Citation
Infrared Luminescence from Spark-Processed Silicon and Erbium-Doped Spark-Processed Silicon

Material Information

Title:
Infrared Luminescence from Spark-Processed Silicon and Erbium-Doped Spark-Processed Silicon
Creator:
KIM, KWANGHOON ( Author, Primary )
Copyright Date:
2008

Subjects

Subjects / Keywords:
Annealing ( jstor )
Applied physics ( jstor )
Electroluminescence ( jstor )
Erbium ( jstor )
Infrared spectrum ( jstor )
Ions ( jstor )
Oxygen ( jstor )
Photoluminescence ( jstor )
Silicon ( jstor )
Wavelengths ( jstor )

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Kwanghoon Kim. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Embargo Date:
12/31/2007
Resource Identifier:
496174503 ( OCLC )

Downloads

This item is only available as the following downloads:


Full Text

PAGE 1

INFRARED LUMINESCENCE FROM SPARK-PROCESSED SILICON AND ERBIUM-DOPED SPARK-PROCESSED SILICON By KWANGHOON KIM A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2005

PAGE 2

Copyright 2005 by Kwanghoon Kim

PAGE 3

This dissertation is dedicated to my beloved family.

PAGE 4

ACKNOWLEDGMENTS It has been a great pleasure working with the faculty, staff, and students at the University of Florida, during my years as a doctoral student. This work would never have been possible if it were not for the freedom I was given to pursue my own research interests, most of all, thanks in large part to the kindness and considerable mentoring provided by Dr. Rolf E. Hummel, my long-time advisor and committee chair. Throughout my doctoral work he encouraged me to develop independent thinking and research skills. He continually stimulated my analytical thinking and greatly assisted me with scientific writing. His witty humor pleased me in the midst of academic endeavors and he always offered himself as a cornerstone to bank on in times of trials. I am also very grateful for having an exceptional doctoral committee and wish to thank Dr. Paul H. Holloway, Dr. David P. Norton, Dr. Wolfgang Sigmund, and Dr. Alexander Angerhofer for their continuous support and encouragement. I would like to thank Thierry A. Dubroca for helping me to build the infrared PL system. I am grateful to Anna M. Fuller, Claus Schllhorn, and Jonathan Hack for the support, the helpful comments, and for correcting my writings. I also appreciate Dr. Young-Woo Heo for helping me to use the PLD system. My parents, Duck-Hun Kim and Kyung-Youn In, deserve my deepest gratitude and love for their dedication and the many years of support that provided the foundation for this work. Their support and encouragement were in the end what made this dissertation possible. iv

PAGE 5

I express gratitude to my parents-in-laws and all other family members for their understanding and concentration. Finally, this dissertation is dedicated to my greatest blessing, my wife, Jin Kyung Lee, the most caring and loving person I have ever known. In spite of her desire to pass quietly through life, unnoticed, all of us who have her in our lives know just how amazing she is in everything that she does. Simply watching her go about her daily life makes me proud to be her husband. I thank her for all her unconditional love and understanding. A lifetime with her will always be too short. Thanks also go to our precious two children, Erwin Kyungbin Kim and Abigail Nayoung Kim, who were born during this odyssey. v

PAGE 6

TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................iv LIST OF TABLES .............................................................................................................ix LIST OF FIGURES .............................................................................................................x ABSTRACT .......................................................................................................................xx CHAPTER 1 INTRODUCTION........................................................................................................1 2 RESEARCH BACKGROUND..................................................................................11 2.1. Electroluminescence of Spark-Processed Silicon................................................11 2.1.1. Morphology of Spark-Processed SiliconThe Presence of Si Nanocrystals.....................................................................................................12 2.1.2. Spray Spark-Processing.............................................................................13 2.1.3. Photoluminescence of Rare-Earth Doped Spark-Processed Silicon in The Visible Region..........................................................................................14 2.2. Er in Si-Based Materials......................................................................................15 2.2.1. Erbium in Crystalline Silicon....................................................................15 2.2.2. Erbium in Silica.........................................................................................16 2.2.3. Erbium in Amorphous SiO x and Amorphous Silicon................................17 2.2.4. Erbium in Nanocrystals Containing SiO x ..................................................20 2.2.5. Erbium in Porous Silicon...........................................................................22 2.3. Summary..............................................................................................................24 3 EXPERIMENTAL PROCEDURES...........................................................................40 3.1. Device Fabrication Procedures............................................................................40 3.1.1. Wafer Preparation for Spark-Processing...................................................40 3.1.2. Erbium on Wafer.......................................................................................40 3.1.3. Spark-Processing.......................................................................................41 3.1.4. Rapid Thermal Annealing (RTA)..............................................................41 3.1.5. Electroluminescence Device Preparation..................................................42 3.1.5.1. Ag and Al metal contacts................................................................42 3.1.5.2. ITO (Indium-Tin-Oxide) contact....................................................43 vi

PAGE 7

3.2. Characterizations.................................................................................................44 3.2.1. Structure and Compositional Analysis......................................................44 3.2.1.1. X-ray diffraction (XRD)..................................................................44 3.2.1.2. High-resolution transmission electron microscopy (HRTEM).......44 3.2.1.3. Scanning electron microscopy (SEM).............................................45 3.2.1.4. Compositional analysis...................................................................45 3.2.2. Optical Characterizations..........................................................................46 3.2.2.1. Differential reflection spectroscopy (DR).......................................46 3.2.2.2. Photoluminescence (PL).................................................................47 3.2.2.3. Electroluminescence (EL)...............................................................48 3.2.3. Electrical and Thermal Characterizations.................................................48 3.2.3.1. Determination of efficiency of the devices.....................................48 3.2.3.2. I-V measurement.............................................................................49 3.2.3.3. Temperature effects.........................................................................49 4 MATERIALS CHARACTERIZATION....................................................................55 4.1. Differential Reflectometry...................................................................................55 4.2. X-Ray Diffraction Measurement.........................................................................57 4.3. High Resolution Transmission Electron Microscopy..........................................58 4.4. Scanning Electron Microscopy and Energy Dispersive Spectroscopy................59 4.5. X-Ray Photoelectron Spectroscopy.....................................................................60 4.6. Secondary Ion Mass Spectrometry......................................................................61 5 RESULTS AND DISCUSSIONS...............................................................................79 5.1. Infrared Photoluminescence of Spark-Processed Silicon....................................79 5.1.1. Infrared Photoluminescence of Spark-Processed Silicon After Annealing.........................................................................................................79 5.1.2. Variation of Parameters.............................................................................81 5.2. Infrared Photoluminescence of Erbium-Doped Spark-Processed Silicon...........83 5.2.1. Annealing of Erbium-Doped Spark-Processed Silicon.............................84 5.2.2. Variation of Processing Parameters...........................................................86 5.2.2.1. Concentration of erbium in spark-processed Si and spark-processing time.........................................................................................86 5.2.2.2. Pressure in spark-processing chamber............................................87 5.2.2.3. Spark-processing gap distance........................................................88 5.2.2.4. Erbium deposition and spark-processing sequence.........................88 5.2.2.5. Spark-processing environment........................................................89 5.2.2.6. Annealing environment...................................................................90 5.2.2.7. Variation in silicon substrate resistance..........................................91 5.2.2.8. Preliminary conclusions..................................................................91 5.2.3. Variation of Measuring Parameters...........................................................92 5.2.3.1. Measuring Temperature..................................................................92 5.2.3.2. Excitation source power and energy...............................................94 5.2.3.3. Emission depth................................................................................95 5.3. Infrared Electroluminescence of Erbium-Doped Spark-Processed Silicon.........95 vii

PAGE 8

5.3.1. Variation of Processing Parameter............................................................97 5.3.1.1. Concentration of erbium in spark-processed silicon.......................97 5.3.1.2. Variation of spark-processing time.................................................97 5.3.1.3. Change in types of substrate............................................................98 5.3.2. Further Variations of Measuring Parameters............................................99 5.3.2.1. Measuring temperature....................................................................99 5.3.2.2. Applied voltage.............................................................................100 5.3.3. Current-Voltage Characteristics of Erbium-Doped Spark-Processed Silicon............................................................................................................101 5.3.4. ITO Contact.............................................................................................102 6 FURTHER DISCUSSIONS AND CONCLUSIONS...............................................156 6.1. Infrared Photoluminescence of Spark-Processed Silicon..................................156 6.2. Erbium Related Photoluminescence from Spark-Processed Silicon.................160 6.2.1. Efficiency of Photoluminescence............................................................161 6.2.2. Role of Oxygen in Efficient Photoluminescence....................................163 6.2.3. Energy Band Considerations...................................................................164 6.2.4. Concluding Remarks...............................................................................165 6.3. Electroluminescence in Erbium-Doped Spark-Processed Silicon.....................165 6.4. Future Work.......................................................................................................172 7 SUMMARY..............................................................................................................181 APPENDIX VISIBLE PHOTOLUMINESCENCE OF RARE-EARTH DOPED SPARK-PROCESSED SILICON...........................................................................................185 A.1. Visible PL of Erbium-Doped Spark-Processed Silicon....................................188 A.2. Visible PL of Cerium-Doped Spark-Processed Silicon....................................194 A.3. Visible PL of Europium-Doped Spark-Processed Silicon................................198 LIST OF REFERENCES.................................................................................................202 BIOGRAPHICAL SKETCH...........................................................................................210 viii

PAGE 9

LIST OF TABLES Table page 6-1 Comparison of near IR PL and visible PL of spark-processed silicon.....................158 6-2 Porous silicon luminescence bands..........................................................................159 ix

PAGE 10

LIST OF FIGURES Figure page 1-1 Schematic for spark-processing..................................................................................7 1-2 Room-temperature normalized PL spectra of sp-Si, prepared under flowing (blue) and stagnant (green) air conditions (excited by the 325nm line of a HeCd laser, 0.3 W cm -2 ). The red peak is observed, when sp-Si is excited by the 488 nm line of an Ar laser.................................................................................................8 1-3 Pure erbium metal. (from “The Elements Collection”)...............................................9 1-4 Energy levels of Er 3+ . Data are shown both for the Er 3+ free ion, and for Er 3+ in a solid host..................................................................................................................10 2-1 Electroluminescence spectrum for sp-Si. Inset: Schematic of an electroluminescence device made from sp-Si..........................................................25 2-2 EL mechanism of spark-processed silicon...............................................................26 2-3 Electromicrographs of sp-Si. (a) TEM and area selected electron diffraction patterns. CR = crystalline, AM = amorphous. (b) Cross-sectional SEM of sp-Si....27 2-4 Raman shifts for c-Si (a) and sp-Si (b) ( exc = 488 nm, 400 W cm -2 spot size 60 m). The peak maximum in (b) is shifted by 7.1 cm -1 and the FWHM increased from 4 to 6.4 cm -1 .....................................................................................................28 2-5 Schematic of spray spark-processing apparatus.......................................................29 2-6 EL intensity of conventional sp-Si and spray sp-Si..................................................30 2-7 PL spectrum of undoped sp-Si in comparison to PL spectra of Si, which was spark-processed with europium, cerium, or terbium................................................31 2-8 Luminescence spectrum of Er-implanted and annealed Si measured at 20K. The half-width of the intense line at 1.53904 m (805.6 meV) is 0.422 meV................32 2-9 Arrhenius plot showing the temperature dependence of the 1.54 m intensity of erbium-implanted crystalline-silicon between 12 and 150 K (open circles)............33 x

PAGE 11

2-10 Energy band diagrams for a-Si:H for EL: (a) at forward bias, (b) at reverse bias...........................................................................................................................34 2-11 PL as a function of excitation photon flux for Er-doped silica and Er-doped silicon-rich silica......................................................................................................35 2-12 Schematic representation of (a) Er excitation model, showing the electronic band structure of Si nanocrystal-doped SiO 2 and the Er 4f energy levels. An optically generated exciton (dotted line) confined in the nanocrystal can recombine and excite Er 3+ . (b) SiO 2 containing Er (crosses) and nanocrystals (bright circles). The nanocrystals that couple to Er (dark circles) show no exciton luminescence...36 2-13 EL intensity as a function of device input power for both forward () and reverse () bias conditions for erbium-doped porous silicon. Inset shows a typical EL spectrum in reverse bias..........................................................................37 2-14 EL intensity of Er-doped porous Si as a function of annealing temperature............38 2-15 Normalized room temperature PL and EL comparison. The spectrum under forward bias was multiplied by a factor of 28 with respect to the reverse bias spectrum...................................................................................................................39 3-1 Schematic representation of (a) erbium-doped spark-processed silicon and (b) a typical EL device utilizing erbium-doped spark-processed silicon. The light escapes through the window at the “bottom” of the device. The Ag contact reflects the EL light back into the device and towards the window.........................50 3-2 Top-view of focused ion beam (FIB) cut erbium-doped spark-processed silicon...51 3-3 Schematic representation of differential reflectometer used in this study................52 3-4 Schematic diagrams of (a) an infrared photoluminescence spectrometer and (b) a visible photoluminescence spectrometer..................................................................53 3-5 Typical spectral response curves for the InGaAs detector.......................................54 4-1 Optical absorption spectrum obtained from 200 to 800 nm by differential reflectometry for silicon on which a 100nm thick erbium layer has been deposited before spark-processing (Stagnant air condition). Spark-processing time was 1 minute. No subsequent annealing..........................................................64 4-2 Optical absorption spectrum from 200 to 800 nm obtained by differential reflectometry for erbium-doped spark-processed silicon prepared in stagnant air condition for 1 minute and subsequently annealed at 900C in air for 15 minutes. Compare to Figure 4-1.............................................................................................65 xi

PAGE 12

4-3 Energy level diagram of Er3+ ion and the intrashell transitions observed by excitation with an electron beam and photons of 325 nm (He-Cd laser) and of 488 nm (Ar-ion laser)...............................................................................................66 4-4 XRD spectra of Er-Si-N-O distribution with various annealing temperatures.........67 4-5 Detailed XRD for erbium-doped spark-processed silicon (a) as deposited (un-activated) and (b) annealed at 700C (activated).....................................................68 4-6 An Er 2 O 3 (222) XRD peak (at 2=29.64 o ) intensity changes with annealing temperature...............................................................................................................69 4-7 HRTEM images of (a) silicon nanocrystals embedded in erbium-doped spark-processed silicon matrix, (b) electron diffraction patterns of silicon polycrystals and (c) diffraction spots of silicon single crystals....................................................70 4-8 STEM images of (a) High angle annular dark-field (HAADF) image: dark areas indicate high-Z elements (Erbium) and bright areas indicate low-Z element (Silicon), (b) bright-filed STEM..............................................................................71 4-9 Scanning electron micrograph of erbium-doped spark-processed silicon, (a) full-view of surface morphology, (b) higher magnification of center area and (c) halo area image of erbium-doped spark-processed silicon..............................................72 4-10 Energy dispersive spectroscopy of erbium-doped spark-processed silicon. (a) EDS measured at the center of erbium-doped spark-processed silicon, (b) EDS measured half way to the edge of the spark-processed area from the center, (c) EDS measured in the halo area. Note that the thin lines indicate the positions of certain elements........................................................................................................73 4-11 XPS spectra of (a) an as-prepared erbium-doped spark-processed silicon, and (b) same, but annealed at 1100C for 15 minutes..........................................................74 4-12 The XPS binding energies peak positions as a function of annealing temperature. (a) Silicon Binding Energy changes after Annealing, (b) Erbium Binding Energy changes after Annealing, and (c) Oxygen Binding Energy changes after Annealing.................................................................................................................75 4-13 SIMS spectra for annealed erbium-doped spark-processed silicon (a) acquired with a 5 keV cesium primary ion beam from the spark-processed area, (b) acquired with a 6 keV oxygen primary ion beam from the perimeter of the spark-processed area................................................................................................77 4-14 SIMS depth profile of typical erbium-doped spark-processed silicon.....................78 5-1 Room temperature infrared photoluminescence spectrum of spark-processed silicon employing a p-type (B-doped, 5 cm) Si wafer and involving 15 minutes of sparkprocessing time utilizing a 1.2 mm gap between the tungsten tip and xii

PAGE 13

the Si wafer under stagnant air conditions. A 488nm Ar laser line was utilized as an excitation source................................................................................................104 5-2 Change of infrared photoluminescence spectra compared to pure spark processed silicon as a function of various annealing temperatures in air for 30minutes. Note that the base line of each curve is shifted upwards for clarity.......................105 5-3 Infrared photoluminescence peak intensity change as a function of annealing temperature (30 minutes, air). The error bars represent data from two independent measurements....................................................................................106 5-4 Infrared photoluminescence peak wavelengths as a function of annealing temperature (30minutes, air)..................................................................................107 5-5 Peak deconvolution of the infrared photoluminescence band of as-processed spark-processed silicon into 3 major peaks having peak maxima near 945, 1009, and 877nm (r 2 =98.5). (a) Experimental data and fitted curve. (b) Fitted curve and deconvoluted peak components.......................................................................108 5-6 Change of the infrared photoluminescence peak intensity as a function of annealing temperature of two deconvoluted peaks................................................109 5-7 Variation of deconvoluted 945 and 1010nm peak maximum wavelengths as a function of annealing temperature..........................................................................110 5-8 Infrared photoluminescence intensity variation of spark-processed silicon as a function of annealing time at 900C in air. Samples used in this experiment were prepared with 1 minute spark-processing under stagnant air conditions...............111 5-9 Infrared photoluminescence intensity variation as a function of spark-processing time for pure spark-processed silicon.....................................................................112 5-10 SEM images of spark-processed silicon using different spark-processing time (a) 10 seconds, (b) 15 minutes, and (c) 1 hour............................................................113 5-11 EDS spectra of spark-processed silicon using different spark-processing times (a) 10 seconds, (b) 15 minutes, and (c) 1 hour.......................................................114 5-12 Infrared photoluminescence intensity as a function of the spark gap between the tungsten tip and silicon substrate...........................................................................115 5-13 Infrared photoluminescence using 3 different excitation wavelengths from an Ar laser (457, 488 and 515nm). The inset shows infrared photoluminescence intensities as a function of the excitation wavelength............................................116 5-14 Photoluminescence spectrum of a control sample: 0.1 m erbium metal was vapor-deposited on a silicon wafer (n-type, 5 cm) and rapid thermal annealed xiii

PAGE 14

at 900C in air for 15 minutes. No spark-processing was utilized. The “spectrum” shows the background noise of the PL system...................................117 5-15 Typical infrared PL spectrum of erbium-doped spark-processed silicon. The sample was prepared by depositing 0.1m erbium on a silicon wafer (n-type, 5 cm) and subsequently spark-processing for 5 minutes under stagnant air, followed by rapid thermal annealing at 900C in air for 15 minutes. The infrared photoluminescence spectrum was obtained by utilizing the 488nm Ar ion laser line as an excitation source.....................................................................................118 5-16 Room temperature photoluminescence spectrum in a limited spectral range (1300 to 1650nm) for an erbium-doped spark-processed silicon sample measured with a spectral resolution of 1nm. A 488nm Ar ion laser was utilized as excitation source. Compare to Figures 5-14 and 5-15.......................................119 5-17 Infrared photoluminescence peak intensity of erbium-doped spark-processed silicon for the 1540nm peak as a function of rapid thermal annealing temperature (15 minutes, air).................................................................................120 5-18 Infrared photoluminescence intensity of erbium-doped spark-processed silicon for the 1540nm peak as a function of annealing time at the optimum annealing temperature of 900C.............................................................................................121 5-19 Infrared photoluminescence intensity of the 1540nm peak of erbium-doped spark-processed silicon as a function of initial erbium layer thickness before spark-processing and as a function of spark-processing time. The samples were annealed at 900C in air for 15 minutes.................................................................122 5-20 Infrared photoluminescence intensity of erbium-doped spark-processed silicon for the 1540nm peak as a function of spark-processing pressure. The samples were prepared using 30 seconds spark-processing time and an erbium layer on silicon of 100nm.....................................................................................................123 5-21 Infrared photoluminescence intensity change of erbium-doped spark-processed silicon as a function of spark-processing time at 200mbar (solid line). The thickness of erbium layer was 75nm and the sample was annealed at 900C for 20minutes. For comparison, IR PL intensity variation as a function of spark-processing time at atmospheric pressure was plotted (dotted line)........................124 5-22 1.54 m infrared photoluminescence intensity variation of erbium-doped spark-processed silicon as a function of the spark gap between the tungsten tip and silicon.....................................................................................................................125 5-23 Schematics of processes to elucidate the most efficient erbium deposition and spark-processing sequence in erbium-doped spark-processed silicon. Set A : 1) clean silicon wafer, 2) spark-processing for 30 seconds, 3) 0.2m of erbium metal deposition, Set B : 1) clean silicon wafer, 2) 0.1m of erbium metal xiv

PAGE 15

deposition, 3) spark-processing for 30 seconds, 4) additional 0.1m of erbium metal deposition, Set C : 1) clean silicon wafer, 2) 0.2m of erbium metal deposition, 3) spark-processing for 30 seconds, *Set D : 1) clean silicon wafer, 2) 0.1m of erbium metal deposition, 3) spark-processing for 30 seconds. * set D is prepared for comparison with other results, see below..................................126 5-24 Infrared photoluminescence spectra for variable process sequences. Intensity at 1.54 m; Set B > Set C > *Set D > Set A..............................................................127 5-25 Infrared photoluminescence spectra of erbium-doped spark-processed silicon processed in air, oxygen and nitrogen ambients respectively................................128 5-26 Infrared photoluminescence spectra of erbium-doped spark-processed silicon using different annealing environments such as air, oxygen and nitrogen ambient...................................................................................................................129 5-27 Comparison of infrared photoluminescence spectra of erbium-doped spark-processed silicon utilizing UHP nitrogen and various annealing ambients. The spectra are compared to “spark-processed in air and annealed in air”...................130 5-28 Infrared photoluminescence of the 1540nm peak of erbium-doped spark-processed silicon as a function of silicon wafer resistivity....................................131 5-29 Temperature dependence of the integrated infrared (1540nm) photoluminescence intensity I(T) of erbium-doped spark-processed silicon from 4 K to elevated temperature plotted in an Arrhenius fashion (square data points). The experimental data were analyzed by using Eq. 5-1 and the fitted results are shown by the solid line...........................................................................................132 5-30 The logarithm of [I 0 /I(T)]-1 is plotted versus photoluminescence measuring temperature.............................................................................................................133 5-31 Infrared photoluminescence intensity of erbium-doped spark-processed silicon as a function of heating and cooling temperature from room temperature (20C) to 200C..................................................................................................................134 5-32 Variation of the photoluminescence intensity as a function of time when erbium-doped spark-processed silicon is exposed to Ar ion laser light (488nm) at room temperature, and at 100C and 200C. The first 20 minute measurements are affected by ramping................................................................................................135 5-33 Infrared photoluminescence intensity variation of erbium-doped spark-processed silicon as a function of the excitation power (488nm Ar ion laser).......................136 5-34 Infrared photoluminescence external output power of erbium-doped spark-processed silicon as a function of the power of the excitation source (488nm Ar ion laser).................................................................................................................137 xv

PAGE 16

5-35 Relationship between the infrared photoluminescence intensity of erbium-doped spark-processed silicon and the excitation wavelength. (Three lines were used, namely 457nm, 488nm, and 515nm.).....................................................................138 5-36 Infrared photoluminescence spectra of erbium-doped spark-processed silicon after different etching times. Samples were annealed at 900C in air for 15 minutes...................................................................................................................139 5-37 The solid line shows a typical infrared electroluminescence room temperature spectrum of erbium-doped spark-processed silicon prepared using 0.1 m erbium metal deposited on a silicon wafer (n-type, 5 cm), a spark time of 30 seconds in air, a spark gap of 2 mm and 900C rapid thermal annealing (RTA) for 15 minutes in air. Aluminum (50nm) on the back-side of the wafer and silver (200nm) on the top of devices was deposited by physical vapor deposition (PVD) as electrical contacts. The dashed line indicates a photoluminescence spectrum of erbium-doped spark-processed silicon for comparison.............................................................................................................140 5-38 Peak deconvolution of the infrared (a) electroluminescence and (b) photoluminescence spectra of erbium-doped spark-processed silicon...................141 5-39 Infrared electroluminescence spectrum of a control device prepared as in Figure 5-37 but without spark-processing.........................................................................142 5-40 Infrared electroluminescence intensity of erbium-doped spark-processed silicon as a function of erbium concentration variation.....................................................143 5-41 Infrared electroluminescence intensity of erbium doped spark-processed silicon monitored at 1550nm as a function of spark-processing time. Initial erbium thickness: 200nm; 900C rapid thermal annealing for 15 minutes in air...............144 5-42 EDS spectra of erbium-doped spark-processed silicon processed with various times (a) 5 seconds, (b) 40 seconds, (c) 70 seconds, and (d) 100 seconds.............145 5-43 Infrared electroluminescence spectra of n-type (dashed curve) and p-type (solid curve) wafers. (100) n-type wafer having 3-5cm resistivity. Applied voltage: 25V. Current: 101mA. 0.0073% output/input efficiency. (100) p-type wafer having 1-3cm resistivity. Applied voltage: 25V. Current: 163mA. 0.0044% output/input efficiency...........................................................................................146 5-44 Reverse current-voltage characteristics of n-type (circles) and p-type (squares) erbium-doped spark-processed silicon EL device..................................................147 5-45 Infrared electroluminescence intensity of erbium-doped spark-processed silicon as a function of operating temperature at 1550nm (peak wavelength) and 1420nm (background wavelength).........................................................................148 xvi

