doctoral program. In addition, I also want to express my gratefulness to those professors who provided instruction in my course work: Drs. Arthur W. Sleight, Wei Kong, Joseph W. Nibler, William H. Warnes, and Milo D. Koretsky. For device testing, everyday interactions, and friendship, I want to extend my appreciation to the present and former members of the Keszler, Tate, Wager, and Chang s groups : Dr. Sangmoon Park, Dr. Cheol-Hee Park, Ji-Eun Yi, Kai Jiang, Dr. Mike Hruschka, Mike Shoemaker, Jeremy Anderson, Jason Stowers, Dr. Peter Hersh, Heather Platt, Stephen Meyer, Bahar Özmen, Dr. Liping Guo, Robert Kykyneshi, Benjamin C. Nielsen, Paul Newhouse, Dr. Hai Chiang, David Hong, Doo-Hyoung Lee, Seung-Yeul Han. I want to thank Drs. K.C. Mishira and M. Raukas at Osram Sylvania for numerous helpful discussions. I am also grateful to Ted Hinke for his contributions in equipment design and Joe Magner for maintenance of research tools. This research work was funded by the U.S. National Science Foundation, Grant Nos, (RSM) and (DAK). Most of all, I deeply appreciate my wife, Mikyeoung, and my son, Jinha, who always stand by me with so much love.

7 CONTRIBUTION OF AUTHORS I am most thankful to my co-workers Dr. Richard S. Meltzer and his graduate student, Yi Zhou, at University of Georgia for measurement of VUV luminescent characteristics and the precious answers to my many questions. Their contributions are significantly represented in Chapter 2-5. Dr. Lev Zakharov provided invaluable assistance in completing the crystal-structure analyses described in chapter 7 and 9.

8 TABLE OF CONTENTS Page CHAPTER 1. INTRODUCTION INTRODUCTION GENERAL CONSIDERATION Selection of Host Compound for Rare Earth Energetic postion of the lowest 5d Level The Stokes shift Energy transfer QUANTUM SPLITTING PCE Dynamics Quantum Splitting by Cross Relaxation Energy Transfer (CRET) DISSERTATION SUMMARY REFERENCES CHAPTER 2. QUANTUM SPLITTING AND ITS DYNAMICS IN GdLiF 4 :Nd INTRODUCTION EXPERIMENT Demonstaration of the Quantum Splitting Excitation spectrum and quantum yield Dynamics of the quantum splitting DISCUSSION CONCLUSION REFERENCES CHAPTER 3. SENSITIZATION OF Gd 3+ AND THE DYNAMICS OF QUANTUM SPLITTING IN GdF 3 :Pr,Eu INTRODUCTION RESULTS AND DISCUSSION... 51

9 TABLE OF CONTENTS (Continued) Page REFERENCES CHAPTER 4. RELAXATION OF THE 4f n-1 5d 1 ELECTRONIC STATES OF RARE EARTH IONS IN YPO 4 AND YBO INTRODUCTION RESULTS AND DISCUSSION CONCLUSION REFERENCES CHAPTER 5. HOST SENSITIZATION OF Gd 3+ IONS ON YTTRIUM AND SCANDIUM BORATES AND PHOSPHATES FOR APPLICATIONS IN QUANTUM SPLITTING INTRODUCTION EXPERIMENTAL RESULTS AND DISCUSSION ScBO YBO ScPO YPO ENERGY TRANSFER RATES CONCLUSIONS...97 REFERENCES...98 CHAPTER 6. LUMINESCENCE OF LANTHANIDES DOPED GdZrF INTRODUCTION EXPERIMENT RESULTS AND DISCUSSION CONCLUSION REFERENCES CHAPTER 7. CRYSTAL STRUCTURE AND Eu 3+ LUMINESCENCE OF

10 TABLE OF CONTENTS (Continued) Page GdMF 7 (M=Hf 4+, Zr 4+ ) INTRODUCTION EXPERIMENT RESULTS AND DISCUSSION Crystal structure luminescence characteristics CONCLUSION REFERENCES CHAPTER 8. THE NEW EFFICIENT UPCONVERSION GREEN PHOSPHOR GdZrF 7 :Yb 3+,Er INTRODUCTION EXPERIMENT STRUCTURAL CHARACTERISTICS LUMINESCENCE CHARACTERISTICS Optimal Er 3+ and Yb 3+ concentration Color purity change vs. Yb 3+ concentrration Effect of Yb 3+ concentration on red emission output Luminescence dependency on the excitation intensity CONCLUSION REFERENCES CHAPTER 9. CRYSTAL STRUCTURE AND LUMINESCENT PROPERTIES OF THE APATITE Gd 4.67 (SiO 4 ) 3 S INTRODUCTION EXPERIMENT RESULTS AND DISCUSSION CONCLUSION REFERENCES

11 TABLE OF CONTENTS (Continued) Page CHAPTER 10. CONCLUSION BIBLIOGRAPHY APPENDICES Appendix A. Luminescent measurement system Appendix B. A HIGH MOBILITY TRANSPARENT THIN-FILM TRANSISTOR WITH AN AMORPHOUS ZINC TIN OXIDE CHANNEL Appendix C. CURRICULUM VITA...202

12 LIST OF FIGURES Figure Page 1.1. Energy level diagram of Pr 3+ ion. PCE process is indicated by two successive transitions from the 1 S 0 level following excitation into the 4f5d band.(adopted from A. P. Vink, P. Dorenbos, C. W. E. Van Eijk, Journal of Solid State Chemistry, 171, (2003)) (a) Emission spectrum of YF 3 :Pr 3+ (b) Schematic of PCE; initial photon emission from 1 S 0 and the second photon emission from 3 P Schematic diagram showing the barycenter shift and crystal-field splitting energy of the 5d levels of an ion in a host compound Configurational coordinate diagram of excitation and emission process Schematic representation of sensitized emission. The energy is absorbed by the sensitizer (S) and then transferred to acceptor (A), which emits Emission spectra of SrAlF 5 : Pr 3+ under x-ray excitation. The dotted line is measured at 100K, the solid line is at 350K (A.P.Vink et. al., Journal of Physics; Condensed Matter, 14, 8889 (2002), with permission from publisher Quantum splitting process by CRET in two lanthanide ions Energy level structure of Gd 3+ in LiYF 4. (adopted from R.T.Wegh, H. Donker, A. Meijerink, Physical Review B 56, 21, (1997)) f n-1 5d levels of free gaseous Ln 3+ ions. ( ) represents spin forbidden 4f-5d transition energy and ( ) for dipole allowed 4f-5d transition energy. (adopted from P. Dorenbos, J. Lumin. 91, (2000)) (Color online) Relative quantum yield of GdLiF 4 :Nd 2% exciting at 160 nm (black, solid curve) and at 351 nm (red, dashed curve). The spectra are normalized on the Nd 3+ 4 D 3/2 and 2 P 3/2 quantum yields (Color online) Energy level diagrams of Nd 3+ and Gd 3+ in GdLiF 4 :Nd with the relevant energy levels labeled. The open box

13 LIST OF FIGURES (Continued) Figure Page represents the 4f 2 5d band of Nd 3+. The boxed areas with horizontal lines represent energy regions with a high density of 4f n levels. ET1 and ET2 indicate resonant energy transfer processes. Labels A, B, and C next to the red (dashed) lines denote three cross relaxation energy transfer processes. Some of the intrinsic lifetimes are indicated (a) Absorption spectrum of YLiF 4 :Nd2% and (b) emission spectrum of YLiF 4 :Gd5% [9] showing significant spectral overlap (Color online) Excitation spectrum of GdLiF 4 containing 1, 2 and 3% Nd 3+ and detecting the Nd 3+ 4 F 3/2 emission using a cutoff filter that transmits for λ>780 nm. Features of the 6 G J, 6 D J and 6 I J levels of Gd 3+ and the 4f 2 5d bands of Nd 3+ are indicated (Color online) Comparison of the excitation spectra of GdLiF 4 :Nd2% detecting only the 4 F 3/2 emission with λ detect >780 nm with that of the case of detection for λ detect < (Color online) Time evolution of the 6 I (281 nm) and 6 P 7/2 (313 nm) emission intensities of Gd 3+ and the 4 D 3/2 and 4 F 3/2 emission intensities of Nd 3+ in a GdLiF 4 :Nd2% sample under 157 nm pulsed laser excitation (Color online) Time evolution of the 6 I (281 nm) and 6 P 7/2 (313 nm) emission intensities of Gd 3+ under 157 nm pulsed excitation in GdLiF 4 :Nd for 1, 2, and 3% Nd concentrations. The dashed lines show the fits using the 6 I decay times shown in the figure. Those same times are used as the rise times in the fits to the 6 P 7/2 emission for the sample with the same Nd 3+ concentration Time evolution of the 4 D 3/2 and 2 P 3/2 emission of Nd 3+ in a sample of GdLiF 4 :Nd2% under 355 nm excitation and the 4 P 3/2 emission under 157 nm excitation. The decay of 2 P 3/2 is the rate limiting state in the feeding of 4 F 3/2. Also plotted as dashed lines are fits to the data using the rise and decay times indicated on the figure....40

14 LIST OF FIGURES (Continued) Figure Page 2.9. (Color online) Time evolution of the 2 P 3/2 and 4 F 3/2 emission in a GdLiF 4 :Nd2% sample under 355 nm and 157 nm excitation. The fits shown on the figure are obtained using the rise and decay times indicated in the legend. They percentage indicates the fraction of population buildup which is contributed by this rise time. The remainder of the population buildup is taken to appear immediately after excitation Energy level diagrams for Pr 3+, Gd 3+, and Eu 3+ showing the various energy transfer pathways labeled a through j. Processes a through d are shown displaced downward by 2500 cm -1 reflecting half the value of the Stoke s shift for LaF 3 for the Pr 3+ 4f5d emission Emission spectra for a sample of GdF 3 containing 0.3% Pr and 0.2% Eu excited at 275 nm ( 6 I state of Gd 3+ ) and 160 nm (4f5d state of Pr 3+ ) Excitation spectra of two samples of GdF 3 :Pr,Eu. Excitation spectra obtained by detecting all wavelengths > 320 nm are referenced to a Na salicylate standard. Excitation spectra obtained with filters selectively for λ>580 nm and λ<560 nm are normalized for the 6 I peak but are not to the scale of the figure Time-resolved emission for 6 I and 6 P 7/2 of Gd 3+ after pulsed excitation at 193 nm showing that the decay of 6 I corresponds to the buildup of 6 P 7/2 and that energy transfer from Pr 3+ predominantly feeds 6 I. The circles are the measurement and the dashed curves are fits using an exponential decay and buildup of 2.4 µs with an initial 20% 6 P 7/2 population Energy level diagrams for Pr 3+, Tm 3+ and Er 3+ in YPO 4 and YBO 3. For Tm 3+ and Er 3+ the 5d levels are split into a lower-energy high spin (HS) and higher energy low spin (LS) states. For Er 3+ the room temperature lifetimes are shown next to the emitting states. Processes labeled A and B for the Pr 3+ -Tm 3+ pair indicate energy conserving cross relaxation paths Excitation (dashed) and emission spectra (solid) for Pr 3+, Er 3+ and Tm 3+ ions in YPO 4 at room temperature. The excitation spectra are

15 LIST OF FIGURES (Continued) Figure Page relative to that of sodium salicylate. For Er 3+ the distinct vertical bars identify the emitting level Excitation spectra at room temperature demonstrating the absence of Pr 3+ to Tm 3+ energy transfer in YPO 4 and YBO 3. None of the features of the Pr 3+ excitation spectra appear in doubly doped samples when only the Tm 3+ emission is detected. The excitation spectra of the doubly-doped samples are not to scale Emission spectra of ScBO 3, excited at160 nm. The instrinsic STE emission is shown amplified by a factor of Emission spectra of the Gd 3+ -doped borates in the red showing the weak Gd 3+ 6 G 6 P emission Excitation spectra of undoped and Gd 3+ -doped ScBO 3 detecting the total emission and measured relative to that of sodium salicylate Time resolved intrinsic emission of undoped and Gd 3+ -doped ScBO 3. The emission was excited at 157 nm and detected at 250 nm. Fitted decay curves are shown by the dashed lines. The fitted values have a 5 ns instrumental contribution Observed decay of the Gd 3+ 6 G 6 P emission in the Gd 3+ -doped borates. The fitted decay curves are shown by the dashed lines with the decay values shown in the legend Fluorescence spectra of YBO 3 excited at160 nm. The intrinsic emission is shown expanded by a factor of Time resolved emission excited at 157 nm and detected at 340 nm. The decay is a double exponential. The short decay component in the figure is lengthened by the 5.9 kω oscilloscope input impedence. Its actual decay time is < 2 ns Time-resolved emission spectra excited at 157 nm. The t=0 spectrum is obtained from the initial intensity of the fast decay component. The spectrum of the slow decay component was obtained from the intensity at 400 ns after the fast component had decayed. It is identical to the time-averaged emission spectrum... 83

16 LIST OF FIGURES (Continued) Figure Page 5.9. Excitation spectra of undoped and and Gd 3+ -doped YBO 3 detecting the total emission and measured relative to that of sodium salicylate Emission spectra of ScPO 4, and YPO 4 excited at 160 nm Excitation spectra of undoped and Gd 3+ -doped ScPO 4.The doped sample is referenced to sodium salicylate (dashed curve). The excitation of the UV portion of the emission is measured relative to sodium salicylate (thin solid curve) while the red portion of the emission is referenced to Y 2 O 3 :5%Eu 3+ (dotted curve). The estimated absolute quantum yield is shown by the bold solid curve Time-resolved emission of undoped and Gd 3+ -doped ScPO 4 excited at 157 nm. The fits are shown by the dashed lines and include a 5ns instrumental contribution Time-resolved emission of ScPO 4 :1%Gd excited at 157 nm and detected at 206nm and 600 nm (Gd 3+ 6 G emission) and 315 nm (Gd 3+ 6 P emission). The inset shows the fit of the 6 G decay Excitation spectra of undoped and Gd 3+ -doped YPO 4 for detection in different wavelength regions showing the dependence of the spectra on detection wavelength Time-resolved emission of undoped (solid curves) and Gd 3+ -doped (dotted curves) YPO 4. The 240 nm emission shows a 55 ns buildup and 380 ns decay while the emission at longer wavelengths (340nm and 460 nm shown in the figure) exhibit a decay with two components. The dashed curves show fits to the 240 nm and 460 nm data for the undoped sample Emission spectra of GdZrF 7 a) undoped, b) doped with 1% Eu 3+, c) doped with 1% Pr 3+, d) double doped with 1% Pr 3+ and 1% Eu 3+ under 160nm excitation XRD patterns of GdZrF7:1%Eu 3+ (a), the reference XRD pattern (b) calculated from single crystal structure data Excitation spectrum of GdZrF 7 :1%Eu 3+ and 3% Eu 3+ samples

17 LIST OF FIGURES (Continued) Figure Page 6.4. Excitation spectra of GdZrF 7 a) undoped, b) doped with 1% Eu 3+, c) doped with 1% Pr 3+, d) double doped with 1%Pr 3+ and 1%Eu 3+ for whole emission above 300nm XRD patterns of (a) GdZrF 7 :1%Pr 3+ (b) GdZrF 7 :1%Pr 3+,1%Eu Emission spectrum of GdZrF 7 doped lanthanides. a) Doped with 1%Eu 3+, b) doped with 1%Ce 3+, c) codoped with 1%Eu 3+ and 1%Tm Emission spectrum of GdZrF 7 :1%Eu 3+,1%Tb 3+ for the whole emission compared to that of GdZrF 7 :1%Eu Excitation spectrum of several GdZrF 7 samples doped with lanthanides, a) doped with 1%Eu 3+, b) doped with 1%Ce 3+, c) co doped with 1%Eu 3+ and 1%Tm 3+. All excitation spectra were measured for whole emission spectrum except b) which was measured excluding the emission peak of 6 P of Gd XRD patterns of (a) GdZrF 7 :Ce 3+ and (b) GdZrF 7 :Eu 3+,Tm Unit cell drawing of GdHfF Two views of the eight coordinated polyhedron of Gd 3+ ion in GdHfF A [001] directional view of GdHfF 7 compound showing Hf 4+ and Gd 3+ ions composing the squares respectively. The crystal coordinate was shown on the picture A [010] directional view showing the top and bottom layers composed of Gd squares and Hf squares. Those layers form slabs of [Gd 2 Zr 2 F 12 ] 2+ with other bottom and top layers of next unit cells along c-direction A fragment of the crystal structure of the two types of zig-zag -Gd- F-Gd-F- chains structure showing the disorder at F5 position a) Experimental XRD pattern of the powder sample of GdHfF 7, b) XRD pattern calculated based on the single crystal structure of GdHfF

18 LIST OF FIGURES (Continued) Figure Page 7.7. Emission spectrum under 160nm excitation. Emission spectra of a) GdHfF 7, b) GdHfF 7 :1%Eu 3+, c) GdZrF 7 :1%Eu 3+, d) Gd(Hf 0.5,Zr 0.5 )F 7 :1%Eu 3+. Y-coordinate is relative emission intensity Excitation spectrum of GdHfF 7 samples compare to other analogous compound. a)undoped GdHfF 7, b) GdHfF 7 :1%Eu 3+, c) GdZrF 7 :1%Eu 3+, d). Gd(Zr,Hf)F 7 :1%Eu Powder XRD data of Gd 0.98-x ZrF 7 :Yb x Er 0.02 samples. The bottom peaks is for x=0.18 and top one is for x=0.98. The x value is increased from x=0.18(bottom one) to 0.22, 0.26, 0.30, 0.34, 0.50 and 0.98(top one). Inset is the magnified one for Fig. 1 in the 2 θ range of degree Reference XRD pattern of GdZrF 7 compound calculated from the single crystal structure solution data Cell parameter change according to the increase of Yb 3+ concentration,(a) cell parameter, (b) cell volume and β angle Raman spectrum of polycrystalline GdZrF 7. The excitation source is He-Ne green laser SEM pictures of Gd 0.74 ZrF 7 :Yb 0.22,Er 0.04 sample at several magnifications, (a) 100, (b) 300 and (c) Upconversion mechanism for green emission under near infrared light excitation showing the energy transfer in Yb 3+ -Er 3+ system. The dotted curve explains the energy transfer from Yb 3+ to Er 3+ via consecutive two or three photon absorption by Er 3+, the downward zigzag line is non-radiative transition, the straight thick downward lines show the radiative transitions.[9] Relative emission output of the Gd 1-x-y ZrF 7 :Yb x Er y samples were measured during 30min compared with the reference one. The Er 3+ concentration was varied as 1%, 2%, 3% and to 4% at three different concentration of Yb 3+. The emission output data for 18% Yb 3+ are on (a), for 22% Yb 3+ on (b), for 26%Yb 3+ on (c). In all graphs the black line marked with black diamond represent the emission output of reference sample. The line with brown triangle marker is for 2% Er 3+, the green cross marker is for 3% Er 3+, the violet square marker is for

19 LIST OF FIGURES (Continued) Figure Page 1% Er 3+ and the black cross maker is for 4% Er 3+ in the downward sequence from the top one Emission output results of GdZrF 7 samples at three concentration levels of Yb 3+ and four concentration levels of Er 3+ collected from the experiment above. The data dispersion on each sample is caused by the emission output increase as time pass by as we mentioned already. The first group of dots express the emission output of reference sample, the next three groups of dots represent the emission output of 1% Er samples, the next three for 2% Er samples, the next three for the 3% Er samples and the last three for the 4% Er samples. At each Er concentration, the first group of dots represent 18% Yb 3+, the second one 22% Yb 3+, and the third one 26% Yb 3+ condition Emission output of Gd 0.98-x ZrF 7 :Yb 3+ x Er samples at further increased concentration of Yb 3+ up to 98% are measured intermittently. The emission output was measured during 50min intermittently and is represented as dots. x-abscise is the Yb 3+ concentration, y-abscise is the relative emission output to that of reference one Emission spectrum at various Yb 3+ concentrations. (a) [Yb] =18%, (b) [Yb] =22%, (c) [Yb] =26%, (d) [Yb] =30%, (e) [Yb] =34%, (f) [Yb] =50% and (g) [Yb] =98%. Every emission spectrum are compared with reference one which is shown by the blue solid line G/R ratio at various Yb 3+ concentrations XRD data of reference up conversion green phosphor with the x-ray pattern of Gd 2 O 2 S from ICDS file Emission spectrum excited by 379nm and 490nm (a) of 22%Yb 3+ sample, (b) 50% Yb 3+ sample and (c) reference one. In all pictures the violet solid lines represent the emission spectrum under 490nm excitation and the blue solid lines are that under the 379nm excitation Unit-cell drawing of Gd 4.67 (SiO 4 ) 3 S

20 LIST OF FIGURES (Continued) Figure Page 9.2. (a) Environment of the free O atom in Ln 4.67 (SiO 4 ) 3 O apatite (b) Environment of S atom in Gd 4.67 (SiO 4 ) 3 S apatite Tricapped distorted trigonal prismatic environment of Gd(2) and seven-coordinate site of Gd(1) Sulfur column along c axis XRD patterns for Gd 4.67 (SiO 4 ) 3 S (a) synthesis in flowing H 2 S(g) (c) prepared in sealed tube and (b) reference pattern calculated from single-crystal structure data Emission spectra for selected concentrations of Tb 3+ in Gd 4.67 (SiO 4 ) 3 S Emission spectra after correction for (a) 7% and (b) 10% Tb 3+ doped Gd 4.67 (SiO 4 ) 3 S Excitation spectrum of 7% and 10% Tb 3+ -doped Gd 4.67 (SiO 4 ) 3 S (λ em = 544nm) Excitation spectrum at liquid helium temperature of (a) 10% Tb 3+ doped Gd 4.67 (SiO 4 ) 3 S and (b) 7% Tb 3+ doped Gd 4.67 (SiO 4 ) 3 O Excitation spectra 4f-4f transitions of 10% Tb 3+ doped Gd 4.67 (SiO 4 ) 3 S Comparison of emission spectra of Gd 4.67 (SiO 4 ) 3 S:10% Tb 3+ under two different excitation wavelengths of 313nm and 370nm. The emission intensity was calibrated with the intensity ratio of those two excitation wavelength using Rhodamin-B

21 LIST OF TABLES Table Page f n-1 5d energy of several Ln 3+ ions doped in LiYF 4 compound Experimental energy transfer rates Wavelengths and decay times of the emission of undoped and Gd 3+ doped scandium and yttrium borates and phosphates Crystal data and some of details of X-ray diffraction experiment and refinement of the crystal structure of GdMF 7 (M=Zr, Hf) Atomic position (x10 4 ) and equivalent isotropic displacement parameters (Å2x 103). U(eq)is defined as one third of the trace of the orthogonalized Uij tensor a) GdHfF Selected bond lengths [Å] in GdMF Selected Bond angles [ ] in GdMF G/R ratio measured from the emission spectrum of each Yb 3+ concentration. Data at two different excitation wavelengths of 490nm and 980nm are shown for two samples of 22%Yb 3+ and 50% Yb Crystal data and details of X-ray diffraction experiment for Gd 4.67 (SiO 4 ) 3 S Atomic positions and equivalent isotropic displacement parameters (Å 2 x 10 3 ) for Gd 4.67 (SiO 4 ) 3 S Bond lengths [Å] Selected Bond angles [ ] The shortest distance between Gd ions in two different sites

22 HIGH QUANTUM-YIELD PHOSPHORS VIA QUANTUM SPLITTING AND UPCONVERSION CHAPTER I 1.1 INTRODUCTION Phosphors are materials that emit light following excitation with electromagnetic energy or high energy particles, e.g., electrons. They are currently used in fluorescent and light-emitting diode lamps as well as variety of displays such as the cathode ray tube (CRT), field emission display (FED), vacuum fluorescent display (VFD), electroluminescent (EL) device, and plasma display panel (PDP). In CRT, FED, and VFD applications, the phosphor is excited by an electron beam at either high or low accelerating voltages, while in PDP and fluorescent lamp applications, the phosphor is excited with high-energy photons in the ultraviolet (UV) or vacuum ultraviolet (VUV) portions of the spectrum. Recently, large-area PDPs have proven to be commercially successful as HDTVs. They operate on the basis of a Xe discharge, which produces the VUV light for excitation of the red, green, and blueemitting phosphors. Similar technology using a Xe discharge is envisioned for producing a fluorescent lamp that is free of Hg. A serious drawback of existing PDP and related VUV-excited phosphors, however, is the low quantum ratio (R ) associated with the high photon energy of the Xe discharge relative to the photon energies of visible light. The average energy of light from a red, green, blue phosphor set corresponds approximately to 500 nm. As shown below, this lead to R ~ 0.3 for a Xe discharge and R ~ 0.5 for a Hg discharge. PDP phosphor excited at 147 nm: R = hν em / hν ex = λ ex / λ em = 147 nm / 500 nm = 0.3 Mercury-based fluorescent lamp phosphor excited at 254 nm: R = 254 nm / 500 nm= 0.5

23 2 Because of the low quantum ratio and the lower efficiency of light production for a Xe discharge relative to a Hg discharge, the energy efficiency of a mercury-free, Xe-based lamp cannot compete with that of a common fluorescent lamp, even with phosphors having unit quantum efficiencies. For the energy efficiency of a Xe lamp to be comparable to that of a conventional Hg-based fluroescent lamp, the quantum efficiency of the phosphors in the lamp must be near 1.5 or higher, i.e., they must be significantly greater than unity. Much of the work presented in this thesis is directed to the development and study of new materials and processes that provide means to realize these high quantum efficiencies under VUV excitation. Inorganic luminescent materials exhibiting quantum efficiencies > 1 are well documented in the literature, but none of these phosphors exhibits the necessary combination of strong absorption at Xe discharge wavelengths, quantum efficiency, color purity, and stability for application in a Hg-free lamp. Much of the work in this thesis is directed to realizing this combination of attributes in a single material. Phosphor quantum efficiencies > 1 have been realized in several ways. One particularly useful method is described as Photon Cascade Emission (PCE), which involves transitions between energy levels of lanthanide ions; PCE is also commonly referred to as quantum cutting, quantum splitting, and multiphoton emission. While quantum-cutting materials are differentiated from normal phosphors by having quantum yields > 1, the physics associated with excitation and emission in a multiphoton material is identical to that of a conventional luminescent substance. Pr 3+ is a representative lanthanide ion exhibiting PCE; its PCE was first reported in 1974 by two separate groups: Piper and co-workers at General Electric and Sommerdijk and co-workers at Philips [1, 2]. As shown in Fig. 1.1, the energy-level structure of the 4f 2 configuration of Pr 3+ affords the opportunity for a two-step emission. The energy of the 1 S 0 state in many compounds is located near 46,500 to 46,900 cm -1, importantly resting below the 4f 1 d 1 levels. An excited electron in the 1 S 0 can relax to the ground state through a multiple-step process with transition from 1 S 0 to 1 I 6 involving emission of a photon as the first step and transition from 3 P 0 to 3 H 4 also involving emission of photon the second step,

24 3 60 4f 1 5d S 0 Energy (*10 3 cm -1 ) I 6 3 P 0 1 D G 4 3 F J 0 3 H J Fig. 1.1 Energy level diagram of Pr 3+ ion. PCE process is indicated by two successive transitions from the 1 S 0 level following excitation into the 4f5d band.(adopted from A. P. Vink, P. Dorenbos, C. W. E. Van Eijk, Journal of Solid State Chemistry, 171, (2003))

25 4 incorporating a nonradiative relaxation process between the 1 I 6 to 3 P 0 levels, cf., Figs As shown in Fig. 1.2 for YF 3 :Pr 3+, emission from the 1 S 0 level produces a photon in the deep blue portion of the spectrum, while emission from 3 P 0 produces photons spread across the blue-green, red, and NIR portions of the spectrum. For the overall process, a quantum efficiency near 140% is observed. Because the initial deep blue photon is positioned at a wavelength of low eye sensitivity, quantum splitting with Pr 3+ has no direct utility in lamp and display applications. (a) Wavelength(nm) Intensity 1 S0-1 I 6 3 P0-3 H 4 3 P0-3 H 3 6, F2 3 P0-3 F 3, (b) 1 st photon 1 S 0 1 I 6 Nonradiative decay 3 P 0 2 nd photon 3 F J 3 H 4 4f 1 5d Excitation ~195 nm ground state Fig. 1.2 (a) Emission spectrum of YF 3 :Pr 3+ (b) Schematic of PCE; initial photon emission from 1 S 0 and the second photon emission from 3 P 0.

26 5 A more recent example of quantum splitting has been demonstrated in the system, LiGdF 4 :Eu 3+ [3]. Here, energy migration and transfer between the ions Gd 3+ and Eu 3+ leads to a measured quantum efficiency of 190%. While this system provides an optimum color purity on the basis of the red Eu 3+ emission, the weak cross sections of the f f transitions of the Gd 3+ do not provide an efficient excitation pathway. The process of quantum splitting in the system is discussed in more detail in Section Another method for realizing dramatically increased quantum efficiencies is through electron multiplication. With high-energy excitation - usually 2.5 times the band-gap energy - excited electrons can relax to energies above the band gap by creating electron-hole pairs via inelastic collisions. These generated electron-hole pairs can lead to photon generation in addition to that of the initially excited electron, leading to enhanced quantum yields. The excitation energy required to observe electron multiplication has been described by P. A. Rodnyi and co-workers [5] by using Eq. 1.1, E t = 9E g 7-(m e /m h ) Eq. 1.1 Where E t is the threshold energy, which is the minimum energy required to ionize atoms in the solid as measured from the bottom of the conduction band, and m e and m h are the effective masses of the electron and hole, respectively. If the energy of an exciting photon hν ex exceeds the value of E g + E t, a secondary electron-hole pair can be created. For m e ~ m h, E t = (3/2) * E g, and for m e >> m h, E t = (9/7) * E g. From experimental results E t ~ 1.5 * E g for semiconductors and E t ~ E g for ionic compounds. As an example, the quantum yield of the phosphor Zn 2 SiO 4 :Mn 2+ is approximately unity for excitation energies slightly exceeding the band gap (E g = 5.5 ev). For excitation energies above 14 ev, the quantum efficiency gradually rises to a value of 1.8 at an excitation energy of 21 ev [5]. The value of 14 ev corresponds to E g + E t.

27 6 Host-sensitized energy transfer is a potentially efficient excitation method for injecting energy into a system exhibiting a high quantum efficiency. One advantage of this method is that a secondary sensitizer ion is not required to increase the absorption of excitation light. Instead, the host is used to strongly absorb the excitation energy, and the resulting host emission (derived from a self-trapped exciton (STE)) is then in turn used to sensitize the activator ion. This excitation process has not yet been adapted in quantum splitting. Sensitizing an activator of interest as a quantum splitter by host emission, however, has been reported for the 3 P J and 1 D 2 emissions of Pr 3+ in SrAlF 5. [12] This host sensitization is examined in this work for excitation of the Gd 3+ 6 G level. To observe quantum splitting from Gd 3+, the host emission energy must be sufficiently energetic and resonant to excite 6 G J of Gd 3+, and the energy transfer rate from the host to Gd 3+ must be faster than the host emission decay rate. 1.2 GENERAL CONSIDERATIONS Several fundamental characteristics of luminescent centers and their interactions in solids must be considered in selecting and synthesizing materials for observation of quantum splitting. A few of these characteristics are considered in this section Selection of Host for Lanthanide In suitable hosts Pr 3+ commonly exhibits quantum splitting via PCE. Pr 3+ is excited via a parity allowed 4f 5d transition rather than directly into the 1 S 0 level, because of the forbidden nature of the 3 H 4 1 S 0 transition. As such, the 5d level must be energetically positioned above the 1 S 0 to sensitize its occupation. The position of the 5d level is determined by two factors, the barycenter and the crystal field splitting parameter Δ. The barycenter shift (centroid shift) is the energy shift in the average energy of the crystal-field-split 5d levels, while the crystal field splitting affects the position of the lowest 5d energy level.

28 7 5d Lowest 5d Free ion Red shift, Depression E D(Q+,A) centroid shift (ε c ) Lowest 5d (4f m, lowest 5d)= lowest 5d of free ion D(Q+,A) 4f m Fig. 1.3 Schematic diagram showing the barycenter shift and crystal-field splitting energy of the 5d levels of an ion in a host compound. A host with a band gap > 7 ev is necessary to prevent overlap of the lowest 4f5d level of Pr 3+ with the conduction band (CB) of the host. As such, fluorides, selected oxides, and chlorides are suitable hosts. The energy gap between the 4f5d and 1 S 0 levels is also important. The free ion has an energy gap E = 11,000 cm -1, but this value is smaller because of the ligand field and nephelauxetic effects around the ion in the host. This gap should not be too large or too small. A large gap leads to direct transition from the 5d level to the ground state rather than the desired nonradiative relaxation from 5d to 1 S 0. For a small gap, excessive thermal population of the 5d level results. As seen from Fig. 1.3, a small centroid shift (ε c ) is required for a high energy position of the lowest 5d level. This shift tends to increase in the order fluoride < sulfate < carbonate < phosphate < borate < silicate < aluminate [6]. A small nephelauxetic effect associated with low ligand polarizability will force the lowest 5d level to a high energy state. The centroid shift is modeled on the basis of Eq. 1.2 [6],

29 8 ε c (cm -1 )= 1.44*10 17 (N α sp /R 6 eff) Eq. 1.2 where N is the number of the nearest neighbor ligands; R eff (pm) is the Ln 3+ - ligand distance; and α sp (Å 3 ) is the spectroscopic polarizability. To produce a high-energy position for the 5d level, a weak crystal field is also needed. This crystal field can be realized with long Pr 3+ -ligand distances and high coordination numbers (CN). In the fluoride YF 3 :Pr 3+, as a result of the small nephelauxetic effect and crystal field, the 5d level (48,900 cm -1 ) is situated above the 1 S 0 level, resulting in PCE. In contrast, the oxide Y 3 Al 5 O 12 :Pr 3+ has a large nephelauxetic effect and crystal field; as a result, the 5d level (33,300 cm -1 ) is situated 13,500 cm -1 lower than the 1 S 0 level. In LaMgB 5 O 10 and LaB 3 O 6 with CN=10 for Pr 3+, the 5d level is located above 1 S 0, 1 S 0 emission is observed, but emission from 3 P 0, cf., Fig. 1.1, does not occur because of the high nonradiative transition rate from 3 P 0 to 1 D 2, which derives from the high phonon energy of the borate. Considering the phonon energy of BO 3-3 at 1450 cm -1 the non-radiative and the radiative rates can be calculated as W nr = 10 7 s -1 and W r of 3 P 0 = 3*10 5 s -1 respectively [7]. Lower frequency phonons are required to observe 3 P 0 Pr 3+ emission in oxides, SrAl 12 O 19 :Pr 3+, for example, has a low phonon energy of 700 cm -1, and both steps of PCE are observed [8]. The nonradiative and radiative transition rates from 3 P 0 are calculated as W nr = 3*10 3 s -1 and W r of 3 P 0 = 3*10 5 s -1, respectively. The nonradiative transition rate is related to the maximum phonon energy and the emission energy (Eq. 1.3) [7]. W nr = β e α( E-2hω) Eq. 1.3 β and α are materials-dependent constants; E is energy gap associated with emission; and 2hω is the maximum phonon energy. For a fixed emission wavelength, nonradiative relaxation rates on the basis of phonon energies increase in the order chloride < fluoride < oxide.