PAGE 17

5-46 Infrared electroluminescence spectra of an erbium-doped spark-processed silicon EL device measured at different voltages..............................................................149 5-47 Infrared electroluminescence intensity of an erbium-doped spark-processed silicon EL device at 1550nm as a function of applied voltages.............................150 5-48 Relation between the infrared electroluminescence intensity of erbium-doped spark-processed silicon device and the applied power...........................................151 5-49 Current-voltage characteristics of erbium-doped spark-processed silicon. A n-type wafer having 5cm has been used. Pulsed d.c. with 200Hz frequency and 30% duty cycle was utilized...................................................................................152 5-50 Current-voltage characteristics of erbium-doped spark-processed silicon as a function of wafer resistivity. Selected wafers having 0.02cm, 1.5cm, 5.2cm, 13cm, 25cm and 40cm have been used..........................................153 5-51 Infrared electroluminescence intensity of erbium-doped spark-processed silicon as a function of wafer resistivity. No EL was obtained for the 0.02cm wafer which had an ohmic characteristic, see Figure 5-50..............................................154 5-52 Infrared electroluminescence spectrum for an erbium-doped spark-processed silicon device where ITO was utilized as a transparent contact which also served as a window............................................................................................................155 6-1 Normalized PL spectra of SiO x thin films having different silicon concentrations, annealed at 1250C for 1 hr. The spectra were measured at room temperature, with a laser pump power of 10 mW.......................................................................174 6-2 Band level diagram for room temperature photoluminescence of spark-processed silicon. Newly introduced infrared emissions are inserted in this diagram. The dotted arrows indicate non-radiative transitions....................................................175 6-3 Schematic representation of first coordination shell surrounding erbium species in Float-zone-Si (left) and Czochralski-Si. The actual spatial orientations of Si and O atoms may not be as depicted, but an attempt is made to show that the optically active species are noncentrosymmetric. Also, the configuration of the host Si atoms in the second coordination shell are not indicated for either erbium species....................................................................................................................176 6-4 Infrared photoluminescence mechanisms of erbium-doped spark-processed silicon. 1.54 m infrared photoluminescence model for direct excitation and absorption from excitation source of the erbium luminescence center..................177 6-5 Energy transfer model for the 1.54 m infrared photoluminescence mechanism of erbium-doped spark-processed silicon...............................................................178 xvii

PAGE 18

6-6 Schematic representation of the possible nano silicon particles (NSP) related Er 3+ excitation mechanism in the EL process of an Au/SRSO:Er/n + -Si system under reverse bias...................................................................................................179 6-7 Suggested model for the 1.54 m infrared EL mechanism. (1) Electrons are injected into the spark-processed silicon by tunneling. (2) Electrons are accelerated to high velocities leading to impact ionization. (3) Impact ionization results in the generation of excited electrons. (4) Excited electrons excite Er 3+ from the ground state to the 4 I 11/2 or 4 I 13/2 state. (5) Electrons relax radiatively to the ground state from the 4 I 13/2 state.......................................................................180 A-1 Visible photoluminescence spectra for erbium, cerium and europium-doped spark-processed silicon. Erbium PL has a characteristic peak near 550nm. Cerium displays a broad peak at 460nm. Europium has a broad maximum near 540nm and a sharp peak near 617nm.....................................................................187 A-2 Visible PL spectra of the erbium-doped spark-processed silicon as a result of HeCd laser (325nm) excitation for short spark-processing times. The 3 arrows indicate characteristic peaks. The inset indicates the wavelength change of the broad maxima as a function of spark-processing time...........................................188 A-3 Visible PL spectra of erbium-doped spark-processed silicon as a function of spark-processing time under HeCd laser (325nm) excitation. The erbium peak disappears after 1 minute spark-processing. For longer spark-processing times (15 minutes, 1 hour) a negative peak is observed. The inset indicates a PL spectrum for erbium-doped spark-processed silicon after 1 hour spark-processing...............................................................................................................189 A-4 Visible PL spectrum of erbium-doped spark-processed silicon after rapid thermal annealed in air for 15 minutes. The sharp peaks are consistent with Er 3+ energy levels ( 2 H 11/2 , 4 S 3/2 ).................................................................................................190 A-5 PL peak intensities as a function of erbium layer thickness. The solid circles () and the solid squares () indicate the sharp erbium peaks and the open triangles () indicate the broad spark-processed silicon related peak. Erbium ion characteristic peaks increase with erbium layer thickness, but at the same time the broad peak decreases........................................................................................191 A-6 Normalized visible PL spectra of erbium-doped spark-processed silicon having various erbium layer thicknesses. As the erbium layers become thicker, the broad peak decreases and the sharp Er 3+ peak increases in intensity.....................192 A-7 Visible PL spectrum for erbium-doped spark processed silicon, spark-processed for 10 seconds with subsequent annealing in pure argon environment for 20 minutes. No Er 3+ characteristic peaks are observed...............................................193 xviii

PAGE 19

A-8 Visible PL spectra of cerium metal, spark-processed cerium, and cerium-doped spark-processed silicon. Cerium-doped spark-processed silicon was prepared using a 100nm cerium layer on a silicon wafer and 20 seconds spark-processing.194 A-9 PL intensity as a function of rapid thermal annealing temperature of cerium-doped spark-processed silicon. The sample was prepared by depositing Ce on Si by PVD, and spark-processing for 1 minute, followed by annealing for 30 minutes in air..........................................................................................................195 A-10 PL intensity of the 460nm peak of cerium-doped spark-processed silicon as a function of spark-processing time..........................................................................196 A-11 PL intensity of cerium-doped spark-processed silicon as a function of distance from the center spot area........................................................................................197 A-12 PL intensity of 611nm europium-doped spark-processed silicon as a function of spark-processing time.............................................................................................198 A-13 PL spectra of europium-doped spark-processed silicon as a function of various rapid thermal annealing temperatures, annealing time of 15 minutes. Europium was deposited by PVD on a silicon substrate and spark-processed for 30 seconds.................................................................................................................................199 A-14 PL intensity ratio as a function of annealing temperature. I 611nm indicates the Eu 3+ characteristic peak intensity and I 550nm indicates the broad spark-processed silicon peak.............................................................................................................200 A-15 PL intensity at 611nm of europium-doped spark-processed silicon as a function spark-processing time.............................................................................................201 xix

PAGE 20

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy INFRARED LUMINESCENCE FROM SPARK-PROCESSED SILICON AND ERBIUM-DOPED SPARK-PROCESSED SILICON By Kwanghoon Kim December 2005 Chair: Rolf E. Hummel Major Department: Materials Science and Engineering Spark-processed silicon has substantial potential as an optical material. In the past 15 years, our group has investigated a multitude of properties of this unique material, concentrating mostly on the visible and near UV spectral region. The present study expands our endeavors to infrared photoluminescence (PL) of undoped spark-processed silicon. A broad infrared photoluminescence peak at around 945 nm under Ar ion laser excitation was observed at room temperature when investigating a spark-processed layer on a silicon wafer. This light emission is interpreted to be the result of energy transfers between certain energy levels involving the spark-processed silicon matrix. The infrared PL intensity of spark-processed silicon was found to be proportional to the excitation energy. However, telecommunication requires presently a light emission near 1.54 m (because fiber-optics “conductors” have a minimum in absorption at this wavelength). This cannot be achieved with pure spark-processed silicon. Therefore spark-processed xx

PAGE 21

silicon needs to be doped with a rare-earth element such as erbium to shift the emission to longer wavelengths. It is known that erbium has a light emission from intrashell energy transition, that is, from 4 I 13/2 4 I 15/2 . Erbium was deposited on a silicon wafer followed by spark-processing, which enables diffusion of some erbium into the SiO x matrix, thus achieving opto-electronically active spark-processed silicon. Rapid thermal annealing enhances the 1.54 m wavelength intensity from erbium-doped spark-processed silicon. The processing conditions that result in the most efficient photoluminescence have been established and will be presented in this dissertation. In contrast to erbium-doped crystalline silicon, whose light emission is highly affected by temperature (10 3 times reduction in intensity when heating from 12 K to 150 K), the intensity of erbium-doped spark-processed silicon decreases by only a factor of 4 when heated from 15 K to room temperature. Based on the most efficient photoluminescence conditions, an erbium-doped spark-processed silicon electroluminescence (EL) device was prepared. The devices studied have metal contacts on top and bottom of the substrate. A window through the metallization at the bottom was formed by masking which enabled the light emission. The electroluminescence spectrum has a peak at around 1.55 m with an output/input power efficiency of about 0.01%. The device works even at elevated temperatures. The experimental findings are interpreted by postulating a photoluminescence mechanism with energy transfer from spark-processed silicon to the Er 3+ ions. Furthermore, a hot electron tunneling-induced electroluminescence mechanism is considered. xxi

PAGE 22

CHAPTER 1 INTRODUCTION The development of integrated circuit (IC) technology is undoubtedly one of the most monumental events of the 20th century. This “miracle” almost exclusively depends on the development of silicon (Si) technology. Silicon, in its crystalline, polycrystalline, or amorphous state, has been the premier semiconducting material for nearly 50 years and will continue to be almost exclusively the material for the semiconductor industry also in the future. Several reasons have made Si the most appropriate material in the semiconductor industry. The decisive reason is its stable oxide (SiO 2 ) which has been used to establish significant basic technologies, including the processes for diffusion, doping, and defining intricate patterns. In addition, from the environmental point of view, silicon is entirely nontoxic, and silica (SiO 2 ), the raw material of silicon, comprises approximately 60% of the minerals in the earth’s crust. Silicon makes up 25.7% of the earth's crust by weight and is the second most abundant element exceeded only by oxygen. It is found largely as silicon oxide in such as sand, quartz, rock crystal, amethyst, agate, flint, jasper, and opal. Silicon is found also in minerals such as asbestos, feldspar, clay, and mica. This implies that the raw materials of silicon can be easily supplied to the IC industry [1]. Even though silicon is the most important material for microelectronics, its indirect band-gap and poor luminescence properties make it essentially ineffective for use in light-emitting devices. Optoelectronic hybrid designs involving the integration of Si with III-V compounds are costly and difficult to fabricate. This provides strong incentives to 1

PAGE 23

2 look for more efficient light emitting forms of Si. A variety of approaches for achieving this goal include alloying with Ge, doping with rare-earth metals like Er, making porous structures by electrochemical etching, and notably by spark-processing. Visible photoluminescence (PL) from porous silicon was reported by L. T. Canham in 1990 [2]. Porous silicon is prepared by anodic etching of a crystalline silicon wafer in an HF aqueous solution. However, this material is not quite satisfactory because it involves a wet method utilizing HF. It is mechanically unstable, and its photoluminescence properties decay rapidly by laser exposure, etching, and annealing. Dry methods for preparing light emitting Si materials are highly desired. It was reported that emissions from Si structures having energies larger than the band-gap were achieved among others by the following dry techniques: spark processing [3-17], gas-evaporation [18, 19], sputtering [20-26], microwave plasma deposition of silane [27-29], laser ablation [10, 30-32], and implantation[33-39]. Spark-processing seems to be most promising because it is a clean, dry, and simple method [17]. This technique involves sparks between an anode tip to which a pulsed high d.c. voltage is applied, and a grounded Si wafer (Figure 1-1). Spark-processing of Si modifies its macroand microstructure and generates a substance that strongly photoluminesces in several bands in the visible spectrum [12, 14]. Strong UV/blue and green photoluminescence was first reported by R. E. Hummel and co-workers when spark-processed silicon (sp-Si) was excited by a 325 nm HeCd laser [14, 16](Figure 1-2). The exact wavelength of emission depends on the preparation conditions. The response time is fast (picoseconds), and the thermal and mechanical stability is high [13]. Microstructure studies reveal a mixture of nanocrystalline Si imbedded in an amorphous

PAGE 24

3 oxide. It should be added in passing that PL with above-band-gap energies was also observed for spark-processed Ge, Sb, Bi, Sn, GaAs, and more materials [8, 9]. Other undoped Si-based materials like crystalline Si, amorphous Si, and porous Si have weaker PL intensities and poorer stability than spark-processed Si. Porous Si has luminescence in the orange/red part of the visible range or, when oxidized, scarcely in the blue region. Its decay times are generally in the s range and its stability towards laser radiation, annealing, and HF etching is low [13]. Another promising alternative approach for a light emitting Si-based device is doping the Si substrate for example with rare-earth metals [40, 41]. The main goal of research in the field of rare-earth doped semiconductors has been the development of optoelectronic amplifiers and sources for optical fiber communication networks operating at 1.54 m. Consequently, the focus of many research endeavors has been on erbium-doped silicon crystals. Erbium ( 68 Er) is the 11 th element in the lanthanide series ( 58 Ce71 Lu). Erbium as an element was first discovered and identified in the Swedish town of Ytterby in 1842. Erbia, the renamed material that Mosander discovered in 1843, is erbium oxide (Er 2 O 3 ), one of erbium's compounds [41]. Erbia has a pink color and is used to color glass and glazes. Pure erbium metal is soft and malleable and has a bright, silvery, metallic luster (Figure 1-3)[42]. As with other rare-earth metals, its properties depend to a certain extent on impurities present. The metal is fairly stable in air and does not oxidize as rapidly as some of the other rare-earth metals. Its electronic structure is [Xe] 4f 12 6s 2 . When erbium is incorporated as an impurity into a dielectric host, it usually takes the trivalent charge state with an electronic configuration of [Xe] 4f 11 . This configuration is unique, as it is

PAGE 25

4 composed of an incompletely filled 4f shell that is shielded from the surrounding host matrix by two filled 5s and 5p shells. The 4f electrons interact by spin-spin and spin-orbit interactions. The two interactions are of similar strength and lead to relatively large differences between the 4f energy levels. Figure 1-4 [43] shows a schematic of the 4f energy levels of Er, both for a free ion and in a solid host where Stark splitting occurs as the degenerate 4f levels split in the electric field caused by the local atomic configuration around the Er 3+ ion. Room temperature Er 3+ luminescence has been obtained in different forms of silicon hosts, either for ion implanted crystalline [44] and porous Si [45, 46], or co-deposited crystalline [47] and amorphous Si [48]. Due to the shielding effect of the outer shells, the exact energies of the 4f states differ only slightly for different hosts in which Er 3+ is incorporated. Er 3+ has become a technologically important ion as the transition from the first excited state to the ground state involves E=0.80 eV, corresponding to =1.54 m. This coincides with the wavelength of the maximum transmission in silica based-optical fibers that are used in optical telecommunication technology. The present revolution in communication technology can be substantially attributed to the fact that erbium-doped optical fiber amplifiers have been developed making long-distance optical communication possible. Since the initial report of weak 4f shell luminescence from erbium-doped crystalline silicon in 1983 [40], erbium has been incorporated in numerous materials [41]. Erbium-doped crystalline silicon structures are usually prepared by high-energy (up to 5 MeV) ion implantation, epitaxial growth, and chemical vapor deposition followed by high temperature annealing. All of these techniques are expensive and time consuming, require specialized equipment, and are limited to very shallow doping profiles.

PAGE 26

5 A unique combination of properties proves to be spark-processed silicon which is a very promising material for optoelectronic applications. The microstructure of spark-processed silicon makes the material an ideal host due to its large surface area which allows easy infiltration of dopant ions into the structure. The spark-processed silicon structure is already oxidized, which provides the large concentrations of oxygen necessary for efficient erbium emission. As already pointed out, light emission from undoped crystalline Si is known to be very poor because of its indirect band-gap. Moreover, erbium ions in crystalline Si are reported to be excited by a fraction of the energy released from the electron-hole pair recombinations in the semiconductor host material [49]. The other well known excitation process of erbium in a silicon-based host is impact ionization by hot electron tunneling. In case of spark-processed Si, impact ionization induced by hot electron tunneling is also believed to be the major electroluminescence (EL) mechanism. Based on the facts that PL in sp-Si is enhanced by rare-earth (Ce, Eu) doping [50, 51], and that sp-Si has better PL properties than other Si based materials, it is speculated that sp-Si is a promising host material for Er doped PL and EL. On the other hand, the nanocrystallites formed in sp-Si silicon can create a direct and wide band-gap due to carrier quantum confinement effects. The present study investigates spark-processed silicon which has been doped with erbium. As mentioned above the spark-processed structures can provide suitable conditions for fast diffusion of foreign atoms into the spark-processed layer. The goals for this research are to investigate the properties of rare-earth ion doped spark-processed silicon, in particular the possibility of fabricating infrared light emitting devices from it, and to develop new effective and simple methods for preparing light emitting silicon

PAGE 27

6 diodes. These results are compared with the infrared luminescence properties of undoped spark-processed Si whose emission wavelengths have been found to be situated at shorter wavelengths. Furthermore, possible mechanisms involved for EL in erbium-doped spark-processed silicon are discussed and the structural characteristics of these materials are investigated.

PAGE 28

7 Figure 1-1 Schematic for spark-processing [8].

PAGE 29

8 Figure 1-2 Room-temperature normalized PL spectra of sp-Si, prepared under flowing (blue) and stagnant (green) air conditions (excited by the 325nm line of a HeCd laser, 0.3 W cm -2 ). The red peak is observed, when sp-Si is excited by the 488 nm line of an Ar laser [11].

PAGE 30

9 Figure 1-3 Pure erbium metal. (from “The Elements Collection” [42])

PAGE 31

10 Figure 1-4 Energy levels of Er 3+ . Data are shown both for the Er 3+ free ion, and for Er 3+ in a solid host [43].

PAGE 32

CHAPTER 2 RESEARCH BACKGROUND 2.1. Electroluminescence of Spark-Processed Silicon The electroluminescence (EL) of spark-processed silicon was studied and characterized by Shepherd [6], Hummel et al. [10], and Shepherd and Hummel [52, 53] and others [3, 4]. Conventional spark-processed Si has a very high electrical resistance, which is in the M range. The EL originates from some radiative recombination processes, which may or may not be the same as that in the PL process, while the process excites radiative recombination centers by the electric field. This is completely different from that in PL, where the excitation is achieved by high-energy photons. Visible EL is obtained from very lightly spark-processed Si. The lightly sp-Si can emit visible light under 7 volts (average) dc, pulsed at 60 Hz with 50% duty cycle to prevent overheating [6, 52]. Electroluminescence of spark-processed silicon appears yellowish to the naked eye and is readily visible in semi-dark to low light conditions. The EL spectrum of spark-processed silicon in the visible range is broad, peaks near 700 nm (1.77 eV), and has a distinct shoulder near 640 nm (1.94 eV) as shown in Figure 2-1. The device from which EL emissions were obtained is a vertical metal-spark-processed Si-semiconductor type structure with a semitransparent metal contact (Ag) on top and a metal Ohmic contact (Al) on the back as shown in the inset of Figure 2-1. The light escapes through the semitransparent Ag contact which attenuates some of the light. 11

PAGE 33

12 The current/voltage characteristics for sp-Si EL devices look similar to a rectifier junction whereby EL only occurs in what is generally identified as the “reverse” biased condition. However, light emission is observed well before the breakdown. Both n and p type silicon can be used as a substrate in the EL device. In both cases, EL is observed only when the top semitransparent metal contact is negatively biased. A visible emission is seen at an applied voltage of as low as 6 V, with currents ranging from 20-40 mA depending on the resistivity of the wafer used. Devices are typically driven with d.c. or 60 Hz pulsed d.c. volts (see above). Based on the results of the EL characterization measurements, a mechanism based on hot electrons was proposed by Shepherd and Hummel which is described in Figure 2-2 [52]. Electrons assisted by field enhancement are injected into the spark-processed layer (SiO x N y ) by tunneling, and are accelerated by the interface field. These injected high-energy electrons generate more electrons by an impact ionization process. Subsequent radiative de-excitation of electrons in spark-processed silicon enables a visible electroluminescence [6]. 2.1.1. Morphology of Spark-Processed SiliconThe Presence of Si Nanocrystals Silicon nanocrystals are known to be involved in the light emission in the visible range [54]. High-resolution TEM micrographs of sp-Si reveal single-crystalline Si particles with diameters in the nanometer range that are imbedded in a SiO 2 matrix. From the transmission electron micrographs in Figure 2-3 (a), it can be noted that {1 1 1} planes of the single-crystalline particles are rotated considerably with respect to each other [12, 14, 15]. Consequently, the electron diffraction pattern of the crystalline area indicates polycrystallinity. The radii in the diffraction pattern correspond to the d-spacing

PAGE 34

13 of silicon. The nanocrystallites are surrounded by an amorphous phase which consists of silicon oxide as identified by Xray energy dispersion spectra. These nanocrystals might be remnants of the original Si bulk crystal that were severely damaged and distorted in order to explain the polycrystalline orientations [14, 15]. It is, however, also possible and more likely that these Si particles are created by nucleation during the condensation of the vapor phase during the spark processing. An epitaxial growth of Si nanoparticles on top of the substrate is then probably hindered by the coexisting an oxide phase [14]. Raman spectroscopy measurements essentially confirm the presence of nanocrystalline Si [7, 15]. Figure 2-4 compares the Raman spectra from bulk Si with those from sp-Si. The Raman shift for crystalline Si peaks at 520.9 cm -1 , with a FWHM of 4 cm -1 . For sp-Si, Raman signals were found to shift to lower energies, accompanied by peak broadening. Peak position and FWHM are inhomogeneously distributed over the sample surface and depend on sample preparation and spot size. Raman peak shifts can be caused by a phonon confinement in microstructures and correlate with the particle size. By assuming spherically shaped particles, sizes of 3 to 5 nm were inferred from the spectra in Figure 2-4. These results are consistent with those from TEM investigations [12, 15]. 2.1.2. Spray Spark-Processing More than 80% of the available spark-processed surface does not participate in EL emission because the center of the sparked spot is considerably rough and unable to be covered by a thin semitransparent metal film. Better surface coverage would lead to increased device currents, larger area emission, and consequently improved EL intensity. A solution consisting of methanol and submicron sized silicon (or silica) is sprayed onto

PAGE 35

14 the Si substrate during spark-processing achieves this goal. Specifically instead of a tungsten tip, the needle of a syringe was used as an anode for the spark apparatus (Figure 2-5)[53]. The syringe was filled with the methanol and submicron silicon particulate suspension. Spark-processing was then conducted by applying moderate pressure to the syringe during application of a voltage. This modification of the spark-processing technique has improved the EL intensity of spark-processed devices by one order of magnitude, as seen in Figure 2-6 [52, 53]. The similarity of the EL spectra for spray sparked and conventionally sparked samples, together with other common features, indicates that the underlying EL mechanism is the same in both cases. 2.1.3. Photoluminescence of Rare-Earth Doped Spark-Processed Silicon in The Visible Region The photoluminescence intensity of some rare-earth metals incorporated into Si during spark-processing is improved up to about one order of magnitude compared to undoped sp-Si and the peak is shifted as shown in Figure 2-7. Solutions of europium, cerium, and terbium salts in methanol have been deposited onto Si substrates before spark-processing [6]. Ce doped sp-Si PL peaks near 455 nm (2.72 eV) whereas Eu doped sp-Si has a PL maximum near 525 nm (2.36 eV). The PL intensity for Ce and Eu is enhanced by a factor of ten and five respectively whereas Tb reduces the PL intensities compared to undoped sp-Si. Spark-processing of erbium on silicon surfaces showed visible PL ascribed to optically active Er 3+ [55]. Varying the concentration of the initial salt solutions which were deposited before spark-processing controlled erbium concentrations in the spark-processed layer.

PAGE 36

15 2.2. Er in Si-Based Materials 2.2.1. Erbium in Crystalline Silicon Er induced IR luminescence has been extensively researched, since photoluminescence of Er-doped Si was firstly reported by Ennen et al. in 1983 (Figure 2-8)[40]. Well-resolved sharply structured luminescence spectra peaking near 1.54 m were observed in erbium-implanted Si. The optical transitions occur between the weakly crystal field split spin-orbit levels, from the first excited state to the ground state, 4 I 13/2 4 I 15/2 , of Er 3 + ([Xe]4f 11 ). Room-temperature electroluminescence is obtained from erbium-doped Si diodes for heavily Erand oxygen-doped material [49, 56]. In Si containing also oxygen, Er was shown to exhibit high quantum efficiencies at low temperatures both for interband excitation and for forward bias in an electroluminescence diode. At temperatures above about 150 K, however, the luminescence yield is strongly quenched, and thus efficient room-temperature operation is very difficult [57, 58]. Such diodes exhibit under reverse bias a 10-20 nm wide emission peak which has a striking similarity to erbium-implanted silica (SiO 2 : Er) in contrast to the sharp line spectra of isolated Er centers seen at lower temperatures. The wide peak is thus attributed to the inhomogeneously broadened emission from Er in amorphous SiO x precipitates. The isolated centers can be transformed into Er-containing precipitates for sufficient Erand O-amounts by proper choice of the annealing temperature after implantation [49, 59]. The main obstacle for commercial applications is the strong thermal quenching of luminescence at temperatures above 150 K. As can be seen in Figure 2-9 the PL intensity quenches by 3 orders of magnitude between 12 and 150 K [60, 61]. This quenching has been explained in terms of two possible mechanisms: (i) the nonradiative de-excitation of

PAGE 37

16 Er in which the energy is transferred to a conduction electron (the so-called Auger de-excitation effect) [62] and (ii) back transfer of the Er excitation energy to the host via a defect level that was assumed to participate in the transfer of energy from the host to the Er 4f shell [58]. The latter defect level is apparently introduced in Er and O co-doped material after empirical optimization of the photoluminescence yield. To avoid the Auger quenching several groups investigated the possibility to excite electroluminescence (EL) in reverse biased p-n junctions [56, 62, 63]. Under such conditions, Er excitation can be achieved by highly inelastic impact with hot electrons (so-called impact excitation) injected to the Er doped region in the tunnel break down regime. It turned out that the temperature quenching in such a device is negligible. The cross sections for the impact excitation mechanism are, however, small and such devices have a rather low efficiency. Furthermore, the lack of luminescence quenching behavior is attributed to the reduced coupling between the Er states and the Si host, which prevents energy back transfer, the mechanism held responsible for isolated Er centers [58]. 2.2.2. Erbium in Silica Er-doped silica glass is finding more and more applications in photonic technology. For example, Er-doped fibers are used as the gain medium in long-distance optical fiber links, operating at 1.54 m. Er-doped silica channel waveguides find applications in planar optical amplifiers that are used in photonic integrated circuits [64], and Er-doped colloidal silica particles might be integrated with polymer technology to fabricate nano-composite waveguide materials [65]. As the quantum efficiency of the luminescent Er transition at 1.54 m can be quite high, Er-doped silica could also find applications in photonic crystals, as a probe of the local optical density of states [66].