30 Energetic position of the lowest 5d Level Because the position of the 5d level in Pr 3+ can be used to predict PCE, it is important to be able to estimate the energy of this level in a given host. This can be done by using extensive compilations of the 5d-level positions of Ce 3+ in various hosts [9]. The Ce 3+ ion has a simple ground-state electron configuration, 4f 1, resulting in excitation only to the 5d level. Emission generally occurs from this level, so an excitation spectrum can be used to assign the position of the 5d level. From examination of many hosts, it has been demonstrated that the energy difference between Pr 3+ and Ce 3+ 5d levels is constant and approximately 12,240 ± 750cm -1 [9]. This relationship can be extended to all of the lanthanides through Eq Lowest 5d level of Ln 3+ = 49,340 cm -1 D (Ln 3+, A) + E Ce3+, Ln3+ Eq. 1.4 The 5d level of the free Ce 3+ ion is positioned at 49,340 cm -1 ; D (Ln 3+, A) represents the crystal-field depression energy in host A; and E Ce3+, Ln3+ is the energy difference between Ce 3+ and the selected Ln 3+. To realize a 5d position above 1 S 0, the excitation 4f 2 4f 1 5d 1 of Pr 3+ must occur atλ ex < 205 nm, corresponding to Ce 3+ having λ ex < 270 nm in a given host..

31 The Stokes Shift As shown in Fig. 1.4, the excited state of an activator ion will have a minimum energy at a cation-ligand separation that differs from the distance in the ground state. The excited state relaxes to the lowest vibrational energy level from which the emission spontaneously occurs. The vertical transition to the ground state results in the occupation of an excited vibrational level, which also subsequently relaxes to the lowest energy state. The energy difference between the absorption and the emission derived from these processes is referred to as the Stokes shift (ΔS). excited state energy ground state V =0 absorption emission V=0 R 0 Fig. 1.4 Configurational coordinate diagram for luminescence process To realize the energetic position of 5d > 1 S 0, not only must the energy of the long wavelength absorption edge (4f-5d) exceed the energy of 1 S 0 (~47000cm -1 ), but S must also be small. A short bond distance is preferable for observing a small

32 Stokes shift, because the shift increases as the square of the average bond length (Eq. 1.5) [10]. 11 S ~ R 2 (R = average bond distance) Eq. 1.5 The position of the minimum of the 5d excited state can be estimated from Eq. 1.6 [10] by considering the Stokes shift. E min (5d) E (4f 5d) - 0.5* S Eq. 1.6 The site symmetry can also affect the position of the 5d energy level. In particular, asymmetry in the site can significantly increase the Stokes shift.

33 Energy Transfer Sensitization via energy transfer provides a means to deliver energy to an activator that inefficiently couples to the excitation source. The sensitizer absorbs the excitation energy and transfers it to the activator through a nonradiative process (Fig. 1.5). Emission Excitation S Energy transfer A Fig Schematic representation of sensitized emission. The energy is absorbed by the sensitizer (S) and then transferred to acceptor (A), which emits This nonradiative energy transfer is generally modeled by the Forster-Dexter theory. In the Dexter model, energy transfer occurs by an exchange interaction, where the electron exchange occurs between sensitizer and activator dopants. Because this exchange involves wave-function overlap of the sensitizer and activator, it occurs only over very short distances. The transfer rate is expressed by Eq. 1.7 [15], W DA = C DA e -2R/L Eq. 1.7 where C DA is the donor acceptor interaction parameter; and R/L is the donor-acceptor distance expressed in the Bohr-radius unit. When the donor and acceptor are separated by large distances corresponding to insignificant orbital overlap, energy transfer can proceed by dipole-dipole interactions (Forster model). The electric field generated by an excited sensitizer (donor) can

34 13 induce a dipole at an acceptor impurity (acceptor). The probability of energy transfer depends inversely on the square of the energy overlap and sixth power of the distance between the donor and acceptor, cf., Eq. 1.8, P AB DD =(1.4*10 24 f A f B S)/( E 2 R 6 ) Eq. 1.8 where f A and f B are the oscillator strengths of the donor and acceptor, respectively; S is the spectral overlap of donor emission and acceptor absorption; E is the transition energy; and R is the distance between the donor and acceptor. As noted, a distinguishing feature of these two mechanisms involves the sensitizer (S)-activator (A) distance. The Dexter model operates only at very short distances, where wave-function overlap is significant. The Forster mechanism is applied to longer S A distances. Here, the transfer rate is associated with a dipoledipole interaction and the oscillator strengths of the S* S and A A* transitions. This contrasts to the Dexter model, where energy transfer is independent of the transition rate. To achieve success in many quantum-splitting schemes, it is essential that the energy-transfer rate is faster than the radiative decay rate of the sensitizer. The radiative decay rate is typically 10 3 to 10 6 s -1 for forbidden 4f-4f transitions and 10 6 to 10 8 s -1 for parity allowed 4f n-1 5d - 4f n transitions. The Forster dipole-dipole transfer rate involving the 4f-4f transitions of the lanthanides is estimated as ~10 5 s -1, assuming 5% impurity concentrations for the sensitizer and activator, S = 0.1 cm -1, 4f-4f oscillator strengths = 10-6, and a transition energy = 3 ev. This result indicates that energy transfer involving the 4f-4f transitions of the lanthanides can lead to efficient transfer. For a 4f n-1 5d -4f transition on the sensitizer and a 4f-4f transition on the activator, the transfer rate is calculated as s -1, assuming a 5% impurity concentration, a 4f n-1 5d -4f oscillator strength = 10-2, and S = Wegh and coworkers, for example, reported an energy transfer rate of 10 9 s -1 between Er 3+ and Gd 3+ [16]. If the transitions of both ions are 4f n-1 5d-4f, then the calculated transfer rate is s -1 for nearest neighbors and 10 8 s -1 at a 1% dopant concentration [17].

35 QUANTUM SPLITTING PCE Dynamics Two mechanisms have been reported for excitation and luminescence of the Pr 3+ ion. One involves excitation (λ ~ 190 nm) of the 1 S 0 level via relaxation from the 5d level. Luminescence under this excitation can show PCE. When excited at 160 nm or higher energies, e.g., by X-rays, host absorption occurs, and the emission spectrum exhibits a strong enhancement of the 3 P 0 and 1 D 2 luminescence transitions. This suggests the existence of an alternate energy-transfer pathway to the 3 P 0 or 1 D 2 levels involving an STE; such luminescence has been described for the doped hosts BaSO 4 [6], SrAlF 5 [12], and LaF 3 -LiF compounds [11]. In BaSO 4 : 1% Pr 3+, the excitation spectrum for 1 D 2 emission contains only the host-absorption band, meaning it is not populated via 1 S 0 but by an STE. In SrAlF 5 :Pr 3+, the STE-mediated energy transfer was observed in the emission spectrum under X-ray excitation. In this case, the emission occurs only from 3 P 0 and 1 D 2 [12], cf., Fig The STE emission. which dominates the emission spectrum at 100 K, becomes weaker with increasing temperature. At 350 K, most of the energy of the STE is transferred to 3 P 0 or 1 D 2. Additional evidence that the 3 P 0 state of Pr 3+ can be populated by energy transfer from a STE was found in the LaF 3 -LiF system [11] The STE and 3 P 0 emissions exhibit similar thermal quenching, which implies that the 3 P 0 state is populated via the energy transfer from the STE. Meanwhile, the 1 S 0 emission intensity rises with temperature to 350 K, suggesting a competitive relationship with STE emission.

36 15 Fig Emission spectra of SrAlF 5 : Pr 3+ under x-ray excitation. (dotted line K; solid line 350 K) (A.P.Vink, P. Dorenbos, J T M De Hass, H Donker, P A Rodnyi, A G Avanesov, C W E van Eijk, Journal of Physics; Condensed Matter, 14, 8889 (2002), with permission from publisher) Quantum Splitting by Cross Relaxation Energy Transfer (CRET) Quantum splitting mechanisms can generally be divided into two categories. One is PCE, where a single ion, e.g., Pr 3+, decays via a multiple-step transition. The other is represented by cross relaxation energy transfer (CRET) between two ions, as observed in LiGdF 4 :Eu 3+, where Eu 3+ receives excitation energy transferred from Gd 3+ [3]. The energy-level diagram in Fig. 1.7 illustrates this process. Because of a resonance between the 6 G J - 6 P J energy of the Gd 3+ ion and the 5 D J - 7 F J energy on the Eu 3+ ion, the excitation energy on the Gd 3+ can be transferred to the Eu 3+ via cross relaxation (1). In the process, the Gd 3+ relaxes to the 6 P 0 level, while the Eu 3+ is excited to the 5 D 0 level. The excited Eu 3+ ion can then emit a photon. The remaining energy on the 6 P level of Gd 3+ is then transferred to Eu 3+ (2), which generates another photon from its 5 D J level. This process of PCE in Gd 3+ provides a useful method for designing quantum cutting phosphors. The multiple-step emission of Gd 3+ can be

37 expected from its 4f energy level structure (Fig. 1.8), and experimentally the emission transitions from 6 G J and 6 P J have been observed in Gd 3+ -doped LiYF 4 [13 ] LiGdF 4 :Eu Energy (cm -1 ) 1 6 G J 6 D J 6 I J 6 P J D J 1 0 Gd 3+ 8 S 7/2 Eu 3+ 7 F J Fig. 1.7 Quantum splitting process by CRET in two lanthanide ions.

38 /2 3/2 11/2, 9/2, 5/2 7/2 6 G J Energy (*10 3 cm -1 ) nm nm 1/2, 7/2, 3/2, 5/2 9/2 11/2, 15/2, 13/2 9/2, 17/2 7/2 7/2 5/2 3/2 6 D J 6 I J 6 P J 311nm 0 8 S 7/2 Fig Energy level structure of Gd 3+ in LiYF 4. (adopted from R.T.Wegh, H. Donker, A. Meijerink, Physical Review B 56, 21, (1997)) In LiGdF 4 :Eu 3+, the internal quantum yield of 190% assumes all of the excitation energy is absorbed into the 6 G J level of the Gd 3+ and converted into emission. The actual external quantum yield is only 32% [4], as a significant portion of the incident light is not absorbed because of the forbidden character of the excitation 8 S 7/2 to 6 G J of Gd 3+. To increase absorption sensitizers operating of the strong 4f-4f5d parity allowed transitions have been examined in this work. Among the lanthanides Pr 3+, Nd 3+, Ho 3+,Er 3+, and Tm 3+ were considered as candidates. These ions exhibit relatively highenergy 4f n - 4f n-1 5d transitions and relatively uncongested 4f levels at high energies, limiting internal relaxation of the 5d energy through these levels. The transition energies from ground 4f state to the 4f n-1 5d excited level are summarized in Fig. 1.9.

39 Gd Lu Eu Yb Energy, cm Nd Pm Sm Dy Dy Ho Ho Er Tm Er Tm Ce Pr Tb Tb allowed forbidden number of 4f electron Fig f n-1 5d levels of free gaseous Ln 3+ ions. ( ) represents spin forbidden 4f-5d transition energy and ( ) for dipole allowed 4f-5d transition energy. (adopted from P. Dorenbos, J. Lumin. 91, (2000)) Krupar and Queffelec have reported the 4f5d positions of Ce 3+, Pr 3+, Nd 3+, Eu 3+, Tb 3+, Dy 3+, Ho 3+, Er 3+, Tm 3+ in LiYF 4 at room temperature, over the range of energies from 5 to 15eV [14]; results are summarized in Table 1.1. The results reproduce the trend observed in Fig The excitation spectra also clearly reveal crystal-field split 4f n-1 5d levels. The five split levels of the 5d 1 configuration are consistent with the site symmetry S 4 (D 2d ) in LiYF 4.

40 19 Table 1.1 4f n-1 5d energy of several Ln 3+ ions doped in LiYF 4 compound Ln 3+ Energy of 5d absorption peaks (ev) *H.A. Ce Pr Nd Eu Tb Dy Ho Er Tm * Host absorption band

41 DISSERTATION SUMMARY To develop efficient excitation processes for quantum splitting systems, several stoichiometric Gd 3+ hosts were examined. These were doped with selected lanthanides (Nd 3+, Pr 3+, Sm 3+, Tm 3+ ) that were expected to exhibit 4f n 4f n-1 5d transitions at sufficiently high energies to populate the 6 G J level of Gd 3+ via CRET. The energy of the absorption bands were determined for Nd 3+ in LiGdF 4 (LGF) and GdPO 4 (GPO); Pr 3+ in GdF 3 and NaGdF 4 ; Sm 3+ in NaGdF 4 and LGF; and Tm 3+ in GdPO 4, and they were then examined as sensitizers under VUV excitation. LGF:Nd 3+ provided an unexpected and unprecedented quantum-splitting process. The details of this process are presented in Chapter 2. Quantum splitting was also observed in the system GdF 3 :Pr 3+,Eu 3+, where Pr 3+ is the sensitizer and Eu 3+ is the activator. Even though quantum splitting is observed in this system, the overall quantum yield is low. A detailed description of the energy transfer and quantum splitting processes are summarized in Chapter 3. In Chapter 4, results of the VUV luminescence of the 4f n-1 5d state of Pr 3+, Tm 3+, Er 3+, and the results of CRET between the Pr 3+ -Tm 3+ pair in YPO 4 and YBO 3 are described. The host intrinsic emission designated as STE was investigated as a new sensitizing method to excite Gd 3+ into 6 G level in oxide hosts. To identify the appropriate host features for STE emission at sufficiently high energy to sensitize Gd 3+, the host cation was varied across the phosphate series YPO 4, LuPO 4, and ScPO 4. The anion also was varied from phosphate to borate and silicate, e.g., ScPO 4, ScBO 3, and Sc 2 Si 2 O 7. Quantum splitting emission was observed in ScPO 4 :Gd 3+ with the quantum yield approaching unity. The experimental results on these compounds are detailed in Chapter 5. Results described in Chapter 2-5 were generated through collaborative work with Dr. Richard S. Meltzer at the University of Georgia (UGA) and Drs. Kailish C. Mishira and Madis Raukas at Osram Sylvania. Materials for study were selected following discussions among the three groups. All samples were synthesized and characterized with respect to structure and UV luminescence at Oregon State

42 21 University. VUV and energy-transfer experiments were conducted at UGA with some assistance from Osram Sylvania. The material GdZrF 7 was prepared as a nearly white phosphor under VUV excitation with high absolute quantum yield by doping with Eu 3+ (Chapter 6), and its crystal structure is described together with that of GdHfF 7 in Chapter 7. The host GdZrF 7 was also developed as an anti-stokes (upconversion) phosphor by codoping with Yb 3+ and Er 3+. Its luminescence, compositional optimization, and particle morphology are described in Chapter 8. While examining new Gd silicate systems for quantum splitting, a new apatite sulfide Gd 4.67 (SiO 4 ) 3 S was synthesized. Results of luminescence characterization of Tb 3+ samples and a single-crystal structure determination are described in Chapter 9. During my tenure at OSU, I also contributed to the development of a variety of electronic materials. The performance of the n-type amorphous oxide semiconductor ZnO x (SnO 2 ) 1-x (0<x<1) in transparent thin-film transistors is summarized in Appendix B. REFERENCES [1] W. W. Piper, J. A. Peluca, F. S. Ham, Journal of Luminescence, 8, (1974) [2] J. L. Sommerdijk, A. Bril, A. W. de JAGER, Journal of Luminescence, (1974) [3] R. T. Wegh, Harry Donker, Koenraad D. Oskam, Andries Meijerink, Science 283, 29, (1999) [4] C.Feldmann, T. Justel, C. R. Rondo, D. U. Wiechert, Journal of Luminescence, 92, (2001) [5] P. A.Rodnyĭ, Optics and Spectroscopy, 89, 4, (2000) [6] E.van der Kolk, P. Dorenbos, A. P. Vink, R. C. Perego, C. W. E. Van Eijk, Physical Review B 64, (2001) [7] P. A. Rodnyĭ, Optics and Spectroscopy 89, 4, (2000)

43 22 [8] A. M. Srivastava, W. W. Beers, Journal of Luminescence, 71, (1997) [9] P. Dorenbos, Journal of Luminescence, 91, (2000) [10] P. A. Rodnyĭ, A. N. Mishin, A. S. Potapov, Optics and Spectroscopy 93, 5, (2002) [11] P. A.Rodnyĭ, A. S. Potapov, A. S. Voloshinovskii, Optics and Spectroscopy 96, 6, (2004) [12] A. P. Vink, P. Dorenbos, C. W. E. Van Eijk, Journal of Solid State Chemistry 171, (2003) [13] R. T.Wegh, H. Donker, A. Meijerink, Physical Review B 56, 21, (1997) [14] J. C. Krupa, M. Queffelec, Journal of Alloys and Compounds 250, (1997) [15] A. J. De Vries, M. F. Hazenkamp, G. Blasse, Journal of Luminescence, 42, (1988) [16] R. T. Wegh, E. V. D. van Loef, A. Meijerink, Journal of Luminescence, 90, , (2000) [17] R. S. Meltzer, private communication.

44 23 CHAPTER 2 QUANTUM SPLITTING AND ITS DYNAMICS IN GdLiF 4 :Nd 3+ W. Jia, Y. Zhou, S.P. Feofilov, R.S. Meltzer Department of Physics and Astronomy University of Georgia Athens, GA J. Y. Jeong and D. Keszler Department of Chemistry 153 Gilbert Hall Oregon State University Corvallis, OR Modified version: Physical Review B: Condensed Matter and Materials Physics (2005), 72(7)

45 24 ABSTRACTS Efficient quantum splitting and sensitization of Gd 3+ is demonstrated for the Gd 3+ -Nd 3+ system in GdLiF 4 :Nd 2%. The quantum splitting results from a two step cross relaxation energy transfer between Gd 3+ and Nd 3+ which first involves a transition 6 G 6 I on Gd 3+ and an excitation within the 4f 3 configuration of Nd 3+ followed by a second cross relaxation energy transfer which brings Gd 3+ to 6 P 7/2. The excited Nd 3+ ion rapidly relaxes, non-radiatively, to the emitting 4 F 3/2 state. The excited Gd 3+ ion then transfer its energy back to Nd 3+ which gives rise to the second photon. The process is studied by emission and excitation spectroscopy. The result is a quantum yield for the emission of IR photons which has its maximum of about 1 ± 0.5, at 175 nm. The dynamics of both the Gd 3+ and Nd 3+ excited states are studied in detail providing information about the mechanisms and rates for the various energy transfer processes. It appears that the second step in the quantum splitting is less efficient than the first. It is found that energy migration among the Gd 3+ ions plays an important role in the quantum splitting and that there is strong evidence that the exchange interaction is the dominant mechanism in the energy transfer. This system provides excellent insights into the quantum splitting process, especially with regard to an evaluation of the details of the dynamics. 2.1 INTRODUCTION It has been suggested that improvements in fluorescent lamps could be realized by replacing the mercury discharge by xenon, thereby removing the deleterious environmental impact of mercury and at the same time improving the energy efficiency. Such innovations require a phosphor that absorbs one vacuum ultraviolet (VUV) photon and emits two or more visible photons, an effect known as quantum splitting or down conversion.[1]

46 25 Quantum splitting can occur either through a process of sequential cascade emission[2] as an excited ion returns to its ground state by first radiating to an intermediate state or by some cross relaxation process which enables the initially excited ion to share its excitation energy with two or more ions, each of which emits a visible photon. Both of these processes have been demonstrated. Cascade emission was first demonstrated in YF 3 :Pr with a 140% quantum efficiency[3]. Cross relaxation induced quantum splitting has been described for GdLiF 4 :Eu with an internal quantum efficiency of 190% [4]. Unfortunately, neither of these schemes has so far yielded a useful phosphor. For the cascade emission, the first photon occurs at 406 nm too far in the deep blue where the sensitivity of the human eye is very low. For the cross relaxation scheme in GdLiF 4 :Eu, the absorption of the VUV photon is too weak to produce a phosphor with high brightness. [5] We attempted to sensitize the absorption by adding Nd 3+ to GdLiF 4 : Eu 3+. We found that Nd 3+ does effectively sensitize the excitation of Gd 3+. However, in addition, Nd 3+ undergoes its own very strong cross relaxation with the Gd 3+ system producing efficient quantum splitting. A similar effect [6] has recently been reported for GdLiF 4 :Tm 3+. In this paper we study, in detail, the quantum splitting process for the singly-doped system, GdLiF 4 :Nd. The result of exciting Nd 3+ into the 4f 2 5d state in the VUV is the appearance of two infrared photons. While this material will not be a commercially viable quantum splitting phosphor since the photons are in the infrared and because of the large energy loss even if two photons were produced per input photon, it does provide important insights into the dynamics and mechanisms of the quantum splitting process. In this paper we (1) demonstrate the existence of the quantum splitting, (2) obtain the actual quantum efficiency of the system relative to the number of input VUV photons, (3) measure and analyze the dynamics of the processes using time-resolved emission, and (4) discuss the mechanisms for the energy transfer.

47 EXPERIMENT Samples of GdLiF 4 :Nd containing 1, 2 and 3 mol% Nd were prepared in powder form. GdF 3 was first synthesized by heating a mixture of 1 Gd 2 O 3 (99.99%, Alfa Aesar) and 8 NH 4 F (99.99%, Alfa Aesar) at 900 C for 1.5 h. The resulting product was then mixed with 1.15 LiF (99.99%, Alfa Aesar), 0.01, 0.02 or 0.03 Nd 2 O 3 (99.99%, Alfa Aesar), and 4 NH 4 F (99.99%, Alfa Aesar) and thoroughly ground. The mixture was then fired at 750 C for 1.5 h in a Pt crucible; the Pt crucible was covered and positioned inside an alumina crucible filled with activated carbon and NH 4 F to limit the exposure of the sample to air. All spectra were obtained at room temperature. Emission spectra were obtained by exciting the sample, contained in vacuum, with a deuterium lamp spectrally filtered with an Acton Model VM-502 VUV monochromator containing a concave grating so that selective excitation could be performed. The visible and UV emission was dispersed with an Acton Spectrapro-150 spectrometer and was detected with a Santa Barbara Instrument Group Model ST-6I CCD camera at the exit focal plane. Emission spectra in the VUV were obtained by exciting the sample with a GAM Laser, Model EX5, pulsed molecular F 2 laser whose output is at 157 nm. The sample emission was focused onto the entrance slit of the VUV monochromator. The emission was detected with a solar blind PMT with a MgF 2 window located at a third slit of the VUV monochromator which was scanned to obtain the spectrum. All emission spectra were corrected for the wavelength dependent response of the detection system. For cw excitation in the UV, a UV-enhanced Ar + laser was used at 351 nm. Excitation spectra were obtained by scanning the VUV monochromator, illuminated by the deuterium lamp, while detecting the emission with a PMT after passing the luminescence through appropriate colored glass or interference filters to select the desired components of the emission. Two PMT detectors were used, both having quartz windows yielding a response in the UV down to 200 nm. One (Hamamatsu R943) had a GaAs photocathode so that emission up to 900 nm could be measured. The other had a photocathode with an S-20 response. The excitation

48 27 spectra of each sample were compared to that of a reference sample of sodium salicylate whose quantum efficiency is assumed to be about 58% and constant over the excitation wavelength range from 140 to 320 nm [7]. The measured quantum yield is relative to input photons rather than absorbed photons since we have not obtained any reflectance measurements for either the samples or the reference. This assumes similar reflectivities of the sample and the sodium salicylate reference. For the time-resolved data, the sample was excited with the pulsed laser at 157 nm (10 ns pulse width), while the emission was detected with the same PMTs described above for the excitation spectra. Temporal resolution was about 20 ns. The emission was selected with a 0.25m monochromator and additional colored glass or interference filters to block light at other wavelengths from entering the monochromator. The bandwidth of the instrument was ~3 nm. The main limitations of the time-resolved spectra were extraneous signals at early times coming either from broadband red/nir emission from atomic fluorine in the laser discharge or from fast decay of defect centers that were excited by the VUV excitation. This red/nir emission was so strong that it was very difficult to do any time resolved spectroscopy from about 620 to 750 nm. For direct excitation of the 4f 3 states of Nd 3+ the third harmonic of a pulsed Nd:YAG laser at 355 nm (10 ns pulse width) was utilized Demonstration of the Quantum Splitting In Fig. 1 the emission spectrum is presented for two different excitation wavelengths, 351 and 160 nm. The emission from 200 nm to 950 nm is dominated by the 4 F 3/2 4 I 9/2 transition. However emission from the 4 D 3/2 and 2 P 3/2 states of Nd 3+ is also observed. Weak emission from the 6 P 7/2 state of Gd 3+ is observed at 313 nm. While it is not evident in this time-averaged spectrum, emission occurs at 281 nm from the 6 I state of Gd 3+. Emission from the 4f 2 5d state of Nd 3+ in the wavelength range of 180 nm to 270 nm, which dominates the spectrum of YLiF 4 :Nd [8], is not observed in GdLiF 4 :Nd suggesting efficient energy transfer from Nd 3+ to Gd 3+, i.e. strong sensitization.

49 28 When the spectra excited at the two different wavelengths are compared, by normalizing them to the 4 D 3/2 and 2 P 3/2 emission, it is seen that under160 nm excitation, the relative intensity of the 4 F 3/2 emission is more than double that observed for 351 nm excitation. This suggests a process which enhances the excitation of 4 F 3/2 in a manner which was used to identify quantum splitting for GdLiF 4 :Eu [4]. This is just the cross relaxation process responsible for quantum splitting. Relative Quantum Yield Gd 3+ 6 P 7/2 GdLiF 4 :Nd 2% Emission Spectra 351nm excitation 160nm excitation Nd 3+ 4 D 3/2 to 4 I J 9/2 11/2 13/2 15/2 Nd 3+ 2 P 3/2 Nd 3+ 4 F 3/2 Nd 3+ 2 P 3/ Wavelength (nm) Fig. 2.1 (Color online) Relative quantum yield of GdLiF 4 :Nd 2% exciting at 160 nm (black, solid curve) and at 351 nm (red, dashed curve). The spectra are normalized on the Nd 3+ 4 D 3/2 and 2 P 3/2 quantum yields. The processes are illustrated in Fig. 2. The diagram shows the relevant 4f 3 and 4f 7 energy levels of Nd 3+ and Gd 3+, respectively. Boxed regions with horizontal lines indicate a high density of states of the two 4f n configurations for which rapid multiphonon relaxation occurs. The open box represents the 4f 2 5d band of Nd 3+. The

50 29 4f 6 5d band of Gd 3+ is off the energy scale and is not relevant here. The long vertical arrow represents the VUV excitation of Nd 3+ into the 4f 2 5d band. Rapid energy transfer to a nearly resonant 4f 7 state of Gd 3+, labeled by ET 1, followed by rapid nonradiative relaxation, populates the 6 G J states of Gd 3+. Cross relaxation energy transfer from the 6 G 7/2 state of Gd 3+ can occur via two paths. One of these, indicated by the red(dashed) arrows labeled A on the energy level diagrams of Gd 3+ and Nd 3+, results in a transition 6 G 7/2 6 P J on Gd 3+, as has been previously observed in the Gd- Eu couple, with a simultaneous 4 I 9/2 4 G 5/2 excitation on Nd 3+. These two transitions have considerable overlap as shown in the room temperature spectra of Fig. 3 where the 6 G J 6 P J emission of Gd 3+ observed [9] in YLiF 4 :Gd is compared to the 4 I 9/2 4 G 5/2 absorption of YLiF 4 :Nd. Subsequently, rapid multiphonon relaxation leads to feeding of the 4 F 3/2 metastable state from which strong IR emission occurs. The second pathway involves a transition 6 G 7/2 6 I J on Gd 3+ coupled with a 4 I 9/2 4 F 5/2, 2 H 9/2 or 4 F 7/2 transition on Nd 3+ as indicated by the red(dashed) arrows labeled B in Fig. 2. Although the spectra are not available for comparison, the transition energies for Nd 3+ in absorption [10] and Gd 3+ predicted for emission [6] are likely to have good resonances. In addition, Peijzel et al. [6] have shown that the reduced matrix elements for this second pathway are about an order of magnitude greater than for the first, making this process about two orders of magnitude faster under the similar resonance conditions. Indeed, as will be shown from studies of the dynamics, the pathway involving the 6 I J levels does dominate the cross relaxation from 6 G 7/2. However, 6 I J can further relax to 6 P J via another cross relaxation process, shown by the red(dashed) arrows labeled C in Fig. 2, that excites the 4 I 13/2 state of Nd 3+. Evidence for this also exists from the dynamical studies discussed below.

51 30 GdLiF 4 :Nd 4f 2 5d 157nm ET 1 2 G 9/2 A B 6 G J A B 2 F 7/2 ET 2 4 D 2 3/2 τ=1µs P 3/2 τ=20µs 4 G 4 5/2 F 5/2 4 F 3/2 τ=400µs C 6 D J 6 I J 6 P J weak direct excitation of Gd 3+ C Nd 3+ 4 I J Gd 3+ 8 S 7/2 Fig. 2.2 (Color online) Energy level diagrams of Nd 3+ and Gd 3+ in GdLiF 4 :Nd with the relevant energy levels labeled. The open box represents the 4f 2 5d band of Nd 3+. The boxed areas with horizontal lines represent energy regions with a high density of 4f n levels. ET1 and ET2 indicate resonant energy transfer processes. Labels A, B, and C next to the red (dashed) lines denote three cross relaxation energy transfer processes. Some of the intrinsic lifetimes are indicated

52 31 Absorption (arb.units) Y L if 4 N d % : 4 I 9 /2 4 G 5 /2 A b s o rp tio n W a v e le n g th (n m ) YLiF 4 :Gd 5% Emission (b) Fig. 2.3 (a) Absorption spectrum of YLiF 4 :Nd2% and (b) emission spectrum of YLiF 4 :Gd5% [9] showing significant spectral overlap.

53 32 The 6 P J states of Gd 3+ then transfer their energy to the nearly resonant 4f 3 states of Nd 3+, as shown by the blue(solid) arrow labeled ET 2. Above the 4 D 3/2 state of Nd 3+ there is a very dense, almost continuous forest of energy levels from the 4f 3 configuration among which the 2 L 17/2 at 32,000 cm -1 is in closest resonance with the 6 P 7/2 states of Gd 3+. [10] Once excited, these will relax almost immediately to the 4 D 3/2 level which lives long enough to produce observable emission. Its decay, whose lifetime is about 1 µs, is dominated by non-radiative relaxation to the 2 P 3/2 level which lives much longer with a lifetime of ~20 µs. These and subsequent multiphonon relaxations ultimately feed the 4 F 3/2 level leading to the emission of a second IR photon. On the other hand, when the 4 D 3/2 state is excited directly at 351 nm, the cross relaxation step is eliminated so that the relative intensity of 4 F 3/2 emission is less than half of that obtained under 157 nm excitation. As described by Wegh et al. [4] for GdLiF 4 :Eu, this is strong evidence for quantum splitting. The dynamics of the system described below will provide further supporting evidence. Finally, it should be noted that the assumption that the initial Nd 3+ Gd 3+ energy transfer (ET1 in Fig. 2) occurs to Gd 3+ states resonant with the 4f 2 5d state of Nd 3+ may not be a good one. Many possible cross relaxation energy transfer processes are equally possible. These could excite many of the lower-lying states of Gd 3+ below the energy of the 4f 2 5d state of Nd 3+ ( 56,000 cm -1 ), shown on the Gd 3+ energy level diagram as the boxed area with many horizontal lines in Fig. 2. For example, cross relaxation processes could leave Nd 3+ in the 4 I J levels J=11/2, 13/2, 15/3 and Gd 3+ in states above 6 G J that conserve the total energy. Note that rapid multiphonon relaxation would still lead to a build up in the population of the 6 G J levels of Gd 3+ as had been assumed. Cross relaxation processes are also possible in which the energy transfer would result in Gd 3+ being excited to 6 D J, 6 I J, or 6 P J by leaving Nd 3+ in its 4 F 9/2 (14,800 cm -1 ), 4 G 7/2 (19,000 cm -1 ), or 4 G 11/2 (21,400 cm -1 ) states, respectively. However, these processes would also still lead to quantum splitting since multiphonon relaxation would populate 4 F 3/2 and the excited Gd 3+ ion would still be capable of transferring its energy to Nd 3+ for producing the second photon. These processes would supplement the energy transfer processes labeled as A and B that were previously discussed.

54 Excitation spectrum and quantum yield The excitation spectra, detecting the 4 F 3/2 4 I 9/2 emission of Nd 3+ at nm, is shown in Fig. 4 for the 1% and 2% and 3% Nd samples. It contains features associated both with Gd 3+ and Nd 3+ as indicated on the figure. One clearly sees the states of the 4f 7 configuration of Gd 3+, namely 6 G 6 J, D J and 6 I J, indicating that energy transfer between Gd 3+ and Nd 3+ occurs, as expected. The 4f 2 5d bands of Nd 3+ are also clearly observed. Relative Quantum Yield Nd 3+ 4f 2 5d GdLiF 4 :Nd Excitation Spectra Detect Nd 3+ Emission (λ>780nm) 6 G J 1% Nd 2% Nd 3% Nd Gd 3+ absorption from 8 S 7/2 to: 6 D J 6 I J Wavelength (nm) Fig. 2.4 (Color online) Excitation spectrum of GdLiF 4 containing 1, 2 and 3% Nd 3+ and detecting the Nd 3+ 4 F 3/2 emission using a cutoff filter that transmits for λ>780 nm. Features of the 6 G J, 6 D J and 6 I J levels of Gd 3+ and the 4f 2 5d bands of Nd 3+ are indicated. The quantum yield relative to that of the reference, sodium salicylate, achieves a maximum of 1.8 in the 2% Nd sample for excitation into the 4f 5d bands of Nd 3+ at 175 nm. This value is obtained by applying a number of corrections to the raw data.

55 34 First, the raw data are corrected for the fact that the relative quantum efficiency of the PMT for the 4 F 3/2 4 I 9/2 emission wavelength of Nd 3+ between 860 and 910 nm is much less than that at the nm emission wavelength range of sodium salicylate. A correction factor for the relative response of the PMT is obtained by convoluting the corrected emission of the sample and sodium salicylate reference each with the quantum efficiency of the PMT and calculating the ratio of these products yielding a correction factor of 20 ± 6. A great deal of effort was made to accurately obtain the relative quantum efficiency of the PMT which, because of the rapid decrease in response in the region above 860 nm, leaves this considerable uncertainty of about ±30%. Secondly, it is estimated that only 33% of the 4 F 3/2 emitted photons occur on the 4 F 3/2 4 I 9/2 transition, based on reported [11] emission spectra of YLiF 4 :Nd and calculations of the branching ratios determined by a Judd-Ofelt analysis [12], implying a further correction of about 3. An actual measurement of the branching ratios obtained from the IR emission spectrum was performed by R. L. Cone at Montana State University using an Applied Detector Corp. 403L Ge detector at the exit slit of a Spex 1000M spectrometer. All spectra were referenced against a tungsten halogen lamp operating at 2800K. The measurement yielded a value of 31.1% for the fraction of the emission occurring to 4 I 9/2, very close to the value calculated. This result produced a correction factor of 3.22±0.3. Finally, there is an uncertainty concerning the relative reflectivities of the samples and sodium salicylate reference. Although these may be somewhat different, they are probably both less than 20% in the strongly absorbing regions of the spectrum of interest. Thus this should add not more than a ±10% error. Using an estimate that the absolute quantum yield of sodium salicylate as 0.58, implies an absolute quantum yield for the 4 F 3/2 emission of about 1.05 ± The estimated uncertainty is based on the accumulated errors discussed above. This value for the quantum yield is about three times the value of 0.32 [5] obtained for GdLiF 4 :Eu. However, it is still well below the theoretical maximum quantum yield of 2 based on the quantum splitting scheme described above. This highlights the fact that even in a system which exhibits highly efficient quantum splitting, other losses can limit the absolute quantum yield. Indeed, measurements of the quantum efficiency of the GdLiF 4 :Eu quantum splitting

56 35 phosphor Error! Bookmark not defined. show that a broad defect absorption reduces the quantum efficiency considerably. A study of the dynamics will allow for an examination of some of the reasons for the reduced quantum yield for GdLiF 4 :Nd. The excitation spectra for detection above and below 780 nm are compared in Fig. 5. The spectra are normalized to the Gd 3+ 6 I transition. The black (dotted) curve is obtained detecting wavelengths λ > 780 nm so that only the Nd 3+ IR emission from 4 F 3/2 is monitored. The red (solid) curve is the excitation spectrum for λ < 780 nm and is dominated by Nd 3+ emission from 4 D 3/2 which is not enhanced by the quantum splitting. Both the 6 G excitation features of Gd 3+ and the 4f 2 5d bands of Nd 3+ are enhanced when detecting the 4 F 3/2 emission supporting the conclusion that quantum splitting plays an important role in the emission. For detection with λ < 780 nm, there is evidence for an impurity or defect absorption band near 200 nm. Excitation Efficiency (arb. units) Nd 3+ 4f 2 5d Excitation Spectrum GdLiF 4 : Nd2% Scaled to Gd 3+ 6 I Gd 3+ 6 Detection: λ > 780 nm Detection: λ < 780 nm Gd W avelength (nm) Fig. 2.5 (Color online) Comparison of the excitation spectra of GdLiF 4 :Nd2% detecting only the 4 F 3/2 emission with λ detect >780 nm with that of the case of detection for λ detect <780.

57 Dynamics of the Quantum Splitting Despite the fact that a great deal of work has been done on quantum splitting due to cross relaxation energy transfer (CRET), there have been, to our knowledge, only two studies [13,14] of the dynamics of this process. The studies considered the Gd 3+ -Eu 3+ couple in GdNaF 4 :Eu 3+ and in GdLiF 4 :Eu 3+. Both the cross relaxation and direct transfer were observed with rates about two orders of magnitude slower than for the Gd 3+ -Nd 3+ couple studied here. As pointed out in Wegh et al. [4], the process achieves its efficiency because of energy migration among the Gd 3+ ions which are stoichiometric in all known successful cross relaxation energy transfer quantum splitters. Dipole-dipole energy transfer or exchange is just too slow except for ions that are near neighbors. The fact that energy migrates within the Gd 3+ ions ensures that the excitation in the 6 G J levels of Gd 3+ gets to spend a portion of its time as a near neighbor of Nd 3+. Thus the dynamics within the Gd 3+ system is expected to play an important role in the process. When a sample of GdLiF 4 containing 2% Nd 3+ is excited at 157 nm with a molecular F 2 laser, one sees a buildup of the 6 P 7/2 transition of Gd 3+ at 313 nm as shown in Fig. 6 by the black (dark solid) curve. This buildup has two components. One is very fast, at a rate which exceeds the time resolution of these experiments (<50ns, limited by some background scattered light from the laser discharge and defect luminescence), which represents about 20% of the population feeding. The second is a slower buildup over several microseconds, representing about 80% of the feeding. The cause of these two components becomes clear from the dynamics of the 6 I emission of Gd 3+ at 281 nm shown by the purple (dotted) curve in Fig. 6. Its decay rate coincides with the 6 P 7/2 population buildup rate. Also shown in Fig. 6 by the red (dot-dashed) curve is the emission at 866 nm from the 4 F 3/2 state of Nd 3+ which also builds up within the temporal resolution of the experiment. Thus we conclude, as suggested based on an earlier discussion of the reduced matrix elements, that cross relaxation process B from Fig. 2 is the dominant one in the quantum splitting. However, the fact that the 6 P 7/2 population does have a very fast component indicates that there may also be a contribution from the cross relaxation energy transfer process

58 37 labeled as A in Fig. 2. The relaxation of Gd 3+ from 6 I to 6 P in a few microseconds is unlikely to occur due to multiphonon relaxation because of the large energy gap ( 3000 cm -1 ) and low phonon energies of the GdLiF 4 host, but rather most likely occurs through the cross relaxation energy transfer process labeled C in Fig. 2. Consistent with this suggestion is the fact that the relaxation is dependent on Nd 3+ concentration as discussed below. In this process a Nd 3+ ion is excited from the 4 I 9/2 ground manifold to 4 I 13/2, for which there is a good resonance match with the 6 I 6 P transitions on Gd 3+. GdLiF 4 :Nd2%, Excitation: λ=157 nm Emission Intensity (arb. units) Gd 3+ 6 P 7/2 (313 nm) Gd 3+ 6 I (281 nm) Nd 3+ 4 F 3/2 (866 nm) Nd 3+ 4 D 3/2 (358 nm) Tim e(µs) Fig. 2.6 (Color online) Time evolution of the 6 I (281 nm) and 6 P 7/2 (313 nm) emission intensities of Gd 3+ and the 4 D 3/2 and 4 F 3/2 emission intensities of Nd 3+ in a GdLiF 4 :Nd2% sample under 157 nm pulsed laser excitation. The behavior of the dynamics of process C and its concentration dependence provides important information on the role of donor-donor energy transfer among the Gd 3+ ions. The dynamics of the 6 I and 6 P emissions are shown as a function of concentration in Fig. 7. The relaxation process is nearly exponential as seen by the

59 38 dashed lines plotted over the 6 I time-resolved emission which are fits to the data assuming an exponential decay of 6 I. The values for the fit are shown on the figure and are summarized in Table 1. The relaxation rate scales nearly linearly with concentration as expected. Also shown are the time-resolved intensity of the 6 P 7/2 emission along with fits to the data using the 6 I decay time as the feeding term in the 6 P 7/2 population. Indeed, the same times describe both the 6 I and 6 P 7/2 emissions. The decay of 6 P 7/2 is also nearly exponential with a rate that depends on Nd 3+ concentration. These rates are also summarized in Table 1. The nearly exponential relaxation processes for all three concentrations suggests that energy migration among the Gd 3+ ions is fast compared to these CRET relaxation rates. In that case the Gd 3+ excitation samples all sites thereby spending a fraction of its time nearby a Nd 3+ ion with which it can undergo CRET. If, after energy transfer from the 4f 2 5d state of Nd 3+ to Gd 3+, the energy remained localized on that Gd 3+ ion, the CRET rates would be highly non-exponential. In addition, without energy migration, CRET process C would be hindered as all of the energy resonances that we have discussed assume that the Nd 3+ ions are in their ground state. However, processes A and B leave the Nd 3+ ion in an excited state for a time roughly equal to the lifetime of the 4 F 3/2 state of about 400 µs. Also, in the absence of rapid Gd 3+ -Gd 3+ energy transfer, some of the possible processes providing the initial Nd 3+ Gd 3+ energy transfer could also leave Nd 3+ in an excited state, as discussed earlier, compromising the CRET processes A and B which also assume that the Nd 3+ ions are in their ground state. Table 2.1 Experimental energy transfer rates. Process Nd 3+ conc. Gd 3+ Nd 3+ Expt ET rate(s -1 ) CRET A All 6 G 6 P 4 I 9/2 4 G 5/2 >2x10 7 CRET B All 6 G 6 I 4 I 9/2 4 F 5/2, 2 H 9/2 >2x10 7 CRET C 6 I 6 P 4 I 9/2 4 I 13/2 1% 3.8x10 5 2% 5.7x10 5 3% 8.0x10 5 Gd 3+ Nd 3+ 6 P 7/2 8 S 7/2 4 I 9/2 2 L 17/2 1% 4.3x10 4 2% 6.7x10 4 3% 9.1x10 4

60 39 GdLiF 4 :Nd, λ ex = 157 nm Emission Intensity (arb. units) 2% Nd 2% Nd τ d ( 6 I)=1.75 µs 3% Nd τ d ( 6 I)=1.25 µs 3% Nd 1% Nd 6 P 7/2 1% Nd τ d ( 6 I)=2.6 µs Time(µs) Fig. 2.7 (Color online) Time evolution of the 6 I (281 nm) and 6 P 7/2 (313 nm) emission intensities of Gd 3+ under 157 nm pulsed excitation in GdLiF 4 :Nd for 1, 2, and 3% Nd concentrations. The dashed lines show the fits using the 6 I decay times shown in the figure. Those same times are used as the rise times in the fits to the 6 P 7/2 emission for the sample with the same Nd 3+ concentration. The excited Gd 3+ ions in the 6 P 7/2 state then undergo energy transfer to the nearly resonant 4f 3 states of Nd 3+ at a rate described by the decay of the Gd 3+ 6 P 7/2 emission. Proof of this second step is seen by monitoring the 4 D 3/2 emission under 157 nm excitation. It is observed that this emission closely follows the Gd 3+ 6 P 7/2 population with a small delay and that it has zero population immediately after the laser excitation (see Fig. 6). This occurs because the intrinsic 4 D 3/2 lifetime ( 1 µs due to multiphonon relaxation to 2 P 3/2 ) is much shorter than the 6 P 7/2 lifetime, as seen from its decay under direct 355 nm excitation into the 4f 3 states just above 4 D 3/2, as shown in Fig. 8. The fact that the 4 D 3/2 population closely follows the excited Gd 3+ population

61 40 demonstrates that energy transfer from Gd 3+ to Nd 3+ does occur, a process which is necessary for the second step of the quantum splitting process. The observation that the 4 D 3/2 emission (spectrally integrated) is more than an order of magnitude greater than the Gd 3+ 6 P 7/2 emission (see Fig. 1) indicates that a significant fraction of the Gd 3+ ions transfer their energy to Nd 3+ since the two populations follow one another because of the short inherent lifetime of 4 D 3/2. Its greater time integrated intensity results from its faster radiative rate than that of 6 P 7/2 which is spin forbidden. Since we do not know the relative radiative rates, it is not possible to estimate from these relative intensities the efficiency of this Gd 3+ Nd 3+ energy transfer. GdLiF 4 :Nd2% Emission Intensity (arb. units) 4 D 3/2 λ ex =355nm τ r = 0 µs τ d = 1 µs 2 P 3/2 λ ex =355nm τ r = 1 µs τ d = 20 µs 2 P 3/2 λ ex =157nm τ r = 4 µs τ d = 22 µs Time (µs) Fig. 2.8 Time evolution of the 4 D 3/2 and 2 P 3/2 emission of Nd 3+ in a sample of GdLiF 4 :Nd2% under 355 nm excitation and the 4 P 3/2 emission under 157 nm excitation. The decay of 2 P 3/2 is the rate limiting state in the feeding of 4 F 3/2. Also plotted as dashed lines are fits to the data using the rise and decay times indicated on the figure

62 41 The 4 D 3/2 state decays non-radiatively to 2 P 3/2 whose population dynamics is also shown in Fig. 8 for both 355 nm and 157 nm excitation. Under 355nm excitation, it builds up at the 4 D 3/2 decay rate and decays in 20 µs, its intrinsic non-radiative lifetime. Under 157 nm excitation it has a slower buildup resulting from the population feeding from 4 D 3/2 whose population is controlled by energy transfer from 6 P 7/2 of Gd 3+. The 2 P 3/2 decay ultimately feeds 4 F 3/2 through multiphonon relaxation down the ladder of states of Nd 3+ from whose radiative decay provides the second photon in the quantum splitting arises. Thus the feeding of 4 F 3/2 for the second step in the quantum splitting continues for 100 µs. The temporal behavior of the 4 F 3/2 emission further supports the presence of quantum splitting. As shown in Fig. 9, when the 4f 3 Nd 3+ states just above 4 D 3/2 are excited directly at 355 nm, such that there is no quantum splitting, the 4 F 3/2 emission builds up with a rise time that is close to the value of the decay time of the 2 P 3/2 Nd 3+ emission (20 µs). The 4 F 3/2 emission under 157 nm excitation, also shown in Fig. 9, shows a much more rapid buildup as expected due to the first step in the quantum splitting, namely the cross relaxation step. However, note that the 4 F 3/2 emission does not immediately begin an exponential decay. Rather its population remains high due to feeding from the second step in the quantum splitting which maintains a feeding term for about 100 µs as 2 P 3/2 decays. Attempts to fit the dynamics presented in Fig. 9 (dashed curves) with an exponential rise and decay indicate that under 355 nm excitation, the 4 F 3/2 emission has both a fast (immediate with respect to the experimental time resolution) followed by an exponential rise with a 12 µs rise time. The latter represents only 33% of the total contribution to the feeding of the 4 F 3/2 population. The source of the fast component is unknown but it suggests the existence of some other channel of relaxation for 355 nm excitation. Under 157 nm excitation there is again a fast component, resulting from the first CRET step due to processes A and B, followed by an additional feeding through 2 P 3/2 for about 100 µs (see Fig. 8). Here the additional feeding contributes only 9% to the 4 F 3/2 population. Under ideal conditions of quantum splitting, this should represent 50% of the contribution to the 4 F 3/2 population through the process labeled ET 2 in Fig. 2. Because of the observation that even under

63 nm excitation there exists an unexplained very fast component to the 4 F 3/2 population, it may be that a somewhat lower value than 50% should be expected. However, the fact that it is only 9% seems to explain, in part, the less than ideal quantum yield. GdLiF 4 :Nd 3+ 2% Emission Intensity (arb.units) 2 P 3/2 λ ex =355nm Expt 4 F 3/2 λ ex =355 nm λ ex =157 nm Fits 4 F 3/2 (λ ex =355nm) τ r =12µs(33%), τ d =360µs 4 F 3/2 (λ ex =157nm) τ r =20µs(9%), τ d =350µs 2 P 3/2 (λ ex =355nm) τ r =0.9µs(100%), τ d =20µs Time(µs) Fig. 2.9 (Color online) Time evolution of the 2 P 3/2 and 4 F 3/2 emission in a GdLiF 4 :Nd2% sample under 355 nm and 157 nm excitation. The fits shown on the figure are obtained using the rise and decay times indicated in the legend. They percentage indicates the fraction of population buildup which is contributed by this rise time. The remainder of the population buildup is taken to appear immediately after excitation. There are a number of potential sources for this reduced contribution including radiative transitions from 4 D 3/2 and 2 P 3/2 that are observed in Fig. 1, radiative transitions from 6 P 7/2 of Gd 3+ prior to energy transfer to Nd 3+, transfer of energy from 6 P 7/2 of Gd 3+ to impurities or defects, and cross relaxation among Nd 3+ ions. In addition, non-radiative processes involving 4 F 3/2 are possible. Indeed, the observed

64 43 lifetimes of the 4 F 3/2 emission are below the low concentration limit of 535 µs in GdLiF 4 :Nd [15] and, in agreement with the results of Zhang et al [15], the 2 and 3% samples exhibit significant non-exponential behavior indicative of Nd 3+ -Nd 3+ cross relaxation (not shown). However, while this would contribute to the reduced quantum yield, it would not explain the lower than expected contribution to the feeding of 4 F 3/ DISCUSSION It is of interest to examine the mechanisms for the cross relaxation energy transfer (CRET) responsible for the quantum splitting. For closely spaced ion pairs, this may occur by dipole-dipole interactions or exchange interactions [16]. For more distant pairs, the exchange will become unimportant because of its rapid decrease with distance. According to Forster-Dexter dipole-dipole energy transfer theory, the transfer rate, P dd AB can written [17] as P dd AB = 1.4 x f A f B S AB E -2 R -6. (1) Here f A and f B, are the oscillator strengths of the transitions on Nd 3+ and Gd 3+, E is the transition energy of each ion (in ev), R is the distance between the two ions (in Angstroms), and, S AB is the spectral overlap (in cm -1 ) of the downward and upward transitions. In Fig. 3 it was shown for CRET process A that there are many 4 I 9/2 4 G 5/2 transitions of Nd 3+ that are nearly resonant with the 6 G J 6 P J transitions of Gd 3+. The oscillator strength of each of these crystal field transitions of Nd 3+ in YLiF 4 are typically about 5 x 10-7 based on spectral analysis of some of the individual crystal field transitions at 20K. However, one can also estimate the oscillator strengths from experimental and calculated values integrated over all transitions in the manifolds by dividing by the number of final states which yields about the same average oscillator strength per crystal field transition [18]. A similar situation holds for process B which involves the 6 G J 6 I J transitions of Gd 3+ and the 4 I 9/2 4, F 2 5/2 H 9/2 or 4 F 7/2 transitions of Nd 3+. These Nd 3+ transitions also have oscillator strengths of about 5 x 10-7.

65 44 The oscillator strengths of the transitions within the 6 G 7/2 6 P J or the 6 G 7/2 6 I J manifolds of Gd 3+ have not been measured but their reduced matrix elements have been calculated [6]. The reduced matrix elements for the 6 G 7/2 6 I J transitions are almost a factor of 10 greater than those of the 6 G 7/2 6 P J transitions, yielding the expectation that under similar resonance conditions, the probability for process B should be one to two orders of magnitude great than for process A. As described earlier, a factor of 5 was observed. The difference may be due to the quality of the energy resonance for the two processes. The Gd 3+ oscillator strengths are calculated based on the reduced matrix elements [6] for Gd 3+ and Judd-Ofelt parameters for Gd 3+ in YLiF 4 [19]. The total oscillator strength to all transitions 6 G 7/2 6 I is 2 x 10-6 and for 6 G 7/2 6 P 7/2 it is 1.5 x Since there are 39 final states in 6 I, each crystal field transition, on average, has an oscillator strength of 5 x It is now possible to estimate the CRET transfer rates for dipole-dipole interactions in process B from Eq. (1). Using typical values of 3 x 10-7 for each transition of Nd 3+ and 5 x 10-8 for each transition of Gd 3+ and assuming a single perfect energy resonance with a linewidth at room temperature of 10 cm -1 (spectral overlap integral = 0.1) one finds a rate of 3.3 x 10 5 s -1 for a nearest neighbor pair separated by 3.73 A. This rate falls to 5 x 10 4 s -1 for a next nearest neighbor pair separated by 5.15 A. To predict what should be observed one has to know whether the donor-donor transfer among the Gd 3+ ions is occurring and whether it is faster than the donor-acceptor CRET rates. The results from the dynamics of process C involving a CRET from 6 I to 6 P suggest, based on the nearly exponential decay of 6 I and rise of the 6 P 7/2 population, that the donor-donor transfer occurs much more rapidly than the observed CRET rate of 6 x 10 5 s -1 in the 2% Nd sample. If one assumes that the same is true for process A where the CRET rates are >2x10 7 s -1, then the predicted rates should take into account the fact that, on average, the excited Gd 3+ excitation spends a fraction, 4x, (x is the fractional concentration of Nd 3+ ) of its time as one of the four nearest neighbors of Nd 3+. Thus for 2% Nd the nearest neighbor rate should be multiplied by a factor of 0.08 yielding a result of 2.7 x 10 4 s -1. This rate is obtained for one resonance between the Gd 3+ 6 G 7/2 6 I and the 4 I 9/2 4, F 2 5/2 H 9/2 or 4 F 7/2 transitions of Nd 3+. Even if one were to assume that all Nd 3+ transitions (11)

66 45 were perfectly resonant with a transition on Gd 3+, which would be an extreme assumption, and if contributions from more distant pairs are added, the maximum predicted rate still would be less than 10 6 s -1. The assumption of rapid energy transfer among the Gd 3+ donors is supported by studies of Gd 3+ -Gd 3+ interactions. Studies of band-to-band exciton transitions in GdCl 3, Gd(OH) 3, and Tb(OH) 3 have shown that exchange interactions among nearest neighbor ions can yield resonant energy transfer rates among nearest neighbors that are as large as to s -1 for resonant energy transfer among Gd 3+ ions in their 6 P 7/2 state or Tb 3+ ions in their 5 D 4 state [20]. These rates correspond to the condition of resonance with homogeneous linewidths at 1.5 K of about 0.1 cm -1. At room temperature, where these linewidths are 10 cm -1, corresponding rates would be 10 8 to 10 9 s -1. Even though the exchange interaction will probably be considerably smaller in fluorides, the expectation that donor-donor transfer rates for the 6 G state of Gd 3+ should exceed 2 x 10 7 s -1 in GdLiF 4 seems quite reasonable. In the limit of no energy transfer among the Gd 3+ ions then the relaxation after the initial energy transfer from Nd 3+ Gd 3+ would occur by interactions between a pair of nearest neighbors. This rate would have a maximum value of ~5 x 10 6 s -1 if all transitions of the two ions were resonant. Even this extreme assumption falls well short of explaining the observed rate of > 2 x 10 7 s -1 and the absence of fast donordonor transfer seems unlikely. Thus the above analysis of the experiments points strongly to the dominant role of exchange interactions in facilitating the CRET responsible for quantum splitting in GdLiF 4 :Nd. It would be interesting to model the full dynamics, taking into account the energy migration of the Gd 3+ excitations in both the 6 G 7/2 and 6 P 7/2 states. Although this problem is a very interesting one, it is not the subject of this paper.

67 CONCLUSION Efficient quantum splitting has been demonstrated for the Gd 3+ -Nd 3+ system in GdLiF 4 :Nd 2%. A VUV photon is absorbed by the Nd 3+ ions whereupon the energy is rapidly transferred to the high-lying excited states of the 4f 7 configuration of Gd 3+ in a time scale of nanoseconds. A rapid and effective cross relaxation energy transfer then occurs in two steps. In the first, a Gd 3+ ion in its metastable 6 G state undergoes a transition to 6 I while a Nd 3+ ions makes a transition 4 I 9/2 4, F 2 5/2 H 9/2 or 4 F 7/2 at a rate >2 x 10 7 s -1. Multiphonon relaxation effectively brings the Nd 3+ ions down to the 4 F 3/2 state where they radiate the first photon. For the remaining excited Gd 3+ ion, there occurs a second cross relaxation energy transfer in which Gd 3+ undergoes a transition 6 I 6 P and Nd 3+ is excited from 4 I 9/2 4 I 13/2. The resulting 6 P 7/2 excitation on Gd 3+ transfers its energy to nearly resonant states of the 4f 3 configuration of Nd 3+ in a time scale of about µs whereby subsequent relaxation brings the population down to 4 F 3/2 of Nd 3+ where the second photon is emitted. This second step appears to be less efficient than the first. The result is a quantum yield for the emission of IR photons which has its maximum of about 1 ± 0.5, under 175 nm excitation. This is considerably below the theoretical value of 2. Nonetheless, this system exhibits the highest quantum yield for quantum splitting based on cross relaxation energy transfer and provides excellent insights into the quantum splitting process, especially with regard to an evaluation of the details of the dynamics and the mechanisms of quantum splitting. An analysis of the dynamics and the theoretical limits of the dipole-dipole contributions, leads to the conclusions that (1) there is rapid donor-donor energy migration among the Gd 3+ ions and (2) that exchange plays the dominant role in the cross relaxation energy transfer responsible for the quantum splitting. Acknowledgements We gratefully acknowledge the financial support of the U.S. National Science Foundation, Grants (RSM) and (DAK). We also express our appreciation to K. Mishra and M. Raukas for helpful discussions and to A. Meijerink

68 47 for a crystalline sample of YLiF 4 :Nd. We are especially grateful to R. L. Cone for measuring the IR emission spectrum of one of our samples. REFERENCES [1] C. Ronda, Journal of Luminescence, 100, 301 (2002). [2] S. Kuck, I. Sokolska, M. Hence, M. Doring, t. Scheffler, Journal of Luminescence, 102/103, 176 (2003). [3] W.W. Piper, J.A. de Luca and F.S. Ham, Journal of Luminescence, 8, 344 (1974). [4] R. T. Wegh, H. Donker, K.D. Oskam, A. Meijerink, Journal of Luminescence, 82, 93 (1999). [5] C. Feldmann, T. Justel, C.R. Ronda, D. U. Wiechert, Journal of Luminescence, 92, 245 (2001). [6] P. S. Peijzel, W.J.M. Schrama, A. Meijerink, Molecular Physics, 102, 11/12, 1285 (2004). [7] J. K. Berkowitz, J. A. Olsen, Journal of Luminescence, 50, 111 (1991). [8] P. W. Dooley, J. Thogersen, J. D. Gill, H. K. Haugen, R. L. Brooks, Optics Communications, B183B, 451 (2000). [9] R. T. Wegh, H. Donker, A. Meijerink, R. J. Lamminmaki, J. Holsa, Physical Review, 56, 13841(1997). [10] C. Gorller-Walrand, L. Fluyt, P. Porcher, A.A.S. Da Gama, G. F. de Sa, W. T. Carnall, G. L. Goodman, Journal of Less Common Metals, 148, 339 (1989). [11] A. L. Harmer, A. Linz and D. R. Gabbe, Journal of Physics and Chemistry of solids, 30, 1483 (1969). [12] J. R. Ryan, R. Beach, Journal of the Optical Society of America B 9, 1883 (1992). [13] H. Kondo, T. Hirai, S. Hashimoto, Journal of Luminescence, 108, 59 (2004). [14] N Takeuchi, S. Ishida, A. Matsumura and Y Ishikawa, Journal of Physical Chemistry, B 108, (2004).

69 48 [15] X. X. Zhang, A. B. Villaverde, M. Bass, B. H. T. Chai, H. Weidner, R. I. Ramotar, R. E. Peale, Journal of Applied Physics, 74, 790 (1993). [16] D. L. Dexter, Physical Review, 108, 630 (1957). [17] T. Kushida, Journal of the Physical Society of Japan, 34, 1318 (1973). [18] O. Guillot-Noel, B. Bellamy, V. Viana and D. Gourier, Physical Review B 60, 1668 (1999). [19] A. Ellens, H. Andres, M. LT. Wegh, A. Meijerink, and G. Blasse, Physical Review B 55, 180 (1997). [20] R. L. Cone and R. S. Meltzer, Phys. Rev. Letts. 30, 859 (1973) and R.L. Cone and R. S. Meltzer, Journal of Chemical Physics, 62, 3573 (1975).

70 49 CHAPTER 3 SENSITIZATION OF Gd 3+ AND THE DYNAMICS OF QUANTUM SPLITTING IN GdF 3 :Pr,Eu S. P. Feofilov b, Y. Zhou a, J. Y. Jeong c, D. A. Keszler c and R. S. Meltzer a, * a Department of Physics and Astronomy, University of Georgia, Athens, GA, USA b Ioffe Physico-Technical Institute, St. Petersburg, Russia c Department of Chemistry, Oregon State University, Corvallis, OR, USA Modified version; Journal of Luminescence, (2007), ,

71 50 ABSTRACT Efforts are reported to sensitize the Gd 3+ excitation with Pr 3+ for application to quantum cutting in GdF 3 doped with Eu 3+ and Pr 3+. Excitation and emission spectra are reported for several samples with different concentrations of Eu 3+ and Pr 3+ at room temperature. In addition, time resolved measurements are performed to obtain the dynamics of the various excited states of Gd 3+. Strong enhancement of the 5 D 0 Eu 3+ emission is observed for excitation of 6 G of Gd 3+ relative to that for other Gd 3+ excited states as expected for quantum cutting. While some enhancement of the J=0 Eu 3+ emission is also observed for excitation into the Pr 3+ 4f5d states the maximum quantum yield is disappointing, falling far below the desired goal of 2. The timeresolved studies indicate that one reason for the ineffective quantum cutting is that the Pr 3+ Gd 3+ energy transfer occurs predominantly to the 6 I state of Gd 3+. Results on the 6 I 6 P relaxation and the Gd 3+ ( 6 P 7/2 ) Eu 3+ energy transfer are described. 3.1 INTRODUCTION Attempts to develop useful lamp phosphors using quantum splitting have involved two routes, cascade emission and a cross relaxation energy transfer (CRET) in which the initially excited ion transfers part of its energy to another ion and each excited ion then emits a photon. Although both processes have been shown to yield internal quantum yields greater than 1, neither of these has lead to a useful phosphor. All of the systems which have exhibited efficient CRET quantum splitting have involved Gd 3+ and either Eu 3+ [1] or Nd 3+ [2] in which the Gd 3+ ion is prepared in the 6 G J states at about 49,000 cm -1. A downward transition on Gd 3+ coupled with an excitation of Eu 3+ or Nd 3+ leave both ions in excited states from which each emits a photon. While several attempts to sensitize the absorption of Gd 3+ using Er 3+ [3] and Tm 3+ [4] have produced quantum splitting, the materials have not proven to be useful phosphors.

72 RESULTS AND DISCUSSION Here we attempt to use the 4f5d absorption of Pr 3+ ions to sensitize the Gd-Eu couple in GdF 3. For Gd 3+ sensitization it is necessary that the sensitizer ion should have a strong absorption to a state that is above the 6 G states of Gd 3+. For Pr 3+ this requires a host with a very small crystal field to minimize the crystal field depression of the 4f5d configuration. GdF 3 satisfies this requirement and is the subject of this study. The energy level diagram along with the relevant energy transfer processes are shown in Fig. 1. When the Gd 3+ is excited to its 6 G state, a cross relaxation energy transfer with Eu 3+ occurs, shown by the dashed arrow labeled e, whereby the Gd 3+ ion undergoes a transition 6 G 6 P while a nearby Eu 3+ ion undergoes a resonant an upward transition 7 F 1 5 D 0. The excited Eu 3+ ion emits the first photon. The Gd 3+ ion in the 6 P 7/2 excited state then transfers its energy to Eu 3+ as shown by the arrow labeled j. This is followed by non-radiative relaxation among the 5 D J levels, all of which emit producing the second photon in the quantum splitting. The emission spectrum of GdF 3 : 0.3Pr, 0.2% Eu is shown in Fig. 2 for excitation at two wavelengths. For 275nm excitation with an Ar + laser, Gd 3+ is excited directly to the 6 I states from which quantum splitting cannot occur because it lies below the 6 G levels. Energy transfer to Eu 3+ still occurs and one sees emission from all the 5 D J levels. Upon excitation at 160nm with a D 2 lamp one again sees emission from all the 5 D J states of Eu 3+. However, the 5 D 0 is greatly enhanced and emission from 6 P 7/2 of Gd 3+ is now observed.

73 CT Energy (cm -1 ) f5d a b c d Half Stokes Shift k e, e' f, f' 6 G 6 D 6 I 6 P g h i j D 2 1 G 4 e' a b c d e D J 0 3 F 3,4 3 H 4 Pr 3+ f' Gd 3+ 8 S f Eu 3+ 7 F J Fig. 3.1 Energy level diagrams for Pr 3+, Gd 3+, and Eu 3+ showing the various energy transfer pathways labeled a through j. Processes a through d are shown displaced downward by 2500 cm -1 reflecting half the value of the Stoke s shift for LaF 3 for the Pr 3+ 4f5d emission.

74 53 GdF 3 :0.3%Pr,0.2%Eu Emission Intensity Excitation Wavelength 275 nm Ar nm D 2 lamp 5 D 3 5 D 2 5 D 0 Gd 3+ 6 P 7/2 5 D Wavelength (nm) Fig. 3.2 Emission spectra for a sample of GdF 3 containing 0.3% Pr and 0.2% Eu excited at 275 nm ( 6 I state of Gd 3+ ) and 160 nm (4f5d state of Pr 3+ ). Excitation spectra, detecting all emission λ > 320 nm, for two samples with different dopant concentrations are shown in Fig. 3. The peaks at ~275 nm are the 6 I states of Gd 3+ while those at 202 nm and 195 nm are the 6 G levels of Gd 3+. Below 190 nm, the 4f5d bands of Pr 3+ are observed. The band between 150 and 160 nm is the charge transfer (CT) band of Eu 3+ ; therefore at 160 nm both Pr 3+ and Eu 3+ are excited. These excitation spectra given by the two solid curves are obtained relative to that of sodium salicylate whose excitation spectrum is nearly independent of wavelength and whose absolute quantum yield is about 0.6. Based on this, the maximum quantum yield for total emission occurs for the 0.3% Pr, 0.2% Eu sample and has a disappointing value of only about 0.2 at about 160 nm. Thus it appears that GdF 3 :Eu sensitized by Pr 3+ is not a useful quantum splitting phosphor. While excitation of the

75 54 4f5d states of Pr 3+ does sensitize excitation of Gd 3+ it does not lead to high quantum yields. Nonetheless, it will be useful to study the dynamics and examine the possible causes for its limited performance. Evidence for strong quantum splitting from the 6 G levels of Gd 3+ is also shown in Fig. 3 for a sample of GdF 3 : 0.3 Pr, 0.03% Eu by comparing excitation spectra obtained by selecting different emission wavelength regions. When only 5 D 0 emission is detected, the 6 G excitation peaks are enhanced by more than a factor of 5 relative to the 6 I peaks as compared to detection wavelengths between nm where only the 5 D J (J>0) emission is detected. Here the excitation spectra are not presented relative to Na salicylate but are normalized to the Gd 3+ 6 I excitation peak. This preferential generation of population in 5 D 0 results from the cross relaxation energy transfer shown by the dashed arrow labeled e in Fig. 1. Under ideal conditions of quantum splitting, the enhancement should be no more than a factor of 2. The observation of a much larger enhancement results from (1) the fact that a large fraction of the Eu 3+ emission occurs from states other than 5 D 0 and the possibility that the Gd 3+ ( 6 P 7/2 ) Eu 3+ energy transfer may be much less than 100% efficient.