PAGE 38

17 The presence of O is known to increase the luminescence from Er-related centers in Si [44]. It can be regarded as proven that O forms complexes with Er directly modify the structural, electrical, and optical properties of Er. Silica has enough oxygen to be incorporated with Er [67]. A technique to deposit thin Er-doped silica films is by using a wet chemical process, in which a solution containing a silica precursor and erbium ions is spin-coated on a substrate followed by a drying step (sol-gel process). Substantial research has been done on Er-doped sol-gel films [68-70]. In most cases, composite silica-based materials were studied or bulk glasses were made rather than thin films, but only a few papers focuses on pure silica sol-gel films. Film thickness (550-750 nm) and composition vary slightly with Er concentration in the range 0.05-1.0 at.%. The refractive index and atomic density of Er doped films is lower than those of pure SiO 2 . After annealing at 900C, room temperature photoluminescence at 1.54 m is observed with a luminescence decay as large as 10-12 ms, for Er concentration up to 1.0 at.%. Annealing was performed in air and led to shorter decay times, possibly caused by a higher O-H content [65]. The sol-gel process in combination with vacuum annealing is ideally suited to obtain Er-doped silica exhibiting long luminescence decay times [71]. 2.2.3. Erbium in Amorphous SiO x and Amorphous Silicon Recently, efficient photoluminescence (PL)[48] and electroluminescence (EL)[72] from erbium in amorphous hydrogenated silica (a-SiO x : H) and silicon (a-Si : H) was reported.

PAGE 39

18 The incorporation of Er 3+ into crystalline silicon (c-Si) has certain drawbacks such as a limited solubility (10 19 cm -3 ), a strong thermal quenching of the room temperature luminescence, and a need for co-doping with oxygen or carbon [47]. Using an amorphous material, especially one that contains large fractions of free oxygen like in amorphous hydrogenated silicon oxides (a-SiO x :H), as a host for erbium can help to reduce these problems. Er concentrations above 10 20 cm -3 can easily be incorporated into a-SiO x :H due to the flexibility of the amorphous network. Furthermore, the presence of electro-negative oxygen provides a strong local crystal field with low symmetry. This phenomenon enhances the PL by making the 4 I 13/2 4 I 15/2 transition partially allowed, thus reducing its lifetime. The optical gap of SiO x can easily be tuned by varying the oxygen content. Therefore, it should be possible to optimize the energy transfer from the host matrix to the luminescent Er ions. The larger band-gap of a-SiO x :H compared to c-Si and the correspondingly deeper localized states also help to reduce the strong thermal quenching known from c-Si [73]. SiO x has an enhanced Er solubility and variable oxygen contents provide favorable erbium environments and reduced excitation back transfer. As the Er PL does not show significant changes as a function of the oxygen concentration, it seems that even smaller amounts of oxygen already satisfy the requirement of an erbium surrounding with low symmetry. However, the biggest benefit of oxygen incorporation is the significant reduction of the thermal quenching of the Er PL, and of the SiO x PL as well, that is due to the widening of the band-gap and the broadening of the band. Optically excited electrons and holes can thermalize into deeper states, which, especially at room temperature, favor radiative recombination or energy transfer to Er and prevent reexcitation into the bands.

PAGE 40

19 The EL intensity of a-Si:H at the wavelength of 1.54 m corresponding to a radiative transition in the internal 4f-shell of the Er 3+ ion is low at 77 K but sharply increases at 220 K and exhibits a maximum near room temperature. Erbium electroluminescence dominated by a narrow line at 1.54 m (810 meV) is observed only in reverse bias when the excitation current through the Al/a-Si:H/n-type Si/Al structure exceeds a threshold value. Under such biasing conditions electrons are injected from the Al contact into the amorphous layer. The energy position of deep f-electron multiple states lying below the valence states of the semiconductor (by about 10 eV) is practically independent of the host matrix due to screening of the f electrons by the outer 5s 2 5p 6 electrons. Excitation energy can be transferred from electrons in the conduction band via Coulomb interactions. In the amorphous matrix there are, in principle, three possible ways of electronic excitation of the f states: (1) impact ionization by mobile carriers, (2) Auger excitation in a band-to-band recombination process, and (3) Auger processes in which a carrier is captured in a localized state in the forbidden gap (Defect Related Auger Excitation, DRAE). The energy released at this transition is transferred to an electron of the internal 4f-shell of the erbium ion exciting it from the ground 4 I 15/2 to the first excited 4 I 13/2 state [74]. DRAE process is reported to be responsible for the Er luminescence in a-Si:H. Energy band diagrams of electroluminescent structures [73] of a-Si:H are shown schematically in Figure 2-10 (a) and (b) for forward and reverse bias, respectively. When forward bias is applied to the structure, holes travel through the amorphous layer to the crystalline silicon substrate (Figure 2-10 (a)) and recombine in the n-type substrate radiatively (1.05 eV). The excess holes are not radiatively captured at defects in undoped

PAGE 41

20 a-Si:H. In the reverse bias mode, electrons can be injected from the aluminum contact into the amorphous layer and excite both erbiumand defect-related luminescence at the wavelength of 1.54 m. No erbium related luminescence feature is observed with the n-type substrate under forward bias [73]. 2.2.4. Erbium in Nanocrystals Containing SiO x Fujii et al. reported that the presence of Si nanocrystals in Er-doped SiO 2 considerably enhances the effective Er absorption cross-section [75]. Silicon nanocrystals were formed in SiO 2 using Si ion implantation followed by thermal annealing [76]. Upon 458 nm laser excitation, a broad nanocrystal-related luminescence peak centered around 750 nm and two sharp Er luminescence lines at 982 and 1536 nm are observed [76]. Recently, silicon-rich silicon oxide (SRSO) has been found suitable as a host material for erbium [77-79], because it contains sufficient free oxygen, and silicon nanocrystals, and has a wide energy gap. Er 3+ in ‘silicon-rich silica’ enhances the PL intensity relative to a stoichiometric silica host (Figure 2-11)[80, 81]. Si nanocrystals embedded in Si dioxide emit light in the visible range due to the recombination of confined excitons within the nanostructure [82, 83]. When rare-earth ions are also introduced in this matrix by a co-doping technique an interaction can occur between the nanocrystal and the rare-earth. The energy is preferentially transferred from the nanocrystal to the rare-earth, which subsequently de-excites radiatively [75, 77, 84]. An Ar-ion laser (488 nm) generates excitons within the Si nanocrystals. Excitons confined in the Si nanocrystal can either give their energy to an intrinsic luminescent center emitting at around 0.85 m or pass this energy to the rare-earth 4f shell, thus exciting the rare-earth ion [78].

PAGE 42

21 A possible excitation model is shown in Figure 2-12 (a), which depicts a schematic band diagram of SiO 2 containing a Si nanocrystal and Er 3+ [76]. First a photon is absorbed by the nanocrystal, which causes generation of an exciton, bound in the nanocrystal. This exciton can recombine radiatively, emitting a photon with an energy that depends on the nanocrystal size. If an Er 3+ ion is present close to the nanocrystal, the exciton can recombine non-radiatively by bringing Er into one of its excited states. The presence of rare-earth ions provides an alternative route for the energy transfer and, as the rare-earth concentration increases, it becomes the main transfer route. This is the mechanism for the decrease of the 0.85 m nanocrystal luminescence. Photoluminescence excitation spectra ascertained that the Er was not excited directly by optical absorption, but indirectly via an energy transfer process involving Si nanocrystals [77, 78, 84]. The exciton-Er energy transfer is fast (<10 -6 sec) and efficient (>60 %). The maximum number of excitable Er ions around a single nanocrystal is one or two, presumably due to an Auger de-excitation or a pair-induced quenching mechanism. However, Si nanocrystals are reported to be less effective in enhancing EL intensity than enhancing PL intensity [85]. The different roles of the Si nanocrystal in EL and PL processes may be responsible for this fact [85]. In the Er 3+ PL process of SRSO:Er films, almost all the photogenerated carriers stem from the Si nanocrystal, in other words the existence of Si nanocrystals affects PL intensity strongly. But in the Er 3+ EL process of SRSO:Er films, the Si nanocrystal-related Er 3+ excitation is just one of the various Er 3+ excitation paths.

PAGE 43

22 Nanocrystals act as very efficient sensitizers for rare-earth ions embedded within silicon dioxide. Moreover, since Auger and back-transfer non-radiative de-excitations can be suppressed, the overall process is also much more efficient than rare-earth luminescence within bulk crystalline silicon 2.2.5. Erbium in Porous Silicon The main PL peak (~1.54 m) of Er-doped porous Si is narrower than or nearly equal to those of Er-implanted SiO 2 [64], or Er-doped a-Si:H [48], though wider than those of Er-doped MBE Si [40] or Er implanted Si [58]. Er ions electrochemically incorporated into porous silicon layers are optically activated only by high temperature annealing in an oxygen-containing atmosphere. The incorporation of Er ions into pores of porous silicon layers is essential for the luminescence. Oxidized Er ions on the surface of crystalline silicon do not become optically active. Both a high temperature (~900C) and the presence of O 2 are essential for the optical activation. Thermal quenching of PL is very small and the intensity at room temperature is still about half of the intensity at 20 K. This temperature dependence is quite different from that of Er-implanted crystalline silicon as already explained in Section 2.2.1. [46]. Stable room-temperature electroluminescence at 1.54 m from erbium-doped porous silicon devices was observed in both forward and reverse bias conditions [86]. It was shown that there is an exponential dependence of the electroluminescence intensity as a function of the applied current and voltage in Figure 2-13. The Er doped p-Si held in forward bias breaks down more often and at lower voltages and is less efficient than the devices held in reverse bias. The stability of the Er doped p-Si showed minimal degradation even after several hours of operation. In reverse bias, the device displays a

PAGE 44

23 lower electroluminescence turn-on input power (25 mW), as well as, higher output power efficiencies. The inset of Figure 2-13 shows a typical electroluminescence spectrum in reverse bias. The line shape and peak position are almost identical for both biases, but the intensity is ~28 times stronger for reverse bias. The electroluminescence intensity is shown as a function of the annealing temperature as shown in Figure 2-14 [87]. The electroluminescence intensity reaches a maximum at an annealing temperature of 800C. Below 800C, the electroluminescence intensity is small due to the low concentration of optically active erbium ions. As the annealing temperature increases above 800C, the number of optically active erbium ions increases, but the transport properties of the device also deteriorate. It is possible that the deterioration is due to the increased oxidation and the suppression of transport channels that would otherwise lead to electroluminescence. Figure 2-15 compares normalized PL and EL spectra of erbium-doped porous silicon. The EL line shape and peak position in forward and reverse biases are almost identical, but compared to the PL they are shifted ~2 nm toward lower energies. The difference in position may be a result of optically and electrically exciting Er centers in different microenvironments within the active layer. In reverse bias, Er doped porous Si showed a large EL temperature quenching dependence that is different from forward bias EL and PL. The EL mechanism of Er doped porous Si in reverse bias was reported[46, 86, 87] as a result of excitation by hot electrons and the mechanism for excitation in forward bias occurring through an electron-hole mediated recombination process as for PL.

PAGE 45

24 2.3. Summary Rare earth metals, in particular erbium, have been used in the past to improve or modify the optical properties of various silicon-containing substrates. In particular, erbium-implanted crystalline-silicon, erbium-doped silica, erbium-embedded in nanocrystalline SiO x , and erbium-doped porous silicon have been targeted. The nanostructured matrix of spark-processed silicon makes this material an ideal host because its large surface area allows easy infiltration of foreign ions into the matrix. The spark-processed structure readily oxidizes, and thus, produces large concentrations of oxygen that are necessary for efficient erbium emission. Doping of spark-processed silicon can be achieved by co-spark-processing of erbium and silicon because it offers the advantages of deeper erbium penetration, lower cost, and simplicity of the process. In contrast, erbium-doped crystalline silicon structures are usually prepared by ion implantation, epitaxial growth, and chemical vapor deposition [88]. All of the latter techniques are expensive and time consuming, require specialized equipment, and are limited to very shallow doping profiles. On the other hand, using spark-processed silicon as the host and doping it with erbium by co-spark-processing produces deeper doping profiles and is simpler and more cost effective, which makes this material and technique appealing.

PAGE 46

25 Figure 2-1 Electroluminescence spectrum for sp-Si. Inset: Schematic of an electroluminescence device made from sp-Si. After [6].

PAGE 47

26 Figure 2-2 EL mechanism of spark-processed silicon. After [52].

PAGE 48

27 (a) (b) Figure 2-3 Electromicrographs of sp-Si. (a) TEM [12, 15] and area selected electron diffraction patterns. CR = crystalline, AM = amorphous. (b) Cross-sectional SEM of sp-Si [15].

PAGE 49

28 Figure 2-4 Raman shifts for c-Si (a) and sp-Si (b) ( exc = 488 nm, 400 W cm -2 spot size 60 m). The peak maximum in (b) is shifted by 7.1 cm -1 and the FWHM increased from 4 to 6.4 cm -1 [14].

PAGE 50

29 Figure 2-5 Schematic of spray spark-processing apparatus [53].

PAGE 51

30 Figure 2-6 EL intensity of conventional sp-Si and spray sp-Si [52].

PAGE 52

31 Figure 2-7 PL spectrum of undoped sp-Si in comparison to PL spectra of Si, which was spark-processed with europium, cerium, or terbium [50].

PAGE 53

32 Figure 2-8 Luminescence spectrum of Er-implanted and annealed Si measured at 20K. The half-width of the intense line at 1.53904 m (805.6 meV) is 0.422 meV [40].

PAGE 54

33 Figure 2-9 Arrhenius plot showing the temperature dependence of the 1.54 m intensity of erbium-implanted crystalline-silicon between 12 and 150 K (open circles) [60].

PAGE 55

34 Figure 2-10 Energy band diagrams for a-Si:H for EL: (a) at forward bias, (b) at reverse bias [73].

PAGE 56

35 Figure 2-11 PL as a function of excitation photon flux for Er-doped silica and Er-doped silicon-rich silica [80].

PAGE 57

36 Figure 2-12 Schematic representation of (a) Er excitation model, showing the electronic band structure of Si nanocrystal-doped SiO 2 and the Er 4f energy levels. An optically generated exciton (dotted line) confined in the nanocrystal can recombine and excite Er 3+ . (b) SiO 2 containing Er (crosses) and nanocrystals (bright circles). The nanocrystals that couple to Er (dark circles) show no exciton luminescence [76].

PAGE 58

37 Figure 2-13 EL intensity as a function of device input power for both forward () and reverse () bias conditions for erbium-doped porous silicon. Inset shows a typical EL spectrum in reverse bias [88].

PAGE 59

38 Figure 2-14 EL intensity of Er-doped porous Si as a function of annealing temperature [88].

PAGE 60

39 Figure 2-15 Normalized room temperature PL and EL comparison. The spectrum under forward bias was multiplied by a factor of 28 with respect to the reverse bias spectrum [88].

PAGE 61

CHAPTER 3 EXPERIMENTAL PROCEDURES 3.1. Device Fabrication Procedures 3.1.1. Wafer Preparation for Spark-Processing Samples were prepared on a 10.5 in 2 (2.541.27 cm 2 ) piece of 400 m thick, n-type, <100>, phosphorous-doped Czochralski grown Si (3-8 cm). The sample is first cleaned which is conducted in an ultrasonic cleaner: a fifteen minute TCE (trichloroethylene) wash, followed by a fifteen minute electronic grade acetone clean, followed by a fifteen minute electronic grade methanol clean, and finally followed by a fifteen minute DI (de-ionized) water rinse. The samples were then dried under N 2 atmosphere at room temperature. 3.1.2. Erbium on Wafer In the second step, an erbium layer was layered down on top of the surface oxidized silicon wafer utilizing a physical vapor deposition (PVD) system with ultimate vacuum capacity in 10 -9 Torr (1.3310 -7 Pa) range. A mechanical pump created a vacuum to 60 milli-Torr, and a turbo molecular drag pump took the system further into the 10 -5 Torr range. Then, utilizing a titanium sublimation pump concomitant with an ion pump established an operating pressure in the 10 -6 Torr range. At this point metal erbium from a high purity (99.9999%) source was deposited as a layer by evaporation. The deposited metal thickness was monitored using an Inficon XTC thickness monitor, which was 40

PAGE 62

41 mounted in the same plane as the samples. In this study, the samples had a 0.1 m thick metal erbium layer, unless otherwise stated. 3.1.3. Spark-Processing A high-frequency, high-voltage, and low current spark was provided by a spark generator. The spark generator consisted of the fly-back transformer of a cathode-ray tube, which typically provides about 10-15 kV, 14-16 kHz, and 5-10 mA of spark voltage, frequency, and current respectively. A tungsten tip was used as an anode with a spark gap, for most samples, of 2 mm. The above-mentioned silicon wafer under the erbium layer was used as the ground for the system. The spark from the tungsten-tip migrates across the silicon substrate. In most cases spark-processing was conducted at atmospheric pressure in air. Spark-processing was performed under various processing parameters, specifically processing time, processing pressure, and spark gap. 3.1.4. Rapid Thermal Annealing (RTA) The intensity of erbium-related luminescence depends on annealing temperature, annealing time, and incorporation of oxygen [86]. The role of oxygen in the activation of Er 3+ ions in silicon based materials has been shown to be important [44] as outlined in Chapter 2. Possibly, the Er 3+ ions are located within the SiO x matrix as well as at the interface of the nano-sized silicon grains. The SiO x coverage of the silicon grains is on the order of several monolayers and effective carrier transport has been observed and attributed to hot electrons [45]. Effective incorporation of oxygen involves rapid thermal processing of the samples in air (or in O 2 ) to activate and diffuse the erbium into the spark-processed silicon. An AG Associate 660 rapid thermal annealing furnace was utilized for this procedure. Various combinations of maximum temperatures and

PAGE 63

42 annealing time were carried out. In most cases, erbium-doped spark-processed silicon was treated in air, at 900 C for 15 minutes, unless otherwise stated. A completed erbium-doped spark-processed silicon sample is depicted in Figure 3-1 (a). 3.1.5. Electroluminescence Device Preparation 3.1.5.1. Ag and Al metal contacts For electroluminescence measurements, an additional metallization process for the contacts is required. As shown in Figure 3-1 (b), the back-side of the sample was roughened using a small grit silicon carbide sand paper and was etched the native oxide away using a commercial buffered HF solution. An aluminum Ohmic contact (50 nm) on the back-side of the wafer was then deposited by physical vapor deposition (PVD). Before this Al deposition, the back-side of the wafer (opposite from the sparked area) was partially masked utilizing an Al foil. After the metal deposition, the mask was removed and the undeposited area was used as a window to allow the electroluminescence light to escape. This procedure has the following rational: In previous work investigating visible EL emission a thin semitransparent silver layer has been deposited on top of the spark-processed area (inset of Figure 2-1). This layer is, however, not very effective, as pointed at in Chapter 2 and in [52, 53]. On the other hand, a thicker metal contact prevents any light emission which was created in the spark-processed area. For devices which operate in the IR, however, Si becomes transparent and light can be made to escape through the above-mentioned window at the bottom of the device. In this case the top metallization can be made thicker and is also used as a reflector towards the bottom. The top metallization (above the spark-processed area)

PAGE 64

43 consisted of a 200 nm silver layer, see Figure 3-1. The metal contacts were vapor deposited at a base pressure of 10 -6 Torr. The deposited metal thickness was monitored using an Inficon XTC thickness monitor. It was mounted in the same plane of the samples. A Tencor P-2 long scan surface profiler was additionally used for checking the accuracy of the crystal monitor. It was found that the two thickness measurements agreed within 97%. 3.1.5.2. ITO (Indium-Tin-Oxide) contact For replacement of the contacts on rare-earth doped spark-processed silicon, an optically transparent and electrically conductive layer was considered. Indium-tin-oxide (ITO) offers the possibility of thicker films and therefore improved coverage, while maintaining good optical transparency. ITO films about 200 nm thick were fabricated by pulsed laser deposition (PLD). An ITO target (99.99% purity) was loaded on the rotating stage in the deposition chamber and samples were placed 3 cm from the target. The system was pumped down to 10 -5 Torr using a conjunction of mechanical and turbo pumps. Then the system was backfilled with oxygen to 10 milli-Torr. A krypton fluoride (KrF) laser having an energy of 350 mJ and a 5 Hz frequency was utilized for the deposition which performed at 100C, at a rate of 140 /min. A Tencor P-2 long scan surface profiler was used for final film thickness measurement. The resultant ITO film had a resistivity of 3x10 -4 cm compared to a resistivity of Ag of 1.59x10 -6 cm.

PAGE 65

44 3.2. Characterizations 3.2.1. Structure and Compositional Analysis The luminescence of rare-earth doped spark-processed silicon is strongly linked to its properties including but not limited to microstructure, crystallinity, composition, and film density. The material characterization of rare-earth doped spark-processed silicon is essential for better understanding of the correlation between structures, properties, and luminescence of rare-earth doped spark-processed silicon. In this work, characterization was performed by a variety of methods. 3.2.1.1. X-ray diffraction (XRD) X-ray diffraction (XRD) analysis has been performed on the samples to examine the phase and structure of the rare-earth doped spark-processed silicon. XRD can also determine atomic arrangement in amorphous materials. XRD is the most sensitive technique for high-Z elements such as rare-earth elements, since the diffracted intensity from these elements is much larger than from low-Z elements such as silicon. The intensity of diffracted X-ray is measured as a function of the diffraction angle 2. A Philips APD 3720 XRD system was used in this study. 3.2.1.2. High-resolution transmission electron microscopy (HRTEM) A thin solid specimen is bombarded in vacuum with a highly-focused, monoenergetic beam of electrons. The beam is of sufficient energy to propagate through the specimen. A series of electromagnetic lenses magnifies the transmitted electron signal. Transmitted electrons form images from small regions of the sample that contain contrast due to several scattering mechanisms associated with interactions between electrons and the atomic constituents of the sample. In this study, high resolution

PAGE 66

45 transmission electron microscopy was used to identify the nanometer size particles in the rare-earth doped spark-processed silicon matrix. A JEOL TEM 2010F was used in this study. A one minute spark-time was used for fabricating the rare-earth doped spark-processed silicon sample which represents the average sample in this study. The sample was thinned utilizing a Dual Beam Focused Ion Beam (FIB) Strata DB 235 to a resulting specimen thickness of approximately 100 to 200 nm. Figure 3-2 shows the actual image of the sample prepared by FIB cutting. 3.2.1.3. Scanning electron microscopy (SEM) Scanning electron microscopy was utilized to obtain high magnification images of the rare-earth doped spark-processed silicon surface and cross-section. A JEOL JSM 6400 and a field emission SEM JEOL JSM-6335F for higher magnification were used in this study. The SEM analysis was performed for surface morphology of the rare-earth doped spark-processed silicon. 3.2.1.4. Compositional analysis Energy dispersive spectroscopy (EDS), Tracor system II spectrometer, allows the quick evaluation of the composition in rare-earth doped spark-processed silicon layers. X-ray photoelectron spectroscopy (XPS) was used to measure the elemental concentration ratio. In particular, the chemical bonding nature was analyzed by XPS. Secondary ion mass spectroscopy (SIMS) was used to confirm the existence of rare earth in the spark-processed silicon and also to measure rare-earths and other elements concentration depth profiles.