76 55 Quantum Yield Relative to Na Sal Eu 3+ - CT to scale Pr f5d GdF 3 :Pr,Eu excitation spectra normalized to 6 I but not to absolute scale Gd G 0.3% Pr,0.2% Eu λ det >320 nm 0.3% Pr,0.03% Eu λ det >320 nm Gd I Excitation Wavelength (nm) 5 D 0 λ det >580 nm 5 D J J>0 λ det <560 nm Fig. 3.3 Excitation spectra of two samples of GdF 3 :Pr,Eu. Excitation spectra obtained by detecting all wavelengths > 320 nm are referenced to a Na salicylate standard. Excitation spectra obtained with filters selectively for λ>580 nm and λ<560 nm are normalized for the 6 I peak but are not to the scale of the figure. The question of sensitization of Gd 3+ is now considered with the aid of Fig. 1. There exist a number of possible CRET routes by which Pr 3+ can transfer energy to Gd 3+. These are shown by the solid arrows labeled a through d. Forster-Dexter energy transfer requires overlap of the emission of the donor and absorption of the acceptor. A number of emissive transitions from Pr 3+ can occur which are nearly resonant with absorptions of Gd 3+. The Pr 3+ emissive transitions labeled a through d on the Pr 3+ energy level diagram in Fig. 1 have been shifted to lower energy relative to the bottom of the 4f5d band by half the Stoke s shift or about 2500 cm -1, reflecting the effect of the large Stoke s shift known for the 5d emission in the isostructural LaF 3

77 56 [5]. One sees that in view of the broad band characteristics of 5d 5f emission ( 2000 cm -1 ), for all processes a-d, near resonances occur with transitions on Gd 3+ from it 8 S 7/2 ground state. Unfortunately, only one of these, CRET labeled a, will generate the desired 6 G population. The relative contribution of the different Pr 3+ Gd 3+ energy transfer processes can be obtained by studies of the dynamics of the Gd 3+ emission shown in Fig. 4 for emission from 6 I (279 nm) and 6 P 7/2 (312 nm) under 193 nm pulsed excitation to the 4f5d state of Pr 3+. One sees that most of the initial population of Gd 3+ appears in 6 I with less than 20% occurring to 6 P 7/2. 6 I then relaxes to 6 P in 2.4 µs as demonstrated by the fits (dashed curves) to the data in Fig. 4 which show that the decay of 6 I describes also the buildup of 6 P 7/2. The fast component of the 6 P emission (<20% of maximum population) results either from energy transfer process a followed by the CRET process e or some other rapid feeding of 6 P. The mechanism of the relaxation process 6 I 6 P in Gd 3+ is unclear as its rate does not depend systematically on either the Eu 3+ or Pr 3+ concentrations. However multiphonon relaxation in a low phonon frequency materials such as GdF 3 is unlikely to explain the 2-5 µs decay time for an energy gap as large as 3000 cm -1. The decay rate of the 6 P 7/2 emission of Gd 3+ increases with Eu 3+ concentration indicating that it results from energy transfer to Eu 3+.

78 57 1E-3 GdF 3 :0.1%Pr,0.03%Eu Excitation 193nm Emission Intensity 1E nm (Gd 3+ 6 P 7/2 ) τ rise =2.4 µs 279 nm (Gd 3+ 6 I) τ decay =2.4 µs 1E Time (µs) Fig. 3.4 Time-resolved emission for 6 I and 6 P 7/2 of Gd 3+ after pulsed excitation at 193 nm showing that the decay of 6 I corresponds to the buildup of 6 P 7/2 and that energy transfer from Pr 3+ predominantly feeds 6 I. The circles are the measurement and the dashed curves are fits using an exponential decay and buildup of 2.4 µs with an initial 20% 6 P 7/2 population. We acknowledge the support of the U.S. National Science Foundation, Grants (RSM) and (DAK).

79 58 REFERENCES [1] R. T. Wegh, H. Donker, K.D. Oskam, A. Meijerink, Journal of Luminescence, 82, 93 (1999). [2] W. Jia, Y. Zhou, S.P. Feofilov, R.S. Meltzer, J. Y. Jeong and D. Keszler, Phys. Rev. B, in press (2005). [3] R. T. Wegh, E. V. D. van Loef, A. Meijerink, Journal of Luminescence, 90, 111 (2000). [4] S. Peijzel, W.J.M. Schrama, A. Meijerink, Molecular Physics, 102, 1285 (2004). [5] P. Dorenbos, Journal of Luminescence, 91, 155 (2000).

80 59 CHAPTER 4 RELAXATION OF THE 4f n-1 5d 1 ELECTRONIC STATES OF RARE EARTH IONS IN YPO 4 AND YBO 3 W. Jia *, Y. Zhou *, D. A. Keszler **, Joa-Young Jeong **, K.W. Jang *** and R.S.Meltzer * * Department of Physics and Astronomy, University of Georgia, Athens,GA USA ** Department of Chemistry, Oregon State University, Corvallis, OR USA *** Dept of Physics, Changwon National University, Chang Won , South Korea Modified version; Physica Status Solidi C: Conferences and Critical Reviews (2005), 2(1), 48-52

81 60 ABSTRACT Large bandgap materials doped with rare earth ions are currently of great interest as new vacuum UV phosphors for lighting and displays. In this report, the optical properties of YPO 4 and YBO 3 doped with Pr, Tm, Er, and Eu are described. The emission resulting from the VUV excitation of the parity allowed 4f n-1 5d 1 states and their quantum efficiencies are reported. Relaxation between the 4f n-1 5d 1 and nearby 4f n excited states is observed for some of these ions and the dynamics of these excited states is reported. In doubly-doped samples, the prospects for quantum cutting using cross relaxation energy transfer in which some of the energy of the initially excited 4f n-1 5d 1 state is transferred to an acceptor ion so that both donor and acceptor ions are left in an excited state from which they each can emit a photon are examined. 4.1 INTRODUCTION The high-lying excited states of rare earth ions are of considerable importance for a wide range of applications including scintillators, vacuum ultraviolet (VUV) phosphors, and UV lasers. In this work the problem of designing efficient VUV phosphors will be considered. The potential utility of a scheme for converting the exciting VUV photon into two visible photons using cross relaxation energy transfer involving a parity allowed 4f n-1 5d 4f n transition on one ion and a transition within the 4f n configuration of a second ion will be tested in the hosts YPO 4 and YBO 3. These materials have large band gaps that fall well into the VUV. So called quantum cutting or multiphoton phosphors offer a significant challenge for their realization. While the idea has been demonstrated in real systems, these systems exhibit serious drawbacks preventing their commercialization. Cascade luminescence from the 1 S 0 state of Pr 3+ in YF 3 was shown to occur with a quantum yield of about 1.4 [1]. Unfortunately, most of the energy in the first step in the stepwise relaxation back to the ground state occurs at about 400nm which is at a wavelength for which the human eye is insensitive. Cross relaxation in LiGdF 4 :Eu 3+

82 61 after initial excitation of the 6 G J levels of Gd 3+ was shown to produce two visible photons from Eu 3+ with 1.9 quantum yield [2]. Here the cross relaxation occurs between Gd 3+ and Eu 3+ involving transitions within the 4f 7 configurations and 4f 6 configurations, respectively. However, it is not possible to excite the 6 G state of Gd 3+ efficiently. Replacing the Eu 3+ dopant in LiGdF 4 by Er 3+ and Tb 3+ provides for efficient excitation and results in a quantum efficiency that is somewhat in excess of 100%. It utilizes a cross relaxation scheme involving a parity allowed 4f 11 5d 4f 12 transition on Er 3+ coupled with a transition within the 4f 7 configuration of Gd 3+ [3]. The demonstration of the latter scheme suggests that the use of parity allowed transitions on one of the ions in the ion-pair should be considered further and this provides motivation for the present work. 4.2 RESULTS AND DISCUSSION The 4f n-1 5d states of Ln 3+ series are rather short-lived. For the first half of the series, these lifetimes are tens of nanoseconds. In the second half of the series, there is a low-lying high spin (HS) state in the 4f n-1 5d configurations whose lifetimes are typically microseconds [4]. For cross relaxation energy transfer to be successful at down converting the energy, its rate must be competitive with these radiative rates. Fortunately, the multipolar interaction responsible for the energy transfer scales with these radiative rates, making possible a competitive cross relation energy transfer rate. The cross relaxation energy transfer scheme is indicated on the energy level diagram shown in Fig. 1 for the ion-pair of Pr 3+ and Tm 3+. After excitation of Pr 3+ into its 4f5d configuration, a cross relaxation can occur as shown by the dashed arrows, labeled A, on both the Pr 3+ and Tm 3+ energy level diagrams. Here the Pr 3+ ion undergoes a downward transition from the 4f5d configuration to its 1 D 2 excited state in the f 2 configuration while simultaneously a Tm 3+ ion undergoes an upward transition from it ground state to its 1 D 2 level, conserving energy and leaving both ions in excited states from which each can emit a photon.

83 LS 4f 11 5d HS 2 F 7/2 60ns 4f5d 4 D 1/2 2.7µs <20ns ns A B A B 2 P 3/2 4.0µs 4 F 4 9/2 I 9/2 4 I 11/2 4 I 13/2 LS 4f 12 5d HS Energy(cm -1 ) excitation: 157nm 3 P 0,1,2 1 D 2 3 P 0,1,2 1 D 2 1 G 4 3 H 5 3 F 4 3 H 4,5,6 0 3 H 6 4 I 15/2 Pr 3+ Tm 3+ Er 3+ Figure 4.1 Energy level diagrams for Pr 3+, Tm 3+ and Er 3+ in YPO 4 and YBO 3. For Tm 3+ and Er 3+ the 5d levels are split into a lower-energy high spin (HS) and higher energy low spin (LS) states. For Er 3+ the room temperature lifetimes are shown next to the emitting states. Processes labeled A and B for the Pr 3+ -Tm 3+ pair indicate energy conserving cross relaxation paths

84 63 The rate for such a cross relaxation can be estimated based on the Forster- Dexter dipole-dipole energy transfer theory. In Eq. 1 the rate, P dd AB is expressed in terms of f A and f B, the oscillator strengths of the two transitions labeled by A in Fig. 1, the transition energies, E, of each ion (in ev), the distance R between the two ions (in Angstroms), and the spectral overlap, S, (in cm -1 ) of the downward and upward transitions [5]. P AB = 1.4 x f A f B S E -2 R -6 (1) If we assume oscillator strengths of 10-2 and 10-6 for the 5d 4f and 4f 4f transitions, respectively, E = 3 ev, and S = 10-3, reflecting the fact that the 5d 4f downward transition is broad (about 1000 cm -1 ), then we find for nearest neighbors at a distance of 3.5 A, a rate of 10 9 s -1 which is ten times greater than the radiative rate of 10 8 s -1. At more typical phosphor dopant concentrations of 2-5%, the energy transfer rate would be expected to be s -1, still competitive with the radiative rate. Exchange mediated energy transfer can be even much faster, but it will only be important for nearest neighbor distances. Samples of YPO 4, YBO3 doped with trivalent rare earth ions were prepared by solid state reaction, following procedure modified relative to those reported previously [6]. For Y 1-x RE x PO 4 samples, appropriate stiochiometric mixtures of Y 2 O 3 (Cerac, 99.99%), (NH 4 ) 2 HPO 4 (Aldrich, 99.99%), Pr 6 O 11 (Alfa aesar, 99.99%), Tm 2 O 3 (Alfa aesar, 99.99%), Er 2 O 3 (Alfa aesar, 99.99%) were ground and then fired at 1150 C for 3h. The fired cakes were then mixed with an additional 3.5 wt% (NH 4 ) 2 HPO 4, and heated again at 1300 C for 3 or 4 hours. For Y 1-x RE x BO 3 samples, similar procedures were followed, except a 4.2wt% excess of H3BO3 (Alfa aesar, 99.99%) was added to each mixture. These samples were also subjected to two heat treatments. Formation of a single phase of the correct crystal structure was confirmed on the basis of powder X- ray diffraction by using a Siemens D5000 diffractometer. Excitation spectra were performed with a deuterium lamp source and VUV monochromator (Acton VM502) to select and scan the excitation wavelength. The excitation spectra are measured relative to that of sodium salicylate whose absolute quantum efficiency is estimated as nearly constant at 55-60% over the wavelength range of interest. The emission was detected with a photomultiplier (PMT) and either glass or interference filters.

85 64 Emission spectra were recorded with a CCD detector (Santa Barbara ST-6B) attached at the focal plane of a spectrometer (Acton Spectra Pro 150). Emission was excited with monochromatic light from the deuterium lamp or using a F 2 gas discharge excimer laser (GAM EX5) emitting at 157nm. Time resolved emission spectra were obtained with this laser which had a temporal pulse width of 10ns. All emission and absorption spectra (except as noted) were fully corrected for the wavelength dependent response of the CCD or PMT. The emission and excitation spectra for 1% Pr 3+ doped YPO 4 is shown at room temperature in the upper trace of Fig. 2, in agreement with what has been previously reported [7]. The quantum yield obtained from the excitation spectrum is quite high, estimated at about 0.5 at the highest peak. The emission is dominated by the 4f5d 4f 2 transitions where the 4f 2 final states are labeled in the figure. The large gap between the 4f5d state and the 3 P levels of Pr 3+ results in dominant 4f5d emission. The quantum yield drops with increasing concentration of Pr 3+ with maxima of approximately 0.25 and 0.08 for 5% and 10% concentrations, respectively. The situation for Tm 3+ doped YPO 4 is quite different as shown in the lower trace of Fig. 1. There is no evidence for emission from the initially excited 4f 12 5d configuration despite the large gap between the lowest 4f 12 5d state and the highest 4f 13 levels. The excitation spectrum shows a maximum quantum yield of just under 50% relative to that of sodium salicylate, implying an absolute yield of about As noted previously,[4] the sharp feature at 160nm is the 4f 12 5d state and the broader feature at about 172nm is the charge transfer (CT) band of Tm 3+ which lies below the 4f 12 5d state. It is perhaps the presence of this low-lying CT state that mediates the efficient relaxation to the 3 P 0 emitting level. In contrast to the case of Pr 3+ the quantum yield is much less sensitive to concentration, dropping only by half, i.e. about 0.13 at 10% Tm 3+ concentrations. The quantum yield for 0.2% Tm 3+ is a bit higher, about 30%, than for 1% Tm 3+ and this may be an underestimate since the absorption has probably dropped significantly at only 0.2% Tm 3+ levels. Er 3+, as for Tm 3+, shows no emission from the 5d configuration inypo 4 and all emission arises from the 4f 12 configuration as shown in the middle spectrum of Fig. 2. This is to be expected for Er 3+ since the 4f 11 5d states lie just above the 4f 12 levels so

86 65 that multiphonon relaxation results in rapid relaxation to the metastable 2 F 7/2 state (see Fig. 1). Because of the relatively high density of levels with smaller energy gaps, multiphonon relaxation also causes non-radiative relaxation to lower-lying levels, some of which are sufficiently long-lived to radiate as well. The emission spectrum thus shows three groups of lines from the three metastable levels indicated by the bold lines in the energy level diagram of Fig. 1. The initial state of each transition in the emission spectrum of Er 3+ in Fig. 2 is identified by a different vertical bar symbol above each peak. The room temperature decay times of each level are indicated next to the level in Fig. 1. Time resolved spectra at 300K show that all three of the emitting levels appear immediately (<20ns) after excitation of the 4f 11 5d state. This is contrary to the expected mulitphonon relaxation process and implies some other nonradiative channel leading directly and rapidly from the 4f 11 5d state to all three of these states of the 4f 12 configuration. The quantum yield relative to sodium salicylate is quite low at room temperature (0.11) as shown on the excitation spectrum on left part of the Er 3+ spectrum. The spectrum, expanded by a factor of 8, shows a weak excitation at 170nm into the high spin (HS) 5d level which occurs for all rare earths with a f n configuration which is more than half filled (n>7). The host absorption feature reported below 150nm at T=10K [4] is not observed at room temperature. The observed quantum yield relative to sodium salicylate falls at lower Er 3+ concentrations, probably due to decreased 5d absorption, suggesting that there is little concentration quenching for Er 3+.

87 66 Figure 4.2 Excitation (dashed) and emission spectra (solid) for Pr 3+, Er 3+ and Tm 3+ ions in YPO 4 at room temperature. The excitation spectra are relative to that of sodium salicylate. For Er 3+ the distinct vertical bars identify the emitting level. We now turn to the issue of quantum cutting using cross relaxation energy transfer from the 4f n-1 5d configuration of the trivalent rare earth ions. As noted in Fig. 1 by the dashed arrows labeled A and B, energy conserving cross relaxation schemes exist for the Pr 3+ -Tm 3+ pair. Process A is very attractive since it leaves both ions in visible emitting excited states. However, as seen for YPO 4 in the upper trace of Fig. 1, emission from 4f5d to 1 D 2 is much weaker than that to the 3 H J and 3 F J levels. Thus the oscillator strength of this transition is probably only about 10-4 reducing expected energy transfer rates below 10 6 s -1 which is insufficient to compete with the radiative rate. While process B would not leave Pr 3+ in a visible emitting state, it can still

88 67 provide a proof of concept demonstration. Tm 3+ absorptions to the 3 P J levels occur in the range of nm. This overlaps with the relatively strong 4f5d emission to 3 F J. However, the excitation spectrum of a doubly doped sample containing 1% Pr 3+ and 5% Tm 3+ (lower half of Fig. 3) shows no evidence of the Pr 3+ absorption when only Tm 3+ emission is detected; only the Tm 3+ excitation features of the CT and 4f 12 5d states are observed. Double doped samples of 1% Pr 3+ and 5% of Tm 3+ in YBO 3 yield a similar disappointing result as shown in the upper trace of Fig. 3. A singly doped sample containing 1% Pr 3+ exhibits, like YPO 4, strong emission from the 4f5d state but with an even larger peak quantum yield relative to sodium salicylate ( 1.2), implying an absolute quantum yield of about 0.6 for excitation between 230 to 250nm. Both the emission and excitation features are shifted by about 20nm to the red relative to YPO 4, as expected, since the 4f5d levels in YBO 3 are predicted to lie about 3100 cm -1 below the corresponding levels in YPO 4 [8]. An excitation spectrum of the doubly doped sample, detecting only the Tm 3+ emission, shows little if any features of the Pr 3+ excitation spectrum indicating the absence of energy transfer. The overlap of the 4f5d emission of Pr 3+ with the absorption from 3 H 6 to 3 P J states of Tm 3+ is even more favorable than in YPO 4 and yet virtually no emission from Tm 3+ is observed when exciting the 4f5d levels of Pr 3+. Cross relaxation energy transfer for the Pr 3+ -Er 3+ couple has also been investigated in YPO 4. The strong 4f5d emission of Pr 3+ between nm overlaps a number of absorptions of Er 3+. Unfortunately these Er 3+ transitions are quite weak with small oscillator strengths. Indeed, excitation of this samples containing 1% Pr 3+ and 10% Er 3+ at 188nm where only Pr 3+ absorbs yields no observable Er 3+ emission. We have also examined the Pr 3+ -Eu 3+ couple in both YPO 4 and YBO 3. For Eu 3+ in these hosts, the CT transition overlaps strongly with the 4f5fd absorption of Pr 3+ so that it is difficult to identify cross relaxation energy transfer. There certainly is overlap between the 4f5d emission of Pr 3+ and the large density of Eu 3+ transitions within the 4f 6 configuration. In samples containing 1% Pr 3+ and 5% Eu 3+ the Pr 3+ emission is strongly quenched in YPO 4 and is totally absent in YBO 3 suggesting that rapid energy transfer from Pr 3+ to Eu 3+ occurs and competes with the radiative decay

89 68 of Pr 3+. However this is most likely to leave the Pr 3+ ion in one of its low-lying levels from which visible emission will not occur. Indeed, very little visible emission from Pr 3+ is seen in either of these samples. Excitation Spectra (Pr to Tm energy transfer?) Quantum Yield Relative to Sodium Salicylate %Pr, 5%Tm 1% Tm 3+ (x2) 1% Pr 3+ YBO 3 1% Pr 3+ YPO 4 1%Pr, 5%Tm Wavelength(nm) Figure 4.3 Excitation spectra at room temperature demonstrating the absence of Pr 3+ to Tm 3+ energy transfer in YPO 4 and YBO 3. None of the features of the Pr 3+ excitation spectra appear in doubly doped samples when only the Tm 3+ emission is detected. The excitation spectra of the doubly-doped samples are not to scale.

90 CONCLUSION The quantum splitting of the high energy 4f n-1 5d electronic states of trivalent rare earth sensitizer ions into shared excitation of lower energy on this sensitizer and another activator ion using cross relaxation energy transfer does not seem promising in the hosts YPO 4 or YBO 3. Of the ions Pr 3+, Tm 3+, Er 3+ and Eu 3+, only Pr 3+ shows emission from its 5d state at room temperature. Attempts to observe cross relaxation energy transfer from Pr 3+ to Tm 3+ and Er 3+ at 5% dopant levels were unsuccessful, despite the fact the energy conserving pathways exist. In order to obtain sufficient energy transfer, it will probably be necessary to invoke exchange interactions and energy migration which will occur only in stoichiometric samples as previously demonstrated for LiGdF 4 :Eu 3+ [2]. For both Er 3+ and Tm 3+ in YPO 4, the 5d emission is quenched and emission from the 4f n configuration appears immediately (<20ns), implying a very fast relaxation from the lowest 4f n-1 5d state to the 4f n states, even when the energy gap between the lowest 4f n-1 5d state and 4f n states are much larger than can be bridged by multiphonon relaxation. We thank the National Science Foundation for their support of this work with Grants (RSM) and (DAK).

91 70 REFERENCES [1] W. W. Piper, J. A. de Luca and F.S. Ham, Journal of Lumininescence, 8, 344 (1974). [2] R. T. Wegh, H. Donker, K. D. Oskam, A Meijerink, Journal of Lumininescence, 82, 93 (1999). [3] K. D. Oskam, R. T. Wegh, H. Donker, E.V.D. van Loef, A. Meijerink, Journal of Alloys and Compounds, 300/301, 421 (2000). [4] L. van Pieterson, M. F. Reid, G.W. Burdick, A. Meijerink, Physical Review B 65, (2002). [5] T. Kushida, Journal of the Physical Society of Japan, 34, 1318 (1973). [6] R. P. Rao, D. J. Devine, Journal of Lumininescence, 87-89, 1260 (2000) [7] L. van Pieterson, M. F., Reid R. T. Wegh, S. Soverna, A. Meijerink, Phys. Rev. B 65, (2002), and T. Justel, P. Huppertz, W. Mayr, D.U. Wiechert, Journal of Lumininescence, 106, 225 (2004). [8] P. Dorenbos, Journal of Lumininescence, 91, 155 (2000).

92 71 CHAPTER 5 HOST SENSITIZATION OF Gd 3+ IONS IN YTTRIUM AND SCANDIUM BORATES AND PHOSPHATES FOR APPLICATIONS IN QUANTUM SPLITTING S.P. Feofilov a, Y. Zhou b, H.J. Seo c, J.Y. Jeong d, D.A. Keszler d and R.S. Meltzer b a Ioffe Physical-Technical Institute, St. Petersburg, Russia b Department of Physics and Astronomy, University of Georgia, Athens, GA c Department of Physics, Pukyoung National University, Pusan , Republic of Korea d Department of Chemistry, Oregon State University, Corvalis, OR 97331, USA Modified version; Physical Review B: Condensed Matter and Materials Physics (2006), 74(8)

93 72 ABSTRACT Energy transfer from the host excitations (STE) excited at λ~160 nm to Gd 3+ impurity ions was observed in yttrium and scandium borates and phosphates. The fluorescence and excitation spectra as well as time-resolved fluorescence data were obtained. For ScPO 4 :1%Gd 3+ efficient energy transfer to the Gd 3+ 6 G state was observed followed by the cascade emission of the visible and UV photons yielding a material exhibiting quantum splitting. For ScBO 3 :Gd and ScPO 4 :Gd, absolute quantum yields approach unity making these potential VUV excited phosphors. A comparison of estimated dipole-dipole energy transfer rates with observations support the importance of energy migration of the intrinsic excitations. 5.1 INTRODUCTION Several schemes have now been demonstrated for implementing quantum cutting which provides a means to obtain two or more photons for each photon absorbed. It therefore serves as a down converting mechanism that offers the prospect for developing materials with quantum efficiency greater than unity and it offers the prospect of providing improved energy efficiency in lighting devices. For example, a new class of fluorescent lamps could be developed by replacing the mercury discharge with xenon provided that phosphors with quantum efficiencies in excess of 150% under vacuum ultraviolet (VUV) excitation could be discovered. Examples of materials that emit two visible photons per absorbed ultraviolet photon were demonstrated in the early 1970s when it was shown that photon cascade emission from the high energy 1 S 0 level of Pr 3+ can yield two visible photons in a sequential two-step radiative process [1,2]. Detailed experimental studies of the quantum efficiency showed that the actual visible ( nm) quantum yield was 127%. It has been shown that Gd 3+ ions can also exhibit photon cascade emission in YLiF 4 [3] and GdBaB 9 O 16 [4] provided Gd 3+ is excited to its 6 G state.

94 73 In a second method using a combination of two lanthanide ions, cross relaxation resonant energy transfer (CRET) in which each ion shares a portion of the energy of the initially absorbed photon, was shown to yield two visible photons. For the Gd 3+ - Eu 3+ couple, an internal quantum efficiency as high as 190% was demonstrated [5]. After initial excitation of the 6 G state of Gd 3+ at about 50,000 cm -1, a CRET occurs whereby the Gd 3+ ion undergoes a non-radiative transition to its 6 P state while the Eu 3+ ion undergoes a transition from it thermally populated 7 F 1 state to its 5 D 0 state [6]. The excited Eu 3+ is responsible for the first visible photon. Resonant energy migration among excited Gd 3+ ions occurs from within the 6 P state until the energy resides nearby another Eu 3+ ion to which it can transfer its energy. The second excited Eu 3+ ion produces the second visible photon, achieving the quantum splitting. The quantum splitting therefore requires that the Gd 3+ be excited into its 6 G excited state at about 50,000 cm -1. However, because of the weak absorption by Gd 3+ resulting from the parity-forbidden and spin-forbidden character of transitions from the ground state to excited states of the f 7 configuration, the direct absorption of this state is weak. Thus a successful phosphor using the Gd-Eu couple will require sensitization of the Gd 3+ excitation. Sensitization of the 6 G state of Gd 3+ using the 4f n-1 5d states of other rare earth ions has been examined. Allowed f-d transitions of many of the rare earth ions occur in the VUV [7]. Tm 3+ and Nd 3+ have both been shown to sensitize Gd 3+ in GdLiF 4 but they each also provide an alternate cross relaxation pathway for energy transfer which is more efficient than the Gd-Eu CRET [8,9]. As a result, after the CRET, the Tm 3+ and Nd 3+ ions are left in low-lying excited states which yield, for the first photon, infrared emission. This of course defeats the goal of a visible quantum splitter. While sensitization of the 6 G state of Gd 3+ by Pb 2+ has also been examined, it was concluded that sensitization of the 6 G state will be difficult with this and other heavy ns 2 ions [10]. Sensitization of Gd 3+ using the intrinsic excited states of the host would be most desirable since it does not require additional doping of ions into the system and the host provides very strong absorption of the VUV excitation light. It requires that the host excited state transfer its energy effectively to 6 G of Gd 3+. This could occur either

95 74 if the host emission were to overlap the Gd 3+ absorption to 6 G at about 204 nm (or higher-lying states of Gd 3+ which relax predominantly to 6 G) or via some efficient intersystem crossing between the self-trapped exciton and the Gd 3+ 6 G potential surface. Excitation of 6 G has been demonstrated in a number of hosts, including GdPO 4, using X-ray excitation [11]. In some sense, this is an example of host sensitization of the 6 G state of Gd 3+ Host sensitization of Gd 3+ to its 6 P levels has been demonstrated in a number of systems [12]. In the present paper we study several materials in which the absorption of light by the host may have the potential for efficient transfer of energy to Gd 3+ ions in the 6 G state, as required for quantum splitting. The main feature that distinguishes these kinds of materials is that they should exhibit intense fluorescence from the intrinsic excitations in the undoped materials. This emission is usually ascribed to selftrapped excitons (STE) [13]. The spectral overlap of the host emission with the absorption of Gd 3+ (or other) ion is the necessary condition of the nonradiative energy transfer according to the Forster-Dexter energy transfer processes [14]. It is also important that the rate of energy transfer from the host to the acceptor ions be faster than the host excitation decay rate. Yttrium and scandium borates and phosphates (YBO 3, ScBO 3, YPO 4, ScPO 4 ) are interesting materials for sensitizing Gd 3+ ions because they have relatively large band gaps and they emit short-wavelength UV fluorescence efficiently when excited in the VUV. Using emission and excitation spectroscopy along with time-resolved emission, we demonstrate host sensitization of the 6 G level of Gd and obtain its efficiency, estimate the absolute quantum yields of both the undoped and Gd-doped samples, and examine the dynamics of both the host to Gd 3+ energy transfer and the dynamics of the photon cascade emission of Gd EXPERIMENTAL Samples of doped YBO 3, ScBO 3, YPO 4, ScPO 4 were prepared in powder form by using Y 2 O 3 (99.999%, Standard Material Corporation), B 2 O 3 (99.99%, Alfa Aesar), Sc 2 O 3 (99.999%, Standard Material Corporation), (NH 4 ) 2 HPO 4 (99.99%, Sigma-

96 75 Aldrich), and Gd 2 O 3 (99.999%, Standard Material Cooperation). The oxides were mixed according to the desired stochiometric ratios of each sample, including a 15 mol % excess of B 2 O 3 or a 10 mol % excess of (NH 4 ) 2 HPO 4. The mixtures were thoroughly ground and fired in alumina crucibles at 1150 C for 3h. For the phosphate samples, the resulting products were ground a second time with an additional 10 mol% excess of the phosphate reagent. This mixture was then heated at 1350 C for 3h. For the borate compounds, the resulting products were ground and mixed two additional times with a 10 mol % excess of B 2 O 3 and fires twice at 1330 C for 4h. All spectra were obtained at room temperature. Emission spectra were obtained by exciting the sample, contained in vacuum, with a deuterium lamp spectrally filtered with an Acton VM-502 VUV monochromator. The visible and UV emission was dispersed with an Acton Spectrapro-150 spectrometer and was detected with a Santa Barbara Instrument Group ST-6I CCD camera. All emission spectra were corrected for the wavelength-dependent response of the detection system. Excitation spectra were obtained by scanning the VUV monochromator, illuminated by the deuterium lamp, while detecting the emission with a PMT after passing the luminescence through colored glass or interference filters. The excitation spectra of each sample were compared to that of a reference sample of sodium salicylate whose quantum efficiency is assumed to be about 58% and constant over the excitation wavelength range from 140 to 320 nm [15]. For the time-resolved data, the samples were excited with a GAM Laser, model EX5 pulsed laser operating with F 2 or ArF whose output was at 157 or 193 nm (pulse width 10 ns). The laser emission was passed through an Acton Research VUV interference filters in order to eliminate other wavelengths from the emission of the laser discharge. The emission was selected with a 0.25 m monochromator and additional colored glass or interference filters. The bandwidth of the instrument was ~3 nm. The emission was detected with a PMT and was averaged and stored in a digital oscilloscope. A temporal resolution of 1-2 ns was obtained.

97 RESULTS AND DISCUSSION ScBO 3 Undoped ScBO 3, under excitation in the VUV at 160 nm emits a broad emission centered at 238 nm which is shown by the bold solid curve in Fig. 1. This emission has been assigned to emission from the STE or to molecular transitions within the BO -3 3 group [16]. When ScBO 3 is doped with Gd 3+, the broad UV emission decreases with an increase in Gd 3+ concentration as shown by the dotted and dot-dashed curves in Fig. 1. Accompanying the decrease of the broad emission is a dramatic increase in the 6 P 8 S emission of Gd 3+ at 313 nm, suggesting efficient energy transfer from the host states to Gd 3+. The energy transfer efficiency is so great that even the undoped samples shows a weak Gd 3+ 6 P emission due to residual Gd 3+ impurities. ScBO 3 λ ex =157nm undoped 1% Gd 5% Gd Emission Intensity STE x20 Gd 3+ 6 P to 8 S Wavelength (nm) Fig. 5.1 Emission spectra of ScBO 3, excited at160 nm. The instrinsic STE emission is shown amplified by a factor of 20.

98 77 As seen in Fig. 2, there is also evidence for weak 6 G 6 P emission around 600 nm which has been previously identified in other materials containing Gd 3+ [3,4]. The time-resolved emission described later add support to this assignment. This means that there is some energy transfer from the host states to 6 G of Gd 3+. The excitation spectra of undoped and 1% Gd and 5% Gd doped samples of ScBO 3 fluorescence (detected at wavelengths longer than 305 nm) are shown in Fig. 3. All three samples show the same spectral excitation features. The quantum yields, measured relative to sodium salicylate, indicate very high quantum efficiencies. Since the absolute quantum yield of sodium salicylate is 0.58, one sees that the 6 P emission of the 5% Gd 3+ sample has an absolute quantum yield of 0.8±0.15. Thus, if efficient energy transfer occurs from Gd 3+ in the 6 P state to another ion that is a good visible emitter, this material could be a highly efficient VUV-excited phosphor. While the broad intrinsic UV emission of the undoped sample has a considerably lower quantum yield, it is still about 0.3. The intrinsic emission decay is shown in Fig. 4. In the undoped ScBO 3 the decay time is 195 ns but it becomes much shorter in the Gd 3+ -doped samples. The data is fitted with exponential curves with the best fitted results shown on the figure. The instrumental response (laser and PMT) yields a 5 ns decay based on studies of ZnO which is known to have a sub-nanosecond lifetime. Thus the estimated decay time of the 1% and 5% Gd 3+ samples is 15 and 3 ns, respectively. These shortened decay times are consistent with the fact that the time-averaged intrinsic emission intensities decrease with Gd 3+ concentration (see Fig. 1). The increased quantum yield in the Gd 3+ -doped samples supports this assertion. The weak emission around 600 nm has a lifetime of 450 and 320 µs in the 1% Gd and 5% Gd samples, respectively as seen in Fig. 5. This is consistent with expectations for the lifetime of the 6 G state. The 6 G decay is dominated by radiative emission to 6 P and is spin-allowed, in contrast to the spin-forbidden 6 P 8 S emission whose lifetime is typically milliseconds. The large energy gap of 5000 cm -1 between 6 G and the next lower manifold, 6 D leads to low multi-phonon relaxation rates and hence to radiative decay.

99 78 Emission Intensity (arb. units) Gd 3+ 6 G -> 6 P Emission YBO 3 : 5% Gd YBO 3 : 1% Gd ScBO 3 :5% Gd ScBO 3 :1% Gd Wavelength (nm) Fig. 5.2 Emission spectra of the Gd 3+ -doped borates in the red showing the weak Gd 3+ 6 G 6 P emission Quantum Yield Relative to NaSal ScBO 3 :5% Gd ScBO 3 :1% Gd ScBO 3 :undoped Excitation Spectra Wavelength (nm) Fig. 5.3 Excitation spectra of undoped and Gd 3+ -doped ScBO 3 detecting the total emission and measured relative to that of sodium salicylate.

100 Time Resolved Emission at 250 nm Intensity (arb. units) ScBO 3-1% Gd 3+ τ=20 ns ScBO 3-5% Gd 3+ τ=8 ns ScBO 3 -undoped τ=195 ns Time (ns) Fig. 5.4 Time resolved intrinsic emission of undoped and Gd 3+ -doped ScBO 3. The emission was excited at 157 nm and detected at 250 nm. Fitted decay curves are shown by the dashed lines. The fitted values have a 5 ns instrumental contribution The relatively weak 6 G emission shows that the Gd 3+ ions are mostly excited to the 6 P state; the excitation to 6 G state is much less efficient in ScBO 3. This may be explained by the insufficient energy of the host excitations. The excitation of the 8 S 6 G transition is at λ=205 nm and corresponds only to the high-energy wing of the host emission so that spectral overlap with the host emission is not ideal. The spectral overlap between host emission and Gd 3+ absorption favors excitation of 6 D and 6 I at 254 and 276 nm, respectively. These states undergo rapid multi-phonon relaxation to 6 P. Despite the strong 6 P emission, it is not possible to study its population buildup with nanosecond resolution as the long lifetime limits the photon emission rate; thus the details of the initial state distributions resulting from the energy transfer from the host states have not be determined.

101 80 Intensity x10-4 Gd 3+ 6 G to 6 P λ em =600 nm Material (τ Fit ) ScBO 3 :5% Gd (320µs) ScBO 3 :1% Gd (450µs) YBO 3 : 5% Gd (200µs) YBO 3 : 1% Gd (260µs) 1x Time (s) Fig. 5.5 Observed decay of the Gd 3+ 6 G 6 P emission in the Gd 3+ -doped borates. The fitted decay curves are shown by the dashed lines with the decay values shown in the legend YBO 3 The case of YBO 3 shows some similarities with that of ScBO 3. The emission spectra of the doped and Gd 3+ -doped samples excited at 157 nm are shown in Fig. 6. One sees a broad emission in the undoped sample (bold solid line) peaking at about 325 nm which is considerably weaker than the emission in ScBO 3. In the Gd 3+ -doped samples a strong 6 P emission appears. The broad emission is not quenched in the 1% Gd 3+ -doped samples but in the 5% Gd 3+ it is reduced significantly.

102 81 Relative quantum yield YBO 3 Gd 3+ 6 P- 8 S Excitation 160 nm undoped 1% Gd 5% Gd X100 STE YBO 3 -undoped YBO 3 :1% Gd YBO 3 :5% Gd Gd 3+ 6 G- 6 P Wavelength (nm) Fig. 5.6 Fluorescence spectra of YBO 3 excited at160 nm. The intrinsic emission is shown expanded by a factor of 100 The time-resolved emission exhibits a fast and slow component for all samples, as shown in Fig. 7. The decay time of the fast component is very short, less than 2 ns, but this is not shown in this figure because of the larger input impedence used to obtain this data which limits the time resolution to about 1 µs. The slow component has a decay time of about 36 µs in the undoped sample. The addition of 1% Gd 3+ has very little affect on the dynamics of the slower decay component, but with 5% Gd 3+ it does decrease somewhat (31 µs). This is in sharp contrast to ScBO 3 which exhibited a single exponential decay which became much faster with the addition of Gd 3+. The spectra of the fast and slow components, shown in Fig. 8, reveal that there are two independent emission features associated with the two decays peaking at 285 nm (fast) and 325 nm (slow). We suggest that the fast component is the intrinsic emission whereas the slow component, which dominates the intensity of the undoped sample,

103 arises from some defect center. The peak of the fast component occurs at a slightly longer wavelength than that of the intrinsic emission of ScBO 3 which peaks 240 nm YBO 3 Excited at 157 nm λ=340 nm Intensity 1x10-4 1x10-5 Undoped 1% Gd 5% Gd Time (µs) Fig. 5.7 Time resolved emission excited at 157 nm and detected at 340 nm. The decay is a double exponential. The short decay component in the figure is lengthened by the 5.9 kω oscilloscope input impedence. Its actual decay time is < 2 ns.

104 83 slow 325 nm YBO 3 λ ex = 157 nm Intensity (arb. units) fast 285 nm undoped 1% Gd 5% Gd t=0 time-avg Wavelength (nm) Fig. 5.8 Time-resolved emission spectra excited at 157 nm. The t=0 spectrum is obtained from the initial intensity of the fast decay component. The spectrum of the slow decay component was obtained from the intensity at 400 ns after the fast component had decayed. It is identical to the time-averaged emission spectrum. The excitation spectra, shown in Fig. 9, indicate that the total emission intensity grows with the introduction of Gd 3+ and that the 5% sample shows an estimated absolute quantum yield of about 0.5 for excitation at 172 nm, about 60% as great as for ScBO 3 containing 5% Gd 3+. The spectral features in the excitation spectra are similar for the doped and undoped samples. Reflectance measurements indicate that the band gap of YBO 3 is at 7.65 ev (163 nm) [17]. This is just the spectral region where the quantum yield drops. It is likely that this is related to the onset of very strong absorption resulting in excitation close to the phosphor particle surface where non-radiative processes can cause a decrease in quantum yield. The peak at 170 nm is perhaps associated with excitation of the molecular BO -3 3 group.

105 84 Quantum Yield Relative to NaSal YBO 3 Excitation Spectra 5% Gd 1% Gd undoped Wavelenth (nm) Fig. 5.9 Excitation spectra of undoped and Gd 3+ -doped YBO 3 detecting the total emission and measured relative to that of sodium salicylate. For excitation at 157 nm the total absolute quantum yield of the undoped sample is about Since approximately 10% of this occurs in the fast emission component, it can be assumed that the short lifetime results from energy transfer to non-radiative killer sites. The radiative lifetime would then be about 300 times the < 2 ns measured lifetime. Thus the radiative rate of the intrinsic emission in YBO 3 is less than 600 ns. Because the quantum yield increases with Gd 3+ concentration due to energy transfer to Gd 3+ (evidenced by the appearance of 6 P emission from Gd 3+ ), the Gd 3+ must compete effectively with the killer sites. Thus the energy transfer to Gd 3+ appears to be faster than in ScBO 3 ; i.e. at room temperature it is likely that more rapid energy migration of the exciton occurs in YBO 3 than in ScBO 3. It is possible that the

106 85 killer sites are associated with the slow component of the emission but this cannot be proven. As seen in Fig. 2, there is some excitation of the 6 G level of Gd 3+ as evidenced by the weak emission at 600 nm, as was the case for ScBO 3 :Gd 3+. The decay of this emission is also shown on Fig. 5 and is similar to that of ScBO 3 with decay times of 260 and 200 µs in the 1% Gd and 5% Gd samples, respectively. This is consistent with the poor overlap of the fast component of the emission with the 6 G absorption which occurs for λ<204 nm ScPO 4 The ScPO 4 :Gd 3+ samples are definitely the most interesting of all studied in the present paper from the point of view of energy transfer to the high energy excited states of Gd 3+ and for enabling quantum cutting. In contrast to the borates, ScPO 4 phosphate does sensitize the 6 G state of Gd 3+ with considerable efficiency. The emission spectra of ScPO 4, undoped and doped with 1% Gd 3+, are shown in Fig.10. The host emission has been ascribed to STEs [18] or intramolecular transitions of the phosphate group[19]. Intrinsic emission from the host excitations, seen in the undoped sample, consists of two broad features whose maxima are at 215 and 320 nm. Emission from undoped ScPO 4 under 140 nm excitation at 10K has been reported at 211, 350 and 470 nm [20]. The 211 nm emission can only be excited with excitation energies above the band gap whereas the other bands can be excited at longer wavelengths, supporting their assignment to impurities or defects. Indeed, single crystals of ScPO 4 have been reported to exhibit emission from a variety of impurities [17]. Energy transfer from the intrinsic excitations to Yb 3+ has been observed [19]. In a sample doped with 1% Gd 3+, strong quenching of the host emission with the appearance of intense emission from Gd 3+ suggests the efficient energy transfer from the host to Gd 3+ ions. Even in the undoped sample, weak 6 P emission from Gd 3+ is observed due to some residual impurity level. In the 1% Gd 3+ sample a very weak intrinsic luminescence persists. Note that in addition to the 6 P emission, strong 6 G 8 S emission is observed at 204 nm along with 6 G 6 P and 6 G 6 I emission near 600

107 and 770 nm, respectively, indicating considerable sensitization of 6 G. The integral 6 G emission intensity is of the same order of magnitude as 6 P 8 S emission at 313 nm. 86 Relative quantum yield (arb. units) pulsed 157 nm YPO 4 undoped YPO 6 G to 8 4 Gd 1% S ScPO 4 undoped ScPO 4 Gd 1% STE cw 160 nm 6 P to 8 S peak at G to 6 P 6 G to 6 I Wavelength (nm) Fig Emission spectra of ScPO 4, and YPO 4 excited at 160 nm. The excitation spectra of Gd 3+ -doped and undoped ScPO 4 samples are shown in Fig.11. Since the emission consists of contributions through the UV, visible and near IR, the excitation spectra were obtained separately in the different spectral regions. The UV and blue emission was isolated with filters and referenced to the sodium salicylate emission and the red and near IR emission was compared to that of Y 2 O 3 :Eu5% whose absolute quantum yield as a function of wavelength is well established [15]. All excitation spectra show the same general spectral features, with a sharp onset at about 180 nm, indicating that they all result from the same initial excitation centers and that the intrinsic center transfers its energy to Gd 3+. The undoped sample (dashed curve in Fig. 11) produces a quantum yield whose maximum value is 0.23 relative to sodium salicylate at 170 nm, or an absolute quantum yield of

108 For the 1% Gd 3+ -doped ScPO 4, the UV emission from 6 P and 6 G (light solid line) shows a maximum quantum yield of 1.2 relative to sodium salicylate, or an absolute quantum yield of 0.7. The quantum yield of the red and near IR emission relative to that of Y 2 O 3 :Eu5%, (dotted line) is about 0.4. The maximum absolute quantum yield of 0.25, based on the known absolute quantum yield of Y 2 O 3 :Eu5% at 170 nm of 0.6, [20] also occurs at 170 nm. The total absolute quantum yield (bold solid line) is then obtained as the sum of these two contributions. It reaches as maximum value of 0.92±0.2. Thus this could be an excellent phosphor if the Gd 3+ ions can transfer their energy to a visible emitting activator. Quantum Yield / Relative Quantum Yield ScPO 4 :1% Gd relative to NaSal (200 nm<λ em <350 nm) Estimated Absolute Quantum Yield ScPO 4 :1% Gd relative to Y 2 O 3 :5% Eu (λ em >550 nm ScPO 4 -undoped relative to NaSal Wavelength (nm) Fig Excitation spectra of undoped and Gd 3+ -doped ScPO 4. The doped sample is referenced to sodium salicylate (dashed curve). The excitation of the UV portion of the emission is measured relative to sodium salicylate (thin solid curve) while the red portion of the emission is referenced to Y 2 O 3 :5%Eu 3+ (dotted curve). The estimated absolute quantum yield is shown by the bold solid curve

109 88 The relative contributions of the 6 G and 6 P emission can be estimated from their relative quantum yields. The result is that about 20% of the total emission occurs from 6 G in ScPO 4 containing 1% Gd 3+. In the ideal circumstance for photon cascade emission, all Gd 3+ ions would be excited to 6 G and would radiate in two steps, first to 6 P and then from 6 P to the 8 S ground state, producing equal contributions in the two spectral regions. If one assumes no non-radiative losses, the observed relative contributions indicate that only about 25% of the Gd 3+ ions start from the 6 G state; the remainder are excited to 6 P, 6 I, or 6 D, where the latter two states rapidly decay to 6 P through multi-phonon emission. Indeed, as noted above, the 215 nm broad intrinsic emission band overlaps both the 6 G and 6 D Gd 3+ absorptions. In addition, the 320 nm feature is strongly resonant with the 6 P and perhaps the 6 I absorptions. Therefore it is to be expected that energy transfer will populate all of these states. The results of time-resolved fluorescence measurements in ScPO 4 :1%Gd 3+ confirms the above conclusions. The decay of the intrinsic emission is shown in are shown in Fig. 12 along with an exponential fit to the data. The decay of the intrinsic emission at 215 nm in the undoped sample is 75 ns after subtracting the instrumental contribution.

110 τ fit =80 ns ScPO 4 Excited at 157 nm λ em =220 nm Undoped Gd 1% Intensity 10-3 τ fit =13 ns Time (ns) Fig Time-resolved emission of undoped and Gd 3+ -doped ScPO 4 excited at 157 nm. The fits are shown by the dashed lines and include a 5 ns instrumental contribution This is much longer than the 9 ns decay time reported in nominally undoped single crystals prepared by flux growth [17]. The decay at 220 nm has been remeasured in single crystals supplied by Lynn Boatner of Oak Ridge National Laboratory and a decay time of 130 ns was observed. This is value is much closer but slightly longer than the 75 ns value obtained in the phosphor powders. In the sample containing 1% Gd 3+ the fitted decay time is 13 ns, yielding an actual measurable decay time of 8 ns. This is assumed to arise from energy transfer to Gd 3+ but the buildup of the Gd 3+ emission cannot be obtained with sufficient resolution. The dynamics of the 6 G and 6 P states of Gd 3+ are shown in Fig. 13 where 6 G is obtained from emission detected at both 206 nm and 600 nm while the dynamics of 6 P is determined from the emission at 313 nm. The 78 µs decay of Gd 3+ 6 G state is due to radiative transitions to the 6 P state: the corresponding build-up of 6 P population may be seen in the 6 P 8 S

111 90 fluorescence kinetics which can be described by the same rate as the 6 G decay. This build-up makes up about 25% of the total 6 P 8 S emission intensity; 75% of the excited population goes to 6 P much faster (ns time scale). This means that the ratio of the numbers of ions excited by energy transfer to 6 G and 6 P states is about 1:3; most of the ions are excited by energy transfer from the host excitations directly to 6 P or to 6 P via 6 D and 6 I. This is in close agreement with the conclusions based on the excitation spectra where the ratio was determined to be 1:4. The experimental results show that ScPO 4 :1%Gd 3+ exhibits quantum cutting due to the Gd 3+ cascade emission process 6 G 6 P 8 S. It is favorable for quantum cutting that the dominant transition from 6 G occurs to the 6 P state, yielding visible emission in the red. However, it is unfortunate that most of Gd 3+ ions are excited by energy transfer from the host states directly to 6 P (or perhaps 6 D or 6 I) preventing the occurrence of a quantum efficiency exceeding unity. In order to obtain a more efficient phosphor material it will be necessary to find a material with a better branching ratio for the transfer of energy to 6 G rather than 6 P population. Since the second photon in the cascade emission occurs in the UV from 6 P, it will also be necessary to incorporate a second ion that emits in the visible which can receive the 6 P energy via an energy transfer process.

112 91 ScPO 4 :1% Gd 6 P--> 8 S Intensity (arb. units) 6 G--> 8 S 315nm 206nm 600nm 10-3 τ fit =78 µs G--> 6 P Time (s) Time (s) Fig Time-resolved emission of ScPO 4 :1%Gd excited at 157 nm and detected at 206nm and 600 nm (Gd 3+ 6 G emission) and 315 nm (Gd 3+ 6 P emission). The inset shows the fit of the 6 G decay YPO 4 YPO 4 also emits intrinsic emission in the UV and deep UV as shown in Fig. 10. As for ScPO 4 there are two broad emission bands, one centered at 240 nm and the other at 400 nm. The emission is considerably weaker than that of ScPO 4. Doping YPO 4 with 1% Gd 3+ completely quenches the 240 nm emission feature and strong 6 P emission from Gd 3+ is evident. A very weak 6 G emission is observed at 600 nm (to 6 P) and 204 nm (to 8 S 7/2 ) whose lifetime is 72 µs. The excitation spectra of the doped and undoped samples, under various detection conditions are reported in Fig. 14. All the excitation spectra show an onset beginning at 180 nm. The excitation spectra for the undoped sample differ somewhat depending on the detection wavelength range.

113 92 There appear to be two bands whose relative strengths are such that the shorter wavelength excitation feature dominates for detection at longer wavelengths and visa versa for detection at shorter wavelengths. Thus there appear to be two independent sources in the excitation spectrum. Two features are also observed in the Gd 3+ -doped YPO 4 fluorescence excitation spectrum. The excitation spectrum of Gd 3+ -doped YPO 4 is in agreement with that reported by Nakazawa [21] who assigned the feature at 152 nm to the host lattice absorption. The 4f 6 5d and charge transfer bands of Gd 3+ are at much higher energies. Relative Quantum Yield (arb. units) 0.12 YPO 4 :1% Gd (λ em > 280 nm) 0.10 YPO 4 (λ em > 190 nm) 0.08 YPO 4 (λ em = nm) YPO 4 (λ em > 280 nm) 0.02 YPO 4 :1% Gd (λ em > 345 nm) Wavelength (nm) Fig Excitation spectra of undoped and Gd 3+ -doped YPO 4 for detection in different wavelength regions showing the dependence of the spectra on detection wavelength For detection wavelengths greater than 280 nm in the Gd 3+ -doped sample, which includes essentially all of its emission, the 152 nm band is stronger; for detection of the weak emission at wavelengths greater than 345 nm, two features of

114 93 nearly equal relative intensity are seen. The two peaks in the excitation spectra probably correspond with the two distinct host states observed in the fluorescence spectra. The strong peak at 152 nm, observed in excitation spectra of the total (Gd 3+ + host) emission may suggest a more efficient energy transfer from the higher energy host excitations to Gd 3+. However, the total quantum yield relative to sodium salicylate is below 0.1 for both the doped and undoped samples. The dynamics of the host emission is shown in Fig. 15. The dynamics are quite unusual. First consider the undoped sample (solid curves). The decay of the emission detected at 240 nm is exponential with a decay time of 380 ns as shown by the fit to the data. This is about five times great than the lifetime in ScPO 4. However, it does not appear immediately but rather has a rise time of 55 ns as shown by the fit to the time dependence of the 240 nm emission (dashed curve). The 400 nm longer wavelength emission feature exhibits a double exponential decay and does not display any detectable rise time. The faster decay component is described by a 55 ns decay time, the same as the 240 nm emission rise time; this is shown by the fit (dashed curve). The longer decay time is 600 ns. The nearly identical dynamics of the 240 nm buildup and 400 nm fast decay component suggest that the center responsible for the 400 nm feature feeds the center corresponding to the 240 nm feature. This does not violate energy conservation provided that the 400 nm emission results from a very large Stokes shift. In this way, its excited state energy may still be above that of the center producing the 240 nm emission. The dynamics of the Gd 3+ -doped sample is nearly identical to that of the undoped sample, except that the 240 nm feature is totally absent. This can occur if the center responsible for the 240 nm emission transfers its energy very efficiently to the Gd 3+ ions. The small affect of Gd 3+ on the 400 nm emission implies that energy transfer from this center to Gd 3+ is inefficient.

115 YPO 4 Excited 157 nm undoped 1% Gd fit(undoped) nm 55 ns rise 380 ns decay Intensity 1E-3 1E nm 460 nm 55 ns decay 600 ns decay 1E Time (s) Fig Time-resolved emission of undoped (solid curves) and Gd 3+ -doped (dotted curves) YPO 4. The 240 nm emission shows a 55 ns buildup and 380 ns decay while the emission at longer wavelengths (340nm and 460 nm shown in the figure) exhibit a decay with two components. The dashed curves show fits to the 240 nm and 460 nm data for the undoped sample. 5.4 ENERGY TRANSFER RATES The decay rates of the intrinsic and Gd 3+ emission for the undoped and doped samples are summarized in Table. 1. It should be possible to compare the observed energy transfer rates from the host to the Gd 3+ ions to estimates based on dipole-dipole mediated Forster-Dexter energy transfer theory. The Forster-Dexter energy transfer rates can be represented by the expression [22,9]. P AB dd = (1.4*10 24 f host f Gd S) /( E 2 R 6 ). Eq. 5.1

116 95 where E is the transition energy in ev, f host and f Gd are the oscillator strengths of the host emission and Gd 3+ absorptions, respectively, S is the spectral overlap expressed in units of cm -1 of the host emission with the Gd 3+ absorption and R is the distance between the host excitation and Gd 3+ expressed in Angstroms. Table 5.1 Wavelengths and decay times of the emission of undoped and Gd 3+ doped scandium and yttrium borates and phosphates ScBO3 YBO3 ScPO4 YPO4 Host emission wavelength 238 nm 285 nm 215 nm (powder) 215 nm (crystal) 320 nm 240 nm 400 nm Host emission (STE) decay time undoped Gd doped 1% 5% 195 ns 15 ns 3 ns < 2 ns <2 ns <2 ns 75 ns 8 ns 130 ns 55 ns rise absent 380 ns decay 55 ns decay 55 ns decay 600 ns decay 900 ns decay Gd 3+ decay time 6 P 6 G 1% 5% 4.0 ms 450 µs 320 µs 4.5 ms 260 µs 200 µs 4.8 ms 78 µs 3.2 ms 72 µs

117 96 An upper limit of the radiative decay rate of the intrinsic emission is given by its observed decay rate. However, since the quantum yield of the intrinsic emission is of order 1 in all but YPO 4, the observed decay rate must, in fact, be close to the radiative rate, typically s -1, based on the data in Table 1. Such a decay rate corresponds to an oscillator strength f A ~10-3 to For Gd 3+ the absorptive transitions are spinforbidden so that oscillator strengths of about 10-6 to 10-7 are to be expected. The overlap can be estimated as 2x10-4 cm -1 based on the observation that the host broad band emission has a bandwidth that is much greater than that of Gd 3+ and is about 5000 cm -1. Dipole-dipole mediated energy transfer rates for typical nearest neighbor cation distances of 3.7 A are then estimated from Eq. 5.1 to lie between 5 x 10 5 to 5 x 10 7 s -1. For a 1% Gd 3+ concentration, the typical nearest neighbor distance between a localized host excitation and Gd 3+ ion is about 7 A yielding estimated energy transfer rates of s -1. Since there are a number of Gd 3+ ions in the vicinity of a host excitation these estimates should probably be increased to the range s -1. The observed rates in these borates and phosphates doped with 1% Gd 3+ are at least 10 8 s -1, a value that is one to three orders of magnitude greater than the rates estimated based on dipole-dipole Forster-Dexter energy transfer. This suggests that the host excitations are mobile allowing them to sample the whole lattice such that they spend a fraction of their time as a nearest neighbor of the Gd 3+ where the dipole-dipole interactions will be about a factor of 100 larger or where much larger exchange interactions can provide an additional energy transfer mechanism.

118 CONCLUSIONS Yttrium and scandium borates and phosphates all exhibit intrinsic emission in the UV and the lifetimes of these intrinsic emissions were determined. Efficient energy transfer from the host excitations to Gd 3+ was observed in the Gd 3+ -doped materials showing that host sensitization occurs in these materials. A comparison of the observed and estimated theoretical rates suggest that the host excitations are mobile at room temperature. For ScPO 4 :1%Gd 3+ about 30% of the energy transfer from the host excitations to Gd 3+ occurs to the 6 G state, demonstrating for the first time host sensitization of the 6 G state of Gd 3+. This excitation is followed by cascade emission of photons making possible quantum cutting in which one visible and one UV photon are emitted. Absolute quantum yields were determined for all samples with measured values of 0.92 and 0.8 in ScPO 4 :Gd and ScBO 3 :Gd, respectively. If the Gd 3+ 6 P excitation can be transferred efficiently to another ion emitting visible radiation, these Gd 3+ -doped materials could be competitive with existing VUV excited phosphors. Acknowledgements We acknowledge the support of the U.S. National Science Foundation, Grants (RSM) and (DAK). We thank Dr. Lynn Boatner of Oak Ridge National Laboratory of the single crystal samples of ScPO 4. We appreciate helpful discussion with Drs. Madis Raukas and Kailash Mishra of OSRAM SYLVANIA.

119 98 REFERENCES [1] Piper, W.W., DeLuca, J.A. and Ham, F.S., Journal of Luminescence, 8, 344, [2] Sommerdijk, J.L., Bril, A.and de Jager, A.W., Journal of Luminescence, 8, 341, [3] Wegh, R.T., Donker, H. and Meijerink, A., Physical Review B , [4] Yang, Z., Lin, J.H., Su, M.Z., Tao, Y. and Wang, W., Journal of Alloys and Compounds. 308, 94, [5] Wegh, R.T., Donker, H., Oskam, K.and Meijerink, A., Science 283, 663, [6] Wegh, R.T., Donker, H., Oskam, K. and Meijerink, A., Journal of Luminescence, 82, 93, [7] Dorenbos, P., Journal of Luminescence, 91, 155, [8] Peijzel, P.S., Schrama, W.J.M. and Meijerink, A., Molecular Physics 102, 1285, [9] Jia, W., Zhou, Y., Feofilov, S.P., Meltzer, R.S., Jeong, J.Y. and Keszler, D., Physical Review B 72, , 2005 [10] Babin, V., Oskam, K.D., Vergeer, P. and Meijerink, A., Radiation Measurements, 38, 767, [11] L.H. Brixner and G. Blasse, Chemical Physics Letters, 157, 283 (1989). [12] Lin J., Su, Q., Journal of Alloys and Compounds, 210, 159 (1994). [13] W. Hayes and A.M. Stoneham, Defects and Defect Processes in Non-metallic Solids, J. Wiley, New York [14] Dexter, D. L., Journal of Chemical Physics, 21, 836, [15] J.K. Berkowitz, and J.A. Olsen, Journal of Luminescence, 50, 111, [16] I. N. Orgorodnikov, V. A. Pustovarov. A. V. Kruchalov, L. I. Isaenko, M. Kirm, G. Zimmerer, Phys, Solid State 42, 464 (2000) A. Meijerink, G. Blasse, M. Glassbeek, Journal of Physics: Condensed Matter, 2, 6303 (1990) [17] A. Mayolet, J.C. Krupa, J. SID, 4, 179 (1996).

120 99 [18] A. Trukhin and L.A. Boatner, Materials Science Forum, , 573 (1997). [19] E. Nakazawa and F. Shiga, Journal of Luminescence, 15, 255 (1977) and 1 X. Wu, H. You, H Cui, X. Zeng, G. Hong, C-H. Kim, C-H. Pyun, B-Y. Yu and C-H. Park, Materials Research Bulletin, 37, 1531 (2002). [20] L.van Pieterson, M.Heeroma, E. de Heer and A. Meijerink, Journal of Luminescence, 91, 177, [21] E. Nakazawa, Journal of Luminescence, 100, ). [22] T. Kushida, Journal of Physical Sociey of Japan, 34, 1334, 1973.

121 100 CHAPTER 6 LUMINESCENCE OF LANTHANIDES DOPED GdZrF 7 Joayoung Jeong and Douglas A. Keszler Oregon State University, Department of Chemistry Corvallis, OR

122 101 ABSTRACT Single phase polycrystalline GdZrF 7 compounds doped with several lanthanides of Eu 3+,Pr 3+,Ce 3+,Tb 3+,Tm 3+ have been synthesized. The VUV luminescence characters have been investigated using optical spectroscopy. GdZrF 7 doped with Eu 3+ and or Pr 3+ shows white emission with quantum yield of almost unity under 180nm excitation. 6.1 INTRODUCTION There is considerable interest in phosphors that can be excited in the vacuum UV (VUV) for applications in mercury-free fluorescent lamps and plasma displays. We report here on a new highly efficient nearly white phosphor. Kolk et al reported a VUV excitation study of Pr 3+ doped LaZrF 7.[1] They demonstrated photon cascade emission (PCE) of Pr 3+ when Pr 3+ is excited to its charge transfer band or to its 4f5d states. However, energy transfer from the STE to Pr 3+ occurred predominantly to 3 P 0 and not 1 S 0 as required for PCE. Here we study the related material, GdZrF 7 doped with various lanthanides. We show that Eu 3+ doped GdZrF 7 produces a nearly white emission with high quantum efficiency in the VUV. While it does not exhibit quantum cutting, which could yield quantum efficiencies of 2, it does perform with quantum efficiency near 1. In this paper we study GdZrF 7 doped with Eu 3+ and with several other lanthanides for their potential as VUV phosphors. The VUV emission and excitation spectra are investigated and quantum yields are determined. 6.2 EXPERIMENT SAMPLE PREPARATION All samples were synthesized by solid state reaction. GdF 3 (Alfa Aesar,99.99%), ZrF 4 (Aldrich,99.99%), NH 4 F (Aldrich,99.99%) and lanthanides of Eu 2 O 3 (Stanford

123 102 Materials Corporation,99.99%), Pr 6 O 11 (Alfa Aesar,99.99%), TbF 3 (Alfa Aesar,99.9%), Tm 2 O 3 (Stanford Materials Corporation,99.99%), Ce(NO 3 ) 3 6H 2 O (Aldrich,99.9%) depending on the kind of dopant were mixed well in stoichiometric ratio and was charged into carbon crucible capped with another carbon crucible to provide the raw mixture less oxygen atmosphere during heating. The ZrF 4 was added 12 mol% in excess than the stoichiometric ratio. The carbon crucibles were put into a bigger aluminium crucible covered with lid and the space between the carbon crucibles and alumina crucible was filled with carbon powder. The heating was carried at C during 1.5-2hs MEASUREMENT All polycrystalline samples were measured in their structural character by x-ray powder diffraction method. The diffraction data was collected at Siemens D5000 Diffractormeter using an in-house program and the λ=1.572 of Cu Kα radiation. The reference x-ray pattern of GdZrF 7 was calculated from the single crystal structure solution data of our other research work. All spectra were measured at room temperature. Emission spectra were obtained by exciting the sample, contained in vacuum, with a deuterium lamp spectrally filtered with an Acton VM-502 VUV monochromator. The visible and UV emission was dispersed with an Acton Spectrapro-150 spectrometer and was detected with a SBIG ST-6I CCD camera. All emission spectra were corrected for the wavelength-dependent response of the detection system. Excitation spectra were obtained by scanning the VUV monochromator, illuminated by the deuterium lamp, while detecting the emission with a PMT after passing the luminescence through colored glass or interference filters. The excitation spectra of each sample were compared to that of a reference sample of sodium salicylate or Y 2 O 3 :5% Eu 3+. The blue and UV parts of the emission spectra were referenced to sodium salicylate and the red regions of the spectra were referenced to Y 2 O 3 :5% Eu 3+. The absolute quantum yields were estimated based on absolute quantum yields of 0.6

124 for both of these reference standards over the UV and VUV for sodium salicylate and at 160 nm for Y 2 O 3 :5% Eu RESULTS AND DISCUSSION GdZrF 7 : undoped Undoped GdZrF 7 emits a strong broad emission band covering much of the visible and the near UV as shown in Fig. 1. We assign this emission to that of the selftrapped exciton (STE). A STE emission was reported for LaZrF 7 by van der Kolk and coworkers at somewhat shorter wavelengths [1] but their LaZrF 7 has a different crystal structure as discussed below. The undoped GdZrF 7 also exhibits a strong Gd 3+ 6 P emission at 313 nm. Since Gd 3+ has only parity and spin-forbidden transitions in the VUV, its appearance indicates energy transfer from the STE to Gd 3+. The overlap of the high energy tail of the STE emission with Gd 3+ 6 P absorption provides a mechanism for the energy transfer and the very poor overlap may explain the incomplete transfer from the STE despite the fact that Gd is stoichiometric in the sample GdZrF 7 :1%Eu 3+ Following energy transfer from the STE to the 6 P level of Gd ion a second energy transfer occurs to the Eu 3+ ions. Fig1 is the emission spectrum of GdZrF 7 doped with Pr 3+ and/or Eu 3+.

125 104 Intensity GdZrF 7 doped and undoped Excited λ= 160 nm, T=300 K a)undoped b)1% Eu c)1% Pr d)1% Eu, 1% Pr 5 D 0 7 F J J= D 3 7 F J 5 D 2 5 D 1 7 F J 7 F J J=1 2 3 J=1 2 3 J=1 2 3 Pr Wavelength (nm) Fig Emission spectra of GdZrF 7 a) undoped, b) doped with 1% Eu 3+, c) doped with 1% Pr 3+, d) double doped with 1% Pr 3+ and 1% Eu 3+ under 160nm excitation. The emission spectrum of the Eu 3+ -doped sample is composed of the STE broad emission band peaking in the blue region and several emission peaks from the 5 D J levels of Eu 3+. The peaks from Eu 3+ were assigned using the energy level structure of trivalent lanthanides in LaF 3 by W.T.Carnel [2]. The strong emission peak at 619nm from the 5 D 0-7 F 2 transition is a sign of the hypersensitivity of Eu 3+ emission at the non symmetric site. Michel Poulain at 1972 [3] solved the crystal structure of SmZrF 7 and explained that SmZrF 7 will be isostructural with lanthanide fluorozirconates LnZrF 7 compounds (Ln=lanthanides, Y) in the space group of P2 1. We recently grew the single crystal and solved the crystal structure of GdZrF 7. Its basic frame of crystal structure is almost same with SmZrF 7 except one type of F - ion is disordered into two positions resulting in P2 1 /m space group [4]. According to our solution the Zr 4+ ion is six fold coordinated in a ZrF 2-6 cluster with O h site symmetry and the Gd 3+ ion is eight

126 105 fold coordinated by F 1- ions in GdZrF 7. However each of the eight Gd-F bond lengths is slightly different such that the Gd 3+ site has a small deviation from inversion symmetry which may be the origin of hypersensitivity of Eu 3+ emission. Many oxide compounds of borate (e.g. YBO 3 :Eu 3+ emitting in red region), silicate or phosphate doped with Eu 3+ do not show a high level of emission from 5 D J (J>1) in the blue, or green regions. On the other hand GdZrF 7 :1% Eu 3+ shows several clear emission peaks in blue and green region together with the red emission. This may be related to the low phonon energy (ħω) of the host compound. Fluorides have a small value for the maximum phonon energy, (typically <500cm -1 [5]) and this retards the non radiative relaxation between two energy levels. The non radiative relaxation rate can be expressed by Eq. 6.1 [6]. W NR = β el exp (-α( E-2ħω max )) Eq. 6.1 where W NR is the non-radiative multiphonon transition probability, β el and α are constants, E=1750cm -1 is the energy difference between 5 D 1 to 5 D 0 level and ħω max is the maximum phonon energy. Schuurumans and van Dijk investigated β el and α in a wide series of crystals and β el and α values of LaF 3 is used here. The calculated non radiative multiphonon transition rates between the 5 D 1 and 5 D 0 levels of Eu 3+ using the maximal phonon energy of 350cm -1 and the β el =1.9*10 7 s -1 and α= 5.3*10-3 cm -1 is 7.2*10 4 s -1 which is much smaller than the value of the borate compound (8.4*10 9 s -1 ) The non-radiative transition rate from 5 D 2 to 5 D 1 and from 5 D 3 to 5 D 2 are calculated to be decreased exponentially into 1.7*10 3 s -1 and 190s -1 respectively. Another factor to be considered is the low Eu 3+ concentration. The excited Eu 3+ can also relax by cross relaxation between two Eu 3+ ions when the Eu 3+ concentration is high. For 3% Eu 3+ in Y 2 O 3, the emission spectrum is dominated by the 5 D 0-7 F J transition [7]. Even the maximal phonon energy of Y 2 O 3 is small (550cm -1 )[6] the high Eu 3+ concentration may remove the transition from high energy levels (J>1). The broad emission in blue region is assigned STE emission as is in undoped GdZrF 7. The XRD results showed the samples are almost pure single phase of GdZrF 7 with very small impurity peaks at 2theta = 23.3 and 28 coming from GdZr 3 F 15 (Fig 2).