PAGE 67

46 3.2.2. Optical Characterizations 3.2.2.1. Differential reflection spectroscopy (DR) Differential reflection spectroscopy is a first derivative technique that provides optical absorption spectra. The theory underlying this technique is given in [89] where it is shown that the peaks in a differential reflectogram correspond to the energies of interband transitions in the sample. The basic concept is that the first derivative of the reflectivity R is proportional to the imaginary part of the dielectric constant, 2 , which characterizes the energy absorption in materials. Figure 3-3 is a schematic representation of a differential reflection spectrometer. One measures the normalized difference between the reflectivities of two slightly different areas of the sample as a function of photon energy. Unpolarized, monochromatic light of continuously varying wavelength (from the IR to the near UV) is alternately deflected to one, then to the other area of a sample by means of a vibrating mirror. The reflected light is detected by a photo multiplier tube (PMT), and the signal thus obtained is electronically processed to yield R/R avg . This ratio is defined as R=R 1 -R 2 that is, the difference in reflectivity between the two areas of the sample and R avg =(R 1 +R 2 )/2 which is the average reflectivity. The electronics of the system facilitates a continuous plot of R/R avg versus the wavelength of the light probe, thereby generating a ‘differential reflectogram’. The characteristic peaks in a differential reflectogram represent the energies which electrons absorb from the probing medium, resulting in their excitation from lower filled energy states into upper, unfilled states. Thus this technique essentially yields optical absorption spectra. It should be noted that simultaneously measuring R 1 and R 2 and forming the ratio R/R avg eliminate possible influences due to

PAGE 68

47 the spectral sensitivity of the PMT, and the spectral reflectivity of the mirrors. Likewise, fluctuations of the line voltage and variations in the spectral output of the light source are removed, resulting essentially noiseless spectrum. 3.2.2.2. Photoluminescence (PL) When the rare-earth doped spark-processed silicon layer is exposed to light of a certain wavelength, an electron of a luminescent center in the material is excited to a higher energy state before it relaxes back to the ground state, by emitting a photon. This process is called photoluminescence (PL) because the excitation source is light. Photoluminescence can be used to characterize both the transitions leading to excitation as well as emission. Figure 3-4 represents a typical photoluminescence spectrometer set-up. In the present study, an Ar ion laser emitting at 488 nm was used to excite the rare-earth doped spark-processed silicon layers. After passing through a narrow band-pass filter, and an iris, which are used to remove the laser sidebands and stray light respectively, the laser beam was focused onto the sample using a quartz lens. Samples were mounted on a XYZ positioner. The light emission by PL was collected by a lens array system and focused onto the entrance slit of a monochromator. For the infrared range PL measurement, a Jobin Yvon TRIAX 550 IR monochromator having a grating with blaze angle of 830 nm was used. A long-pass filter with 50% transmission cutoff wavelength at 520 nm for an Ar ion laser was placed at the entrance to the monochromator in order to prevent any scattered laser light from entering. After the PL emission traversed the monochromator, the light impinged upon a TE cooled (-30C) Electro Optical Systems IGA5 InGaAs detector connected to a Stanford Research Systems SR810 lock-in amplifier coupled to a light chopper. The response characteristics

PAGE 69

48 of the detector (as obtained by the manufacturer) are shown in Figure 3-5. For the PL measurements in the visible range (see Appendix), a 325 nm HeCd laser, and an “Instruments SA, Inc. HR-320 monochromator” with a grating of blaze angle of 660 nm coupled with a Hamamatsu R943-02 GaAs photomultiplier tube were used. 3.2.2.3. Electroluminescence (EL) When a rare-earth-doped spark-processed silicon layer is exposed to an electric field, the electric field excites electrons in this layer, which then emit the excess energy as photons. This process is called electroluminescence (EL) because the excitation source is an electric field. In this study, a pulsed direct current (dc) square wave having a frequency of 200 Hz and a 30% duty cycle has been used. It was generated by a combination of function generator and dc power supply as a voltage source. Measurements of the EL spectra of rare-earth doped spark-processed silicon were conducted using the infrared spectrometer system discussed in the PL section, while the low-pass filter at the entrance to the monochromator was removed. 3.2.3. Electrical and Thermal Characterizations 3.2.3.1. Determination of efficiency of the devices In the following sections the external power efficiency will be given. This is determined by measuring the input power (i.e. applied voltage times current through the sample) and divided by the light output power. In order to obtain the latter information, a calibrated light source was utilized, which was placed at a given position specified by the manufacturer (50 cm between light source and 1cm 2 opening, driven by a 1 amp calibrated power supply). The calibration curve provided by the manufacturer plots the “irradiance” given in mW/cm 2 nm versus the wavelength. This plot allows converting

PAGE 70

49 “arbitrary units” (in volts) into irradiance units. In our case the emission curves of the PL or EL devices were then integrated between 1300 and 1650 nm (the sensitive range of our detector). Finally the active area of the PL or EL device was measured and was set in relation to 1 cm 2 . In actual calibrated curve is shown in Figure 5-43. 3.2.3.2. I-V measurement In this study, a dc voltage source was used for current-voltage characterization. The I-V characteristics of rare-earth doped spark-processed silicon were obtained by connecting a digital multimeter for current readings in series with the sample and a dc power supply, and another digital multimeter for voltage readings across the sample. 3.2.3.3. Temperature effects The effects of elevated temperature on PL and EL intensity were investigated by mounting the sample on a variable power heating stage, which could be controlled by an applied voltage during measurements. The temperatures of the sample surface were obtained by a “Fluke 5K/J” digital thermometer in connection with a chromel-alumel thermocouple. Dry ice and a peltier cooler were utilized when the effect of low temperatures was to be studied for EL. An APD Cryogenics HC-2 was utilized for temperatures down to the liquid helium temperature (4 K) and up to room temperature (300 K).

PAGE 71

50 (a) (b) Figure 3-1 Schematic representation of (a) erbium-doped spark-processed silicon and (b) a typical EL device utilizing erbium-doped spark-processed silicon. The light escapes through the window at the “bottom” of the device. The Ag contact reflects the EL light back into the device and towards the window.

PAGE 72

51 Figure 3-2 Top-view of focused ion beam (FIB) cut erbium-doped spark-processed silicon.

PAGE 73

52 Figure 3-3 Schematic representation of differential reflectometer used in this study. After [90]

PAGE 74

53 (a) (b) Figure 3-4 Schematic diagrams of (a) an infrared photoluminescence spectrometer and (b) a visible photoluminescence spectrometer.

PAGE 75

54 Figure 3-5 Typical spectral response curves for the InGaAs detector.

PAGE 76

CHAPTER 4 MATERIALS CHARACTERIZATION Those readers who are mostly interested in the PL and EL results may skip these sections and turn to Chapter 5. This chapter describes several properties of erbium-doped spark-processed silicon as found during this study. Absorption spectra obtained by differential reflectometer provide insight into the electronic energy structure of erbium doped spark processed silicon and also provides a suggestion of selecting the most efficient excitation light source for the photoluminescence measurement. X-ray diffraction analysis yields information about the structural change occurring during the fabrication process, especially heat treatment, and the activation process. The high resolution transmission electron microscopy was carried out to study the microstructure of erbium-doped spark-processed silicon. The scanning electron microscopy provides images of surface and cross-sectional morphology and dimensions. Compositional research is accomplished by energy dispersive spectroscopy and electron microprobe analysis. X-ray photoelectron spectroscopy provides surface concentration and bonding information. Trace element analysis and existence of rare-earth constituents in the spark-processed area is performed by secondary ion mass spectroscopy. 4.1. Differential Reflectometry Visible optical absorption spectra were obtained by differential reflectometry. Samples were prepared for examining the absorption changes related to the Er activation 55

PAGE 77

56 process in the spark-processed silicon matrix. Figure 4-1 represents the spectrum from 200 to 800 nm (6.2 to 1.5 eV), obtained from a sample that had been spark processed for 1 minute in stagnant air, and which initially had 100 nm of erbium metal on the silicon wafer. In the UV range, an absorption band from 200 to 280 nm (6.2 to 4.4 eV) is obtained that has a shoulder at 215 nm (5.8 eV), and peaks at 228 nm (5.6 eV), 255 nm (4.8 eV), and 271 nm (4.6 eV). Other small absorption peaks near 443 nm (2.8 eV), 532 nm (2.3 eV), and 600 nm (2.1 eV) are also discernible. Depicted in Figure 4-2 is the optical absorption spectrum for the same sample which has been subsequently annealed in air at 900C for 15 minutes. At first glance it is observed that the formerly weak peaks above 300 nm are now substantially enhanced. As before; an absorption band from 200 to 240 nm (6.2 to 5.2 eV) with a shoulder at 215 nm (5.7 eV) and a peak at 228 nm (5.6 eV) are observed. These peaks are known to originate from pure silicon [91]. Additional absorption peaks at 260 nm (4.77 eV) and 278 nm (4.46 eV) are also observed. The 215 nm (5.7 eV) and 260 nm (4.77 eV) absorption bands have been reported to originate from normal spark-processed silicon [6]. These peaks are common to both absorption spectra. In addition to pure silicon and spark-processed silicon absorption peaks, peaks at 372 nm (3.4 eV), 386 nm (3.2 eV), 415 nm (3.0 eV), 458 nm (2.7 eV), 496 nm (2.5 eV), 526 nm (2.35 eV), 550 nm (2.25 eV), 658 nm (1.9 eV), and 800 nm (1.55 eV) are detected. The energy level diagram of the Er 3+ ion and its intrashell transitions are shown in Figure 4-3. This also indicates that erbium in the spark-processed silicon matrix is activated and exists in the Er 3+ ionic state. Each absorption peak has the same energy value as an Er 3+ energy level in the visible range. Unlike absorption data obtained by photoluminescence excitation measurement, differential reflectometry can measure all absorption levels from

PAGE 78

57 a wide range of excitation energies. In other words all possible absorption levels in erbium-doped spark-processed silicon can be and are reported in Figure 4-2. It is concluded from these measurements that annealing (for example at 900C for 15 minutes) substantially activates the Er 3+ levels for spark-processed Si. This observation is paramount when considering EL and PL devices. 4.2. X-Ray Diffraction Measurement X-ray diffraction measurement was used to investigate the elemental distribution of erbium in spark-processed silicon. Samples were prepared by spark-processing a Si/100 nm erbium sample under stagnant air condition for 1 minute with subsequent annealing in air for 15 minutes at different annealing temperatures. As mentioned above (Section 4.1.) annealing favors the distribution and activation of erbium in the host matrix of spark-processed silicon. Figure 4-4 illustrates the changes of the Er-Si-N-O distribution with annealing temperature. The post-annealing XRD spectrum of erbium-doped spark-processed silicon shows the presence of Er 4 Si 2 N 2 O 7 and an Er-O phase with the Er 2 O 3 (222) structure as the most dominant complex. It also reveals the presence of an Er-Si compound whose best-fit structure is Er 5 Si 3 (112) (Figure 4-5). In contrast, a phase of Er 2 Si 3 N 4 O 3 and metal erbium (101) was dominant in the XRD spectrum of samples, before annealing. With increasing annealing temperature, a redistribution of Er in spark-processed silicon can be attained. In a solid host matrix in which O and Er are present, Er accepts O atoms forming Er-O complexes with various structures, such as ErO 4 , ErO 6 , or Er 2 O 3 [57, 92-94]. The crystal field splitting of the 4f states is influenced by the presence of oxygen in various coordination configurations.

PAGE 79

58 Figure 4-6 exhibits the Er 2 O 3 (222) peak (at 2=29.64 o ) intensity as a function of annealing temperature. An initially rapid increase can be observed which levels off above about 1000C. This observation indicates that Er 2 O 3 (222) is increasingly formed by heat treatment up to a certain annealing temperature which coincides with the luminescence variation of erbium-doped spark-processed silicon with annealing temperature. 4.3. High Resolution Transmission Electron Microscopy During spark-processing and rapid thermal annealing, a nanometer size crystalline phase was expected to precipitate in the matrix of erbium-doped spark-processed silicon. High resolution TEM was performed to elucidate this point. A sample was prepared by spark-processing a 100 nm erbium layer deposited on a silicon wafer under stagnant air for 1 minute and then annealing it in air for 15 minutes at 900C. Cross-sectional TEM images were prepared. In Figure 4-7 (a), the HRTEM image shows that nc-Si is identifiable in the sample with diameters in the nanometer range that are imbedded in a SiO x matrix. Note that the nano-sized particles are observed even at short (1 minute) spark-processing times. The electron diffraction patterns taken from this area show diffraction circles of Si polycrystals (Figure 4-7 (b)) and diffraction spots of Si single crystals (Figure 4-7 (c)). It is estimated that the size of the Si nanocrystals ranges from 2-6 nm in diameter with an average diameter of approximately 4.5 nm. The diffraction pattern of nc-Si corresponds to the (100) plane of silicon. It is postulated that these silicon crystal particles are formed during spark-processing or during the RTA process. A high angle annular dark-field (HAADF) image is shown in Figure 4-8 (a), which is compared with a conventional bright field image (Figure 4-8 (b)). The dark areas indicate a high-Z element, such as erbium, and the bright areas indicate a low-Z element,

PAGE 80

59 such as silicon. It is concluded from these images that large clusters of Er metal are still discernible even after annealing the spark-processed Si. These clusters are estimated from Figure 4-8 to be about 80 nm in diameter that is, they are at least one order of magnitude larger than the Si nanocrystals seen in Figure 4-7 (a). 4.4. Scanning Electron Microscopy and Energy Dispersive Spectroscopy SEM was performed to characterize the surface morphology of erbium-doped spark-processed silicon. A spark time of 1 minute in air was utilized for this experiment. The spark-processed area is characterized by large pits probably caused by the flash evaporation process. The flash evaporation of the erbium/silicon surface is a reaction of the erbium and silicon plume with oxygen and nitrogen products that form globule-like particles onto the surfaces during the off time of the spark. At higher magnification the surface is observed to be extremely rough and pitted, with globules of condensed and redeposited material randomly arranged. The density of the pitted area is decreased in the halo area. The formed globules in the halo region range in size from a few microns to sub-micron dimensions. To obtain a semi-quantitative chemical composition of the erbium incorporation along with silicon and oxygen, energy dispersive spectroscopy (EDS) was carried out on the spark-processed area of the sample. The center of the spark-processed area in Figure 4-9 yields an EDS spectrum as depicted in Figure 4-10 (a). It demonstrates that silicon and oxygen are dominant in this region, and that erbium is dispersed in the spark-processed silicon. Measuring from the center towards the halo side, the EDS spectra show that the erbium concentration is increasing. This is because during the spark-processing, flash evaporation of erbium and silicon was occurring more on the center

PAGE 81

60 than the periphery region of the spark processed area. A more detailed chemical analysis has been performed applying additional methods, see below. 4.5. X-Ray Photoelectron Spectroscopy XPS was carried out using Mg K radiation (1253.6 eV) as a photon source. A pass energy of 89.45 eV was used for all scans, giving an overall resolution of 0.5 eV. The acquisition was performed at a fixed angle of 45 from the surface. The XPS data were corrected using binding energy values of the C 1s line, because binding energies are typically shifted during the measurements due to sample charging. The peaks observed for different atomic species were Si 2p, Er 4d and O 1s. Figure 4-11 shows the XPS spectra of Si 2p, Er 4d and O 1s for the sample before and after annealing at 1100C for 15 minutes. The ratio of the Er, Si, and O concentrations on the surface are calculated by the atomic sensitivity factor method. Erbium atoms located in the silicon oxide layer exist in the form of isolated erbium atoms or as Er 2 O 3 . The erbium 4d peak concentration is seen to be about 5.4 at%. The erbium concentration is assumed to be distributed evenly in the spark-processed area. Figure 4-12 (a) depicts the silicon 2p peaks as a function of annealing temperature. The silicon 2p peaks are observed at the binding energy near 102.1 eV, which indicates that sample contains SiO x :Si bonds. Among all the erbium peaks, the 4d signal was the most ideal one because it is located far enough from other characteristic lines. The erbium 4d binding energy does not change much with annealing temperature as seen in Figure 4-12 (b). The erbium 4d binding energy is about 168.5 – 168.9 eV, which is known to be an Er-O bond having the published binding energy of 168.7 eV. Also in the O 1s peak, it is possible to distinguish (Figure 4-12 (c)) different components having

PAGE 82

61 binding energies of 529.1 and 531.5 eV. The oxygen 1s peak does change with annealing temperature. Before annealing, Oxygen 1s has only one peak at the binding energy of 531.5 eV which indicates Er-Si-O bonds. After annealing an additional binding energy peak feature is observed at 529.1 eV which is known to be related to the Er-O bond. It can be postulated that an Er-O bond is created and increases by annealing with the Er-Si-O bond which is originally formed during spark-processing in the erbium-doped spark-processed silicon structure. A known SiO 2 bond is located at the binding energy of 532 eV which is higher than what is observed in this study. This is because spark-processing is not creating SiO 2 but SiO x :Si. It can be suggested that an Er-Si-O bond is formed during spark-processing and during heat treatment. Erbium is incorporated into the Er-Si-O structure in the form of Er 2 O 3 or Er-O for erbium-doped spark-processed silicon. 4.6. Secondary Ion Mass Spectrometry Secondary Ion Mass Spectrometry (SIMS) analysis was performed to detect the existence of erbium and verify the form of the erbium compound present in the spark-processed area. A beam of primary ions with energies ranging from 1 to 10 keV is used to sputter the sample surface. An oxygen primary ion beam is used to sputter the electropositive elements and a cesium primary ion beam for the electronegative elements. Positive and negative secondary ions formed during the sputtering process are extracted from the sample and analyzed by a quadrupole mass spectrometer. The samples were as usual prepared by spark-processing the 100 nm erbium deposited silicon under stagnant air condition for 1 minute and subsequently annealing them in air for 15 minutes at 900C. Figure 4-13 (a) was obtained by applying a 5 keV cesium primary ion beam directed towards the spark-processed area. Er at 166 AMU, ErO at 179.6, 180.6, 181.5,

PAGE 83

62 and 183.4 AMU and ErSi at 194.7, 195.6, 196.6, and 198.4 AMU were found in the spectrum. Figure 4-13 (b) is acquired with a 6 keV oxygen primary ion beam from the perimeter of the spark-processed area. Er at 164, 165, 166, and 168 AMU, ErO at 180, 181, 182, and 184 AMU and ErSi at 195, 196, 197, 198, and 199 AMU were found in the spectrum. The main difference of the two sets of measurements is that one stems from the spark-processed region and the other is not processed. From the spark-processed region only trace amounts of erbium metal (166 AMU) were found, whereas from the outside of the spark-processed region, an intense erbium metal peak was found. Both areas show ErO peaks between 179 to 185 AMU. ErSi peaks were more distinguishable from the spectrum of the spark-processed area than that outside of the spark processed area. In order to distinguish the Er-O-Si complex in the spark-processed region, a SIMS depth profile analysis was collected (Figure 4-14). The depth profile from the spark-processed area shows that ErO is formed at the surface which initially increases with depth after which its amount decreases again with depth. Erbium and silicon have a lower concentration at the free surface but eventually start to increase for higher depth values after which the Si concentration remains constant when the Er concentration decreases with depth. The Er, ErO, and SiO show maximal amounts near the interface region between erbium-doped spark-processed silicon and unprocessed silicon. It is interesting to note that the SiO profile has the same trend as the Er profile. The depth profile from the unprocessed area shows that the Er profile is uniform at the interface with the silicon wafer and there is a clear distinguishable interface between the two regions. From these two profile data, it can be postulated that the erbium at the free surface is evaporated during the spark-processing and forms an Er-O compound during re-deposition. Likewise

PAGE 84

63 silicon is mixing with Er and O during spark-processing and by diffusion during annealing.

PAGE 85

64 Figure 4-1 Optical absorption spectrum obtained from 200 to 800 nm by differential reflectometry for silicon on which a 100nm thick erbium layer has been deposited before spark-processing (Stagnant air condition). Spark-processing time was 1 minute. No subsequent annealing.

PAGE 86

65 Figure 4-2 Optical absorption spectrum from 200 to 800 nm obtained by differential reflectometry for erbium-doped spark-processed silicon prepared in stagnant air condition for 1 minute and subsequently annealed at 900C in air for 15 minutes. Compare to Figure 4-1.

PAGE 87

66 Figure 4-3 Energy level diagram of Er3+ ion and the intrashell transitions observed by excitation with an electron beam and photons of 325 nm (HeCd laser) and of 488 nm (Ar-ion laser). After [92]

PAGE 88

67 2030405060 1100oC1000oC900oC800oC700oCIntensity (a.u)2(Degrees)As deposited Figure 4-4 XRD spectra of Er-Si-N-O distribution with various annealing temperatures.

PAGE 89

68 (a) as deposited (b) annealed at 700 o C Figure 4-5 Detailed XRD for erbium-doped spark-processed silicon (a) as deposited (un-activated) and (b) annealed at 700C (activated).

PAGE 90

69 0600700800900100011001200 102103 XRD Intensity (a.u.)Annealing Temperature (oC)Er2O3 (222)2 Figure 4-6 An Er 2 O 3 (222) XRD peak (at 2=29.64 o ) intensity changes with annealing temperature.

PAGE 91

70 (a) (b) (c) Figure 4-7 HRTEM images of (a) silicon nanocrystals embedded in erbium-doped spark-processed silicon matrix, (b) electron diffraction patterns of silicon polycrystals and (c) diffraction spots of silicon single crystals.

PAGE 92

71 (a) (b) Figure 4-8 STEM images of (a) High angle annular dark-field (HAADF) image: dark areas indicate high-Z elements (Erbium) and bright areas indicate low-Z element (Silicon), (b) bright-filed STEM.

PAGE 93

72 (a) (b) (c) Figure 4-9 Scanning electron micrograph of erbium-doped spark-processed silicon, (a) full-view of surface morphology, (b) higher magnification of center area and (c) halo area image of erbium-doped spark-processed silicon.

PAGE 94

73 (a) (b) (c) Figure 4-10 Energy dispersive spectroscopy of erbium-doped spark-processed silicon. (a) EDS measured at the center of erbium-doped spark-processed silicon, (b) EDS measured half way to the edge of the spark-processed area from the center, (c) EDS measured in the halo area. Note that the thin lines indicate the positions of certain elements.

PAGE 95

74 (a) XPS of as-prepared erbium doped spark-processed silicon Binding Energy (eV)N(E)Min: 5370Max: 353287 1000900 800 700 600 500 400 300 200 100 0 Si 2p38.0 % Si 2s Er 4d2.6 % C 1s50.6 % Er 4s O 1s35.7 % Cr 2p31.0 % Fe 2p32.2 % Fe 2p1 O KVV C KVV (b) XPS of erbium-doped spark-processed silicon after 1100C annealing Binding Energy (eV)N(E)Min: 6650Max: 327663 1000900 800 700 600 500 400 300 200 100 0 Si 2p37.1 % Er 4d5.4 % C 1s38.4 % Er 4p O 1s44.5 % Cr 2p30.6 % F 1s1.8 % Fe 2p32.1 % O KVV C KVV Figure 4-11 XPS spectra of (a) an as-prepared erbium-doped spark-processed silicon, and (b) same, but annealed at 1100C for 15 minutes.

PAGE 96

75 (a) -1000600700800900100011001200101.2101.4101.6101.8102.0102.2102.4 Binding Energy (eV)Annealing Temperature (oC) Silicon (b) 0600700800900100011001200167.5168.0168.5169.0169.5 Erbium Binding Energy (eV)Annealing Temperature (oC) Figure 4-12 The XPS binding energies peak positions as a function of annealing temperature. (a) Silicon Binding Energy changes after Annealing, (b) Erbium Binding Energy changes after Annealing, and (c) Oxygen Binding Energy changes after Annealing.

PAGE 97

76 (c) -1000600700800900100011001200528.4528.8529.2529.6530.0530.4530.8531.2531.6 Binding Energy (eV)Annealing Temperature (oC) O-Si bond O-Er bond Figure 4-12 Continued.

PAGE 98

77 (a) 160170180190200210100101102103 198.4196.6195.6194.7186.2183.4181.5Log CpsMass (AMU)Er 166179.6180.6ErOErSi (b) 160170180190200210100101102103104 187182180160162168165ErErO164178166184185Log CpsMass (AMU) Figure 4-13 SIMS spectra for annealed erbium-doped spark-processed silicon (a) acquired with a 5 keV cesium primary ion beam from the spark-processed area, (b) acquired with a 6 keV oxygen primary ion beam from the perimeter of the spark-processed area.

PAGE 99

78 020040060080010001200 100101102103104105106 Region I (Surface)Region II Log Cps Sputtering time (s)Er 166CsSiO 177CsSi 161ErO 182 Region III Figure 4-14 SIMS depth profile of typical erbium-doped spark-processed silicon.

PAGE 100

CHAPTER 5 RESULTS AND DISCUSSIONS 5.1. Infrared Photoluminescence of Spark-Processed Silicon As lined out in previous chapters, spark-processing is a treatment that transforms silicon into a material with light emitting properties in the visible range. So far, light emission of spark-processed silicon in the infrared range has not been studied. This is the goal of the present investigations. Specifically, the spark-processing conditions that result in the brightest infrared photoluminescence are to be explored. Figure 5-1 depicts the infrared photoluminescence spectrum employing a p-type (B-doped, 5 cm) Si wafer that was spark-processed for 15 minutes using a 1.2 mm gap between a tungsten tip and the Si wafer under stagnant air conditions. A continuous wave 488nm Ar ion laser was utilized as the excitation source. The photoluminescence spectrum was measured from 700 to 1800 nm, with 5 nm increments and an integration time (time over which the signal is collected per data point) of 3 seconds was utilized. A broad, strong infrared photoluminescence peak in the 945 nm (1.31 eV) spectral range is observed. The 945 nm photoluminescence emission is different than what is commonly reported for anodic-etched, porous silicon and other silicon based materials with light emitting features. This phenomenon will be discussed in the following chapter. 5.1.1. Infrared Photoluminescence of Spark-Processed Silicon After Annealing Figure 5-2 depicts the change of the infrared photoluminescence spectra for pure spark-processed silicon which was annealed in air for 30 minutes at the temperatures 79

PAGE 101

80 given on the curves. It is observed that the peak intensities increase with increasing temperatures in the temperature range between 350 and 500C (Figure 5-3). This feature is commonly attributed to the conversion of chemisorbed water to oxide-related silanol-type groups (Si-O-H), which are also known to contribute to visible PL enhancement of spark-processed silicon after 450C annealing [50, 95]. At temperatures above 450C silanol desorbs or decomposes, eventually resulting in a dehydrogenated surface. At lower temperatures (<350C), the Si-OH groups are covered by water and therefore inefficient for PL, whereas after annealing at higher temperatures (>600C) they are no longer present. Accordingly, the PL intensity decreases after peaking near 500C annealing, as seen in Figure 5-3. Eventually, the infrared PL band increases again significantly in intensity after annealing above 600C and essentially shows a PL intensity maximum when annealed near 900C. Figure 5-4 shows the variation of the peak wavelength due to annealing. It is observed that this peak initially increases from 945 to 1030 nm with increasing annealing temperature up to about 500C after which it decreases again to the room temperature value. This observation is different from that for the visible PL of spark processed silicon whose peak position stays essentially constant after various annealings. Furthermore, the PL spectrum near its peak position becomes asymmetrical after annealing, as shown in Figure 5-2. It can be presumed that the 945 nm infrared PL emission band is caused by more than one mechanism. In order to investigate this hypothesis further a peak deconvolution was performed assuming 3 major peaks such as near at 945, 1010, and, 877 nm, see Figure 5-5. The variation of two of these peaks (945 nm and 1010 nm) with respect to

PAGE 102

81 peak intensity and peak wavelength were then again plotted as a function of the annealing temperature (Figures 5-6 and 5-7). Specifically, Figure 5-6 presents the infrared PL peak intensity as a function of annealing temperature. Both peak intensities show a similar trend for annealings between 300C and 600C, as observed in Figure 5-3. This result suggests, as before, that both PL peaks (945 nm and 1010 nm) are affected by silanol-type groups. It is worth noting that the 1010 nm peak has a higher IR PL intensity after the 450C annealing and a lower PL intensity after the 900C anneal compared to the equivalent 945 nm peaks. This may indicate that the 1010 nm peak is more affected by silanol than the 945 nm peak. The variations of the deconvoluted 945 and 1010 nm peak wavelengths are shown in Figure 5-7. The fluctuation of the wavelength of the peak maxima after annealing for both of the deconvoluted peaks is quite small (2~3 nm). This result is consistent with the behavior of the visible PL peak position change of spark-processed silicon. The infrared PL intensity of spark-processed silicon as a function of annealing time is depicted in Figure 5-8. Samples used in this experiment were prepared applying 1 minute spark-processing under stagnant air. Sets of these samples were subjected to annealing in air at 900C for various times. No significant intensity change with annealing time was observed. 5.1.2. Variation of Parameters In order to investigate the infrared PL response of spark-processed silicon with spark-processing time, a series of pertinent experiments were conducted. It was visually observed that a longer spark-processing time caused a larger affected area. Figure 5-9 depicts the infrared PL intensity as a function of spark-processing time for spark

PAGE 103

82 processed silicon. PL intensities were observed to increase with increasing spark-processing time. The infrared PL intensity increases abruptly after a threshold of about 15 minutes of spark-processing time to around 10 times its original intensity. To investigate this further, EDS and SEM were utilized to examine the differences between various pertinent samples. The samples were spark-processed for 10 seconds, 15 minutes and 1 hour respectively. Figure 5-10 shows that as spark-processing time increases, the size and number of the globule-like surface features and holes (pores) increases. EDS spectra in Figure 5-11 indicate that the relative O to Si increases as the spark-processing time increases. This is interpreted to mean that the thickness of the SiO 2 /SiO x surface layer increases with spark-processing time. Figure 5-12 depicts the IR PL intensity of spark-processed silicon as a function of the gap between tungsten tip and silicon substrate. It is observed that the infrared PL intensity decreases with increasing spark-processing gap. Specifically, the most efficient infrared PL was achieved for a small spark-processing gap distance of 1 mm. In order to investigate the infrared PL dependency with excitation energy, various emission lines of the Argon ion laser were utilized, specifically the 457, 488, and the 515 nm lines (Figure 5-13). The pertinent experiments were conducted under the same output power of 15 mW for each laser line. It is observed that the peak wavelength positions are nearly identical with the change of excitation wavelength. Furthermore, a gradual decrease in intensity with laser wavelength is observed (inset of Figure 5-13). It is concluded that the lowest excitation wavelength of the Ar ion laser (488 nm in our case) yields the largest IR PL output.

PAGE 104

83 5.2. Infrared Photoluminescence of Erbium-Doped Spark-Processed Silicon As can be seen from the results above, pure spark-processed silicon does not photoluminescence at 1540 nm which is, as outlined above, the preferred wavelength for telecommunication purposes. Erbium doping of spark-processed silicon may shift the emission wavelength into this critical range. Thus, the infrared photoluminescence of erbium-doped spark-processed silicon was investigated. As outlined in Section 3.1.3. spark-processing causes high energy ablation of the silicon surface, which is additionally oxidized. When present on the surface, erbium centers can become trapped in the local condensed layer during spark-processing. As before, an Ar ion laser was used as an excitation source since its emission is resonant with the 4 I 15/2 4 F 7/2 absorption transitions in Er 3+ . Depending on the exact shape of the absorption spectrum, which varies with the host materials, a specific laser line (e.g. 476, 488, 496, 502, or 515 nm) was used by others to obtain the optimum emission [44]. In the present case the Ar ion laser excitation was performed using near normal incidence of the laser beam to the sample surface. Further, in most of the cases the Ar laser 488 nm emission line was utilized as an excitation source. In all experiments, a standard single-grating monochromator was used to disperse the luminescence light onto a detector which consisted of InGaAs and which is sensitive in the 0.7-1.8 m wavelength range. First of all, in order to investigate if erbium on silicon without spark-processing may have specific emission characteristics in the infrared range, the following control experiment was performed. 0.1 m of erbium metal was vapor deposited onto a silicon wafer (n-type, 5 cm) with subsequent rapid thermal annealing (RTA) at 900C in air for 15 minutes. No infrared PL emission was observed from this control sample as seen

PAGE 105

84 in Figure 5-14. In other words, the infrared PL emission of erbium-doped spark-processed silicon which will be shown below is dependent on the spark-processed layer. Typical infrared PL spectra of erbium-doped spark-processed silicon are presented in Figures 5-15 and 5-16. The detailed experimental set-up is described in Chapter 3 and sketched in Figure 3-1 (a). Figure 5-15 shows a infrared PL spectrum when the sample was prepared by depositing 0.1 m of erbium a silicon wafer (n-type, 5 cm) and spark-processing for 5 minutes under stagnant air condition, followed by rapid thermal annealing at 900C in air for 15 minutes. The infrared PL spectrum was obtained utilizing the 488 nm Ar ion laser as excitation source. The infrared PL spectrum has peaks around 950-1000 nm (1.30-1.24 eV) and around 1540 nm (0.82 eV). It is postulated applying the results from previous sections, which the peaks around 950-1000 nm originate from the spark-processed silicon itself. However, new peaks around 1540 nm are discernible in Figure 5-15 which are considered to originate from the erbium-doped spark-processed silicon. Figure 5-16 shows a room temperature photoluminescence spectrum for an erbium-doped spark-processed silicon sample measured with a spectral resolution of 1 nm. A sharp PL feature is observed having a main peak near 1545 nm and side peaks at 1559, 1567, and 1587 nm. This feature is characteristic of the first excited state 4 I 13/2 to the ground state 4 I 15/2 transitions in Er 3+ . The spectral width of the main peak (14 nm FWHM) and wide tails of the spectrum are a result of Stark splitting of the excited and ground state in the host, which causes additional broadening. 5.2.1. Annealing of Erbium-Doped Spark-Processed Silicon Annealing is one of the most important processes of activating the luminescence centers of rare-earth doped host materials. In order to determine the optimum annealing

PAGE 106

85 conditions for the erbium-doped spark-processed silicon, the following experiments were performed. Spark-processed erbium-metal-deposited silicon samples were annealed individually at 700, 800, 900, 1000, and 1100C for 15 minutes in air. Figure 5-17 depicts the infrared PL peak intensity at 1540 nm as a function of the rapid thermal annealing temperature. It is observed that upon annealing up to 700C essentially no PL intensity change occurs. Starting slightly above this temperature, the IR PL intensities raise sharply by about one order of magnitude to reach a maximum near 900C. Above this temperature a decrease in infrared PL intensity is observed, which can be explained by assuming the precipitation of erbium [44]. XRD data in Figure 4-6 show a similar profile of the Er 3+ concentration with regard to the annealing temperature. The increased photoluminescence at high annealing temperatures is a good indication of the increased number of optically active erbium ions. Figure 5-18 shows the infrared PL peak intensity for the 1540 nm peak as a function of annealing time at the optimum annealing temperature of 900C. It is observed that the infrared PL intensity initially increases sharply with increasing time in the range of 0 to 5 minutes of annealing and nears optimization after 5 minutes where it reaches a plateau up to about 15 minutes (900 seconds) after which the PL intensity decreases again. It is also noted that there is approximately a 30% decrease in PL intensity after 30 minutes annealing at the optimum temperature. Unless otherwise stated, the following experiments will be conducted at the optimal conditions for PL, namely the erbium-doped spark-processed silicon, heat treated in air at 900C for 15 minutes. As a final remark; Annealing did not lead to a change in spectral shape or peak position.

PAGE 107

86 5.2.2. Variation of Processing Parameters 5.2.2.1. Concentration of erbium in spark-processed Si and spark-processing time In order to study the concentration dependence of erbium-doped spark-processed silicon, samples with six different initial thicknesses of erbium layers were prepared which were subsequently spark-processed. Figure 5-19 shows the infrared PL intensity as a function of the initial erbium layer thickness and also as a function of spark-processing time. It is observed that there is a steep increase in the infrared PL intensity up to about 100 nm thickness followed by a smaller increase above that thickness. The smaller increase of the infrared PL intensity at higher erbium concentration suggests that there is a limit to the concentration of Er 3+ ions that can be optically activated in the spark-processed silicon matrix. A second explanation involves precipitation of erbium, as can also be seen in the high angle annular dark field scanning tunneling electron micrographs (HAADF STEM) in Figure 4-8. It is likely that these precipitates do not contribute to the optical transitions, i.e. PL or EL. Figure 5-19 also reveals the relationship between the infrared PL intensity and the spark-processing time at a given erbium layer thickness. It is observed that with increasing spark-processing time, the infrared PL intensity at first increases and then peaks in the 30-60 seconds ranges after which it decreases again. Unless otherwise stated, most of the following experiments were performed using an initial 100 nm thick erbium layer followed by 30 seconds of spark-processing time. At first glance, 60 seconds spark-processing of a 600 nm erbium-on-silicon layer would have increased the 1540 nm PL intensities by a factor of about 2. However, samples whose initial erbium thicknesses are above 300 nm were excluded from these

PAGE 108

87 experiments since they showed structural problems, specifically, cracking and peeling of the top erbium layer after annealing. 5.2.2.2. Pressure in spark-processing chamber The effect of ambient gas pressure during spark-processing on the infrared photoluminescence from erbium-doped spark-processed silicon was the next processing parameter investigated. Figure 5-20 shows the infrared PL intensity of erbium-doped spark-processed silicon as a function of spark-processing chamber pressure. All samples were prepared using a 30 seconds spark-processing time. It should be noted that as the chamber pressure was reduced, the spark nature appeared to be changed. Specifically, as the processing pressure is reduced, the bottom end of the spark cone plume widens. Further, the spark character transforms from a ‘vigorous’ discharge to a ‘sluggish’ discharge as the pressure is reduced. It is observed that the infrared PL intensity initially increases slightly with a decrease in pressure from 1013 mbar to 800 mbar. Further decreases in pressure result in a decrease in PL intensity. This slight increase of the PL intensity by reducing the pressure in the spark-processing chamber was not considered to be significant enough to go through this inconvenience. As mentioned above, as processing pressure is reduced, the bottom of the spark column widens. This is due to the fact that in an atmosphere lacking sufficient candidates for ionization, the spark plasma consumes a larger volume of substrate to maintain its quasi-equilibrium state. As a result, the area of the sample surface increases. Concomitant with a decrease in pressure is a decrease in the available oxygen, which is required for the formation of the luminescence centers in spark-processing. This results in a decrease of infrared PL intensity at a processing pressure lower than 800 mbar.

PAGE 109

88 Figure 5-21 depicts the infrared PL intensity of erbium-doped spark-processed silicon as a function of spark-processing time at 200 mbar. It is observed that as spark-processing time increases, the infrared PL intensity also increases, peaking near 180 seconds. Further increases in spark-processing time result in a decrease of infrared PL intensity. The dotted line in the Figure 5-21 depicts the infrared PL intensity variation with increasing spark-processing time at atmospheric pressure. In comparison to the spark-processing time results obtained at the atmospheric pressure, infrared PL intensity at 200 mbar needs a longer spark-processing time to reach the highest intensity point. 5.2.2.3. Spark-processing gap distance In order to find an optimal spark-processing gap distance, the infrared PL intensities were recorded using different gap distances. The most efficient infrared photoluminescence was achieved when a spark-processing gap distance of 1 to 2 mm was applied as depicted in Figure 5-22. Increases in spark-processing gap distance beyond 1 or 2 nm resulted in a decrease of infrared PL intensity. 5.2.2.4. Erbium deposition and spark-processing sequence In a further attempt to achieve the most efficient way for infrared PL, the following sets of experiments were performed. As depicted in Figure 5-23, set A was prepared utilizing the following sequence: 1) preparing clean silicon wafer, 2) spark-processing for 30 seconds on a silicon substrate, 3) 0.2 m of erbium metal deposition on spark-processed silicon. Set B: 1) preparing clean silicon wafer, 2) 0.1 m of erbium metal deposition on silicon wafer, 3) spark-processing for 30 seconds, 4) deposition of another 0.1 m of erbium metal. Set C: 1) preparing clean silicon wafer, 2) 0.2 m of erbium

PAGE 110

89 metal deposition on silicon wafer, 3) spark-processing for 30 seconds. Set D: 1) preparing clean silicon wafer, 2) 0.1 m of erbium metal deposition on silicon wafer, 3) spark-processing for 30 seconds. All sets were annealed at 900C for 15 minutes under air. The infrared PL spectra were all recorded using identical spectrometer settings. The results obtained are depicted in Figure 5-24. Set A shows an extremely small characteristic peak at 1540 nm. Set B has the most efficient infrared PL emission. Set C shows almost 80% of the maximum PL intensity of the set B at 1540 nm. Set D shows half of the intensity of the set C. The results are summarized by stating the PL intensity to be Set B > Set C > Set D > Set A. From these results, it is postulated that Er 3+ ions in set A are barely incorporated within or activated in the spark-processed silicon matrix. Er 3+ ions in set B, C, and D are considered to diffuse well into the spark-processed silicon matrix. In set B, additional erbium is filling the pores of erbium-doped spark-processed silicon and additional luminescence centers of Er 3+ ions seem to be developed and activated. Set B is the most desirable sequence of erbium deposition and spark-processing for erbium-doped spark-processed silicon. Most of the experiments performed in the following sections utilize, however, sequence set C, because of its simplicity and comparable PL intensity. It is also assumed that repetitive erbium deposition and spark-processing would enable the achievement of an even better infrared photoluminescence. 5.2.2.5. Spark-processing environment In order to find out the most appropriate gas environment for spark-processing, the following experiments were executed. Three sets of samples were spark-processed in UHP oxygen, in UHP nitrogen, and in air respectively. All samples were spark-processed

PAGE 111

90 for 30 seconds at the processing pressure of 1013 mbar. The samples were then annealed at 900C for 15 minutes in air. The spectrometer settings were identical for all measurements of the infrared PL. Figure 5-25 depicts the infrared PL spectra of erbium-doped spark-processed silicon processed in air, oxygen, and nitrogen. It is observed that the infrared PL intensity is highest for the sample spark-processed in air. Oxygen is known to be essential for enhancing infrared PL, but excessive oxygen supplied during spark-processing may quench the visible PL intensity [50, 54, 96]. Likewise excessive and sparse amounts of oxygen during the spark-processing causes the quenching of infrared PL intensity of erbium-doped spark-processed silicon. 5.2.2.6. Annealing environment Annealing in air, oxygen and nitrogen were carried out, in order to find an optimal annealing environment. Each sample was spark-processed for 30 seconds under stagnant air and annealed in a given environment for 15 minutes at 900C. Figure 5-26 shows the infrared PL spectra of erbium-doped spark-processed silicon using different annealing environments. The highest infrared PL intensity was observed for the sample annealed in air, and the weakest for the sample annealed in nitrogen. Samples spark-processed in nitrogen and annealed in nitrogen display only 25% of the infrared PL peak intensity of similar samples processed in air and annealed in air, as shown in Figure 5-27. The peak intensities of samples spark-processed in air and annealed in nitrogen yield 65% and samples spark-processed in nitrogen and annealed in air only 45% of similar samples, however spark-processed in air and annealed in air.

PAGE 112

91 From this result, it is concluded that more oxygen associates with erbium and silicon during spark-processing than during annealing. All taken, the optimal condition for maximal PL emission is spark-processed and annealed in air. 5.2.2.7. Variation in silicon substrate resistance The infrared PL dependence of the wafer resistivity caused by different doping is depicted in Figure 5-28. P-type silicon wafers having resistivities of 0.02, 5.2, 13, 25, and 40 cm were utilized. All other parameters were kept identical. It is observed that the infrared PL intensities are smaller for both, low and high resistivities whereas a maximal PL intensity for the 1540 nm peak is seen for a wafer resistivity near 13 cm. 5.2.2.8. Preliminary conclusions It was learned from the investigations presented in this section that the maximal IR PL intensity for erbium-doped spark-processed silicon is achieved for the following conditions: 1) Processing sequence B in Figure 5-23: erbium deposition first, then spark-processing, followed by an additional erbium deposition. 2) As the initial erbium layer thickness is increased the infrared PL intensity increases likewise. 3) The strongest infrared (1540 nm) PL can be obtained by spark-processing times between 30-60 seconds. 4) The optimum spark-processing pressure is near atmospheric pressure. 5) The erbium-doped spark-processed silicon processed in air shows the highest intensity of the infrared PL.

PAGE 113

92 6) The desirable spark-processing gap distance is between 1 and 2 mm. 7) Annealing erbium-doped spark-processed silicon at 900C for 15 minutes in air best activates the Er 3+ ions and thus yields the most efficient PL. 8) The highest infrared PL intensity of erbium-doped spark-processed silicon was observed for a silicon substrate resistivity of 13 cm. Unless otherwise stated, the subsequent experiments were performed utilizing the aforementioned parameters leading to the most efficient PL. 5.2.3. Variation of Measuring Parameters 5.2.3.1. Measuring Temperature Figure 5-29 depicts the temperature dependence of the infrared (1540 nm) PL intensity I(T) of erbium-doped spark-processed silicon. The luminescence peak intensity, plotted in an Arrhenius fashion, is found to decrease linearly upon increasing the temperature from 15 K to 100 K with activation energy of 0.174 meV. At temperatures larger than 100 K the IR PL intensity decreases rapidly and eventually reaches again a linear slope up to about 500 K with activation energy of 96.6 meV. It is observed that the peak wavelength of the infrared PL stays fixed at approximately 1540 nm. Although the PL intensity decreases as the temperature increases, the degree of thermal quenching is small. At lower temperatures (15 K) the PL intensity of erbium-doped spark-processed silicon is enhanced by a factor of 4 compared to room temperature as seen in Figure 5-29. This value is much smaller than that reported (about two or three orders of magnitude) for an erbium-doped bulk silicon crystal [40]. The PL intensity continuously increases with decreasing temperature. Above 250 K thermal

PAGE 114

93 quenching of I(T) dominates, which can be described by an exponential dependence on 1/T. The dependence of the luminescence intensity on temperature, I(T), can be expressed in the empirical form [96] nrrTTcTII00exp1)( 00exp1)(TTcITI (Equation 5-1) where c is a constant that is independent of the temperature T. Equation 5-1 describes the temperature dependence of PL in disordered solids in terms of a characteristic temperature T 0 , which is a measure of the system disorder and is related to the tail width of the density of localized states into the gap region. The experimental data in Figure 5-29 (squares) were analyzed by using Equation 5-1 and the fitted results are shown by the solid line in Figure 5-29. In Figure 5-30 the logarithm of [I 0 /I(T)]-1 is plotted versus temperature. The good fit by a straight line indicates the presence of exponential bandtails as assumed in Equation 5-1. A typical value for T 0 for erbium-doped spark-processed silicon is found by extrapolation to be 90 K. For comparison, T 0 for spark-processed silicon varies between 70 and 110 K [50, 96]. This leads to the conclusion that erbium-doped spark-processed silicon has exponential bandtails and considerable disorder. The infrared PL intensity as a function of the heating and cooling temperature is displayed in Figure 5-31. It is seen there that with increasing temperatures the infrared PL intensity decreases. From 200C back to room temperature, with decreasing temperatures

PAGE 115

94 the infrared PL intensities increase back to the original position (intensity). No hysteretic effect of annealing is observed. The variation of the PL intensity for erbium-doped spark-processed silicon as a function of exposure time to Ar ion laser light (488 nm) is depicted in Figure 5-32 for different measuring temperatures. It is observed that the PL intensity at room temperature stays essentially constant over time. However when the same substances are measured at 100C and 200C, they lose some of their infrared PL intensities (30% and 65% respectively) within the first 20 minutes of ramping time after which they stabilize. 5.2.3.2. Excitation source power and energy Figure 5-33 depicts the increase in infrared PL intensity as a function of the excitation power (488 nm Ar ion laser). A nearly exponential increase of the infrared PL intensity is observed with increasing laser power. The infrared PL external output power as a function of input power of the excitation source is displayed in Figure 5-34, as well. It is observed that the change of the sample output power is again an exponential function and thus closely related to the change of laser input power. Figure 5-35 depicts the relationship between the infrared PL intensity and the excitation wavelength that is, the excitation energy. It is observed that the infrared PL intensity is largest for the 488 nm excitation wavelength (2.54 eV) and lower for the 457 nm and 515 nm wavelengths. As shown on Figure 4-3, Er 3+ has a 4 F 7/2 energy level at 2.54 eV which coincides with the 488 nm excitation wavelength. It can be therefore postulated that erbium-doped spark-processed silicon has the best PL efficiency when the excitation energy resonates with the Er 3+ ion energy level.

PAGE 116

95 5.2.3.3. Emission depth In order to investigate the infrared PL response of erbium-doped spark-processed silicon on HF etching, the following experiments were conducted. A buffered oxide etch consisting of NH 4 F 40% 6 parts and HF 1 part was utilized. Figure 5-36 depicts several infrared PL spectra of erbium-doped spark-processed silicon as a function of different etching times. The specimens show a relatively constant peak position (1540 nm) throughout the experiments. However the infrared PL peak intensity reveals a decrease with longer etching times. This experiment can be interpreted to reveal the PL emission is strongest near the surface of the spark-processed layer whereas deeper layers contribute only smaller amounts to the overall emission. 5.3. Infrared Electroluminescence of Erbium-Doped Spark-Processed Silicon Electroluminescence is known to be the conversion of electrical energy to optical energy. Since spark processing is a treatment that transforms erbium-deposited silicon into a material with photoluminescing properties in the infrared range, it is speculated that spark-processing will also yield an electroluminescing device. For starters the conditions which yield the highest PL intensity (see above) will be utilized for EL. On application of a voltage to the device structure explained in Chapter 3 and depicted in Figure 3-1(b), infrared electroluminescence was indeed observed as depicted in Figure 5-37 (solid line). As shown in Figure 3-1(b), the negative electrode of the power supply is connected to the top, metal film. Electron injection into the erbium-doped spark-processed silicon is crucial for achieving an effective infrared electroluminescence. Thus no light emission or electroluminescence is observed under opposite biasing conditions. The sample preparation is given in Chapter 3 and the caption of Figure 5-37. For

PAGE 117

96 comparison, Figure 5-37 depicts also a PL spectrum (dotted line). The infrared EL spectrum is observed to display a major peak near 1555 nm (0.81 eV) and a FWHM of 41 nm. The spectrum has a characteristic light emission of the first excited state 4 I 13/2 to ground state 4 I 15/2 transition in the Er 3+ ion. It is further observed that the EL spectrum is a convolution of several peaks, probably as a result of Stark splitting of the excited states and ground states. Compared to the EL spectrum, the PL spectrum is broader (FWHM=44 nm) and slightly shifted to shorter wavelengths having a peak wavelength near 1540 nm. This may be an indication of different microenvironments for the Er 3+ ions that can be excited optically and electrically. Moreover, the EL peak intensity is about 35% smaller than that for PL for the voltage used (25 V dc, pulsed at 200 Hz with a 30% duty cycle). Figures 5-38 (a) and 5-38 (b) show the results of peak deconvolution of the infrared EL and PL spectra of erbium-doped spark-processed silicon respectively. The major distinguishable peaks for PL are around 1530, 1540, 1550, and 1580 nm. Similar peaks are observed in the infrared EL spectrum with divergent intensities. Specifically, the infrared EL spectrum displays major peaks near 1540, 1550, 1567, and 1578 nm. It is further observed that the 1540 nm peak for EL is smaller than the corresponding peak for PL whereas this ratio is inversed for the 1550 nm peak. Figure 5-39 depicts the infrared electroluminescence spectrum of the control sample, otherwise identically prepared but without spark-processing. No light emission or electroluminescence is observed from this control sample, under either biasing conditions. In other words, spark-processing is the key process to generate infrared EL for erbium-doped silicon.