127 106 Intensity (a) GdZrF7:Eu1% θ Intensity (b) θ Fig. 6.2 XRD patterns of GdZrF7:1%Eu 3+ (a), the reference XRD pattern (b) calculated from single crystal structure data. The quantum yield of GdZrF 7 :Eu 3+ samples are shown in Fig 3 and 1% Eu 3+ doped one is estimated to have absolute quantum yield of almost unity under 180nm excitation [14].

128 Excitation Spectra (Estimated Absolute Yield) Estimated Absolute Quantum Yield GdZrF 7 :Eu Eu 1% Eu 3% Wavelength (nm) Fig. 6.3 Excitation spectrum of GdZrF 7 :1%Eu 3+ and 3% Eu 3+ samples GdZrF 7 :Pr 3+, (Eu 3+ ) The emission spectrum was shown in Fig 1. Curve c) is the extended emission spectrum of 1% Pr 3+ doped sample and d) is the extended emission spectrum of double doped sample with 1% Pr 3+ and 1% Eu 3+. The emission spectrum of Pr 3+ doped sample is also composed of a broad band and several sharp peaks. The broad emission band peaking in the blue region is similar to that observed in both Eu 3+ doped sample and in the undoped sample while the sharp emission peaks represent the 4f-4f transitions of Pr 3+ energy states. In LaZrF 7 :Pr 3+ the emission spectrum showed the transition not only from 1 S 0 level but also from 3 P 0 level of Pr 3+ resulting in PCE [1]. In Fig 1, the peaks at 480nm, 602nm and 640nm on curve c) are assigned to transitions from 3 P 0 -to 3 H 4, 3 H 6 and 3 F 2 of Pr 3+ respectively while the emission peaks from 1 S 0

129 108 level are completely absent. As in LaZrF 7,:Pr 3+, the 1 D 2 emission appears to be quenched. Usually the absence of emission from 1 S 0 of Pr 3+ is explained by the position of the lowest 5d level of Pr 3+. For the occurrence of 1 S 0 emission, the lowest 5d level of Pr 3+ should be located above the 1 S 0 energy level and under this circumstance one can obtain PCE. Dorenbos [8] has surveyed the crystal depression energy (red shift) for a number of compounds and has provided a very useful equation to evaluate the lowest 5d level of lanthanides in those compounds. However his paper does not include this information for GdZrF 7. Instead we tried to estimate its 5d position from the excitation spectrum data. From the excitation spectra obtained by detecting all the emission above 300nm, one can see clearly in both samples doped with Pr 3+ that an additional shoulder appears near 207 nm which can be regarded as the absorption peak of the lowest 5d level of Pr 3+. This can be compared with the excitation spectrum of LaZrF 7 : Pr 3+ where the lowest absorption peak occurred at 205nm for the lowest 5d level of Pr 3+ in the excitation spectrum for detection of either the 1 S 0 or 3 P 0 emissions [1]. The crystal field splitting of a lanthanide ion depends predominantly on the coordination number and on the type of nearest neighbor anion. The higher is the coordination number the smaller is the crystal field splitting. In addition, there is a dependence on the type of nearest neighbor cation such that the higher is the charge to size ratio of the nearest neighbor cation the smaller is the crystal field splitting. If the crystal structure of LaZrF7 is assumed isostructural with GdZrF7 the main difference between GdZrF 7 and LaZrF 7 is that the cation size of Gd 3+ (1.19Å) is smaller than La 3+ (1.3Å). The distance between Gd 3+ and F 1- can be predicted to be smaller in GdZrF 7. Hence the Pr 3+ (1.27Å), located at the Gd 3+ site in GdZrF 7, should feel a stronger interaction with its F 1- neighbors than in LaZrF 7. As a result, in GdZrF 7 the Pr 3+ ion will experience a higher crystal field splitting such that the lowest 5d level of Pr 3+ in GdZrF 7 will be located at a lower energy than in LaZrF 7. However actual data show just a small difference of 2 nm in their lowest 5d levels and the lowest 5d level of Pr 3+ in GdZrF 7 is still appears to be 1560cm -1 higher than 1 S 0 level. The x-ray data of LaZrF 7 in Kolk s paper is different with that of GdZrF 7 even the crystal structure was expected to be isostructural and this may explain the reason why the estimated result about the lowest 5d level above was not repeated quite well

130 109 in the experimental data in both compounds. Therefore in GdZrF 7 compound the embedment of the lowest 5d level of Pr 3+ and 1 S 0 level into the strong absorption band arising from 225nm (Fig 4) will be the reason for the absence of 1 S 0 emission. The other distinguishing feature of the Pr 3+ doped sample is the strong emission peak from the 6 P level of the Gd ion. The strong 6 P emission suggests that the energy transfer from Gd 3+ ion to Pr 3+ ion is very inefficient. This is in sharp contrast with Eu 3+ doped GdZrF 7 where the low intensity of the 6 P emission suggests the efficient energy transfer from Gd 3+ to Eu 3+. Only lanthanide ions such as Eu 3+ [9], Tb 3+ [10,11] and Dy 3+ [12] are known as acceptor ions for energy transfer from Gd 3+ via the 6 P or 6 I levels. 1.0 Relative excitation efficiency (sodium saliclyate reference) a)undoped (arb. scale) b)1% Eu c)1% Pr d)1% Eu, 1% Pr Excitation Spectra: GdZrF 7 Detection: λ > 280 nm, T=300 K Wavelength (nm) Fig. 6.4 Excitation spectra of GdZrF 7 a) undoped, b) doped with 1% Eu 3+, c) doped with 1% Pr 3+, d) double doped with 1%Pr 3+ and 1%Eu 3+ for whole emission above 300nm

131 110 To our knowledge there are no studies of energy transfer from Gd 3+ to Pr 3+ and no energy gaps between 6 G level and 6 P levels in Gd 3+ has the resonance with any of the excitation energy into 3 P J (J=0,1,2), 1 I 6 and 3 D 0 levels of Pr 3+ from ground state. Actually this process is irrelevant because the population of 6 G level Gd 3+ was not observed. We think there should be other pathway to populate the 3 P 0 level of Pr 3+ creating emissions at 480nm, 602nm and 640nm. The STE emission band of GdZrF 7 overlaps with the excitation absorptions to the 3 P J (J=0, 1, 2) and 6 I 0 levels of Pr (20935~22690 cm -1, [2]) and energy transfer from the STE could excite Pr 3+ to those levels. The weakness of the Pr 3 + emission can be explained by the small amount of Pr 3+ which requires energy transfer over relatively large distances. The weak direct excitation from 5d absorption of Pr 3+ could be another explanation for this Pr 3+ emission. The STE emission band also overlaps with the absorption of the 6 P level of Gd and excites Gd into 6 P states. However Gd 3+ is a stoichiometric element so that rapid energy transfer competes effectively with the STE radiative emission favoring a high excitation probability and strong emission intensity. In case of double doped sample GdZrF 7 :Pr 3+, Eu 3+, the total quantum yield still remained almost the same as that of GdZrF 7 :Eu 3+ (Fig 4). Based on the relative excitation efficiency in Fig 4 we can conclude the quantum yield of GdZrF 7 :1%Pr 3+ and GdZrF 7 :Eu 3+, 1%Pr 3+ are all near unity by comparison with that of GdZrF 7 :Eu 3+ which was shown to have absolute quantum yield of almost unity. However, in the case of GdZrF 7 :Pr 3+, much of the emission is in the UV ( 6 P 8 S of Gd 3+ ) so that the visible quantum yield is somewhat reduced. The excitation spectrum in Fig 4 showed that there is no quantum splitting in Eu 3+ emission in GdZrF 7 :Eu 3+ compound. GdZrF 7 :Eu 3+ does not have the absorption peaks of 6 G of Gd at around 200nm or may have weak peaks which is buried by the strong broad absorption band while LiGdF 4 :Eu 3+ or LiGdF 4 :Nd 3+, known as a nice quantum cutter has distinct absorption peaks of the 6 G level of Gd 3+ [9,13]. The emission from Eu 3+ in GdZrF 7 :Eu 3+ can not be explained by the energy transfer via the 6 G level of Gd as was the case in LiGdF 4 compound. The Eu 3+ emission in GdZrF 7 :Eu 3+ will need another excitation source like STE emission mentioned already.

132 111 The x-ray diffraction results showed that Pr 3+ doped sample and Pr 3+ and Eu 3+ codoped sample are almost single phase of GdZrF 7 compound (Fig 5) Intensity (a) GdZrF7:Pr 1% θ Intensity (b) GdZrF7:Pr1%,Eu1% θ Fig. 6.5 XRD patterns of (a) GdZrF 7 :1%Pr 3+, (b) GdZrF 7 :1%Pr 3+,1%Eu GdZrF 7 :Ce 3+ or Tb 3+ or Tm 3+ Studies of energy transfer between Ce 3+ and Gd 3+ have shown this to be efficient. Ce 3+ was used as a sensitizer ion to absorb the excitation light efficiently and deliver it to an activator that can not be excited directly by the excitation light. The emission spectrum of GdZrF 7 :Ce 3+ was essentially identical to that of the undoped GdZrF 7 and was composed of a broad emission band peaking in the blue and a strong

133 112 emission peak from 6 P of Gd 3+ ion at 313nm. Doping with Ce 3+ did not provide any improved quantum yield nor was there any 5d emission of Ce 3+ under 160nm excitation (Fig 6). This can be understood by estimating the wavelength of absorption of the lowest 5d level of Ce 3+ in GdZrF 7 using the previously determined position of the lowest 5d state of Pr 3+ at 275nm [8]. The STE emission would then have little overlap with the lowest 5d absorption of Ce 3+ so that energy transfer to Ce 3+ would not be possible, thereby explaining the absence of 5d emission of Ce 3+. As a result, this also explains its inability to sensitize Gd 3+. In Fig 9 curve a) shows the XRD data of GdZrF 7 :Ce 3+ The addition of Tm 3+ or Tb 3+ to GdZrF 7 :Eu 3+ produced very little change in the emission spectrum. The Tb 3+ was added with the intension of modifying the color index of GdZrF 7 : Eu 3+ by creating additional emission in the green region. Tb 3+ is well known to accept energy from Gd via 6 I level (about 274nm) [10]. However co-doping of Tb 3+ added only two weak emission peaks at 383nm and 544nm assigned as the transition from 5 D 3-7 F 6 and 5 D 4-7 F 5 (Fig 7). The co-doping of 1%Tm 3+ to GdZrF 7 :1%Eu 3+ produced only a loss of quantum efficiency (Fig 8). It did not create new emission peaks from Tm 3+ and instead decreased the intensity of Eu 3+ emission a little bit (Fig 6). The expected blue emission from the 1 D 2 levels of Tm 3+ was not observed in the emission spectrum. The x-ray data in Fig 9 shows Tm 3+ co-doped sample is single phase of GdZrF 7 compound.

134 GdZrF 7 Emission Spectra Excitation λ = 160 nm, T=300 K a)1% Eu b)1% Ce c)1% Eu 1% Tm Intensity Wavelength (nm) Fig. 6.6 Emission spectrum of GdZrF 7 doped lanthanides. a) Doped with 1%Eu 3+, b) doped with 1%Ce 3+, c) codoped with 1%Eu 3+ and 1%Tm 3+

135 114 Relative quantum yield (arb. units) GdZrF Excited at 160nm CCD Slit: 400µm VUV Slit: 3mm No filter Exposure time: 20s Corrected a) Eu 1% b) Eu 1%, Tb 0.5% W avelength (nm) Fig. 6.7 Emission spectrum of GdZrF 7 :1%Eu 3+,1%Tb 3+ for the whole emission compared to that of GdZrF 7 :1%Eu 3+.

136 115 Relative excitation efficiency GdZrF7 VUV Slit: 400µm PMT=900V Filter=WG320 Dark current deducted uncorrected a) Eu 1% b) Ce 1% c) Tm1%, Eu1% Wavelength (nm) Fig. 6.8 Excitation spectrum of several GdZrF 7 samples doped with lanthanides, a) doped with 1%Eu 3+, b) doped with 1%Ce 3+, c) co doped with 1%Eu 3+ and 1%Tm 3+. All excitation spectra were measured for whole emission spectrum except b) which was measured excluding the emission peak of 6 P of Gd 3+.

137 (a) GdZrF7:Ce3+ Intensity θ (b) GZF:1%Tm,1%Eu Intensity θ Fig. 6.9 XRD patterns of (a) GdZrF 7 :Ce 3+ and (b) GdZrF 7 :Eu 3+,Tm CONCLUSION GdZrF 7 was synthesized using the solid state reaction and its VUV luminescence characters were investigated for several lanthanide dopants. In GdZrF 7 which is known to be isostructural with SmZrF 7 Gd 3+ ion locates at the site coordinated with eight F 1- ions. The small distance difference among each of Gd 3+ -F 1- bonding can explain the hypersensitivity of Eu 3+ emission which shows the strong emission peak of 5 D 0-7 F 2 transition in GdZrF 7 :Eu 3+. The quantum yield of this phosphor was nearly unity. The Pr 3+ in this compound did not show the PCE because the lowest 5d level of Pr 3+ and 1 S 0 states are included in the strong broad absorption

138 117 band. The emission of undoped GdZrF 7 includes a broad band in blue region which is assigned as STE emission. From the experiment of the several lanthanides of Pr 3+, Tb 3+, Tm 3+ and Ce 3+ as dopant we concluded that the STE emission formed from band to band excitation transfers the excitation energy to 6 P level of Gd. Only in the case of Eu 3+ was there efficient transfer from Gd 3+. The GdZrF 7 :1%Pr 3+ and GdZrF 7 :1%Pr 3+,1%Eu 3+ also showed high quantum yield of almost unity similar to GdZrF 7 :1%Eu 3+ but they did not significantly change the color coordinates of the material. Acknowledgements Acknowledge National Science Foundation. REFERENCES [1] E. Van der Kolk, P.Dorenbos, C.W.E. van Eijk, Optics Communications. 197, (2001) [2] Energy level structure and transition probabilities of the trivalent lanthanides in LaF3, W.T.Carnel, Hannah Crosswhite, H.M.Crosswhite. [3] Michel Poulain, Marcel Poulain et Jacques Lucas, Material Research Bulletin, 7, , (1972) [4] Joayoung Jeong, L.N. Zakharov, Y. Zhou, R.S. Meltzer, D.A. Keszler, in praperation [5] A.P. Vink, P.Dorenbos, C.W.E. van Eijk, Journal of Solid State Chemistry, 171, (2003) [6] M.F.H.Schuurmans, J.M.F. van Dijk, Physica B+C, 123, 131 (1984) [7] G.Blasse, B.C.Grabmaier, Springer,Verlag Berlin Heidelberg, Luminescent Materials, 1994 [8] P.Dorenbos, Journal of Luminescence, 91, (2001)

139 118 [9] Rene T.Wegh, Harry Donker, Koenrrad D.Oskam, Andries Meijerink, Science, 282, (1999) [10] M.J.J.Lammers, G.Blasse, Physical Status Solidi. (b), 127, 663 (1985) [11] A.J. De Vries, M.F.Hazenkamp, G.Blasse, Journal of Luminescence, 42, (1988) [12] Yuji Saito, Takashi Kumagai, Shinji Okamoto, Hajime Yamamoto, Takashi Kunimote, Japanese Journal of Applied Physics, 43,6A, (2004) [13] Jia, W.; Zhou, Y.; Feofilov, S. P.; Meltzer, R. S.; Jeong, J. Y.; Keszler, D., Physical Review B: Condensed Matter and Materials Physics, 72(7), (2005) [14] Y. Zhou, J.Y. Jeong, D.A. Keszler, S.P. Feofilov and R.S. Meltzer, 16th International Conference on Dynamical Processes in Excited States of Solids, DCP 07.

140 119 CHAPTER 7 CRYSTAL STRUCTURE AND Eu 3+ LUMINESCENCE OF GdMF 7 (M=Hf 4+, Zr 4+ ) Joayoung Jeong, L.N. Zakharov and Douglas A. Keszler Oregon State University, Department of Chemistry Corvallis, OR

141 120 ABSTRACT The crystals of GdMF 7 (M=Hf 4+, Zr4 4+ ) has been grown by solid state reaction. The crystal structure of GdHfF 7 was determined by single crystal X-ray diffraction method. Four Gd 3+ ions and four M 4+ ions form interlaced square planars respectively along the [001] direction. The Gd 3+ ion at one corner of the Gd 3+ square locate at the center of the M 4+ square and the M 4+ ion at M 4+ square locate at the center of that Gd 3+ square making those two squares interlaced each other. Another layer composing of Gd 3+ square and M 4+ square at the same pattern locates below this top layer along c-direction. However Gd 3+ square locates below M 4+ square of the top layer and M 4+ square locates below the Gd 3+ square of the top layer. This bottom layer forms a slab with the top layer of next lower unit cell. The VUV emission of GdMF 7 doped with Eu 3+ was investigated under 160nm excitation. 7.1 INTRODUCTION The phase diagram research in mixed compounds of LnF 3 and HfF 4 was carried out by Fedorov, P. P. and coworkers (Ln=La, Nd, Sm, Gd, Ho, Er, Yb) [1]. They showed the phases of LnHfF 7 (Ln=La-Lu,Y), LnHf 2 F 11 (Ln=La-Nd) and LnHf 3 F 15 (Ln=Sm-Lu,Y) were formed in this binary systems. It was found that the LnHf 2 F 11 compounds when the Ln is light lanthanide are isostructural with the analogous fluorozirconate compounds with orthorhombic space group Ibam. Koreneve Yu. M and coworkers indexed the powder x-ray diffraction patterns of the LnHfF 7 in a monoclinic lattice and provided their lattice parameters (Ln=La-Lu,Y) [2]. Michel Poulain and coworkers solved the crystal structure of SmZrF 7 and provided the cell parameter of GdZrF 7 from the powder XRD result [3]. However structural establishment of single crystal of GdHfF 7 or GdZrF 7 has not been provided yet. In luminescence study, there has been no research in gadolinium heptafluorohafnate based host compound. In this work we grew the single crystals of GdHfF 7 and GdZrF 7 and their crystal structures were determined. The polycrystalline forms of these compounds

142 were also prepared and the VUV luminescence characteristics of Eu 3+ in GdHfF 7 was investigated compared to that of Eu 3+ doped GdZrF 7 from the polycrystalline samples EXPERIMENT Sample preparation For the powder samples the raw chemicals of 1 mole GdF 3 (Alfa Aesar, 99.99%), 1. 2 mole HfCl 2 O 8H 2 O (Alfa Aesar, 98+%) were mixed in small amount of alcohol and dried at oven. The dried raw mixture was ground with NH 4 F (Aldrich, 99.99%) and charged into carbon crucible with lid. This carbon crucible was again placed inside a bigger alumina crucible covered with Alumina lid and the space between the two crucibles was filled with activated carbon powders. The carbon powders were used to help prevent the mixtures from exposed to oxidizing atmosphere. This crucible set was fired at 600 C 2hs and the fired cake was ground. The ground powder was heated again at 850 C 2h in the same atmosphere. For the GdZrF 7 compound 1.12 mole ZrF 4 (Alfa Aesar, 99.9%) was mixed instead and the uniformly ground mixture was heated in the same atmosphere above at 750 C 1.5h. The final powder samples of GdMF 7 (M=Hf 4+, Zr4 4+ ) prepared by heat treatment were charged in φ=3mm stainless (sus#316) tube. This tube was sealed when it was vacuumed. The sealed tube was heated at 950 C 2h followed by very slow cooling of 6 C/hr till 750 C and cooled along the cooling speed of furnace X-ray crystallography. Two types of crystals were found; one showed the unit cell of GdF 3, the other type was close to that of SmZrF 7 which crystal structure was determined previously [3]. It can be expected that compounds of GdMF 7 are isostructural with SmZrF 7. The full single crystal diffraction studies were carried out with the latter type crystals and it confirmed that the investigated compounds are GdMF 7. (M=Hf 4+, Zr 4+ ).

143 122 X-Ray diffraction data were collected on a Bruker Smart Apex diffractometer at 173(2) K using Mo Kα radiation (λ= Å). Absorption corrections were made by SADABS [4]. The structure was solved using direct methods and refined with fullmatrix least-squares methods based on F 2. Crystal data and some of details of X-ray diffraction experiment and refinement of the crystal structure of GdMF 7 are given in the Table 1. All calculations were performed using the SHELXTL (v. 6.10) package [5]. In contrast to the structure of SmZrF 7 obtained in space groups P2 1 [3] and P2 1 /n [6], the crystal structure of GdMF 7 were solved and refined in space group P2 1 /m. It was found that one of F atoms (a µ 2 -bridge connecting Gd atoms) is disordered over two positions related by a mirror plane. Thus in the crystal there are two zig-zag -Gd-F-Gd-F- chains with different positions of the bridging F atoms, but with the same occupations. Refinement of the crystal structure of GdMF 7 in space group P2 1 with different occupations for these two zig-zag -Gd-F-Gd-F- chains show that such variant of the structure is not so good as the structure in the space group P2 1 /m.

144 123 Table 7.1 Crystal data and some of details of X-ray diffraction experiment and refinement of the crystal structure of GdMF 7 (M=Zr, Hf) Empirical formula GdZrF 7 GdHfF 7 Formula weight u u Temperature 173(2) K 173(2) K Wavelength Å Å Crystal system monoclinic monoclinic Space group P2 1 /m P2 1 /m Unit cell dimensions a = (7)Å b = (7)Å c = (10)Å α= 90 β= (2) γ= 90 a = 6.097(2)Å b = (18)Å c = 8.229(3)Å α= 90 β= (5) γ= 90 Volume (6) Å (15) Å 3 Z 2 2 Density (calculated) g/cm g/cm 3 Absorption coefficient mm mm -1 F(000) Theta range for data 2.54 to to collection Index ranges -7 h 7, -7 k 7, -10 l 10-8 h 8, -7 k 7, -10 l 10 Reflections collected Independent reflections 700 [R int = ] 721 [R int = ] Absorption correction Semi-empirical from equivalents Semi-empirical from equivalents Max. and min and and transmission Refinement method Full-matrix least-squares on F 2 Full-matrix least-squares on F 2 Data / restraints / 700 / 0 / / 0 / 53 parameters Goodness-of-fit on F Final R indices [I > 2σ (I)] R1 a = , wr2 b = R1 = , wr2 = R indices (all data) R1 = , wr2 = R1 = , wr2 = Extinction coefficient 0.077(3) Largest diff. peak and and e/å and e/å 3 hole a R1= Σ ( F 0 - F c ) / Σ F 0 b wr2 = [ Σ {w(f 2 0 -F 2 c )} 2 2 / Σ wf 0 ] 1/2

145 luminescence measurement The VUV emission spectra were obtained by exciting the sample, contained in vacuum, with a deuterium lamp spectrally filtered with an Acton VM-502 VUV monochromator. The visible and UV emission was dispersed with an Acton Spectrapro-150 spectrometer and was detected with a SBIG ST-6I CCD camera. The emission spectra were not corrected for the wavelength-dependent response of the detection system. VUV Excitation spectra were obtained by scanning the VUV monochromator, illuminated by the deuterium lamp, while detecting the emission with a PMT after passing the luminescence through colored glass or interference filters. The excitation spectra of each sample were calibrated with a reference sample of sodium salicylate. All spectra were measured at room temperature 7.3 RESULTS AND DISCUSSION Crystal Structure Both GdHfF 7 and GdZrF 7 compounds have the same crystal structure with tiny variations in matching bond lengths and bond angles between two compounds. A view of unit cell of GdHfF 7 compound is shown in Fig 1. In both structures the Gd 3+ ion is eight-coordinated with F 1- ions forming distorted square antiprismatic structure (Fig. 2) and Hf (Zr) 4+ ion is six-coordinated with F 1- ions forming an octahedron around the Hf (Zr) 4+ atom. In the polyhedron of Gd 3+ ion the Gd-F distances are being in the ranges 2.205(8)-2.374(6) Å (Hf) and 2.206(6)-2.380(4) Å (Zr). The octahedron of Hf (Zr) 4+ ions have small variations of the Hf (Zr)-F distances and F- Hf (Zr)-F angles (table 3 and 4). Both GdZrF 7 and SmZrF 7 compounds have different structure against lanthanide fluorozirconate compounds with small size Ln like Er,Tm,Yb and Lu forming cubic (ReO 3 structure) when they are quenched [7].

146 Fig. 7.1 Unit cell drawing of GdHfF

147 Fig. 7.2 Two views of the eight coordinated polyhedron of Gd 3+ ion in GdHfF

148 127 Fig. 7.3 A [001] directional view of GdHfF 7 showing Hf 4+ and Gd 3+ ions composing the squares respectively. In the view of [001] direction in Fig 3 the four Gd 3+ ions are locating at each corner of a square and four of Hafnium ions are also forming a square. The Gd 3+ ion at each corner of a square is locating at the center position of the Hf 4+ square as if Gd 3+ ion is face centered in the Hafnium square and the Hf 4+ ion at each corner of a square locates vice versa. However those two squares are not exactly on a plane as is shown in Fig 4. Below this top layer of Gd 3+ and Hf 4+ square planar structure there is bottom layer composed of squares of Gd 3+ and Hf 4+. The different point from the top layer is that Hf 4+ square in the bottom layer locates beneath the Gd 3+ square of the top layer and Gd 3+ square in the bottom layer locates beneath the Hf 4+ square. The [001] directional view in Fig 3 shows that each of the Gd 3+ polyhedron and Hf 4+ octahedron are connected by corner sharing in the plane of those layers via the fluorine atoms with forming those two squares of Gd 3+ and Zr 4+ to be interlaced. The bottom layer form slabs of [Gd 2 Hf 2 F 12 ] 2+ with other top layer of the next lower unit cell along c-direction and the slabs are connected by [F 2 ] 2- layers as shown in Fig 4. This structure can be

149 128 described as a succession of [Gd 2 Hf 2 F 12 ] 2+ layers and [F 2 ] 2- layers equiberating charges Fig. 7.4 A [010] directional view showing the top and bottom layers composed of Gd squares and Hf squares. Those layers form slabs of [Gd 2 Hf 2 F 12 ] 2+ with other bottom and top layers of next unit cells along c-direction.

150 129 c Fig. 7.5 A fragment of the crystal structure of the two types of zig-zag -Gd-F-Gd-Fchains structure showing the disorder at F5 position.. b As was found in [2,3] the GdHfF 7 is isostructural with monoclinic SmZrF 7. It could be that GdHfF 7 compound shall be crystallized in monoclinic system with space group of P2 1. However it was found in this research that the position of one of the F 1- ion, F5 atoms connecting two Gd 3+ ions is disordered over two possible positions related by mirror plane. This is different point from the analogous of F 1- ion whose position is not disordered in SmZrF 7 determined in P2 1. It indicates that in the crystal structure of GdHfF 7 there are two types of zig-zag -Gd-F-Gd-F- chains with different orientations of the bridging F atoms (Fig. 5) in ratio 1:1. Only one possible orientation of the similar Sm-F-Sm-F- chains was found in SmZrF 7 (P2 1 ). In another crystal structure of SmZrF 7 determined in space group P2 1 /n there are also two types of chains just as GdHfF 7 does. However there is regular order in the packing of the two zig-zag- Sm-F- Sm-F- chains while in the GdHf(Zr)F 7 positions of the two zig-zag -Gd-F-Gd-Fchains are random. Both Gd-F-Gd and Sm-F-Sm angles are same as 146.2(4). The ionic size of Hf 4+ and Zr 4+ (0.85 Å and 0.86 Å respectively) is so similar that replacing Hf 4+ with Zr 4+ would not affect the crystal structure a lot.

151 130 Table 7.2 Atomic position and equivalent isotropic displacement parameters (Å2x 103). U(eq)is defined as one third of the trace of the orthogonalized Uij tensor. a) GdHfF 7 x y z U(eq) Occupancy Hf(1) (1) 1/ (1) 6(1) 0.5 Gd(1) (1) 1/ (1) 5(1) 0.5 F(1) (11) 1/ (8) 17(2) 0.5 F(2) (11) 1/ (8) 12(1) 0.5 F(3) (12) (12) (7) 41(2) 0.5 F(4) (14) (17) (7) 66(3) 1.0 F(5) (14) (16) (10) 11(2) 0.5 b) GdZrF 7 x y z U(eq) Occupancy Zr(1) (1) 1/ (1) 7(1) 0.5 Gd(1) (1) 1/ (1) 7(1) 0.5 F(1) (8) 1/ (5) 17(1) 0.5 F(2) (7) 1/ (5) 11(1) 0.5 F(3) (8) (8) (5) 42(1) 0.5 F(4) (9) (12) (5) 66(2) 1.0 F(5) (9) (11) (7) 11(1) 0.5 Some of the important bond distances between Gd 3+ ion and F 1- ion and between M 4+ ion and F 1- ion and bond angles including in both GdHfF 7 and GdZrF 7 compounds are listed in table 3 and table 4.

152 Table 7.3 Selected bond lengths [Å] in GdMF 7. GdHfF 7 GdZrF 7 Hf(1)-F(1) 2.001(6) Zr(1)-F(1) 1.998(4) Hf(1)-F(2) 1.995(6) Zr(1)-F(2) 2.002(4) Hf(1)-F(3) 1.978(5) Zr(1)-F(3) 1.984(4) Hf(1)-F(4) 1.971(6) Zr(1)-F(4) 1.981(4) Gd(1)-F(1) 2.334(6) Gd(1)-F(1) 2.342(4) Gd(1)-F(2) 2.374(6) Gd(1)-F(2)# (4) Gd(1)-F(3) 2.294(6) Gd(1)-F(3)# (4) Gd(1)-F(4) 2.324(6) Gd(1)-F(4)# (4) Gd(1)-F(5) 2.205(8) Gd(1)-F(5)# (6) Gd(1)-F(5) )# (9) Gd(1)-F(5)# (6) #1 x,-y+1/2,z #2 -x+1,-y,-z #4 -x+2,y+1/2,-z+1 #6 -x+1,-y,-z+1 #8 x,y,z-1 131

153 Table 7.4 Selected Bond angles [ ] in GdMF 7 GdHfF 7 GdZrF 7 F(4)-Hf(1)-F(4)#1 93.5(7) F(4)-Zr(1)-F(4)#1 93.6(4) F(4)-Hf(1)-F(3)# (4) F(4)-Zr(1)-F(3)# (3) F(4)-Hf(1)-F(3) 89.8(4) F(4)#1-Zr(1)-F(3)#1 90.1(3) F(3)#1-Hf(1)-F(3) 87.0(5) F(3)#1-Zr(1)-F(3) 86.2(3) F(4)-Hf(1)-F(2) 89.1(2) F(4)#1-Zr(1)-F(2) 89.18(14) F(3)-Hf(1)-F(2) 91.3(2) F(3)-Zr(1)-F(2) 91.10(14) F(4)-Hf(1)-F(1) 89.8(2) F(4)-Zr(1)-F(1) 89.75(15) F(3)-Hf(1)-F(1) 89.8(2) F(3)-Zr(1)-F(1) 90.04(14) F(2)-Hf(1)-F(1) 178.4(3) F(1)-Zr(1)-F(2) (18) F(5)-Gd(1)-F(5)# (13) F(5)-Gd(1)-F(5)# (9) F(5)-Gd(1)-F(3)# (3) F(5)-Gd(1)-F(3)# (18) F(3)#4-Gd(1)-F(3)#5 80.5(4) F(3)#4-Gd(1)-F(3)#5 81.5(3) F(5)-Gd(1)-F(4)#6 72.8(3) F(5)-Gd(1)-F(4)# (18) F(3)#4-Gd(1)-F(4)# (17) F(3)#4-Gd(1)-F(4)# (13) F(3)#5-Gd(1)-F(4)#6 92.0(3) F(3)#5-Gd(1)-F(4)#6 91.5(2) F(5)-Gd(1)-F(4)# (3) F(5)-Gd(1)-F(4)# (2) F(4)#6-Gd(1)-F(4)#7 74.6(5) F(4)#6-Gd(1)-F(4)#7 74.7(4) F(5)-Gd(1)-F(1) 144.7(3) F(5)-Gd(1)-F(1) (17) F(3)#4-Gd(1)-F(1) 72.89(18) F(3)#4-Gd(1)-F(1) 73.11(12) F(4)#6-Gd(1)-F(1) 72.12(19) F(4)#6-Gd(1)-F(1) 71.92(13) F(5)-Gd(1)-F(2)#8 74.0(3) F(5)-Gd(1)-F(2)# (18) F(3)#4-Gd(1)-F(2)# (18) F(3)#4-Gd(1)-F(2)# (12) F(4)#6-Gd(1)-F(2)# (2) F(4)#6-Gd(1)-F(2)# (15) F(1)-Gd(1)-F(2)# (2) F(1)-Gd(1)-F(2)# (15) Hf(1)-F(1)-Gd(1) 174.3(3) Zr(1)-F(1)-Gd(1) 174.4(2) Hf(1)-F(2)-Gd(1)# (3) Zr(1)-F(2)-Gd(1)# (2) Hf(1)-F(3)-Gd(1)# (3) Zr(1)-F(3)-Gd(1)# (2) Hf(1)-F(4)-Gd(1)# (3) Zr(1)-F(4)-Gd(1)# (2) Gd(1)-F(5)-Gd(1)# (4) Gd(1)-F(5)-Gd(1)# (3) Symmetry transformations used to generate equivalent atoms: #1 x,-y+1/2,z #2 -x+1,-y,-z #3 -x+1,y+1/2,-z #4 -x+2,y+1/2,-z+1 #5 -x+2,-y,-z+1 #6 -x+1,-y,-z+1 #7 -x+1,y+1/2,-z+1 #8 x,y,z-1 #9 x,y,z+1 132

154 133 The powder X-ray diffraction method was used to check the polycrystalline samples of this compound obtained at different conditions. The XRD pattern for the polycrystalline sample of GdHfF (Fig 6 a) prepared according to the synthesis process explained above is in excellent agreement with XRD pattern calculated based on the single crystal structure solution (Fig 6 b). So the powder sample obtained at such conditions is pure. The difference in peak intensity ratio is attributed to the preferred orientation of particles into (002) plane a) Intensity θ Intensity b) θ Fig. 7.6 a) Experimental XRD pattern of the powder sample of GdHfF 7, b) The XRD pattern calculated based on the single crystal structure of GdHfF 7

155 Luminescence Characteristics We investigated the luminescence characters of GdMF 7 :Eu 3+. The 1% Eu 3+ doped and undoped polycrystalline samples of gadolinium heptafluorohafnate,gdhff 7 were prepared. In Fig 7 a), and 7 b) the VUV emission spectra under 160nm excitation are shown and compared to the emission spectra of polycrystalline samples of gadolinium fluoroheptazirconate, GdZrF 7. Inset of Fig 7 is the extended one. Relative quantum yield (arb. units) a) GdHfF 7 undoped b) GdHfF 7 : Eu 1% c) GdZrF 7 : Eu 1% d) Gd(Hf,Zr)F 7 : Eu 1% Wavelength (nm) Fig. 7.7 Emission spectrum under 160nm excitation. Emission spectra of a) GdHfF 7, b) GdHfF 7 :1%Eu 3+, c) GdZrF 7 :1%Eu 3+, d) Gd(Hf 0.5,Zr 0.5 )F 7 :1%Eu 3+.