PAGE 118

97 5.3.1. Variation of Processing Parameter 5.3.1.1. Concentration of erbium in spark-processed silicon Figure 5-40 presents the EL luminescence of erbium-doped spark-processed silicon as a function of the initial thickness of the erbium layer. It is observed that the infrared EL intensity steeply increases up to a layer thickness of about 100 or 200 nm, after which it moderately increases further. The slope change in the infrared EL intensity above 100 nm, indicates that there is a limit to the concentration of Er 3+ ions that can be optically activated in the spark-processed silicon matrix. It is also suggested that precipitation of erbium may take place, as also seen in the infrared photoluminescence of erbium-doped spark-processed silicon. 5.3.1.2. Variation of spark-processing time Figure 5-41 depicts the infrared EL intensity at 1550 nm as a function of spark-processing time. Each sample was prepared as follows: 200 nm of erbium metal was deposited on a silicon substrate, spark-processed at a given time, and then rapid thermal annealed in air for 15 minutes. All samples were measured under identical spectrometer conditions. It is observed that with increasing spark-processing time, the infrared EL intensity increases in its early stages sharply, and peaks in the range of 30-40 seconds. Spark-processing for more than 50 seconds resulted in a decrease in infrared EL intensity. Above 80 seconds of spark-processing caused a barely detectable EL. In order to further investigate the influence of spark-processing time, EDS was utilized to examine the differences between various samples which were spark-processed for 10, 40, 70, and 100 seconds respectively. Figure 5-42 shows the spectra obtained. It is observed that essentially no erbium can be detected after 70 and 100 seconds of spark

PAGE 119

98 processing time whereas a moderate erbium concentration is seen after 40 seconds spark-processing. The largest erbium concentration is seen after only 5 seconds spark-processing. In this case the erbium is probably still deposited on the free surface, but not in the SiO x matrix. It is further suggested that the increase in infrared EL intensity in the 1-40 second range is due to a number of factors. First, the increase in area of the halo region results in an increase in the light emitting area, and therefore an increase in EL intensity. Second, as the spark-processing time is increased in this range, the number of emitting centers, the surface morphology, and the thickness of surface oxide layer are optimized, which all result in an enhancing EL intensity. It is additionally suggested that for higher spark-processing times, the thicker oxide layer starts to negatively affect the electrical conduction and thus spark-processing efficiency. 5.3.1.3. Change in types of substrate In order to investigate the influence of the wafer type to infrared EL, the following experiments were performed. The infrared EL devices were fabricated using n-type and p-type silicon with resistivities in the same range. The infrared EL devices were prepared with a Czochralski grown (100) 3-5 cm n-type silicon wafer and alternatively using a Czochralski grown (100) 1-3 cm p-type silicon wafer. They were processed utilizing identical spark-processing procedures and metallizations. The devices were measured under identical spectrometer settings at 25 V, 30% duty cycle excitation. Electroluminescence was observed in both cases when the top contact was negatively biased. No EL was observed in either n-type or p-type devices with a positively biased top contact. Figure 5-43 shows the infrared EL spectra of n-type

PAGE 120

99 (dashed curve) and p-type (solid curve) wafers. No significant EL peak position changes are discernible between the obtained spectra. It is observed that the infrared EL device fabricated on n-type silicon has a 30% higher efficiency and 1550 nm infrared EL intensity than the p-type device. The devices fabricated on n-type and p-type have output power/input power (external) efficiencies of 0.0073% and 0.0044% respectively. Figure 5-44 shows the reverse current-voltage characteristics for n-type (circles) and p-type (squares) erbium-doped spark-processed silicon EL devices. The current-voltage curves for the erbium-doped spark-processed silicon devices reveal increasing currents when negative voltages are applied, which are characteristics for quasi-rectifying devices. It is observed that the infrared EL devices fabricated on p-type silicon have slightly higher currents than the n-type at the same voltages. In summary, n-type silicon is slightly more efficient for the erbium-doped spark-processed silicon infrared EL device. 5.3.2. Further Variations of Measuring Parameters 5.3.2.1. Measuring temperature The infrared EL intensity at 1550 nm under 25 V excitation was monitored, as a function of operating temperature of the erbium-doped spark-processed silicon EL device. The infrared EL intensity as a function of temperature at 1550 nm (peak wavelength) and 1420 nm (background wavelength) are presented in Figure 5-45. It is observed that the infrared EL intensity decreases with increasing temperature from room temperature to 200C. In contrast, the EL intensity at 1420 nm remains essentially unchanged. It is postulated that only the EL peaks related to Er 3+ ions are affected by the measuring temperature, and no discernible intensity changes are found in other wavelength regions.

PAGE 121

100 It is suggested that a higher device temperature results in an increase in the non-radiative recombination rate of generated phonon processes. During the heating and cooling of the device, no hysteretic effect of annealing is observed. No spectral shift of the infrared EL was observed with increased device temperature. The temperature dependence of the reverse current-voltage characteristics is depicted in the inset of Figure 5-45. A positive temperature coefficient is observed. That is, the breakdown voltage increases with increasing temperature. It is known that impact-ionization has also a positive temperature coefficient. That is, the device current increases at a given voltage with decreasing temperature. This suggests that impact ionization plays a decisive role in EL for erbium-doped spark-processed silicon. 5.3.2.2. Applied voltage Figure 5-46 represents infrared EL spectra of an erbium-doped spark-processed silicon EL device measured at different excitation voltages. It is observed as expected that the infrared EL intensity increases with increasing driving voltage. No spectral shift of the maximal peak wavelength is observed. The infrared EL intensity at 1550 nm as a function of applied voltage is depicted in Figure 5-47. It is observed that beyond a threshold voltage of about 3 V the infrared EL intensity increases rapidly with applied voltage, and eventually levels off beginning at 25 V. The saturation of the infrared EL intensity beyond 25 V is believed to be due to the finite number of available states into which electrons can be excited. Further increases in applied voltage results in a breakdown of the device, which is indicated by an abrupt increase in current (see Figure 5-49).

PAGE 122

101 Figure 5-48 shows the relationship between the infrared EL intensity of erbium-doped spark-processed silicon and the applied power. It is observed that the infrared EL intensity sharply increases initially and then levels off. For explanation of this behavior the same arrangements as above can be put forward. 5.3.3. Current-Voltage Characteristics of Erbium-Doped Spark-Processed Silicon The electrical properties of erbium-doped spark-processed silicon were further investigated by measuring the current-voltage characteristics. When semiconductors including silicon are contacted with metals, they display electrical features known to be a rectifying contact or an ohmic contact. Figure 5-49 depicts the current-voltage characteristics of an erbium-doped spark-processed silicon device. It is similar to the rectifying characteristics of a diode with the exception that the device current is larger than for an ideal rectifying reverse biased device (ideally zero current). The current increases with an applied negative voltage. Thus an erbium doped-spark-processed silicon EL device has a quasi-rectifying I-V characteristics. At large values of reverse bias the current increases dramatically, and eventually displays breakdown. This breakdown voltage is seen in Figure 5-49 to be about 52 V. At this voltage a current of 180 mA is flowing. The current-voltage dependence on wafer resistivity is depicted in Figure 5-50. Selected wafers having resistivities of 0.02 cm, 1.5 cm, 5.2 cm, 13 cm, 25 cm, and 40 cm respectively have been used. Each sample was prepared having the same processing parameters and measuring condition. It is observed that as the wafer resistivity increases, the reverse device current decreases rapidly. The infrared EL device fabricated with 0.02 cm silicon is observed to have Ohmic-like I-V characteristics.

PAGE 123

102 With increasing resistivity, current-voltage characteristic of the erbium-doped spark-processed silicon gradually changes from Ohmic type to rectifying type. In fact, the device prepared with 40 cm silicon wafer depicts rectifying-like I-V characteristics. The infrared EL intensity dependence on wafer resistivity is presented in Figure 5-51. Identically spark-processed samples were excited with the same voltage (15 V). It is observed that the infrared EL intensity sharply decreases with increasing wafer resistivity. Correlation between infrared EL intensity and device current is observed from the results. No EL was observed from the sample fabricated using the 0.02 cm wafer which had ohmic characteristics, see Figure 5-50. This is identical with the observations made for EL device for undoped spark-processed silicon which emits in the visible spectral range [6]. 5.3.4. ITO Contact In order to study an alternative EL contact material, a device was fabricated utilizing ITO (indium-tin-oxide) for the top transparent contact in order to allow light emission through the top surface. The optical properties of indium tin oxide (ITO) at 1550 nm are: n=1.27, k=0.12 where n is the index of refraction and k is the damping constant. ITO films about 200nm thick were layered down by pulsed laser deposition (PLD). Figure 5-52 shows the infrared EL spectrum for an erbium-doped spark-processed silicon device where ITO was utilized as transparent contact. Again, infrared EL was observed while applying a negative excitation voltage of 20 V on the top contact. It is observed that the spectrum is modified compared to the before-mentioned metal contacts. Specifically, the peak is lower by around 70% and the peak position is shifted from 1550

PAGE 124

103 nm to 1535 nm. The ITO-containing device is less stable than the device with a metal contact it displays a higher device current and an abrupt change in device current.

PAGE 125

104 80010001200140016001800 0.02.0x10-54.0x10-56.0x10-58.0x10-51.0x10-4 IR PL Intensity (a.u.)Wavelength (nm) Figure 5-1 Room temperature infrared photoluminescence spectrum of spark-processed silicon employing a p-type (B-doped, 5 cm) Si wafer and involving 15 minutes of sparkprocessing time utilizing a 1.2 mm gap between the tungsten tip and the Si wafer under stagnant air conditions. A 488nm Ar laser line was utilized as an excitation source.

PAGE 126

105 800900100011001200 0200400600800100012001400160018002000 25oC200oC350oC300oC400oC450oC500oC600oC700oC800oC900oC1000oC Relative Intensity (a.u.)Wavelength (nm)1100oC Figure 5-2 Change of infrared photoluminescence spectra compared to pure spark processed silicon as a function of various annealing temperatures in air for 30minutes. Note that the base line of each curve is shifted upwards for clarity.

PAGE 127

106 02004006008001000 0.02.0x10-44.0x10-46.0x10-48.0x10-41.0x10-3 IR PL Intensity (a.u.)Annealing Temperature ( oC) Figure 5-3 Infrared photoluminescence peak intensity change as a function of annealing temperature (30 minutes, air). The error bars represent data from two independent measurements.

PAGE 128

107 020040060080010001200 920960100010401080 Annealing Temperature ( oC)IR PL Peak Position (nm) Figure 5-4 Infrared photoluminescence peak wavelengths as a function of annealing temperature (30minutes, air).

PAGE 129

108 700800900100011001200 0.01.0x10-52.0x10-53.0x10-5 700800900100011001200 0.01.0x10-52.0x10-53.0x10-5 1.7eV1.6eV1.5eV1.4eV1.3eV1.2eV1.1eV 8771009 Wavelength (nm)945(b) IR PL Intensity (a.u.)(a) Figure 5-5 Peak deconvolution of the infrared photoluminescence band of as-processed spark-processed silicon into 3 major peaks having peak maxima near 945, 1009, and 877nm (r 2 =98.5). (a) Experimental data and fitted curve. (b) Fitted curve and deconvoluted peak components.

PAGE 130

109 020040060080010000.01.0x10-42.0x10-43.0x10-44.0x10-45.0x10-46.0x10-47.0x10-48.0x10-49.0x10-41.0x10-3 IR PL Peak Intensity (a.u.)Annealing Temprature (oC)945nm peak 1010nm peak Figure 5-6 Change of the infrared photoluminescence peak intensity as a function of annealing temperature of two deconvoluted peaks.

PAGE 131

110 020040060080010001200 940950100010101020 945nm peak IR PL Peak Position (nm)Annealing Temperature (oC)1010nm peak Figure 5-7 Variation of deconvoluted 945 and 1010nm peak maximum wavelengths as a function of annealing temperature.

PAGE 132

111 020040060080010001200140016001800 0.05.0x10-61.0x10-51.5x10-52.0x10-52.5x10-53.0x10-53.5x10-5 IR PL Intensity (a.u.)Annealing Time (s) Figure 5-8 Infrared photoluminescence intensity variation of spark-processed silicon as a function of annealing time at 900C in air. Samples used in this experiment were prepared with 1 minute spark-processing under stagnant air conditions.

PAGE 133

112 101001000 0.02.0x10-54.0x10-56.0x10-58.0x10-51.0x10-41.2x10-41.4x10-41.6x10-41.8x10-42.0x10-4 IR PL Intensity(a.u.)Spark Processing Time (s) Figure 5-9 Infrared photoluminescence intensity variation as a function of spark-processing time for pure spark-processed silicon.

PAGE 134

113 (a) (b) (c) Figure 5-10 SEM images of spark-processed silicon using different spark-processing time (a) 10 seconds, (b) 15 minutes, and (c) 1 hour.

PAGE 135

114 02468101214161820050100 02468101214161820050100 02468101214161820050100 cpsEnergy (keV)SiO(a)O/Si = 0.1O/Si = 0.3O/Si = 0.2(b) SiOcpsEnergy (keV)(c) SiOcpsEnergy (keV) Figure 5-11 EDS spectra of spark-processed silicon using different spark-processing times (a) 10 seconds, (b) 15 minutes, and (c) 1 hour.

PAGE 136

115 1234567 2.0x10-52.5x10-53.0x10-53.5x10-54.0x10-5 IR PL Intensity (a.u.)Spark processing gap (mm) Figure 5-12 Infrared photoluminescence intensity as a function of the spark gap between the tungsten tip and silicon substrate.

PAGE 137

116 800100012001400160018000.01.0x10-42.0x10-43.0x10-44.0x10-45.0x10-46.0x10-47.0x10-48.0x10-4 450460470480490500510520 0.01.0x10-42.0x10-43.0x10-44.0x10-45.0x10-46.0x10-47.0x10-48.0x10-4 IR PL Intensity (a.u.)Excitation Wavelength (nm)488nm515nm IR PL Intensity (a.u.)Wavelength (nm)457nm Figure 5-13 Infrared photoluminescence using 3 different excitation wavelengths from an Ar laser (457, 488 and 515nm). The inset shows infrared photoluminescence intensities as a function of the excitation wavelength.

PAGE 138

117 80010001200140016001800 0.05.0x10-61.0x10-51.5x10-52.0x10-52.5x10-53.0x10-53.5x10-5 IR PL Intensity (a.u.)Wavelength (nm) Figure 5-14 Photoluminescence spectrum of a control sample: 0.1 m erbium metal was vapor-deposited on a silicon wafer (n-type, 5 cm) and rapid thermal annealed at 900C in air for 15 minutes. No spark-processing was utilized. The “spectrum” shows the background noise of the PL system.

PAGE 139

118 800100012001400160018000.05.0x10-61.0x10-51.5x10-52.0x10-52.5x10-53.0x10-5 1.61.41.210.8 Visible IR PL Intensity (a.u.)Wavelength (nm)NearInfrared Energy (eV) Figure 5-15 Typical infrared PL spectrum of erbium-doped spark-processed silicon. The sample was prepared by depositing 0.1m erbium on a silicon wafer (n-type, 5 cm) and subsequently spark-processing for 5 minutes under stagnant air, followed by rapid thermal annealing at 900C in air for 15 minutes. The infrared photoluminescence spectrum was obtained by utilizing the 488nm Ar ion laser line as an excitation source.

PAGE 140

119 13001350140014501500155016001650 0.01.0x10-42.0x10-43.0x10-44.0x10-45.0x10-46.0x10-47.0x10-4 0.920.880.840.80.76 Wavelength (nm) IR PL Intensity (a.u.) Energy (eV) Figure 5-16 Room temperature photoluminescence spectrum in a limited spectral range (1300 to 1650nm) for an erbium-doped spark-processed silicon sample measured with a spectral resolution of 1nm. A 488nm Ar ion laser was utilized as excitation source. Compare to Figures 5-14 and 5-15.

PAGE 141

120 02004006008001000 0.01.0x10-42.0x10-43.0x10-44.0x10-45.0x10-46.0x10-4 IR PL Intensity (a.u.)Annealing Temperature (oC) Figure 5-17 Infrared photoluminescence peak intensity of erbium-doped spark-processed silicon for the 1540nm peak as a function of rapid thermal annealing temperature (15 minutes, air).

PAGE 142

121 020040060080010001200140016001800 0.02.0x10-64.0x10-66.0x10-68.0x10-61.0x10-51.2x10-51.4x10-5 IR PL Intensity (a.u.)Annealing Time (s) Figure 5-18 Infrared photoluminescence intensity of erbium-doped spark-processed silicon for the 1540nm peak as a function of annealing time at the optimum annealing temperature of 900C.

PAGE 143

122 0100200300400500600 0.05.0x10-51.0x10-41.5x10-42.0x10-42.5x10-43.0x10-43.5x10-4 60s spark processing30s spark processing300s spark processingIR PL Intensity (a.u.)Erbium layer thickness (nm)10s spark processing Figure 5-19 Infrared photoluminescence intensity of the 1540nm peak of erbium-doped spark-processed silicon as a function of initial erbium layer thickness before spark-processing and as a function of spark-processing time. The samples were annealed at 900C in air for 15 minutes.

PAGE 144

123 02004006008001000 0.02.0x10-54.0x10-56.0x10-58.0x10-51.0x10-41.2x10-41.4x10-41.6x10-4 IR PL Intensity (a.u.)Processing Pressure (mbar) Figure 5-20 Infrared photoluminescence intensity of erbium-doped spark-processed silicon for the 1540nm peak as a function of spark-processing pressure. The samples were prepared using 30 seconds spark-processing time and an erbium layer on silicon of 100nm.

PAGE 145

124 050100150200250300 6.0x10-67.0x10-68.0x10-69.0x10-61.0x10-51.1x10-51.2x10-5 IR PL Intensity (a.u.)Spark Processing Time (s)Spark processing time variation at atmospheric pressure (x0.083) Figure 5-21 Infrared photoluminescence intensity change of erbium-doped spark-processed silicon as a function of spark-processing time at 200mbar (solid line). The thickness of erbium layer was 75nm and the sample was annealed at 900C for 20minutes. For comparison, IR PL intensity variation as a function of spark-processing time at atmospheric pressure was plotted (dotted line).

PAGE 146

125 1234567 1.1x10-51.2x10-51.3x10-51.4x10-51.5x10-51.6x10-51.7x10-51.8x10-5 IR PL Intensity (a.u.)Spark Processing Gap (mm) Figure 5-22 1.54 m infrared photoluminescence intensity variation of erbium-doped spark-processed silicon as a function of the spark gap between the tungsten tip and silicon.

PAGE 147

126 Figure 5-23 Schematics of processes to elucidate the most efficient erbium deposition and spark-processing sequence in erbium-doped spark-processed silicon. Set A : 1) clean silicon wafer, 2) spark-processing for 30 seconds, 3) 0.2m of erbium metal deposition, Set B : 1) clean silicon wafer, 2) 0.1m of erbium metal deposition, 3) spark-processing for 30 seconds, 4) additional 0.1m of erbium metal deposition, Set C : 1) clean silicon wafer, 2) 0.2m of erbium metal deposition, 3) spark-processing for 30 seconds, *Set D : 1) clean silicon wafer, 2) 0.1m of erbium metal deposition, 3) spark-processing for 30 seconds. * set D is prepared for comparison with other results, see below.

PAGE 148

127 130013501400145015001550160016500.002.50x10-55.00x10-57.50x10-51.00x10-41.25x10-41.50x10-41.75x10-4 Set ASet BSet CSet D 0.002.50x10-55.00x10-57.50x10-51.00x10-41.25x10-41.50x10-41.75x10-4 IR PL Intensity (a.u.) Set DSet BSet CIR PL Intensity (a.u.)Wavelength (nm)Set A Figure 5-24 Infrared photoluminescence spectra for variable process sequences. Intensity at 1.54 m; Set B > Set C > *Set D > Set A

PAGE 149

128 130013501400145015001550160016500.02.0x10-54.0x10-56.0x10-58.0x10-51.0x10-41.2x10-41.4x10-4 IR PL Intensity (a.u.)Wavelength (nm)AirO2N2Erbum .2mspark 30secRTA 900oC 15min in Air Figure 5-25 Infrared photoluminescence spectra of erbium-doped spark-processed silicon processed in air, oxygen and nitrogen ambients respectively.

PAGE 150

129 130013501400145015001550160016500.01.0x10-52.0x10-53.0x10-54.0x10-55.0x10-56.0x10-5 IR PL Intensity (a.u.)Wavelength (nm)AirO2N2Er .2mspark 30sRTA 900oC Figure 5-26 Infrared photoluminescence spectra of erbium-doped spark-processed silicon using different annealing environments such as air, oxygen and nitrogen ambient.

PAGE 151

130 130013501400145015001550160016500.00.10.20.30.40.50.60.7 Sparked in N2, Annealed in N2Sparked in N2, Annealed in AirNormalized Intensity (a.u.)Wavelength (nm)Sparked in Air, Annealed in N2* Compared to Sparked in Air and Annealed in Air Figure 5-27 Comparison of infrared photoluminescence spectra of erbium-doped spark-processed silicon utilizing UHP nitrogen and various annealing ambients. The spectra are compared to “spark-processed in air and annealed in air”.

PAGE 152

131 0102030402.5x10-53.0x10-53.5x10-54.0x10-54.5x10-55.0x10-55.5x10-56.0x10-56.5x10-5 IR PL Intensity (a.u.)Wafer Resistivity (cm) Figure 5-28 Infrared photoluminescence of the 1540nm peak of erbium-doped spark-processed silicon as a function of silicon wafer resistivity.

PAGE 153

132 102030405060701E-51E-4 Temperature (K)20010050 1000/Temperature (1/K) IR PL Intensity (a.u.)25 Figure 5-29 Temperature dependence of the integrated infrared (1540nm) photoluminescence intensity I(T) of erbium-doped spark-processed silicon from 4 K to elevated temperature plotted in an Arrhenius fashion (square data points). The experimental data were analyzed by using Eq. 5-1 and the fitted results are shown by the solid line.

PAGE 154

133 01002003000.1110 [I0/I(T)]-1Temperature (K) T0=90.42K Figure 5-30 The logarithm of [I 0 /I(T)]-1 is plotted versus photoluminescence measuring temperature.

PAGE 155

134 0204060801001201401601802002201.0x10-41.5x10-42.0x10-42.5x10-43.0x10-43.5x10-4 IR PL Intensity (a.u.)Measuring Temperature (oC) heatingcooling Figure 5-31 Infrared photoluminescence intensity of erbium-doped spark-processed silicon as a function of heating and cooling temperature from room temperature (20C) to 200C.

PAGE 156

135 0501001502002503003500.05.0x10-51.0x10-41.5x10-42.0x10-42.5x10-43.0x10-43.5x10-4 200 oC IR PL Intensity (a.u.)Time (minutes)Room temperature100 oC Figure 5-32 Variation of the photoluminescence intensity as a function of time when erbium-doped spark-processed silicon is exposed to Ar ion laser light (488nm) at room temperature, and at 100C and 200C. The first 20 minute measurements are affected by ramping.

PAGE 157

136 141618202224262830324.0x10-56.0x10-58.0x10-51.0x10-41.2x10-41.4x10-41.6x10-41.8x10-42.0x10-42.2x10-42.4x10-4 IR PL Intensity (a.u.)Ar Laser Power (mW) Figure 5-33 Infrared photoluminescence intensity variation of erbium-doped spark-processed silicon as a function of the excitation power (488nm Ar ion laser).

PAGE 158

137 141618202224262830320.040.060.080.100.12 Device Outout Power (mW)Ar Laser Input Power (mW) Figure 5-34 Infrared photoluminescence external output power of erbium-doped spark-processed silicon as a function of the power of the excitation source (488nm Ar ion laser).

PAGE 159

138 4504604704804905005105207.00x10-68.00x10-69.00x10-61.00x10-51.10x10-51.20x10-5 IR PL Intensity (a.u.)Excitation Wavelength from Ar laser (nm) 10% Figure 5-35 Relationship between the infrared photoluminescence intensity of erbium-doped spark-processed silicon and the excitation wavelength. (Three lines were used, namely 457nm, 488nm, and 515nm.)

PAGE 160

139 13001350140014501500155016000.02.0x10-54.0x10-56.0x10-58.0x10-51.0x10-4 No etching IR PL Intensity (a.u.)Wavelength (nm)1 minute 5 minutes Figure 5-36 Infrared photoluminescence spectra of erbium-doped spark-processed silicon after different etching times. Samples were annealed at 900C in air for 15 minutes.

PAGE 161

140 130013501400145015001550160016500.02.0x10-54.0x10-56.0x10-58.0x10-51.0x10-41.2x10-41.4x10-4 PL IR PL Intensity (a.u.)Wavelength (nm)EL Figure 5-37 The solid line shows a typical infrared electroluminescence room temperature spectrum of erbium-doped spark-processed silicon prepared using 0.1 m erbium metal deposited on a silicon wafer (n-type, 5 cm), a spark time of 30 seconds in air, a spark gap of 2 mm and 900C rapid thermal annealing (RTA) for 15 minutes in air. Aluminum (50nm) on the back-side of the wafer and silver (200nm) on the top of devices was deposited by physical vapor deposition (PVD) as electrical contacts. The dashed line indicates a photoluminescence spectrum of erbium-doped spark-processed silicon for comparison.

PAGE 162

141 130013501400145015001550160016500.02.0x10-54.0x10-56.0x10-58.0x10-5 0.88 eV0.84 eV0.80 eV0.76 eV Wavelength (nm) IR EL Intensity (a.u.) (a)r2=99.7% 130013501400145015001550160016500.05.0x10-51.0x10-41.5x10-4 0.88 eV0.84 eV0.8 eV0.76 eV Wavelength (nm) IR PL Intensity (a.u.)r2=99.5%(b) Figure 5-38 Peak deconvolution of the infrared (a) electroluminescence and (b) photoluminescence spectra of erbium-doped spark-processed silicon.

PAGE 163

142 800100012001400160018000.02.0x10-64.0x10-66.0x10-68.0x10-61.0x10-51.2x10-51.4x10-5 EL Intensity (a.u.)Wavelength (nm) Figure 5-39 Infrared electroluminescence spectrum of a control device prepared as in Figure 5-37 but without spark-processing.

PAGE 164

143 01002003004005006000.02.0x10-54.0x10-56.0x10-58.0x10-51.0x10-4 Erbium Thickness (nm)IR Intensity (a.u.) Figure 5-40 Infrared electroluminescence intensity of erbium-doped spark-processed silicon as a function of erbium concentration variation

PAGE 165

144 020406080100120 0.02.0x10-54.0x10-56.0x10-58.0x10-51.0x10-41.2x10-4 IR EL Intensity (a.u.)Spark Processing Time (s)peak intensities are picked at 1550nm Figure 5-41 Infrared electroluminescence intensity of erbium doped spark-processed silicon monitored at 1550nm as a function of spark-processing time. Initial erbium thickness: 200nm; 900C rapid thermal annealing for 15 minutes in air.

PAGE 166

145 02468101214161820 020406080100 02468101214161820 020406080100 02468101214161820 020406080100 02468101214161820 020406080100 cpsEnergy (keV)SiO(d) cpsEnergy (keV)ErErSiO(c) cpsEnergy (keV)ErSiOEr(b) cpsEnergy (keV)ErSiOEr(a) Figure 5-42 EDS spectra of erbium-doped spark-processed silicon processed with various times (a) 5 seconds, (b) 40 seconds, (c) 70 seconds, and (d) 100 seconds.

PAGE 167

146 13001350140014501500155016000.02.0x10-54.0x10-56.0x10-58.0x10-51.0x10-4 P-type Wavelength (nm)Irradiance (mW/cm2nm)N-type Figure 5-43 Infrared electroluminescence spectra of n-type (dashed curve) and p-type (solid curve) wafers. (100) n-type wafer having 3-5cm resistivity. Applied voltage: 25V. Current: 101mA. 0.0073% output/input (external) efficiency. (100) p-type wafer having 1-3cm resistivity. Applied voltage: 25V. Current: 163mA. 0.0044% output/input efficiency.

PAGE 168

147 403020100 5004003002001000 Reverse Voltage (V) Current (mA)N-typeP-type Figure 5-44 Reverse current-voltage characteristics of n-type (circles) and p-type (squares) erbium-doped spark-processed silicon EL device.

PAGE 169

148 020406080100120140160180200 0.02.0x10-56.0x10-58.0x10-51.0x10-4 -40-30-20-100 4003002001000 194oC Current (mA) Reverse voltage (V)20oC90oC EL intensity change at 1420nmIR Intensity (a.u.)Temperature (oC)EL intensity change at 1550nm Figure 5-45 Infrared electroluminescence intensity of erbium-doped spark-processed silicon as a function of operating temperature at 1550nm (peak wavelength) and 1420nm (background wavelength)

PAGE 170

149 130013501400145015001550160016500.01.5x10-53.0x10-54.5x10-56.0x10-57.5x10-59.0x10-5 1V dc5V dc10V dc20V dc IR EL Intensity (a.u.)Wavelength (nm)30V dc Figure 5-46 Infrared electroluminescence spectra of an erbium-doped spark-processed silicon EL device measured at different voltages.

PAGE 171

150 0102030400.02.0x10-54.0x10-56.0x10-58.0x10-51.0x10-4 Applied Voltage (V)IR Intensity (a.u.) Figure 5-47 Infrared electroluminescence intensity of an erbium-doped spark-processed silicon EL device at 1550nm as a function of applied voltages.

PAGE 172

151 024681012140.02.0x10-54.0x10-56.0x10-58.0x10-5 Applied Power (W)IR Intensty (a.u.) Figure 5-48 Relation between the infrared electroluminescence intensity of erbium-doped spark-processed silicon device and the applied power.

PAGE 173

152 -60-40-2002040-200-1000100200300400500 Current (mA)Voltage (V) Figure 5-49 Current-voltage characteristics of erbium-doped spark-processed silicon. A n-type wafer having 5cm has been used. Pulsed d.c. with 200Hz frequency and 30% duty cycle was utilized.

PAGE 174

153 -50-40-30-20-1001020-800-600-400-2000200400600800 40cm25cm13cm5.2cm1.5cmCurrent (mA)Voltage (V)0.02cm Figure 5-50 Current-voltage characteristics of erbium-doped spark-processed silicon as a function of wafer resistivity. Selected wafers having 0.02cm, 1.5cm, 5.2cm, 13cm, 25cm and 40cm have been used.

PAGE 175

154 0510152025303540451.0x10-52.0x10-53.0x10-54.0x10-55.0x10-56.0x10-57.0x10-58.0x10-59.0x10-5 IR Intensity (a.u.)Wafer resistiity (cm)No EL was obtained from 0.02cm wafer Figure 5-51 Infrared electroluminescence intensity of erbium-doped spark-processed silicon as a function of wafer resistivity. No EL was obtained for the 0.02cm wafer which had an ohmic characteristic, see Figure 5-50.

PAGE 176

155 146014801500152015401560158016000.02.0x10-64.0x10-66.0x10-68.0x10-61.0x10-51.2x10-51.4x10-51.6x10-51.8x10-52.0x10-5 IR Intensity (a.u.)Wavelength (nm) Figure 5-52 Infrared electroluminescence spectrum for an erbium-doped spark-processed silicon device where ITO was utilized as a transparent contact which also served as a window.

PAGE 177

CHAPTER 6 FURTHER DISCUSSIONS AND CONCLUSIONS 6.1. Infrared Photoluminescence of Spark-Processed Silicon This section briefly summarizes the experimental finding reported in previous chapters and explains these observations using existing models and new insights. This section also compares infrared spark-processed silicon results with those gained in the visible region and with photoluminescence obtained by other techniques. Spark-processing causes high-energy ablation of the silicon surface, which oxidizes to form a highly defected Si-rich SiO x layer. Spark-processed silicon emits light at near-infrared and visible regions due to the band-to-band transition between energy levels (or quasi-energy bands) formed by these layers [11]. The infrared photoluminescence of spark-processed silicon is broad and displays a peak near 945 nm (1.31 eV). This infrared photoluminescence from spark-processed silicon was newly discovered and reported here for the first time. The infrared PL of spark-processed silicon was found to have several similar properties with visible PL of spark-processed silicon. The 945 nm infrared peak intensity of spark-processed silicon increases with increasing spark-processing time but has a fixed peak wavelength. The photoluminescence in the visible range shows a similar tendency when changing the spark-processing time [96, 97]. The optimum spark-processing gap between Si substrate and counter electrode found in this research is around 1-2 mm. This result also coincides with the visible PL intensity dependence on gap distance of spark-processed silicon [50, 97]. The excitation wavelength dependence of the 945 nm near 156

PAGE 178

157 infrared PL shows a decrease of intensity and a fixed peak wavelength (Figure 5-13). Similarly, for visible PL light emission the wavelengths were found to vary as a function of excitation wavelength, that is, of excitation energy [16]. Actually the 945 nm infrared emission is not an emission from a single level. This statement is supported by the result of a peak deconvolution analysis which yields that the 945 nm infrared photoluminescence of spark-processed silicon consists of two separate emissions (945 nm, 1010 nm) and these two emissions compete with each other with different heat treatment conditions. The infrared PL intensity variations as a function of annealing temperatures of spark-processed silicon show striking similarities with the 525 nm green PL of spark-processed silicon. Specifically, there are distinguishable PL increases observed after annealing in the range of 350 to 500C. This feature is commonly attributed to the conversion of chemisorbed water to oxide-related silanol-type groups (Si-O-H), which are known to contribute to the visible PL enhancement of spark-processed silicon after 450C annealing [95]. At temperature above 450C silanol desorbs or decomposes, eventually resulting in a dehydrogenated surface. At lower temperatures (<350C) the Si-OH groups are covered by water and therefore inefficient for PL, whereas after annealing at higher temperatures (>600C) they are no longer present. Furthermore, the 945nm as well as the 1010 nm deconvoluted peak wavelengths are essentially constant (2~3 nm) after annealing. This result is consistent with the behavior of visible PL of spark-processed silicon. The infrared and visible photoluminescence findings are compared in Table 6-1.

PAGE 179

158 Table 6-1 Comparison of near IR PL and visible PL of spark-processed silicon Near IR PL Visible PL Peak wavelength 945 nm 385 nm, 525 nm [50] Spark-processing time Positive dependence Positive dependence [96, 97] Spark-processing gap 1-2 mm 1 mm [50, 97] Excitation wavelength Negative dependence Negative dependence [16] Annealing behavior Max. at 450C Max. at 450C (green peak) [95] Another infrared emission of spark-processed silicon peaking near 800nm was reported by our group before [98]. This infrared emission, having the highest intensities after annealing at 900C, displayed features that were similar to the infrared photoluminescence found in strongly oxidized porous silicon [96]. However, this 800 nm infrared band is known to be activated only after annealing above 500C. In contrast, the 945 nm infrared band of the present study does not need previous annealing. In short, the 945 nm near infrared photoluminescence from spark-processed silicon has unique peak wavelength and different features compared to the already known near infrared light emission of spark-processed silicon. Porous silicon PL is known to be attributed to the low dimensionality of the surviving silicon skeleton [54] which cause quantum-size effects, or to the dangling bonds [99]. Porous silicon has also wide spectral tunability of PL from the near infrared through the visible range to the near UV, which must be activated in many cases by oxidizing heat treatments. Their emission energies and stability are however much smaller compared to spark-processed silicon. Such a broad range of emission energies arises from a number of distinct luminescent bands [100]. Table 6-2 presents the luminescent bands for porous silicon.

PAGE 180

159 Table 6-2 Porous silicon luminescence bands Spectral range Peak wavelength Lifetime Label PL EL UV ~350 nm 1 ns UV band Yes No Blue-green ~470 nm 1 ns F band Yes No Blue-red 400~800 nm 100 s S band Yes Yes Near IR 1100~1500nm IR band Yes No From Cullis et al. [100] As one can see from Table 6-2, no emission bands are observed for porous silicon between 800 and 1100 nm. In particular, there is no emission in the 945 nm infrared range. This suggests that spark-processed silicon has different luminescence centers than porous silicon in the infrared range. Another difference needs to be mentioned: The intensity for porous silicon PL decreases in aged samples [54]. In contrast, the infrared PL of spark-processed silicon shows little degradation even after several hours of operation or years of storage at ambient environment [50]. Si nanocrystals having different Si contents display PL at room temperature in the 700-1100 nm range as shown in Figure 6-1 [101]. The photoluminescence signal shows a blue shift with decreasing Si content presumably as a result of a smaller size of the silicon nanocrystals [101]. SiO x films with 44 at.% Si content display a similar infrared PL emission as spark-processed silicon. However, in contrast to spark-processed silicon, the PL of Si nanocrystals originates from a single emission level as comparison of Figures 6-1 and 5-5 demonstrate. It is therefore suggested that the emission structure seen in Figure 5-5 originates from a number of SiO x matrices which were produced by spark-processing. The visible PL of spark-processed silicon has been explained by interactions between the various luminescent levels and multiple de-excitation paths in the range of 1.7-2.8 eV (440-730 nm) and a broad absorption band between 3.2 and 6.2 eV (388 and

PAGE 181

160 200 nm)(Figure 6-2)[11]. Because of these closely spaced levels, a variation of the excitation wavelength results in a change of the emission wavelength in the visible range. These levels originate from Si-O clusters having nitrogen inclusions [11]. It is possible to introduce new emission levels by defects which occur through the spark-processing of silicon. For the infrared emission, new levels near 1.31 eV (945 nm) are introduced. Through non-radiative de-excitation paths the excited electrons can revert back into the lower infrared emission levels. Slightly spark-processed or virgin silicon may also form levels near the band-gap of silicon (1.12 eV, 1100 nm). There may be also some contributions from nanoparticles in spark-processed silicon. Based on the energy level diagram proposed by Hummel et al. [11], infrared emission levels discovered in this research are inserted in Figure 6-2. It is proposed that the electrons which have been pumped by the Ar ion laser from the ground state into a quasi band consisting of closely spaced absorption levels between 1.7 and 2.8 eV revert back to the lower levels at 1.31 and 1.22 eV by non-radiative transitions from which they revert to the ground state by emitting the observed IR radiation, as indicated in Figure 6-2. 6.2. Erbium Related Photoluminescence from Spark-Processed Silicon Erbium-doped spark-processed silicon photoluminesces efficiently at 1.54 m as shown in Figure 5-16. This can be explained by knowing that rare-earth ions such as erbium in their trivalent state show luminescence from intra-4f-shell transitions the wavelength of which is almost independent of the host matrix. Widening of the band-gap of the host material enables erbium-related photoluminescence to be enhanced. Spark-processing of silicon/erbium layers was found to result in efficient sensitizing of Er 3+ ions. Similar effects have been reported for erbium-doped porous silicon [102-104] and

PAGE 182

161 hydrogenated amorphous silicon [103, 105, 106] which have been used because of their wide band-gap in comparison to crystalline silicon. As reported above erbium-doped spark-processed silicon yields a well discernible peak near 1.54 m along with several other smaller peaks at longer wavelengths when exposed to 488 nm Ar laser light. This is interpreted to be due to transitions from the erbium first excited state 4 I 13/2 to the ground state 4 I 15/2 (see Figure 2-3). However, because of Stark splitting sublevels are generated, which are somewhat different for spark-processed silicon than others. This leads to similar but not identical PL spectra for erbium-doped spark-processed silicon from other techniques. 6.2.1. Efficiency of Photoluminescence The attainment of efficient luminescence from erbium in silicon requires four conditions: (1) the erbium must be incorporated in the optically active 3+ state; (2) high concentrations of erbium must be incorporated into the silicon matrix; (3) the useful transition process needs to be more efficient than competing routes of carrier recombination such as back transfer or Auger recombination; and (4) the excited Er 3+ should decay radiatively [54, 107]. In order to achieve condition (1) above, rapid thermal annealing is applied. This process increases the number of optically active trivalent erbium ions in erbium-doped spark-processed silicon. In other words, rapid thermal annealing yields optical activation of erbium centers, thus forming a coordination shell with oxygen atoms which inhibit non-radiative pathways for Er 3+ ions [108]. Moreover, annealing reduces the number of defects which decrease the carrier lifetime and act as non-radiative decay centers for Er 3+ . Furthermore, annealing may change the excitation probability and luminescence

PAGE 183

162 efficiency, and finally annealing can aid in transporting erbium to optically active sites. However, excessive annealing decreases the infrared PL intensity which may be the result of excessive oxidation, the suppression of transport channels, and an activation of chemical impurities introduced or activated in erbium-doped spark-processed silicon. Second, it has been shown in Figure 5-19 that the 1.54 m infrared photoluminescence intensity increases to a certain degree as a function of the initial erbium metal thickness deposited on the surface. However, at higher Er 3+ concentrations, self quenching of the infrared emission occurs due to erbium clustering or precipitation (Figure 4-8). Similar erbium precipitates have been observed to form during Er ion implantation at high doses, and have been interpreted to act as sites for non-radiative recombinations [108]. Specifically, excessive erbium seems to be trapped at the interface between spark-processed silicon and crystalline silicon during spark-processing. Indeed, the highest erbium concentrations were observed near the interface in SIMS depth profiles (Figure 4-14). However, the major part of the erbium at the interface is not activated as luminescence centers, that is, it just remains in metal form or precipitates, as dynamic SIMS depth profiles indicate (Figure 4-14). Third, a widening of the band-gap is postulated to result in small thermal quenching of the infrared PL in erbium-doped spark-processed silicon. A large band-gap may be due to high oxygen content. On the other hand, thermal quenching has been explained by others to be due to an increase in the probability of phonon assisted back transfer of the energy from Er 3+ to the host silicon crystals [109]. The band-gap of erbium-doped spark-processed silicon is much larger than that of crystalline silicon, and thus the back transfer becomes less likely, because the latter requires multiple phonons.

PAGE 184

163 Erbium-doped spark-processed silicon shows a 4 times increase in IR PL intensity when cooled from room temperature to 15 K. Similar small thermal quenching has been reported for erbium-doped porous silicon [110] and semi-insulating polycrystalline silicon (SIPOS) [109]. In particular, some authors claim that erbium-related PL from erbium-doped porous silicon is due to erbium atoms incorporated into amorphous Si:H:O regions in porous silicon [109, 111]. 6.2.2. Role of Oxygen in Efficient Photoluminescence Moreover, it needs to be noted that oxygen plays a key role on the activation and concentration of erbium. Specifically, it was found that the addition of oxygen to erbium-doped silicon enhances the concentration of luminescent erbium centers, and reduces the quenching of the PL at elevated temperatures [99, 107, 108, 112-115]. The activation of erbium in silicon demonstrates that large amounts of active erbium ions can be incorporated into a silicon matrix provided that the oxygen concentration is increased as well. It has been found that an optimal silicon to oxygen ratio in SiO x :Er is about 4 to 6, see Figure 6-3 [115]. However, spark-processed silicon has already significant amounts of oxygen in its matrix. Thus, more oxygen does not always guarantee a higher PL intensity in spark-processed silicon. In order to investigate this point, spark-processing (Figure 5-25) and annealing (Figure 5-26) was performed in oxygen. The result was that when spark-processing was carried out in pure oxygen, the infrared PL intensity was actually lower than the one for spark-processing in air. This means that the energy transfer from sp-Si processed in pure oxygen to erbium luminescence center is less efficient than when processed in air. Moreover, non-radiative silicon-erbium oxides resulting from excessive oxygen lower the overall efficiency of photoluminescence.

PAGE 185

164 One more point needs to be considered in this context: The efficiency of excitation may be connected with the presence of discrete and quasi-discrete electron states for Er 2 O 3 clusters in the matrix. These Er 2 O 3 clusters may be also a source of the erbium-related emission in erbium-doped spark-processed silicon. Indeed, Figure 4-5 indicated that Er 2 O 3 is present in erbium doped spark-processed silicon. 6.2.3. Energy Band Considerations Erbium may be excited in spark-processed silicon by direct absorption or through a photo-carrier-mediated process, that is, by the direct absorption of the excitation light (Figure 6-4) or by the energy transfer from spark-processed silicon (Figure 6-5). Excitation of Er 3+ occurs through the transition from the 4 I 15/2 state to the 4 F 7/2 level of Er 3+ , i.e. through a direct transition which resonates with the 488 nm (2.54 eV) wavelength of the incident photons (Figure 6-4). Subsequently, the ions decay non-radiatively to the first excited state ( 4 I 13/2 ) and then to the ground state ( 4 I 15/2 ), emitting photons at 1.54 m (Figure 6-4). Additionally, there may be also indirect excitation process involving spark-processed silicon (Figure 6-5). The off-resonant excitations (457 nm (2.71 eV), 515 nm (2.40 eV)) to Er 3+ levels are still efficient [116]. The agreement with the fine structures between the resonant (488 nm) and non-resonant (457 nm, 515 nm) excitations means that both PL emissions originate from the same kind of Er 3+ ions, that is, excitation levels. The just-mentioned energy transfer from spark-processed silicon to Er 3+ levels is now explained in detail. The excitation light is absorbed mainly by spark-processed silicon and the pumped electrons are experiencing radiative and/or non-radiative transitions. A part of the transition energy is then horizontally transferred to Er 3+ (Figure

PAGE 186

165 6-5). The amount of the energy transfer increases as the erbium concentration increases. This results in an increase of the 1.54 m peak and a quenching of visible spark-processed silicon peaks with increasing erbium concentration. This has been indeed observed in Figure 5-15 that is, two infrared PL peaks corresponding to emissions from energy levels in spark-processed silicon and involving intra 4f transition in Er 3+ are observed simultaneously but with variable intensity. Incidentally, the indirect pumping mechanism was reported to be much more efficient than the direct excitation of Er 3+ ions [117]. 6.2.4. Concluding Remarks The room temperature photoluminescence of erbium-doped spark-processed silicon in this study was found to have an efficiency of about 0.2-0.3%. On the other hand, other investigations reported for erbium-doped crystalline silicon efficiencies of around 0.05% [56]. Erbium in spark-processed silicon benefits from the advantages of both, silicon (efficient excitation) and SiO 2 (weak non radiative process), while it avoids their disadvantages (low excitation efficiency in SiO 2 and strong non-radiative processes in silicon). The coupling of the two approaches has been demonstrated to produce quite interesting effects. 6.3. Electroluminescence in Erbium-Doped Spark-Processed Silicon Preparing a device which provides electroluminescence in erbium-doped spark-processed silicon is somewhat difficult since it requires the right balance between good optical properties and good electrical conduction. It was succeeded to accomplish this

PAGE 187

166 task as described in Section 5.3. Under reverse biased * conditions, specifically in the breakdown regime, infrared electroluminescence of erbium-doped spark-processed silicon is detected. Erbium-doped silicon rich silicon oxide (SRSO), silicon nanoclusters and silica also display electroluminescence in the breakdown regime and also only under reverse bias [54, 85]. In contrast to this erbium-doped crystalline silicon, porous silicon, and silicon nanocrystals reveal electroluminescence under both forward and reverse bias [46, 54, 56, 86, 88, 118]. Even though electroluminescence under forward bias is possible, the intensities of the latter are 10-20 times less than that under reverse bias with impact excitation [56]. Additionally, electroluminescence under reverse bias is more stable [88]. Luminescence mechanisms involving carrier-mediated processes and resonant Auger recombination [56, 119] as well as exciton recombination [56, 120] have been proposed as possible mechanisms. Figure 6-6 depicts a schematic representation of the possible nano-silicon particles (NSP) related Er 3+ excitation mechanism in the EL process of an Au/SRSO:Er/n + -Si system under reverse bias. A similar mechanism could be considered for erbium-doped spark-processed silicon. However this mechanism occurs under carrier injection mode at forward bias. Thus, recombination of excitons is not the suggested mechanism for EL in erbium-doped spark-processed silicon. Instead impact excitation may be the preferred pumping mode under reverse bias condition for erbium-doped spark-processed silicon. The emission spectra in Figure 5-46 show that for erbium-doped spark-processed silicon neither a shift of the peak at 1.55 m nor other erbium related peaks around 1.55 * A reverse bias in this study means that the n-type silicon substrate was biased positively.

PAGE 188

167 m are observed in the working voltage range. Indeed, most of the parameters that increase the efficiency of electroluminescence coincide with the efficiency parameters for photoluminescence. The observed increase in photoluminescence as a function of the parameters cited below is a good indication of the enhancement of the number of optically active erbium ions [88]. The process parameters that resulted in the most efficient electroluminescence of erbium-doped spark-processed silicon were found to be 100nm of initial erbium layer thickness, 30 seconds of spark-processing time, n-type wafer, and 3 cm wafer resistivity. In electroluminescence devices, the wafer resistivity plays an important role as current pathway. Consequently a lower resistivity wafer enables a higher electroluminescence. Specifically, the infrared electroluminescence device intensities were observed to decrease with increasing wafer resistivity, indicating that the electrical characteristics of the bulk silicon are relevant to both the conduction and the electroluminescence performance. In contrast, in the case of photoluminescence, the silicon wafer resistivity does not play a major role to affect the emission intensity. The voltage, V B , for avalanche breakdown in silicon generally follows the equation [1]: V B = F 2 / 2 q N D (Equation 6-1) where is the dielectric permittivity, F is the breakdown field, q is the electron charge, and N D is the dopant concentration. The electroluminescence intensity dependence on wafer resistivity exhibits this behavior for avalanche breakdown. Further, the temperature

PAGE 189

168 dependence of the reverse characteristics exhibits a positive temperature coefficient as is typically observed for impact ionization. Figure 5-49 shows a typical I-V characteristic of an erbium-doped spark-processed silicon device. The overall shape is semi-symmetrical with signs of rectification in reverse bias. In reverse bias, tunneling of electrons into the spark-processed silicon may occur, especially at higher bias voltages, thus contributing to the current. The “softness” of the I-V characteristics for erbium-doped spark-processed silicon electroluminescence devices is possibly due to field enhancement, caused by the surface features that constitute the halo region around the sparked spot. Indeed, the low threshold voltage for electroluminescence in erbium-doped spark-processed silicon supports this suggestion. The higher conductivity of the erbium-doped spark-processed silicon compared to the native oxide (SiO x /SiO 2 ), could also be a contributing factor for current flow. The electroluminescence peak position of erbium-doped spark-processed silicon was found to be different from that for photoluminescence. The difference in peak wavelength between infrared photoluminescence and electroluminescence may be the result of optically and electrically excited erbium centers in different microenvironments within the active layer. It has been pointed out when displaying Figure 5-37 that electroluminescence of erbium-doped spark-processed silicon is less luminescent than photoluminescence of the same substance. The different roles of the spark-processed silicon in electroluminescence and photoluminescence processes may be responsible for this fact. For photoluminescence almost all photo-generated carriers stem from the spark-processed silicon. Thus, the photoluminescence intensity is strongly affected by spark-processing. In contrast, for electroluminescence of erbium-doped spark-processed silicon,

PAGE 190

169 the spark-processed silicon related Er 3+ excitation is just one of the various Er 3+ excitation paths. Other possible Er 3+ excitation processes for electroluminescence are: (1) injected electrons and holes. They can be captured by defects in Si and spark-processed silicon, and their non-radiative recombination can excite Er 3+ ; (2) the hot carriers generated in the breakdown process of the spark-processed silicon impact Er 3+ and excite them directly. One could argue that a generation of electron-hole pairs would occur due to the high electric field in the erbium-doped layer. The energy released through recombination of these electron-hole pairs could in turn excite erbium atoms. However, no emission bands related to the recombination of electron-hole pairs in the wavelength range from 0.8 to 1.7 m was observed. Thus the above suggestion is excluded. Moreover, the fact that no electroluminescence is observed in devices with ohmic characteristics makes it unlikely that a recombination of electron-hole pairs in nanoparticles is primarily responsible for the observed infrared light emission. The 1.55 m electroluminescence is instead interpreted to be caused by erbium ions that are excited through collisions with energetic electrons and accelerated by the electric field. In other words, the excitation mechanism in these devices is very likely to be due to impact by energetic carriers. Moreover, the observation that a rectifying contact is necessary for electron injection into the spark-processed layer is indicative that a hot electron process is active. Whether erbium is excited directly by hot carriers impact or through energy transfer from spark-processed silicon excited by hot carriers, is not distinguishable. In any case spark-processed silicon has a role in allowing a high current density to exist and hence an erbium excitation.