156 135 GdZrF 7 doped with Eu 3+ was shown to be a high quantum yield phosphor under 180nm excitation with white emission in our other work [8]. The STE created by the host intrinsic absorption transfer the excitation energy to the 6 P level of Gd 3+ and the second energy transfer from Gd 3+ to Eu 3+ is followed generating Eu 3+ emission. The emission spectrum of 1% Eu 3+ doped GdZrF 7 in Fig. 7 c) is composed of a broad band emission assigned as STE, strong emission peak from 6 P of Gd 3+ and those sharp peaks in visible range from 5 D J - 7 F j emission of Eu 3+ [8]. In case of 1% Eu 3+ doped GdHfF 7, the 5 D J - 7 F J emission of Eu 3+ is strong as much as GdZrF 7 :Eu 3+ while the STE emission is decreased a lot in its emission intensity and blue shifted (Fig 7 b). This shift of STE is shown more clearly in the emission spectrum of undoped GdHfF 7 sample in Fig 7 a) where we can see one additional peak at 278nm designated as an emission of 6 I of Gd 3+ together with the very strong emission peak at 312nm of 6 P of Gd 3+. The 6 I state of Gd 3+ is probably populated by energy transfer from the STE which is blue shifted and becomes overlapped with 6 I absorptions. The two emissions, especially the 6 I emission from Gd 3+ ion, are almost disappeared when Eu 3+ is doped meaning the energy transfer from both 6 I and 6 P of Gd 3+ to Eu 3+ are very efficient in GdHfF 7. The emission spectrum of the intermediate compound Gd(Zr,Hf)F 7 between heptafluorozirconate and heptafluorohafnate, shows that the STE has the same characters of heptafluorozirconate with about half emission intensity while it does not show the emission peak at 278nm observed in pure Hafnate sample (Fig 7 d) Fig 8 shows the excitation spectrum in VUV region measured for the whole emission spectrum relative to sodium salicylate. We can compare the relative excitation efficiency compared to that of GdZrF 7 :1%Eu (curve c) which was already shown to be almost unity in quantum yield [8]. It is clear that the quantum yield is decreased as the amount of Hf 4+ is increased and the pure heptafluorohafnate (curve a) shows the lowest quantum yield. However the excitation spectrum of the pure heptafluorohafnate is shifted to high energy consistent with the idea that the STE of this is blue shifted in emission spectrum.

157 136 Relative excitation efficiency a) GdHfF 7 undoped b) GdHfF 7 : Eu 1% c) GdZrF 7 : Eu 1% d) Gd(Hf, Zr)F 7 : Eu 1% Wavelength (nm) Fig. 7.8 Excitation spectrum of GdHfF 7 samples compare to other analogous compound. a)undoped GdHfF 7, b) GdHfF 7 :1%Eu 3+, c) GdZrF 7 :1%Eu 3+, d). Gd(Zr,Hf)F 7 :1%Eu CONCLUSION The single crystal structure of gadolinium heptafluorohafnate, GdHfF 7 was determined by X-ray diffraction methods. The Gd 3+ squares and Hf 4+ squares interlace each other and form top and bottom layers in a unit cell along c-axis direction. The bottom layer has Hf 4+ square underneath Gd 3+ square of top layer and Gd 3+ square underneath the Hf 4+ square of top layer. These layers form slabs of [Gd 2 Zr 2 F 12 ] 2+ with other layers of next neighboring unit cells along c-axis and the slabs are connected by [F 2 ] 2- layers. The luminescence characters of GdHfF 7 are investigated when it is doped with Eu 3+. The 1%Eu 3+ doped GdHfF 7 shows a broad weak emission band in blue region and 5 D J - 7 F J emission of Eu 3+ under 160nm excitation. By comparing with the emission and excitation spectrum of undoped GdHfF 7, 1%Eu 3+ doped GdZrF 7 and 1%Eu 3+ doped Gd(Zr,Hf)F 7, we can conclude

158 that the STE emission of the GdHfF 7 is blue shifted than the GdZrF 7 and the energy transfer from Gd 3+ to Eu 3+ is very efficient. 137 Acknowledgements We appreciate the help of Hinke Ted, staff in chemistry department of OSU, for preparation of the vacuum sealed stainless tubes. REFERENCES [1] Korenev, Yu. M.; Antipov, P. I.; Novoselova, A. V.; Fedorov, P. P.; Sobolev, B. P., Zhurnal Neorganicheskoi Khimii (2000), 45(2), [2] Fedorov, P. P.; Val'kovskii, M. D.; Bondareva, O. S.; Sobolev, B. P. Zhurnal Neorganicheskoi Khimii, 38(10), (1993), [3] Michel Poulain, Marcel Poulain et Jacques Lucas, Material Research Bulletin, 7, (1972), [4] G. M. Sheldrick, Bruker/Siemens Area Detector Absorption Correction Program, Bruker AXS, Madison, WI, [5] SHELXTL-6.10 "Program for Structure Solution, Refinement and Presentation" BRUKER AXS Inc., 5465 East Cheryl Parkway, Madison, WI USA [6] Graudejus, O.;Schroetter, F.;Mueller, B.G.;Hoppe, R., eitschrift fuer Anorganische und Allgemeine Chemie, 620, (1994), [7] Michel Poulain, Bruce C. Tofield, Journal of Solid State Chemistry, 39 (1981) [8] Y. Zhou, J.Y. Jeong, D.A. Keszler, S.P. Feofilov and R.S. Meltzer, 16th International Conference on Dynamical Processes in Excited States of Solids, DCP 07.

159 138 CHAPTER 8 THE NEW EFFICIENT UPCONVERSION GREEN PHOSPHOR GdZrF 7 :Yb 3+,Er 3+ Joayoung Jeong and Douglas A. Keszler Oregon State Univeisity, Department of Chemistry Corvallis, OR

160 139 ABSTRACT The studies of Yb 3+ and Er 3+ doped GdZrF 7 shows that its emission output is almost twice as high as that of Yb 3+ and Er 3+ doped Gd 2 O 2 S upconversion phosphor when it was pumped with near infrared laser source (980nm). It also has better color purity because of the high green to red emission ratio and the shift of the red emission band to higher energy side. 8.1 INTRODUCTION There are growing interests in converting near infrared light into shorter visible emission as the solid state laser technology is developed and the new application area like bio imaging appears. The up converting concept was introduced at 1966 by Auzel in Yb 3+ -Er 3+ doped glasses for laser. Ovsyankin and Feofilov observed the upconversion process via cooperative sensitization in Yb 3+ -Tm 3+ system [1]. To date several compounds such as LaF 3 [2], YF 3 [3], BaYF 3, NaYF 4 [4], KY 3 F 10 [5], YOCl, Y 2 O 2 S [6], La 2 O 2 S and Gd 2 O 2 S [7] are reported as efficient hosts for upconversion of infrared light. Among them the hexagonal structure NaYF 4 host is known to show the highest luminescence efficiency. The oxysulfide host also has attraction for the bio application purpose because of its lack of toxicity and stability. The small phonon energy of the host compound is promising for high upconversion efficiency by suppressing the non-radiative energy loss. In this aspect the GdZrF 7 compound could be better than NaYF 4 having the phonon energy of cm -1 [8]. The GdZrF 7 compound has heavier elements of Gd and Zr compared to Y and Na ions in NaYF 4 that probably induce lower phonon energy. Fluoride compounds are difficult to synthesize as a pure phase and usually some specialized tube furnace is required with the toxic HF gas. All previous fluorides were synthesized through this difficult process to get pure phase. In this paper GdZrF 7 compound which is pure enough to act as a nice upconversion host was prepared by more simple method explained in experimental part.

161 EXPERIMENT sample preparation For the polycrystalline samples the raw chemicals of 1 mole GdF 3 (Alfa Aesar, 99.99%), 1.12 mole ZrF 4 (Alfa Aesar, 99.9%), YbF 3 (Alfa Aesar, 99.99%), ErF 3 (Aldrich, 99.99%) were mixed with 2 mole of NH 4 F (Aldrich, 99.99%) in mortar uniformly. The mixture was charged and packed into a carbon crucible with a carbon lid. This carbon crucible was capped by an alumina crucible and was positioned inside a bigger alumina crucible covered with an alumina lid. The space between the two crucibles was filled with carbon powders. The carbon powder was used to help prevent the mixtures from becoming oxidized. This crucible set was fired at 760 C for 1.5h and ground with additional small amount of ZrF 4 and 2mol of NH 4 F. The ground powder was heated again at 860 C for 1.5h in same atmosphere. Our preparation method for fluorides is similar to normal solid state reaction process except that we use carbon crucible and carbon powders X-ray crystallography All polycrystalline samples were structurally charactericterized by powder x-ray diffraction. The reference x-ray pattern of GdZrF 7 was calculated with Findit software using the single crystal structure solution data of our previous research [9]. The diffraction data were collected with a Siemens D5000 Diffractormeter using an inhouse program and λ=1.572 of Cu Kα radiation luminescence measurement The visible emission spectra were obtained by exciting the sample with a 100mW 980nm near infrared laser diode (Word Star tech, UH5-100G-980). The visible emission from sample was dispersed with an Acton Spectrapro-150 spectrometer and was detected with a PMT tube (Hamamatsu, R636-10) and the

162 141 electric signal is delivered to computer. The emission spectra were not corrected for the wavelength-dependent response of the detection system. All spectra were measured at room temperature. The emission output is measured by the same PMT tube without dispersing the emission output into monochromator under the fixed intensity of near infrared laser. 8.3 STRUCTURAL CHARACTERISTICS All polycrystalline samples were characterized by XRD confirming that the crystal system of Yb 3+ and Er 3+ co-doped GdZrF 7, even at very high Yb 3+ concentration, is monoclinic with P2(1)/m space group [8]. There is one Gd 3+ site with point group symmetry C 2v which can be occupied by the Yb 3+ or Er 3+ ions. The powder x-ray results at various concentration of Yb 3+ from low concentration to almost pure YbZrF 7 are shown in Fig. 1. As is clear from the inset picture of Fig. 1 showing 2θ =21 to 2θ =24, as the concentration of Yb 3+ is increased, the peaks are shifted to high 2θ values which can be expected when the smaller size cation of Yb 3+ replaces the bigger Gd 3+ ion. The other note is that YbZrF 7 sample shows the same crystal structure with GdZrF 7. The difference in the peak intensity ratio compared with the reference pattern (Fig. 2), which is calculated from our single crystal structure solution, results from the preferred orientation of our polycrystalline samples into 002 planes. Fig. 3 shows the cell parameter variation with the change in Yb 3+ concentration from the XRD data. As the Yb 3+ concentration is increased, the cell parameter and cell volume decrease linearly as expected from the refraction peaks shift at XRD data. The relatively faster cell shrinkage along the a and c axis are accompanied by the increase of the β angle.

163 Yb18 Yb20 Yb22 Yb24 Yb19 Yb21 Yb23 Xray results at vaious [Yb] intensity [Yb] 60 Fig. 8.1 Powder XRD data of Gd 0.98-x ZrF 7 :Yb x Er 0.02 samples. The bottom peaks is for x=0.18 and top one is for x=0.98. The x value is increased from x=0.18(bottom one) to 0.22, 0.26, 0.30, 0.34, 0.50 and 0.98(top one). Inset is the magnified one for Fig. 1 in the 2 θ range of degree. Intensity GdZrF7-calculated θ Fig. 8.2 Reference XRD pattern of GdZrF 7 compound calculated from the single crystal structure solution data.

164 Cell parameter change as [Yb] y = x Distance(Å) (a) a b c 6.5 y = x y = x [Yb], mol% Cell volume and beta agnle change y = x Volume(Å) (b) β V beta angle y = x [Yb], mol % Fig 8.3 Cell parameter change according to the increase of Yb 3+ concentration, (a) cell parameter, (b) cell volume and β angle.

165 144 The phonon energy of GdZrF 7, which will be important to realize high upconversion efficiency, can be assumed smaller than that of NaYF 4 considering the bond distances and bond angles. As the bond distance increases the phonon energy of the bond will be decreased. The bond distances between rare earth ion and F ion are distributed in a wider range from Å to Å in the case of GdZrF 7 than the range from Å to Å in the case of NaYF 4. The average distances of those bonds in GdZrF 7 and NaYF 4 are Å and Å respectively, and the difference is rather small. The rare earth ion in both compounds are eight fold coordinated. However the bonding angle in those two compounds and the point symmetry of the rare earth ions are quite different. None of the bonding angles of F - - Gd 3+ -F - in GdZrF 7 approach 180 degrees while all the bonding angles of F - -Y 3+ -F - in NaYF 4 are near to 180 degrees which means the phonon energy of GdZrF 7 compound is mostly composed of bending mode whereas the higher ones in phonon energy of NaYF 4 depends mostly on the stretching vibration mode. The Gd 3+ ion locating in a distorted square anti-prismatic structure in GdZrF 7 is on the non-centro symmetric site while the Y 3+ site in NaYF 4 has inversion symmetry. It would be realistic to assume that the inversion symmetry at the Y 3+ site in NaYF 4 can be attributed to higher phonon energy through stretching vibration mode even the bond distance is similar to that of GdZrF 7. Fig. 4 is the Raman spectrum of GdZrF 7 measured at WITec alpha300r Raman spectroscopy with He/Ar laser 514nm. The dominant peaks appear in cm -1 which is similar to the known values, cm -1 in NaYF 4 [9]. The heavierness of atomic weight of Gd and Zr than that of Y and Na will be another factor determining the lower phonon energy of GdZrF 7.

166 145 Fig. 8.4 Raman spectrum of polycrystalline GdZrF 7. The excitation source is He-Ne green laser. The particle morphology of GdZrF 7 is shown in the SEM pictures of Fig. 5 at several magnifications. On the left side of Fig. 5(c) is the agglomeration reaction shown between the small particles resulting in a few tens um size particle which is bigger than the reference sample having particle size of 4-5um. The commercial product of green upconversion phosphor, PTIR545F obtained as a sample from Phosphor Technology company was used as a reference sample in this paper. The reference one is assumed as Gd 2 O 2 S:Yb 3+, Er 3+ from the XRD data shown later in this paper

167 146 (a) 100 (b) 300

168 147 (continued) (c) 1250 Fig. 8.5 SEM pictures of Gd 0.74 ZrF 7 :Yb 0.22, Er 0.04 sample at several magnifications, (a) 100, (b) 300 and (c) LUMINESCENCE CHARACTERISTICS The up conversion emission mechanism of Yb 3+ -Er 3+ system has been previously studied as a representative green emitting ion pair in many host materials. Fig. 6 is the energy level diagram explaining the upconversion mechanism in Yb 3+ - Er 3+ system [9].

169 148 4 G 11/ H 9/2 2 Energy, 10 3 cm F 7/2 4 S 3/2 4 F 9/2 4 I 9/ I 11/2 2 F 5/2 4 I 13/ I 15/2 2 F 7/2 Er Er 3+ Yb 3+ Yb Fig. 8.6 Upconversion mechanism for green emission under near infrared light excitation showing the energy transfer in Yb 3+ -Er 3+ system. The dotted curve explains the energy transfer from Yb 3+ to Er 3+ via consecutive two or three photon absorption by Er 3+, the downward dotted line is non-radiative transition, the straight thick downward lines show the radiative transitions.(adopted from J. F. Suyber, J. Grimm, M. K. Van Veen, D. Biner, K. W. Krämer, H. U. Güdel, Journal of Luminescence, 117, 1-12 (2006)) The green upconversion mechanism is explained by the two photon absorption process into 4 F 7/2 energy state of Er 3+ which is populated by excited state absorption at 4 I 11/2 level of the second photon. The fast non-radiative relaxation from the 4 F 7/2 state populates the 2 H 11/2 or 4 F 9/2 energy states in a short time followed by subsequent transition of green or red emission to the ground state of Er 3+. We investigated the visible emission output and emission spectrum at several different concentrations of sensitizer, Yb 3+ and activator, Er 3+ to ultimately find the optimal composition at this new host compound of GdZrF 7.

170 Optimal Er 3+ and Yb 3+ concentration Er 3+ concentration was varied from 1% to 4% by increment of 1% at each three concentration level of Yb 3+ (18 %, 22% and 26%). Fig. 7 (a), (b), (c) show the emission output change at various dopant concentrations mentioned above. 140% Emission output at 18% Yb 120% Relative intensity 100% 80% 60% 40% 20% 0% (a) commercial [Er]1s [Er]2s [Er]3s [Er]4s time(30m15s) % Emission potput at 22%Yb 140% Relative Intensity 120% 100% 80% 60% 40% 20% 0% (b) commercial [Er] 1% [Er] 2% [Er] 3% [Er] 4% time(30m20s) 1st 2nd 3rd 4th 5th 6th 7th 8th 9th 10th 11th 12th 13th

171 150 (continued) 250% Emission output at 26%Yb 200% Relative Intensity 150% 100% 50% 0% (c) commercial [Er] 1% [Er] 2% [Er] 3% [Er] 4% time(30m15s) Fig. 8.7 Relative emission output of the Gd 1-x-y ZrF 7 :Yb x Er y samples were measured during 30min compared with the reference one. The Er 3+ concentration was varied as 1%, 2%, 3% and to 4% at three different concentration of Yb 3+. The emission output data for 18% Yb 3+ are on (a), for 22% Yb 3+ on (b), for 26%Yb 3+ on (c). In all graphs the black line marked with black diamond represent the emission output of reference sample. The line with brown triangle marker is for 2% Er 3+, the green cross marker is for 3% Er 3+, the violet square marker is for 1% Er 3+ and the black cross maker is for 4% Er 3+ in the downward sequence from the top one. The emission output was measured intermittently under 980nm infrared laser diode excitation with the reference sample to compare. After one cycle of measurement was finished for all samples, the same measurement process was repeated for all samples during 30min. As it is clear the emission output of GdZrF 7 at every Er 3+ concentration and at every Yb 3+ concentration are maintained above their initial emission output while the reference one shows a slight decrease as time passes. At all three concentration of Yb 3+, the 2% Er 3+ condition shows the highest result which is represented as brown triangle markered lines in Fig. 7 (a), (b) and (c). The highest emission output is about two orders higher than that of reference one at the condition of 2% Er 3+ and 26% Yb 3+ concentration.

172 151 Fig. 8 collects all the emission output data of Fig.7 (a), (b), and (c) into one picture. It is more clear to figure out at 2% Er 3+ concentration we can get the highest emission output and at 2% Er 3+ concentration the emission output of GdZrF 7 materials keeps increasing as Yb 3+ concentration is increased in this experimental range. We increased the Yb 3+ concentration further to higher concentration to see the saturation point. 250% Emission output vs. [Yb] and [Er] 200% Relative light output 150% 100% 50% [Er]=1 [Er]=2 [Er]=3 [Er]=4 [Yb], [Er], mol% 0% Fig. 8.8 Emission output results of GdZrF 7 samples at three concentration levels of Yb 3+ and four concentration levels of Er 3+ collected from the experiment above. The data dispersion on each sample is caused by the emission output increase as time pass by as we mentioned already. The first group of dots express the emission output of reference sample, the next three groups of dots represent the emission output of 1% Er samples, the next three for 2% Er samples, the next three for the 3% Er samples and the last three for the 4% Er samples. At each Er concentration, the first group of dots represent 18% Yb 3+, the second one 22% Yb 3+, and the third one 26% Yb 3+ condition. The further increase in Yb 3+ concentration was carried at 30%, 34%, 50% and 98% Yb 3+ concentration. The 98% Yb 3+ sample means the pure YbZrF 7 having 2% Er 3+. In Fig. 9 the emission output results at all Yb 3+ concentration range including the former data set are shown. It maybe realistic assuming that the saturation concentration

173 152 inyb 3+ be around 34% even we do not have data between 34% and 50% Yb 3+ concentration. 250% Emission output vs. [Yb] 200% Relative light output 150% 100% 50% [Yb],mol% 0% Fig. 8.9 Emission output of Gd 0.98-x ZrF 7 :Yb 3+ x Er samples at further increased concentration of Yb 3+ up to 98% are measured intermittently. The emission output was measured during 50min intermittently and is represented as dots. x-abscise is the Yb 3+ concentration, y-abscise is the relative emission output to that of reference one Color purity change vs. Yb 3+ concentration The color purity is another important characters as to be a quality phosphor and this can be interpreted from the emission spectrum. Fig. 10 (a) to (g) shows the emission spectrum for the same GdZrF 7 sample set in Fig. 9. We calculated the green to red emission ratio (G/R ratio, the ratio between the maximum peak intensity in green emission band and red emission band) to figure out the color purity change by the variation in Yb 3+ concentration quantitatively. We assumed that the maximum peak intensity ratio can represent the emission ratio. The calculated G/R ratio is graphed into Fig. 11. The emission spectrum show increase in their emission output until the Yb 3+ concentration reaches up to 34%. At 50% Yb 3+ the green emission

174 153 decreased drastically to the similar level of 18% Yb 3+ sample which has the same concentration difference from the optimal Yb 3+ condition (34%) as 16%. However the red emission at 50% Yb 3+ is much stronger. This is related to the energy back transfer from Er 3+ to Yb 3+ with populating the 4 F 3/2 level and resulting in red emission. This will be explained later in detail. 3.50E-08 emission spectra at various [Yb] 3.50E-08 emission spectra at various [Yb] Relative Intensity E E E E-08 (a) [Yb] 18 commercial Relative Intensity E E E E-08 (b) [Yb] 22 commercial 1.00E E E E E (Å) E (Å) E-08 emission spectra at various [Yb] 3.50E-08 emission spectra at various [Yb] Relative Intensity E E E E-08 (c) [Yb] 26 commercial Relative Intensity E E E E-08 (d) [Yb] 30 commercial 1.00E E E E E (Å) E (Å) 6900

175 154 (continued) 3.50E-08 emission spectra at various [Yb] 3.50E-08 emission spectra at various [Yb] Relative Intensity E E E E-08 (e) [Yb] 34 commercial Relative Intensity E E E E-08 (f) [Yb] 50 commercial 1.00E E E E E (Å) E (Å) E-08 emission spectra at various [Yb] Relative Intensity E E E E-08 (g) [Yb] 98 commercial 1.00E E E (Å) 6900 Fig Emission spectrum at various Yb 3+ concentrations. (a) [Yb] =18%, (b) [Yb] =22%, (c) [Yb] =26%, (d) [Yb] =30%, (e) [Yb] =34%, (f) [Yb] =50% and (g) [Yb] =98%. Every emission spectrum are compared with reference one which is shown by the blue solid line. The G/R ratio is continuously decreased in the order of third power of Yb 3+ concentration as shown by the fitting equation on the trend line in Fig. 11. For the optimal condition in Yb 3+ and Er 3+ concentration the emission output data of Fig. 9 and the color purity data of Fig. 10 are considered together and the optimal condition for GdZrF 7 :Yb 3+, Er3 + is expected to be near 2% Er 3+ and 34% Yb 3+ concentration. At

176 155 34% Yb 3+ even the G/R ratio is not the highest, it is still high enough to emit purer green color compared to the reference (G/R ratio of reference sample is 1.34) as is shown in the emission spectrum of Fig. 10 (e) G/R ratio vs Yb concent. G/R Poly. (G/R) G to R ratio y = -5E-06x x x R 2 = [Yb] Fig G/R ratio at various Yb 3+ concentrations. The best result in this paper shows more than two orders of emission output and higher color purity than the reference one. The reference is assumed to be Gd 2 O 2 S doped with Yb 3+ and Er 3+ by the XRD data. Fig. 12 is the XRD data of reference sample with the x-ray pattern of Gd 2 O 2 S in ICDS file.

177 156 Intensity Reference θ Intenisty Gd2O2S-trigonal θ Fig XRD data of (a) reference up conversion green phosphor and (b) of Gd 2 O 2 S from ICDS file. The G/R ratio is decreased as Yb 3+ concentration is increased. In the range of 18% Yb 3+ to 34% Yb 3+ not only the red emission output keeps increasing but also the portion of red emission in the whole emission flux is increased. However at very high Yb 3+ concentration like 50% the red emission output is decreased even the portion of red emission in the whole emission output is increased There is another feature in the emission spectrum of GdZrF 7 compound. The emission spectrum is shifted to high energy side compared to the reference one which would contribute to higher color purity. The shift of emission band position to lower energy side in the reference sample shows the nephelauxetic effect of oxysulfide compound relative to fluoride compound.

178 Effect of Yb 3+ concentration on red emission output We investigate the reason for the increase in the portion of red emission output as the concentration of Yb 3+ is increased. There are two possible ways to populate the 4 F 9/2 energy level of Er 3+ from which red emission can be generated. The first one is two photon absorption process marked with (1) and the second one is the three photon absorption process marked with (2) in Fig. 6. The first process can generate both green and red emission by transition from 2 H 11/2 or 4 S 3/2 and 4 F 9/2 levels respectively to the ground state after non-radiative relaxation from 4 F 7/2 level. In the second process, the energy back transfer to Yb 3+ of the transition energy from 2 H 9/2 to 4 F 9/2 in Er 3+ can populate the 4 F 9/2 energy state of Er 3+. We measured the emission spectrum of two chosen samples, one of which is 22% Yb 3+ and the other one is 50% Yb 3+, under 379nm excitation (this can excite Er 3+ into 4 G 11/2 energy level) and 490nm excitation (this can excite Er 3+ into 4 F 7/2 energy level) and compared the green and red emission intensity in the emission spectrum. The reference one is also investigated under the same excitation sources. 2.50E E-08 (a) 2.09E-08 Intensity E E E E E E E wavelength(a)

179 158 (continued) 2.50E-08 (b) 2.00E E-08 Intensity E E E E E E E wavelength(a) 4.00E E-07 (c) E E-07 Intnesity E E E E E E E E E wavelength(a) Fig Emission spectrum excited by 379nm and 490nm (a) of 22%Yb 3+ sample, (b) 50% Yb 3+ sample and (c) reference one. In all pictures the violet solid lines represent the emission spectrum under 490nm excitation and the blue solid lines are that under the 379nm excitation. When both 22% and 50% Yb 3+ samples are excited with 490nm, the G/R ratio in the emission spectrum is similar (8.2 for 22% Yb 3+ and 7.47 for 50% Yb 3+ sample). This means that second photon absorption process is not the main reason for the increase in

180 159 red emission at 50% Yb 3+ sample. Meanwhile the G/R ratio of both samples under the 379nm excitation is quite different (0.86 for 22% Yb 3+ and 0.28 for 50% Yb 3+ ) and the 50% sample shows stronger red emission than 22% Yb 3+ sample. This stronger red emission at high concentration of Yb 3+ under 379nm excitation could mean the three photon absorption process is participating for that emission increase. As the concentration of Yb 3+ is increased the portion of three photon absorption process relative to the two photon absorption process in the excitation process becomes more important and at high Yb 3+ concentration like 50% the three photon absorption process becomes mainly responsible to the emission. The emission spectrum of reference one (Fig. 13 (c)) shows that it relies more seriously on the three photon absorption process for its emission. Its lower G/R ratio under 490nm excitation will be the results of the high phonon energy and the high non-radiative relaxation rate between 4 G 11/2 and 2 H 9/2 levels of oxysulfide compound Luminescence dependency on the excitation intensity In the previous research, the green upconversion emission of Yb 3+ -Er 3+ system was explained by the two photon excitation process from luminescence intensity dependency result on excitation laser intensity. The relationship is expressed with the equation of I emit α (I exci. ) n, n 2 However in the GdZrF 7 compound it probably is not correct to say the emission is attributed to two photon excitation process. According to the emission spectrum data of our experiment under the different excitation wavelength of 379nm and 490nm that can differentiate the two photon absorption excitation and three photon absorption excitation process exclusively, more than two photon absorptions are expected to participate in the excitation process of GdZrF 7 compound. If the green emission in GdZrF 7 compound under 980nm excitation is determined only by the two photon absorption process in excitation then the G/R ratio in emission spectrum should be similar to the value when it is excited by 490nm light. This value in GdZrF 7 compound will be high because of the small phonon energy compared to the energy

181 160 gap between 4 G 11/2 and 2 H 9/2 levels and actual result in this experiment is high. However the G/R ratio data under 980nm excitation in the whole experimental concentration range of Yb 3+ show lower value than the expected ones from 490nm excitation.(table 1) Table 8.1 G/R ratio measured from the emission spectrum of each Yb 3+ concentration. Data at two different excitation wavelengths of 490nm and 980nm are shown for two samples of 22%Yb 3+ and 50% Yb 3 excitation [Yb]18% [Yb]22% [Yb]26% [Yb]30% [Yb]34% [Yb]50% [Yb]98% 490nm nm As is clear here, the red emission is much stronger under 980nm excitation than the expected red emission from the transition branching ratio of 4 F 7/2 energy level into which Er 3+ will be excited by the two photon absorption process (the same excited state by 490nm excitation). 8.5 CONCLUSION The optimal concentration of Yb 3+ and Er 3+ was investigated in GdZrF 7 compound as a new upconversion green phosphor. The best emission output is more than two orders higher than that of commercial Gd 2-x-y O 2 S:Yb 3+ x, Er 3+ y sample. It also has high G/R ratio imparting better color purity. The increase in red emission as Yb 3+ concentration is increased can be interpreted with the increased three photon absorption process in excitation process which results in population of the 4 F 9/2 level responsible for red emission. The XRD data shows that unit cell shrinks as the Yb 3+ concentration is increased at faster rate along a and c axes than along the b axis and a continuous increase of β angle is observed.

182 161 Aknowledgement I appreciate Phosphor Technology in England for the Gd 2 O 2 S based commercial upconversion phosphor samples. REFERENCES [1] V. V. Ovsyankin, and P. P. Feofilov, Soviet Phys.-JETP Letters, 4, , (1966) [2] J. L. Sommerdijk, Journal of Luminescence, 4, , (1971) [3] N. M. P. Low, A. L. Major, Journal of Luminescence, 4, , (1971) [4] N. Menyuk, K. Dwight, and J. W. Pierce, Applied Physics Letters, 21(4), , (1972) [5] Alexandra Lapaport, Janet Milliez, Journal of Display Technology, 2(1), 68-78, (2006) [6] Xi-xian Luo, Wang-he Cao, Materials Letters, 61, , (2007) [7] P. N. Yacom, J. P. Wittke, and I. Ladany, Metallugical Transations, 2, , (1971) [8] J. F. Suyber, J. Grimm, M. K. Van Veen, D. Biner, K. W. Krämer, H. U. Güdel, Journal of Luminescence, 117, 1-12 (2006) [9] Joayoung Jeong L. N. Zakharov Y. Zhou, R. S. Meltzer and D. A. Keszler, Crystal structure and Eu 3+ luminescence of GdMF 7 (M=Hf,Zr) in preparation

183 162 CHAPTER 9 CRYSTAL STRUCTURE AND LUMINESCENT PROPERTIES OF THE APATITE Gd 4.67 (SiO 4 ) 3 S Joayoung Jeong, L.N. Zakharov and Douglas A. Keszler Oregon State University, Department of Chemistry Corvallis, OR To be submitted to Solid State Sciences

184 163 ABSTRACT The defect apatite material Gd 4.67 (SiO 4 ) 3 S has been prepared by solid state reaction, and its crystal determined by single crystal X-ray diffraction methods. The compound crystallizes in space group of P6 3 m. One of the Gd atoms, Gd(2), located at the center of a tri-capped trigonal prism with nine oxygen atoms. The other Gd atom, Gd(1), occupies a distorted seven coordinate environment of five of oxygen and two sulfur atoms. The distorted tetrahedral SiO 4 groups are connected to the Gd atoms by sharing vertices. The sulfur is coordinated by six Gd(1) atoms in a site having S 6 symmetry. The luminescent properties of the Tb 3+ -doped compound have been studied and compared with those of the oxide analog Gd 4.67 (SiO 4 ) 3 O:Tb INTRODUCTION Apatite materials typified by compositions such as Ca 3 -(PO 4 ) 3 Cl and Sr 5 - (PO 4 ) 3 F, have been widely adopted for luminescent and laser materials. As phosphor materials, the alkaline earth halophosphates such as Ca 5 (PO 4 ) 3 Cl doped with antimony and manganese[1] have been used in fluorescent lamps. Apitites are also known as chemically diverse materials; the Ca 2+ in Ca 5 (PO 4 ) 3 F can be replaced by Sr 2+, Ba 2+, Y 3+, and lanthanides, and PO 3-4 can be replaced by SiO 4-4 and VO 3-4. The halide anion can also be replaced with OH 1-, O 2-, or S 2-. The structure of apatite derivative 7Gd 2 O 3 9SiO 2 was reported in 1969[2], and the luminescence characteristics of Prand Ce- doped samples were described by A. J. de Vries and co-workers[3]. They examined emission from the lowest 5d energy levels of Pr 3+ and Ce 3+. As solid-state laser materials, A. M. Anderson and J. C. Wang reported the efficient Yb- and Ndbased lasers in the phosphate apatites X 5 (PO 4 ) 3 Z (X = Ca,Sr,Ba; Z = F,Cl)[4]. In this work we reported the crystal growth and structure of the apatite Gd 4.67 (SiO 4 ) 3 S. The powder form of this compound is also prepared for characterization Tb 3+ luminescence.

185 EXPERIMENTAL Sample Preparation For crystal growth the chemicals Gd 2 O 3 (Stanford Materials Corp., %), SiO 2 (CERAC, 99.5%) were mixed in the molar ratio 2.335:3 with 10 wt% LiF (CERAC, 99.9%) and 100 wt% precipitated S (Fisher Scientific). The mixture was ground and placed in a covered carbon crucible. This crucible was placed in a covered alumina crucible with the space between the two crucibles filled with mixture of powdered sulfur and carbon. This crucible set was fired at 1200 C for 3h and then cooled to1000 C at 5 C/min and to 850 C at 10 C/min; the power to the furnace was then turned off for rapid cooling to room temperature. Small crystals having needle shapes were extracted and analyzed via X-ray diffraction in detail, vide infra. Powder samples were synthesized according to the formula derived from the structural analysis on the needle-shaped single crystal. A mixture of 2.33 Gd 2 O 3, and 3 SiO 2 with 5 wt% LiF was heated at 1150 C for 3h, the resulting mixture was then heated under flowing H 2 S (g) at 1200 C for 2h. Powder samples were also synthesized by grinding stoichiometric mixtures of Gd 2 O 2 S, Gd 2 O 3, and S in vacuum, sealing said mixtures in evacuated silica tubes, and heating at 1150 C for 2h. The Gd 2 O 2 S was presynthesized by heating a mixture of Gd 2 O 3, Na 2 CO 3, and S in an alumina crucible at 1250 C for 3h. The Gd 2 O 2 S was extracted by dissolving the flux with water X-ray diffraction analysis X-Ray diffraction data were collected on a Bruker Smart Apex diffractometer at 173(2) K by using Mo Kα radiation. The structure was solved with direct methods and refined with full-matrix least-squares methods based on F 2 by using the SHELXTL (v. 6.10) package [6]. Crystal data and details of the experiment are given in the Table 1. Absorption corrections were made by using the computer program SADABS [5].

186 Table 9.1 Crystal data and details of X-ray diffraction experiment for Gd 4.67 (SiO 4 ) 3 S Empirical formula Gd 4.67 (SiO 4 ) 3 S Formula weight u Temperature 173(2) K Wavelength Å Crystal system Hexagonal Space group P6 3 /m Unit cell dimensions a = Å b = Å c = Å Volume (9) Å 3 Z 2 Density (calculated) g/cm 3 Absorption coefficient mm -1 F(000) 453 Theta range for data collection 2.42 to Index ranges -12 h 11, -8 k 12, -8 l 8 Reflections collected 3436 Independent reflections 482 [R int = ] Completeness to θ = % Absorption correction Semi-empirical from equivalents Max. and min. transmission and Refinement method Full-matrix least-squares on F 2 Data / restraints / parameters 482 / 0 / 40 Goodness-of-fit on F Final R indices [I > 2σ(I)] R1 a = , wr2 b = R indices (all data) R1 = , wr2 = Extinction coefficient (4) Largest diff. peak and hole and e/å 3 a R1= Σ ( F 0 - F c ) / Σ F 0 b wr2 = { Σ [w(f 2 0 -F 2 c )] 2 2 / Σ wf 0 } 1/2 165

187 Luminescence measurements To record the emission spectra, a Xenon lamp (Oriel 300W) was used in conjunction with a Carry model-15 prism monochromator to select a suitable excitation wavelength. The emission from the sample was passed through a filter to eliminate second-order effects and then dispersed by a grating monochromator (Oriel 22500) before detection with a PMT tube (Hamamatsu, R636-10). The measured emission data were corrected by using a calibrated tungsten lamp (Eppley Laboratories, Inc.) Excitation spectra were recorded on the same system by scanning the excitation monochromator at a fixed emission wavelength. Depending on the wavelength range, the spectra were corrected with sodium salicylate or Rhodamin-B. Low temperature excitation spectra were measured by using a flow cryostat and liquid helium. Powder samples were spread directly on a copper plate in the cryostat. 9.3 RESULTS AND DISCUSSION Final positional parameters for Gd 4.67 (SiO 4 ) 3 S are listed in Table 2. The compound is similar to the apatite derivatives Ln 4.67 (SiO 4 ) 3 O (Ln = La [7] and Gd [2]) that crystallize in space group P6 3 /m. The position of the S atom in Gd 4.67 (SiO 4 ) 3 S, however, differs from that of the O atom in Ln 4.67 (SiO 4 ) 3 O. A view of the unit-cell contents of Gd 4.67 (SiO 4 ) 3 S is given in Fig. 1. The Gd(1) and Gd(2) atoms occupy positions on the mirror plane (Wykoff 6h position) and on the three-fold axis (Wykoff 4f position), respectively. Refinement of the occupation factors for these positions reveals that the position 6h is fully occupied, while the position 4f is partially occupied at 83.2%.

188 Table 9.2 Atomic positions and equivalent isotropic displacement parameters (Å2 x 103) for Gd 4.67 (SiO 4 ) 3 S. x y z U eq Occupancy Gd(1) (1) (1) 1/4 6(1) Gd(2) 1/3 2/ (1) 12(1) 0.832(7) S (1) Si (3) (3) 1/4 6(1) O(1) (7) (7) 1/4 9(1) O(2) (5) (6) (8) 17(1) O(3) (10) (8) 3/4 22(2) 167 The Si atom has a distorted tetrahedral coordination by O atoms. The Si atom and two O atoms (O(1) and O(3)) in the tetrahedron are on mirror planes and the two other O atoms are in general positions. Thus the structure of Gd 4.67 (SiO 4 ) 3 S has a framework similar to that in Ln 4.67 (SiO 4 ) 3 O; the Gd(1) and Gd(2) atoms form parallel to the c axis interconnected via SiO 4 bridges (Figure 1). However positions of the free O atom in Ln 4.67 (SiO 4 ) 3 O and the S atom are different. The free O atom in Ln 4.67 (SiO 4 ) 3 O is at a symmetrical µ 3 -bridge between three Gd(1) atoms, occupying the geometrical center of the Gd(1) triangle (Figure 2a). The distance Gd-O is Å [2]. The S atom in Gd 4.67 (SiO 4 ) 3 S is not inside the above-mentioned Gd triangle, but between them (Figure 2b). Thus six Gd(1) atoms form a distorted octahedron around the S atom with the Gd-S distance of (4) Å. The S atom locates outside of the Gd triangle because of the larger radius of S 2- (Shannon radii, 1.84 Å) in comparison to that of O 2- (Shannon radii, 1.40 Å). The calculated distance of Gd(1)-center of Gd triangle in Gd 4.67 (SiO 4 ) 3 S is about 2.4 Å which is somewhat longer than that of Gd(1)- O in Ln 4.67 (SiO 4 ) 3 O. However this value is much smaller than the sum ~2.98 Å of radii for S and Gd. The middle value of between 6 coordinated and 8 coordinated atomic radii was chosen as atomic radius of Gd 3+.

189 Fig. 9.1 Unit-cell drawing of Gd 4.67 (SiO 4 ) 3 S 168

190 169 (a) (b) Fig. 9.2 (a) Environment of the free O atom in Ln 4.67 (SiO 4 ) 3 O apatite (b) Environment of S atom in Gd 4.67 (SiO 4 ) 3 S apatite

191 170 Fig. 9.3 Tricapped distorted trigonal prismatic environment of Gd(2) and sevencoordinate site of Gd(1). Fig. 3 shows the Gd(2) atom located in a tricapped distorted trigonal prism with coordination by nine oxygen atoms. Three O(1) atoms form one trigonal face, and three O(3) atoms form the other face. These faces are rotated one relative to the other slightly. The three O(2) are positioned in capping sites beyond the three rectangular faces of the prism. The distances Gd(2)-O(1) and Gd(2)-O(3) at 2.432(4) and 2.339(5), respectively, are normal while the Gd(2)-O(2) distance is long, cf., Table 3. Relevant bond angles are listed in Table 4.. Gd(1) has seven coordination with two types of oxygen atoms (O(1),four of O(2)) and two sulfur atoms and files up along c direction as Gd(2) atom does.(fig 1) The distances between Gd(1) and each of the coordinating atoms are listed in Table 3. The Gd(1)-O(2) distance is normal, and the Gd(1)-S distance is comparable to the sum of crystal radii ( 2.8 Å). As mentioned earlier the sulfur atom is located between the two trigonal Gd(1) planes in a site with S 6 symmetry. The sulfur atoms are arranged in columns extending along the c axis, cf., Fig. 4.

192 171 Fig. 9.4 Sulfur column along c axis. Table 9.3 Bond lengths [Å]. Gd(1)-O(1) 2.306(6) Gd(2)-O(1) 2.432(4) Gd(1)-O(2)# (5) Gd(2)-O(1)# (4) Gd(1)-O(2) 2.311(5) Gd(2)-O(1)# (4) Gd(1)-O(2)# (5) Gd(2)-O(2)# (5) Gd(1)-O(2)# (5) Gd(2)-O(2)# (5) Gd(1)-S(1)# (4) Gd(2)-O(2)# (5) Gd(1)-S(1) (4) Si(1)-O(3)# (7) Gd(2)-O(3)# (5) Si(1)-O(2)# (5) Gd(2)-O(3) 2.339(5) Si(1)-O(2)# (5) Gd(2)-O(3)# (5) Si(1)-O(1) 1.637(6) Symmetry transformations used to generate equivalent atoms: #1 x,y,-z+1/2 #2 x-y+1,x,-z #3 x-y+1,x,z+1/2 #4 -x+2,-y+2,z+1/2 #5 -x+y,-x+1,z #6 -y+1,x-y+1,z #7 -x+1,-y+2,-z #8 y-1,-x+y,-z #13 -x+1,-y+2,z+1/2

193 Table 9.4 Selected Bond angles [ ] O(1)-Gd(1)-O(2)# (14) O(2)#2-Gd(1)-S(1) 70.77(12) O(1)-Gd(1)-O(2) 85.65(14) O(2)#3-Gd(1)-S(1) (13) O(2)#1-Gd(1)-O(2) 123.1(3) S(1)#4-Gd(1)-S(1) (12) O(1)-Gd(1)-O(2)# (17) Gd(1)#2-S(1)-Gd(1)#9 180 O(2)#1-Gd(1)-O(2)# (19) Gd(1)#2-S(1)-Gd(1) (9) O(2)-Gd(1)-O(2)# (8) Gd(1)#9-S(1)-Gd(1) (9) O(1)-Gd(1)-O(2)# (17) Gd(1)#2-S(1)-Gd(1)# (8) O(2)#1-Gd(1)-O(2)# (8) Gd(1)#9-S(1)-Gd(1)# (9) O(2)-Gd(1)-O(2)# (19) Gd(1)-S(1)-Gd(1)# (8) O(2)#2-Gd(1)-O(2)#3 62.7(2) Gd(1)#2-S(1)-Gd(1)# (9) O(1)-Gd(1)-S(1)# (7) Gd(1)#9-S(1)-Gd(1)# (9) O(2)#1-Gd(1)-S(1)# (14) Gd(1)-S(1)-Gd(1)# (8) O(2)-Gd(1)-S(1)# (12) Gd(1)#10-S(1)-Gd(1)# O(2)#2-Gd(1)-S(1)# (13) Gd(1)#2-S(1)-Gd(1)# (9) O(2)#3-Gd(1)-S(1)# (12) Gd(1)#9-S(1)-Gd(1)# (9) O(1)-Gd(1)-S(1) (7) Gd(1)-S(1)-Gd(1)# O(2)#1-Gd(1)-S(1) (12) Gd(1)#10-S(1)-Gd(1)# (8) O(2)-Gd(1)-S(1) 72.96(14) Gd(1)#11-S(1)-Gd(1)# (9) Symmetry transformations used to generate equivalent atoms: #1 x,y,-z+1/2 #2 x-y+1,x,-z #3 x-y+1,x,z+1/2 #4 -x+2,-y+2,z+1/2 #5 -x+y,-x+1,z #6 -y+1,x-y+1,z #7 -x+1,-y+2,-z #8 y-1,-x+y,-z #9 -x+y+1,-x+2,z #10 -y+2,x-y+1,z #11 y,-x+y+1,-z #12 -x+2,-y+2,-z #13 -x+1,-y+2,z+1/2 #14 x,y,-z-1/2 172 X-ray powder diffraction analysis was used to monitor the powder form of this compound obtained under selected synthesis conditions. The XRD pattern for the sample (Fig. 5(a)) prepared by heating at 1200 under H 2 S(g) agrees will the pattern calculated on the basis of the single-crystal data (Fig. 5(b)). So the powder sample obtained at such conditions is single phase. In contrast, the powder sample synthesized by mixing Gd 2 O 2 S, Gd 2 O 3, and S in a sealed tube at 1150 is not pure. The XRD pattern of the powder sample obtained using this way (Fig. 5(c)) has additional peaks

194 in comparison to the calculated pattern. Several weak peaks around 2θ = 30 are likely to derive from unreacted starting materials. 173 intensity (a) Gd4.67(SiO4)3S-powder θ (b) Gd4.67(SiO4)3S-reference Intensity θ intensity (c) anealing in sealed tube θ Fig. 9.5 XRD patterns for Gd 4.67 (SiO 4 ) 3 S (a) synthesis in flowing H 2 S(g) (c) prepared in sealed tube and (b) reference pattern calculated from single-crystal structure data.

195 174 We investigated the Gd 4.67 (SiO 4 ) 3 S apatite as a luminescent host by doping with Tb 3+ at concentrations of 2, 4, 7, and 10 mol%. In Fig 6, emission spectra as a function of Tb 3+ concentration are summarized. The emission intensity continually increases as the concentration of Tb 3+ increases. No concentration quenching up to 10 mol% is observed and probably more space is left for increasing the emission intensity by adding more Tb 3+. Corrected emission spectra of 7% and 10% Tb 3+ doped samples are shown at Fig. 7. The emission transitions are assigned to those from 5 D 4 7 F J as usual for Tb 3+ emission. 2.00E-08 relative intensity E E E E E E E E E-09 [Tb]2% [Tb]4% [Tb]7% [Tb]10% 0.00E wave length(å) Fig. 9.6 Emission spectra for selected concentrations of Tb 3+ in Gd 4.67 (SiO 4 ) 3 S.

196 175 relative intensity E E E E E E E E E E Å 5 D 3-7 F Å 5 D 3-7 F Å 5 D 3-7 F Å 5 D 3-7 F Å 5 D 3-7 F Å 5 D 3-7 F 3 a) [Tb]7% b) [Tb]10% 0.00E wavelength(å) Fig. 9.7 Emission spectra after correction for (a) 7% and (b) 10% Tb 3+ doped Gd 4.67 (SiO 4 ) 3 S. M. J. J. Lammers and co-workers reported Gd 3+ absorption peaks at 250, 280, and 315nm in the UV excitation spectrum of Gd 4.67 (SiO 4 ) 3 O:Tb 3+ apatite[8]. Our data exhibited a broad excitation band with no distinct Gd absorption peaks (Fig. 8). To understand this difference in excitation spectra, the low temperature excitation spectra for Gd 4.67 (SiO 4 ) 3 S:10%Tb 3+ and Gd 4.67 (SiO 4 ) 3 O:7%Tb 3+ were measured at liquid helium temperature; the results are given in Fig. 9. As note, Gd 4.67 (SiO 4 ) 3 O:Tb exhibits the sharp absorption peaks attributed to 6 D J, 6 I J, and 6 P J states of Gd while Gd 4.67 (SiO 4 ) 3 S:Tb 3+ has broad absorption bands near 320 nm without the sharp Gd absorption. As a result, no or little energy migration between Gd ions is expected in Gd 4.67 (SiO 4 ) 3 S. For an effective energy transfer involving Gd, the distance limit between donor and acceptor is less than ~ 4Å for an exchange process. The distances between Gd ions of each Gd site in Gd 4.67 (SiO 4 ) 3 O and Gd 4.67 (SiO 4 ) 3 S compounds are shown in Table 5. Considering the distance limit in energy transfer, we can assume that Tb ion should replace the Gd(2) at 4f site to enable the energy transfer among Gds or between Gd and Tb in Gd 4.67 (SiO 4 ) 3 O, while it should replace Gd(1) at 6h site in Gd 4.67 (SiO 4 ) 3 S. The long distance over 4Å between Gd ions in 6h site at Gd 4.67 (SiO 4 ) 3 S should limit efficient energy transfer between Gd ions or Gd and Tb

197 176 ions and result in no Gd absorption peaks in the excitation spectrum. If Tb 3+ ion occupies the Gd(2) site as we thought intuitively from the partial occupancy of Gd(2) site the excitation spectrum should have several Gd absorption peaks because Gd(2) site in Gd 4.67 (SiO 4 ) 3 S has the same coordination of O atoms with Gd 4.67 (SiO 4 ) 3 O. relative intensity [Tb]7% [Tb]10% to 5L9,5D wavelenth(å) Fig. 9.8 Excitation spectrum of 7% and 10% Tb 3+ -doped Gd 4.67 (SiO 4 ) 3 S (λ em = 544nm).

198 E E-09 GSiS:10%Tb GSiO:7%Tb relative intensity E E E E E E wavelength(å) Fig. 9.9 Excitation spectrum at liquid helium temperature of (a) 10% Tb 3+ doped Gd 4.67 (SiO 4 ) 3 S and (b) 7% Tb 3+ doped Gd 4.67 (SiO 4 ) 3 O. Table 9.5 The shortest distance between Gd ions in two different sites. Gd site 6h Between 6h-4f 4f Gd(1)- Gd(1) Gd(1)- Gd(2) Gd(2)- Gd(2) Gd 4.67 (SiO 4 ) 3 S 4.099* (3.383)** Gd 4.67 (SiO 4 ) 3 O 4.099* * The inter-plane Gd(1)-Gd(1) distance is listed here in consideration of energy transfer. ** Two different distances between Gd(2) exist in the sulfide. The 4f 2-4f5d absorptions of Tb 3+ are calculated at 209 nm (spin allowed) and 244 nm (spin forbidden); they are beyond the range of the excitation measurements. The absorption increase high energy area beyond 260 nm in excitation spectrum shows the onset of the 5d band. These 4f5d energy levels were calculated by using the equation suggested by P. Dorenbos to predict the lowest 5d level of lanthanides in

199 178 many hosts and the crystal depression data of Gd 4.67 (SiO 4 ) 3 O[9]. The absorption band arising from 320 nm in Gd 4.67 (SiO 4 ) 3 S is assumed to be the charge-transfer (CT) state of Tb-S. The excitation spectrum in Fig. 10 shows absorption peaks associated with 4f-4f transitions of Tb 3+ in the range from 330 to 500 nm. The features are assigned to excitations into the 5 L J levels from the ground state of Tb E-09 relative intensity E E E E-10 7 F 6-5 L 7 7 F 6-5 L 9 7 F 6-5 L 10 7 F 6-5 G E-10 7 F 6-5 D E wavelength(å) Fig Excitation spectra 4f-4f transitions of 10% Tb 3+ doped Gd 4.67 (SiO 4 ) 3 S. The emission spectra of 10% Tb 3+ doped sample measured under two excitation wavelengths at 313 and 370 nm are compared in Fig. 11. The emission intensity under both excitations are calibrated to compensate for the intensity difference of excitation light using the intensity ratio of Rhodamin-B between these two wavelengths. The emission under 370 nm excitation created by 4f-4f transition from 5 L 10 level to ground state is weak whereas the emission under 313 nm excitation showed much stronger intensity consistent with the proposed CT nature of the absorption.

200 179 relative intensity E E E E E E nmexcitation 313nmexcitation 4.00E E E wavelength(å) Fig Comparison of emission spectra of Gd 4.67 (SiO 4 ) 3 S:10% Tb 3+ under two different excitation wavelengths of 313nm and 370nm. The emission intensity was calibrated with the intensity ratio of those two excitation wavelength using Rhodamin- B. 9.4 CONCLUSION The framework of the apatite sulfide Gd 4.67 (SiO 4 ) 3 S has been established to be the same as that found in Gd 4.67 (SiO 4 ) 3 O. The position of S atom, however has been determined to be different from that of free O atom in Gd 4.67 (SiO 4 ) 3 O. The S resides in a distorted octahedral environment of Gd atoms, while the O atoms rest in the center of a Gd triangle. The Gd(2) atom occupies a distorted tricapped prism. The Gd(1) has seven coordination with five oxygen and two sulfur atoms. Both Gd(2) and Gd(1) files up along lines parallel to the c axis. The SiO 4 distorted tetrahedral units bridge the two types of Gd atoms. The possibility of Gd 4.67 (SiO 4 ) 3 S apatite as a luminescent host was investigated by doping with Tb 3+. The intensity of green luminescence increases with Tb 3+ concentration up to the maximum tested concentration of 10 mol%. Strongest absorption is associated with S Tb CT transitions at λ < 320.

201 180 Acknowledgements Acknowledge National Science Foundation. REFERENCES [1] Shigeo Shinoya,William M. Yen, Phosphor handbook, CRC press, Boca Raton, 394. [2] Smolin, Yu. I.; Shepelev, Yu. F. Izvestiya Akademii Nauk SSSR, Neorganicheskie Materialy 5(10), (1969). [3] De Vries, A. J., Blasse, G. Materials Research Bulletin 21(6), (1986). [4] A. M. Anderson, J. C. Wang, Advanced Solid State Lasers 26, (1999). [5] G. M. Sheldrick, Bruker/Siemens Area Detector Absorption Correction Program, Bruker AXS, Madison, WI, [6] SHELXTL-6.10 "Program for Structure Solution, Refinement and Presentation" BRUKER AXS Inc., 5465 East Cheryl Parkway, Madison, WI USA [7] Toropov, N. A., Kougiya, M. V., Doklady Akademii Nauk SSSR 182(3), (1968). [8] M. J. J. Lammers, G. Blasse, Journal of Electrochemcal Society, 134(8), (1987). [9] P. Dorenbos, Journal of Luminescence, 91, (2000).

202 181 CHAPTER 10 CONCLUSION AND FUTURE WORK Several systems have been examined in the development of new quantumsplitting phosphors. For the first time, efficient sensitization of the high-enery levels of Gd 3+ ion has been demonstrated by using Nd 3+. The strong 4f 3 4f 2 5d 1 highenergy absorption transition of Nd 3+ has been used to deliver energy into the densely packed 4f levels of Gd 3+ above the 6 G state in LiGdF 4 :Nd. This energy on Gd 3+ is back-transferred to the Nd 3+ ion, resulting in a unique quantum-splitting process with emission near 980 nm. Because of the rapid back transfer of energy from Gd 3+ to Nd 3+, it is not possible to transfer the energy efficiently from the Gd 3+ to another ion that emits in the visible. The investigation on the energy transfer processes from sensitizer to Gd 3+ and from Gd 3+ to Nd 3+ provided good information on the luminescence dynamics of quantum splitting process. In Pr 3+,Eu 3+ :GdF 3, Pr 3+ acts as a sensitizer to trigger the quantum splitting. Unfortunately, the effect is observed only at very low concentrations of Pr 3+ (0.3%) and Eu 3+ (0.2%), and a correspondingly low absolute quantum yield of 20% results. Quantum splitting via PCE on a single ion was also examined with the ions Tm 3+ and Er 3+. Their emission spectra are characterized by f-f transitions without emission from the 5d state; no quantum splitting was observed. However, in the case of Tm 3+, the non-radiative transition from the 5d level to the next lower 4f level ( 3 P J or 1 I 6 ) may be a possible first transition for CRET to a second ion. An experiment to convert the emission color in near UV transition 1 S 0 1 I 6 of Pr 3+ into visible range was unsuccessful for Sm 3+ co-doping, even though a resonance exists between the energy levels of the two ions. Host intrinsic emission via STEs was tested as a means to sensitize Gd 3+ into the 6 G level. Gd 3+ -doped oxides were examined, and factors affecting the STE emission were discussed. The fact that Gd 3+ was excited beyond 6 G level by the host intrinsic emission in ScPO 4 peaking at 215 nm provides a new possible excitation method for the quantum splitting phosphor. The ScPO 4 :1%Gd 3+ shows high absolute quantum

203 182 yield of 0.9 ± 0.2 following excitation at 170nm. The kind of cation in its atomic radii and ionization energy is shown to have very strong relationship with host intrinsic energy. Smaller cation radii or higher ionization energies lead to higher STE emission energy in phosphate compounds. Several new compounds are synthesized. Single crystals of the material GdZrF 7 were grown for structure refinement by using the flux method. The up-conversion process involving the ion pair Yb 3+ -Er 3+ was investigated in this host. Compositionally optimized samples in this system were demonstrated as potentially useful upconversion phosphors with almost 200% of the light output of the commercial material Gd 2 O 2 S;Yb,Er; relative to the commercial sample, the fluoride sample also exhibits a better green color purity. Continued development of this material with emphasis on the production of nanoparticles could extend use into the bioimaging field as a probe material. The luminescence of several lanthanides in this compound was also investigated; among them the Eu 3+ doped GdZrF 7 was found to be a nearly whiteemitting phosphor with an absolute quantum yield near 0.9 under VUV excitation. The crystal structure of the new apatite compound Gd 4.67 (SiO 4 ) 3 S was also solved from single-crystal X-ray diffraction data; luminescence from Tb 3+ doped samples revealed a bright-green emission.

204 183 BIBLIOGRAPHY A. Ellens, H. Andres, M. LT. Wegh, A. Meijerink, and G. Blasse, Physical Review B 55, 180 (1997). A. J. De Vries, M. F. Hazenkamp, G. Blasse, Journal of Luminescence, 42, (1988) A. J. De Vries, G. Blasse, Materials Research Bulletin 21(6), (1986). A. L. Harmer, A. Linz and D. R. Gabbe, Journal of Physics and Chemistry of solids, 30, 1483 (1969). A. M. Anderson, J. C. Wang, Advanced Solid State Lasers 26, (1999). A. M. Srivastava, W. W. Beers, Journal of Luminescence, 71, (1997). A. Mayolet, J. C. Krupa, J. SID, 4, 179 (1996). A. Meijerink, G. Blasse, M. Glassbeek, Journal of Physics: Condensed Matter, 2, 6303 (1990). A. P. Vink, P. Dorenbos, C. W. E. Van Eijk, Journal of Solid State Chemistry, 171, (2003). A. Trukhin and L.A. Boatner, Materials Science Forum, , 573 (1997). Alexandra Lapaport, Janet Milliez, Journal of Display Technology, 2(1), (2006) C. Gorller-Walrand, L. Fluyt, P. Porcher, A.A.S. Da Gama, G. F. de Sa, W. T. Carnall, G. L. Goodman, Journal of the Less Common Metals 148, 339 (1989). C. Ronda, Journal of Luminescence 100, 301 (2002). C. Feldmann, T.Justel, C. R. Rondo, D. U. Wiechert, Journal of Luminescence, 92, (2001). D. L. Dexter, Journal of Chemical Physics, 21, 836, (1953). D. L. Dexter, Physical Review, 108, 630 (1957). E. Nakazawa and F. Shiga, Journal of Luminescence, 15, 255 (1977)

205 184 E. Nakazawa, Journal of Luminescence, 100, 89 (2002). E. van der Kolk, P.Dorenbos, C.W.E. van Eijk, Optics Communications. 197, (2001) E. van der Kolk, P. Dorenbos, A. P. Vink, R. C. Perego, C. W. E. Van Eijk, Physical Review B 64, (2001) Energy level structure and transition probabilities of the trivalent lanthanides in LaF3, W.T.Carnel, Hannah Crosswhite, H.M.Crosswhite. G. M. Sheldrick, Bruker/Siemens Area Detector Absorption Correction Program, Bruker AXS, Madison, WI, G. Blasse, B. C. Grabmaier, Springer,Verlag Berlin Heidelberg, Luminescent Materials, (1994). H. Kondo, T. Hirai, S. Hashimoto, Journal of Luminescence, 108, 59 (2004). I. N. Orgorodnikov, V. A. Pustovarov. A. V. Kruchalov, L. I. Isaenko, M. Kirm, G. Zimmerer, Physics of the Solid State, 42, 464 (2000). J. C. Krupa, M. Queffelec, Journal of Alloys and Compounds 250, (1997). J. F. Suyber, J. Grimm, M. K. Van Veen, D. Biner, K. W. Krämer, H. U. Güdel, Journal of Luminescence, 117, 1-12 (2006). J. K. Berkowitz, J. A. Olsen, Journal of Luminescence, 50, 111 (1991). J. L. Sommerdijk, Journal of Luminescence, 4, , (1971). J. Lin, Q. Su, Journal of Alloys and Compounds, 210, 159 (1994). J. R. Ryan, R. Beach, Journal of theoptical Society of America, B 9, 1883 (1992). J. Y. Jeong et. al. Crystal structure and Eu 3+ luminescence of GdMF7(M=Hf,Zr) in preparation J. K. Berkowitz, and J. A. Olsen, Journal of Luminescence, 50, 111, (1991). J. L. Sommerdijk, A. Bril, A. W. de Jager, Journal of Luminescence, 8, (1974). K. D. Oskam, R. T. Wegh, H. Donker, E.V.D. van Loef, A. Meijerink, Journal of Alloys and Compounds. 300/301, 421 (2000).

206 185 L. van Pieterson, M. F. Reid, G. W. Burdick, A. Meijerink, Physical Review B 65, (2002). L. van Pieterson, M. F., Reid, R. T. Wegh, S. Soverna, A. Meijerink, Physical Review B 65, (2002). L. H. Brixner and G. Blasse, Chemical Physics Letters, 157, 283 (1989). L. van Pieterson, M. Heeroma, E. de Heer and A. Meijerink, Journal of Luminescence, 91, 177, (2000). M. J. J. Lammers, G. Blasse, Journal of the Electrochemcal Society, 134(8), (1987). M. F. H. Schuurmans, J.M.F. van Dijk, Physica B+C, 123, 131 (1984). M. J. J. Lammers, G.Blasse, Physica Status Solidi. (b), 127, 663 (1985). Michel Poulain, Bruce C. Tofield, Journal of Solid State Chemistry, 39, (1981). Michel Poulain, Marcel Poulain et Jacques Lucas, Materials Research Bulletin, 7, , (1972). N Takeuchi, S. Ishida, A. Matsumura and Y Ishikawa, Journal of Physical Chemistry B, 108, (2004). N. A. Toropov, M. V. Kougiya, Doklady Akademii Nauk SSSR 182(3), (1968). N. M. P. Low, A. L. Major, Journal of Luminescence, 4, , (1971). N. Menyuk, K. Dwight, and J. W. Pierce, Applied Physics Letters, 21(4), , (1972). O. Guillot-Noel, B. Bellamy, V. Viana and D. Gourier, Physical Review B60, 1668 (1999). O. Graudejus, F. Schroetter, B. G. Mueller, R. Hoppe, eitschrift fuer Anorganische und Allgemeine Chemie, 620, (1994). P. A. Rodnyĭ, A. N. Mishin, A. S. Potapov, Optics and Spectroscopy 93, 5, (2002). P. A. Rodnyĭ, A. S. Potapov, A. S. Voloshinovskii, Optics and Spectroscopy 96, 6, (2004).

207 186 P. A. Rodnyĭ, Opitcs and Spectroscopy 89, 4, (2000). P. Dorenbos, Journal of Luminescence, 91, (2000). P. N. Yacom, J. P. Wittke, and I. Ladany, Metallugical Transations, 2, (1971). P. P. Fedorov, M. D. Val'kovskii, O. S. Bondareva, B. P. Sobolev, Zhurnal Neorganicheskoi Khimii, 38(10), (1993). P. S. Peijzel, W.J.M. Schrama, A. Meijerink, A., Molecular Physics, 102, 1285, (2004). P. W. Dooley, J. Thogersen, J. D. Gill, H. K. Haugen, R. L. Brooks, Optics Communications, B183B, 451 (2000). R. L. Cone and R. S. Meltzer, Physical Review Letters, 30, 859 (1973). R.L. Cone and R. S. Meltzer, Journal of Chemical Physics, 62, 3573 (1975). R. P. Rao, D. J. Devine, Journal of Luminescence, 87-89, 1260 (2000). R. T. Wegh, E. V. D. van Loef, A. Meijerink, Journal of Luminescence, 90, , (2000). R. T. Wegh, H. Donker, K. D. Oskam, A Meijerink, Journal of Luminescence, 82, 93 (1999). R.T.Wegh, H. Donker, A. Meijerink, R. J. Lamminmaki, J. Holsa, Physical Review B 56, 21, (1997). S. Kuck, I. Sokolska, M. Hence, M. Doring, t. Scheffler, Journal of Luminescence, 102/103, 176 (2003). SHELXTL-6.10 "Program for Structure Solution, Refinement and Presentation" BRUKER AXS Inc., 5465 East Cheryl Parkway, Madison, WI USA Shigeo Shinoya, William M. Yen, Phosphor handbook, CRC press, Boca Raton, 394. T. Justel, P. Huppertz, W. Mayr, D.U. Wiechert, Journal of Luminescence, 106, 225 (2004). T. Kushida, Journal of the Physical Society of Japan, 34, 1318 (1973). V. Babin, K. D. Oskam, P. Vergeer, A. Meijerink, Radiation Measurements, 38, 767, (2004).

208 V. V. Ovsyankin, and P. P. Feofilov, Soviet Phys.-JETP Letters, 4, , (1966). W. Hayes and A.M. Stoneham, Defects and Defect Processes in Non-metallic Solids, J. Wiley, New York (1985). W. Jia, Y. Zhou, S. P. Feofilov, R. S. Meltzer, J. Y. Jeong, D. Keszler, Physical Review B: Condensed Matter and Materials Physics, 72(7), (2005). W. Jia, Y. Zhou, S.P. Feofilov, R.S. Meltzer, J. Y. Jeong and D. Keszler, Physical Review B, in press (2005). W. W. Piper, J. A. DeLuca, and F. S. Ham, Journal of Luminescence, 8, 344, (1974). X. X. Zhang, A. B. Villaverde, M. Bass, B. H. T. Chai, H. Weidner, R. I. Ramotar, R. E. Peale, Journal of Applied Physics, 74, 790 (1993). Xi-xian Luo, Wang-he Cao, Materials Letters, 61, , (2007). X. Wu, H. You, H Cui, X. Zeng, G. Hong, C-H. Kim, C-H. Pyun, B-Y. Yu and C-H. Park, Material Research Bulletin, 37, 1531 (2002). Y. Zhou, J.Y. Jeong, D.A. Keszler, S.P. Feofilov and R.S. Meltzer, 16th International Conference on Dynamical Processes in Excited States of Solids, DCP 07. Yu. M. Korenev, P. I. Antipov, A. V. Novoselova, P. P. Fedorov, B. P. Sobolev, Zhurnal Neorganicheskoi Khimii, 45(2), , (2000). 187 Yuji Saito, Takashi Kumagai, Shinji Okamoto, Hajime Yamamoto, Takashi Kunimote, Japanese Journal of Applied Physics, 43,6A, (2004). Yu. F. Smolin, Yu. F. Shepelev, Izvestiya Akademii Nauk SSSR, Neorganicheskie Materialy 5(10), (1969). Z. Yang, Lin, H. J., M. Z. Su, Y. Tao, W. Wang, Journal of Alloys and Compounds. 308, 94, (2000).

209 APPENDIX 188

210 189 Appendix A 1. UV/VIS Spectroscopy LUMINESCENT MEASUREMENT SYSTEM Excitation monochromator Carry model-15 Water filter Xe lamp Oriel 300W Oriel Emission monochromator PMT Hamamatsu, R Reader Data processing

211 VUV Luminescent Measurement spectroscopy ( Dr. Richard S. Meltzer s Group at UGA)