PAGE 191

170 The spark-processed layers can be used to generate hot electrons that are accelerated into the erbium-doped spark-processed silicon by the large applied field. These electrons enable efficient impact excitation of Er 3+ ions incorporated into the spark-processed silicon. Assuming that the voltage is applied across the high resistive layer underneath the electrodes (about 200 nm), the electric field in this region is estimated to be about 100 KV/cm at the threshold voltage of 5 V, which is large enough for electron impact excitation of Er atoms to occur. Taking the presented results in account, it is proposed in Figure 6-7 that in areas with sufficient field enhancement, facilitated by the existing surface features, electrons are injected with sufficient energy into the spark-processed silicon, where they generate excited electrons by impact ionization. These excited electrons then return to the ground state via radiative and non-radiative pathways. These energies are transferred to the Er 3+ luminescence center and excite electrons from the ground state ( 4 I 15/2 ) to the excited state ( 4 I 11/2 or 4 I 13/2 ). Finally, the energies dissipate radiatively to the ground state with a 1.54 m emission. Above some critical field, the fixed positive charge created by impact ionization causes significant enhancement of the cathode field. This results in an increased electron injection and impact ionization, which eventually leads to a runaway current and destructive breakdown. Erbium-doped spark-processed silicon electroluminescence devices have been found in this research to have a 0.007% external power efficiency. This value is about 2 orders of magnitudes smaller than what has been reported for erbium-doped crystalline silicon devices [87]. The most successful results claim a room-temperature emission

PAGE 192

171 with an external quantum efficiency of 0.1% in a megahertz-modulated Er 3+ based LED [121]. In order to obtain efficient electroluminescence, not only the current should easily flow through the device and excite the optically active erbium centers, but also the medium itself should not interfere with the emission. To improve the emission intensity, various attempts for finding better suited materials as emission windows were made. Spark-processed silicon has a rough and porous surface. Thus, it is impossible to make an opening, which is not covered with electric contact materials at the spark-processed side of sample. A conducting material is needed, which can cover the entire area of light emission. ITO was considered as a contact electrode material, because of its transparency and reasonable conductivity. But this transparency is situated mostly in the visible range. For =1550 nm, the index of refraction for ITO is n=1.28 and the damping constant is k=0.12. Using Equation 6-2 [90], k W 4 (Equation 6-2) it can be calculated that ITO has a penetration depth, W, of 1.02 m at 1550 nm. For comparison, silicon with n=3.42 and k0 at 1550 nm, the penetration depth is essentially “infinity”: that is, silicon is transparent in this spectral range. Thus, silicon is utilized as an emission window in this research. In conclusion of this section, the electroluminescence properties of light emitting devices based on erbium-doped spark-processed silicon were studied. They were stable, exhibit an electroluminescence at room temperature, and a weak temperature dependence of the electroluminescence signal. Further improvements in efficiency are needed before

PAGE 193

172 erbium-doped spark-processed silicon devices can be commercially used, but present and expected improvements are very promising. In comparison to erbium-doped crystalline silicon devices prepared by ion implantation, epitaxy, or chemical-vapor deposition, erbium-doped spark-processed silicon devices are simpler and of lower cost because they do not require specialized and expensive equipment to produce. This will make erbium-doped spark-processed silicon attractive and desirable for applications. As a final remark, it should be noted that the processes described for erbium-doped spark-processed silicon can be also utilized for several other rare earths such as Ce, Eu, Er and Tm (see Appendix and [122]). 6.4. Future Work 1) Further research on the metal contact and the current flow of the EL devices is suggested. The Current applied from the source are delivered to the device through metal contacts. Some of the current does not flow through the erbium-doped spark-processed silicon but instead is side-tracked through the metal surface or the un-processed silicon. Thus, further research for better and more efficient ways to supply the current to the erbium-doped spark-processed silicon needs to be done. 2) A time-resolved luminescence study is suggested which can lead to a precise decay dynamics in erbium-doped spark-processed silicon. 3) Photoluminescence and electroluminescence of other rare-earth-doped spark-processed silicon should be studied. Due to the various energy levels of rare-earth ions, light emission having various colors (from UV to IR) involving spark-processed silicon is possible. Also, color combination by using more than one kind of rare-earth ion would be interesting to study.

PAGE 194

173 4) Other transparent and conductive electrode materials need to be found. This project would involve absorption studies for rare-earths-doped spark-processed silicon. 5) Research on utilizing various spark-processing voltages, frequencies, and currents to fabricate a different depth profile of rare-earth-doped spark-processed silicon is suggested. Investigations for finding spark-processing conditions to make smoother and more regular surfaces of the spark-processed silicon should be performed as has been previously done by utilizing spray spark-processing [52, 53]. 6) Another suggestion related to this research is applying spark-processing as a doping technique which is not limited to silicon and rare-earths, but to Mn-doped spark-processed graphite, or Ga/N-doped spark-processed Zn. 7) Eventually micro-light sources for on-chip analysis, such as bio-chips, biosensors, micro fluorescent displays and efficient silicon based micro-light sources need to be developed which can be integrated into silicon integrated circuits.

PAGE 195

174 Figure 6-1 Normalized PL spectra of SiO x thin films having different silicon concentrations, annealed at 1250C for 1 hr. The spectra were measured at room temperature, with a laser pump power of 10 mW. After [101]

PAGE 196

175 Figure 6-2 Band level diagram for room temperature photoluminescence of spark-processed silicon [11]. Newly introduced infrared emissions are inserted in this diagram. The dotted arrows indicate non-radiative transitions.

PAGE 197

176 Figure 6-3 Schematic representation of first coordination shell surrounding erbium species in Float-zone-Si (left) and Czochralski-Si. The actual spatial orientations of Si and O atoms may not be as depicted, but an attempt is made to show that the optically active species are noncentrosymmetric. Also, the configuration of the host Si atoms in the second coordination shell are not indicated for either erbium species. After [115].

PAGE 198

177 0.00.51.01.52.02.53.03.54.0 4I15/24I11/24I13/24I9/24F9/24S3/22H11/24F7/24F5/24F3/22H9/22G11/2Energy (eV) 1.54m488nm Figure 6-4 Infrared photoluminescence mechanisms of erbium-doped spark-processed silicon. 1.54 m infrared photoluminescence model for direct excitation and absorption from excitation source of the erbium luminescence center

PAGE 199

178 Figure 6-5 Energy transfer model for the 1.54 m infrared photoluminescence mechanism of erbium-doped spark-processed silicon.

PAGE 200

179 Figure 6-6 Schematic representation of the possible nano silicon particles (NSP) related Er 3+ excitation mechanism in the EL process of an Au/SRSO:Er/n + -Si system under reverse bias. After [85].

PAGE 201

180 Figure 6-7 Suggested model for the 1.54 m infrared EL mechanism. (1) Electrons are injected into the spark-processed silicon by tunneling. (2) Electrons are accelerated to high velocities leading to impact ionization. (3) Impact ionization results in the generation of excited electrons. (4) Excited electrons excite Er 3+ from the ground state to the 4 I 11/2 or 4 I 13/2 state. (5) Electrons relax radiatively to the ground state from the 4 I 13/2 state.

PAGE 202

CHAPTER 7 SUMMARY Three major sub-projects involving infrared light emission, based on spark-processed silicon, were investigated in this work. First, the infrared photoluminescence emission of spark-processed silicon containing no other constituents was investigated. A broad, strong infrared photoluminescence which peaks in the 945 nm (1.31 eV) spectral range was observed for the first time. The 945 nm photoluminescence is different from the photoluminescence commonly reported for anodic etched porous silicon and other silicon based materials. The PL intensity increases moderately after annealing between 350 and 500C in air and then increases even more after annealing above 600C. No noticeable change in the peak wavelength is observed with annealing time and temperature. The PL intensity, however, increases with increasing spark-processing time. The silicon to oxygen ratio was investigated with EDS. It was found that the most efficient PL occurs for a Si/O ratio of 0.3. On the other hand, the most efficient infrared PL was achieved when a smaller spark gap distance of about 1mm was applied. Excitations with an Argon ion laser were accomplished using wavelengths such as 457, 488, and 515 nm. The most efficient pumping wavelength was the 457 nm line. A model for the infrared PL is proposed which mirrors that for visible PL. Specifically, it is proposed that the electrons which have been pumped by the laser from the ground state into a broad quasi absorption band or closely-spaced absorption lines between 1.7 and 2.8 eV revert back to lower infrared levels at 1.31 and 1.22 eV by non181

PAGE 203

182 radiative transition from where they revert radiatively to the ground state by emitting the observed 945 nm and 1010 nm light.. Second, the infrared photoluminescence of erbium-doped spark-processed silicon was studied. The samples were prepared by depositing a 100 nm thick Er layer on cleaned, surface oxidized Si wafers that were then spark-processed for 30 seconds in air. Subsequently, the resulting samples were annealed at 900C for 15 minutes in air. Spark-processing causes the high-energy ablation of the surface, which in turn causes erbium centers to become trapped in the condensed layer during the off-time of spark-processing. A room temperature photoluminescence spectrum for erbium-doped spark-processed silicon is observed that has a main peak near 1545 nm and side peaks at 1559, 1567, and 1587 nm. The spectrum is characteristic for the first excited state 4 I 13/2 to the ground state 4 I 15/2 transitions in Er 3+ . The infrared PL intensity increases when the thickness of the Er layer is increased. It was further found that the spark-processing time which yielded the optimal PL intensity in the IR was in the 30-60 second range. Spark-processing times beyond 60 seconds resulted in a moderate decrease in infrared PL intensity. Spark-processing in air at atmospheric pressure, and annealing in air at 900C yielded the most efficient PL. The highest infrared PL intensity of erbium-doped spark-processed silicon was obtained utilizing a silicon substrate resistivity near 13 cm. At low temperatures down to helium temperature the PL intensity of erbium-doped spark-processed silicon is enhanced by a factor of 4 compared to room temperature. Infrared PL showed a high stability at room temperature and even at elevated temperatures. The infrared PL intensifies with increasing excitation power. The infrared PL intensity has its maximum intensity by using the 488 nm line, which is resonant with Er 4 F 7/2 levels. When etched in

PAGE 204

183 HF for 5 minutes, erbium-doped spark-processed silicon still had 50% of its original PL intensity. For explanation of the infrared PL of erbium-doped spark-processed silicon it is suggested that the excitation light is absorbed mainly by the spark-processed silicon. The excited electrons are then partially transferred to Er 3+ resulting the 1.54 m peak and a quenching of visible silicon peaks with increasing erbium concentration. Emission centers are proposed to be surface or interface related, akin to oxygen defects. Third, EL devices have been prepared by utilizing identical spark-processing procedures as for PL samples, and depositing additional Ag and Al metallizations to form a window at the bottom of the wafer. The resulting infrared EL spectra are observed to have peaks near 1550 nm (0.81 eV). The spectra display the characteristic light emission of the first excited state 4 I 13/2 to ground state 4 I 15/2 transition for the Er 3+ ion. The EL spectrum is observed as a convolution of several peaks. EL is affected by erbium concentration and spark-processing time similar to PL. Both, n-type and p-type silicon substrates utilized in the devices displayed EL, however the n-type devices were more efficient. The increasing photoluminescence is a good indication of increasing the number of optically active erbium ions. Most of the conditions which have been found to increase the PL efficiency also enhance the EL of erbium-doped spark-processed silicon. However, the EL is affected more by the resistivity of the silicon substrate than PL. The device current dependence on wafer resistivity and the temperature dependence of the reverse current of erbium-doped spark-processed silicon both exhibit a positive temperature coefficient as is known to occur for impact ionization. Thus, it is suggested that the spark-processed layers generate hot electrons that are accelerated into the erbium

PAGE 205

184 doped spark-processed silicon by the large applied field (100 KV/cm). These hot electrons enable an efficient impact excitation of Er 3+ ions the latter of which are incorporated into the spark-processed silicon. The erbium-doped spark-processed silicon electroluminescence device developed in this work was found to have an external power efficiency of 0.007%. The erbium-doped spark-processed silicon devices were found to be robust and stable. They show promising results, but further improvements are needed before they can be incorporated in commercial applications. In comparison to erbium-doped crystalline silicon devices prepared by ion implantation, epitaxy, or chemical-vapor deposition, erbium-doped spark-processed silicon devices are simpler and can be fabricated at lower cost because they do not require specialized and expensive equipment. This may make erbium-doped spark-processed silicon attractive and desirable for applications.

PAGE 206

APPENDIX VISIBLE PHOTOLUMINESCENCE OF RARE-EARTH DOPED SPARK-PROCESSED SILICON So far, the infrared PL of erbium-doped spark-processed silicon was presented only in the infrared. In this chapter, the visible PL of some rare-earth-doped park-processed silicon sample is shown. Specifically, erbium, cerium, and europium on silicon were investigated. As before, rare-earth-doped spark-processed silicon was prepared using rare-earth depositions on a silicon wafer of around 100 nm, and applying 30 seconds of spark-processing, followed by annealing at high temperatures. A HeCd 325 nm laser excitation was utilized for the visible PL measurements. Erbium-doped spark-processed silicon displays sharp PL peaks superimposed on a broad spark–process silicon spectrum. This sharp emission originates from transitions between the trivalent Er energy levels of 2 H 11/2 (525 nm) and 4 S 3/2 (558 nm) to the ground level. As the thickness of erbium layer increases, the PL peak from erbium ions increases in intensity, and the broad spark-processed silicon peak decreases. The optimal spark-processing time is 10-30 seconds after which the erbium peak decreases and the spark-processed silicon peaks increases. Longer spark-processing time result in “negative” peaks superimposed on the broad spark-processed silicon spectrum. Trivalent erbium ions are considered to absorb the light emission from the spark-processed silicon. Cerium-doped spark-processed silicon displays a broad peak near 460 nm, that is a shift of the PL peak wavelength known for virgin spark-processed silicon (385 nm). As 185

PAGE 207

186 spark-processing times increase, this peak eventually reverts back to the virgin spark-processed silicon PL peak wavelength of 385 nm. Europium-doped spark-processed silicon displays similar characteristics as erbium-doped spark-processed-silicon. Europium causes a sharp peak superimposed on the broad spark-processed silicon peak. These sharp peak stems from the trivalent Eu ions 5 D 0 7 F 2 (611 nm) energy levels. The europium-doped spark-processed silicon PL peak intensity decreases with spark processing time after 20 seconds. The samples shown in this Appendix display various characteristics a s described in the figure captions. They are self-explanatory and will not be further described in this section for brevity.

PAGE 208

187 3504004505005506006507000.00.20.40.60.81.0 EuropiumCerium Normalized PL Intensity (a.u.)Wavelength (nm)Erbium Figure A-1 Visible photoluminescence spectra for erbium, cerium and europium-doped spark-processed silicon. Erbium PL has a characteristic peak near 550 nm. Cerium displays a broad peak at 460 nm. Europium has a broad maximum near 540 nm and a sharp peak near 617 nm.

PAGE 209

188 A.1. Visible PL of Erbium-Doped Spark-Processed Silicon 35040045050055060065070050001000015000200002500030000350004000045000500005500060000 05101520 360370380390400410420430440 Peak Position (nm)Spark Processing Time (s) PL Intensity (a.u.)Wavelength (nm)1s3s5s10s20s Figure A-2 Visible PL spectra of the erbium-doped spark-processed silicon as a result of HeCd laser (325 nm) excitation for short spark-processing times. The 3 arrows indicate characteristic peaks. The inset indicates the wavelength change of the broad maxima as a function of spark-processing time.

PAGE 210

189 3504004505005506006507000100000200000300000400000500000600000 510515520525530535540 240000245000250000255000260000265000270000275000 PL Intensity (a.u.)Wavelength (nm) PL Intensity (a.u.)Wavelength (nm)1min20s10s5min15min1hour Figure A-3 Visible PL spectra of erbium-doped spark-processed silicon as a function of spark-processing time under HeCd laser (325 nm) excitation. The erbium peak disappears after 1 minute spark-processing. For longer spark-processing times (15 minutes, 1 hour) a negative peak is observed. The inset indicates a PL spectrum for erbium-doped spark-processed silicon after 1 hour spark-processing.

PAGE 211

190 350400450500550600650700 600080001000012000140001600018000200002200024000 3.43.232.82.62.42.221.8 558nm537nm2H11/24S3/2 PL Intensity (a.u.)Wavelength (nm)525nm Energy (eV) Figure A-4 Visible PL spectrum of erbium-doped spark-processed silicon after rapid thermal annealed in air for 15 minutes. The sharp peaks are consistent with Er 3+ energy levels ( 2 H 11/2 , 4 S 3/2 ).

PAGE 212

191 203040506070010000200003000040000500006000070000 Erbium thickness (nm)PL Intensity (a.u.)425nm524nm552nm Figure A-5 PL peak intensities as a function of erbium layer thickness. The solid circles () and the solid squares () indicate the sharp erbium peaks and the open triangles () indicate the broad spark-processed silicon related peak. Erbium ion characteristic peaks increase with erbium layer thickness, but at the same time the broad peak decreases.

PAGE 213

192 3504004505005506006507000.00.51.01.52.02.53.0 20nm50nm PL Intensity (a.u.)Wavelength (nm)70nm Figure A-6 Normalized visible PL spectra of erbium-doped spark-processed silicon having various erbium layer thicknesses. As the erbium layers become thicker, the broad peak decreases and the sharp Er 3+ peak increases in intensity.

PAGE 214

193 350400450500550600650700500010000150002000025000300003500040000 wavelength (nm)PL Intensity (a.u.) Figure A-7 Visible PL spectrum for erbium-doped spark processed silicon, spark-processed for 10 seconds with subsequent annealing in pure argon environment for 20 minutes. No Er 3+ characteristic peaks are observed.

PAGE 215

194 A.2. Visible PL of Cerium-Doped Spark-Processed Silicon 350400450500550600650700020000400006000080000100000120000140000160000 PL Intensity (a.u.)Wavelength (nm)spark processed ceriumcerium doped spark processed siliconcerium Figure A-8 Visible PL spectra of cerium metal, spark-processed cerium, and cerium-doped spark-processed silicon. Cerium-doped spark-processed silicon was prepared using a 100 nm cerium layer on a silicon wafer and 20 seconds spark-processing.

PAGE 216

195 02004006008001000100000120000140000160000180000200000220000240000260000280000300000 PL Intensity (a.u.)Temperature (C) Figure A-9 PL intensity as a function of rapid thermal annealing temperature of cerium-doped spark-processed silicon. The sample was prepared by depositing Ce on Si by PVD, and spark-processing for 1 minute, followed by annealing for 30 minutes in air.

PAGE 217

196 010020030040050060060008000100001200014000160001800020000 PL Intensity (a.u.)Sparking time (sec) Figure A-10 PL intensity of the 460 nm peak of cerium-doped spark-processed silicon as a function of spark-processing time.

PAGE 218

197 0123456380400420440460480 Peak Position Wavelength (nm)Distance from the spark center (mm) Figure A-11 PL intensity of cerium-doped spark-processed silicon as a function of distance from the center spot area.

PAGE 219

198 A.3. Visible PL of Europium-Doped Spark-Processed Silicon 010203040506002000004000006000008000001000000 350400450500550600650700750 020000040000060000080000010000001200000 60sec spark processedPL Intensity (a.u.)Wavelength (nm)20sec spark processedunprocessed PL Intensity (a.u.)Spark Processing Time (s) Figure A-12 PL intensity of 611 nm europium-doped spark-processed silicon as a function of spark-processing time.

PAGE 220

199 3504004505005506006507000500010000150002000025000300003500040000 25oC 200oC 300oC 400oC 500oC 600oC 700oC 800oC 900oC 1000oC 1100oCPL Intensity (a.u.)Wavelength(nm) Figure A-13 PL spectra of europium-doped spark-processed silicon as a function of various rapid thermal annealing temperatures, annealing time of 15 minutes. Europium was deposited by PVD on a silicon substrate and spark-processed for 30 seconds.

PAGE 221

200 0200400600800100012000246810121416 I611nm020040060080010001200 05000100001500020000250003000035000 PL Intensity (a.u.)Annealing Temperature (oC)020040060080010001200 02000400060008000100001200014000160001800020000 PL Intensity(a.u.)Annealing Temperature (oC) I611nm / I550nmAnnealing Temperature(oC)I550nm Figure A-14 PL intensity ratio as a function of annealing temperature. I 611nm indicates the Eu 3+ characteristic peak intensity and I 550nm indicates the broad spark-processed silicon peak.

PAGE 222

201 010203040506045678 PL Intensity (a.u.)Spark Processing Time (min) Figure A-15 PL intensity at 611 nm of europium-doped spark-processed silicon as a function spark-processing time.

PAGE 223

LIST OF REFERENCES 1. S. M. Sze, Semiconductor Devices Physicsand Technology, John Wiley & Sons, New York, 1985. 2. L. T. Canham, Applied Physics Letters 57 (1990) 1046-1048. 3. J. Yuan, D. Haneman, I. Andrienko, R. Siegele, P. Evans, Semiconductor Science and Technology 13 (1998) 615-621. 4. J. Yuan, D. Haneman, Journal of Applied Physics 86 (1999) 2358-2360. 5. J. Yuan, D. Haneman, Applied Physics Letters 67 (1995) 3328-3330. 6. N. D. Shepherd, Visible Electroluminescience in Spark-Processed Silicon, Department of Materials Science and Engineering, Ph.D. dissertation, University of Florida, Gainesville, 2001, p. 150. 7. S. Rupp, J. Quilty, H. J. Trodahl, M. H. Ludwig, R. E. Hummel, Applied Physics Letters 70 (1997) 723-725. 8. M. H. Ludwig, R. E. Hummel, M. Stora, Thin Solid Films 255 (1995) 103-106. 9. M. H. Ludwig, R. E. Hummel, S. S. Chang, Journal of Vacuum Science & Technology B 12 (1994) 3023-3026. 10. R. E. Hummel, M. E. Stora, N. Shepherd, S. Yu, F. Fajardo, Journal of Porous Materials 7 (2000) 131-134. 11. R. E. Hummel, N. Shepherd, D. Burton, Applied Physics Letters 79 (2001) 3218-3220. 12. R. E. Hummel, A. Morrone, M. Ludwig, S. S. Chang, Applied Physics Letters 63 (1993) 2771-2773. 13. R. E. Hummel, M. H. Ludwig, S. S. Chang, P. M. Fauchet, J. V. Vandyshev, L. Tsybeskov, Solid State Communications 95 (1995) 553-557. 14. R. E. Hummel, M. H. Ludwig, S. S. Chang, Solid State Communications 93 (1995) 237-241. 15. R. E. Hummel, M. H. Ludwig, Journal of Luminescence 68 (1996) 69-76. 202

PAGE 224

203 16. R. E. Hummel, M. Ludwig, S. S. Chang, G. Latorre, Thin Solid Films 255 (1995) 219-223. 17. R. E. Hummel, S. S. Chang, Applied Physics Letters 61 (1992) 1965-1967. 18. H. Morisaki, H. Hashimoto, F. W. Ping, H. Nozawa, H. Ono, Journal of Applied Physics 74 (1993) 2977-2979. 19. M. Romand, M. Charbonnier, Journal of Adhesion Science and Technology 15 (2001) 945-965. 20. M. F. Cerqueira, J. Jansen, G. J. Adriaenssens, J. A. Ferreira, Thin Solid Films 296 (1997) 118-121. 21. M. F. Lambrinos, R. Valizadeh, S. J. Colligon, Nuclear Instruments & Methods in Physics Research Section B-Beam Interactions with Materials and Atoms 127 (1997) 369-374. 22. C. Ciofi, A. Diligenti, A. Nannini, G. Pennelli, N. Vannucci, F. Fuso, M. Allegrini, Solid State Phenomena 54 (1997) 109-118. 23. O. Hanaizumi, K. Ono, Y. Ogawa, Applied Physics Letters 82 (2003) 538-540. 24. E. C. Cho, J. Xia, A. G. Aberle, M. A. Green, Solar Energy Materials and Solar Cells 74 (2002) 147-154. 25. M. Medvedeva, I. Wojciechowski, B. J. Garrison, Surface Science 505 (2002) 349-357. 26. L. Khomenkova, N. Korsunska, T. Torchynska, V. Yukhimchuk, B. Jumayev, A. Many, Y. Goldstein, E. Savir, J. Jedrzejewski, Journal of Physics-Condensed Matter 14 (2002) 13217-13221. 27. F. S. Pool, J. M. Essick, Y. H. Shing, R. T. Mather, Thin Solid Films 240 (1994) 152-156. 28. K. Luterova, P. Fojtik, A. Poruba, J. Dian, J. Valenta, H. Stuchlikova, J. Stepanek, J. Kocka, I. Pelant, Journal of Non-Crystalline Solids 266 (2000) 583-587. 29. K. Luterova, A. Poruba, J. Dian, O. Salyk, P. Horvath, P. Knapek, J. Valenta, J. Kocka, I. Pelant, Journal of Porous Materials 7 (2000) 135-138. 30. K. M. A. Elkader, I. Ulrych, V. Chab, J. Oswald, P. Kubat, P. Engst, J. Stuchlik, A. Fejfar, R. Cerny, I. Pelant, J. Kocka, Thin Solid Films 255 (1995) 302-304. 31. T. Makino, Y. Yamada, N. Suzuki, T. Yoshida, S. Onari, Applied Surface Science 197 (2002) 594-597.

PAGE 225

204 32. V. Craciun, N. Bassim, R. K. Singh, D. Craciun, J. Hermann, C. Boulmer-Leborgne, Applied Surface Science 186 (2002) 288-292. 33. G. A. Kachurin, I. E. Tyschenko, K. S. Zhuravlev, N. A. Pazdnikov, V. A. Volodin, A. K. Gutakovsky, A. F. Leier, W. Skorupa, R. A. Yankov, Nuclear Instruments & Methods in Physics Research Section B-Beam Interactions with Materials and Atoms 122 (1997) 571-574. 34. S. Coffa, G. Franzo, F. Priolo, S. Libertino, R. Mosca, E. Gombia, C. Spinella, Nuclear Instruments & Methods in Physics Research Section B-Beam Interactions with Materials and Atoms 120 (1996) 74-80. 35. M. Tajima, Journal of Crystal Growth 237 (2002) 324-329. 36. M. Tajima, S. Ibuka, S. Arai, Materials Science and Engineering B-Solid State Materials for Advanced Technology 91 (2002) 10-15. 37. N. A. Sobolev, Materials Science and Engineering B-Solid State Materials for Advanced Technology 91 (2002) 189-191. 38. R. H. Magruder, R. A. Weeks, R. A. Weller, R. A. Zuhr, Journal of Non-Crystalline Solids 304 (2002) 224-232. 39. F. Xu, Z. S. Xiao, G. A. Cheng, Z. Z. Yi, T. H. Zhang, L. L. Gu, X. Wang, Thin Solid Films 410 (2002) 94-100. 40. H. Ennen, J. Schneider, G. Pomrenke, A. Axmann, Applied Physics Letters 43 (1983) 943-945. 41. H. Ennen, U. Kaufmann, G. Pomrenke, J. Schneider, J. Windschief, A. Axmann, Journal of Crystal Growth 64 (1983) 165. 42. M. Winter, WebElementsTM, the periodic table on the WWW, http://www.webelements.com/webelements/elements/text/Er/key.html, 1993. 43. A. Polman, Physica B-Condensed Matter 300 (2001) 78-90. 44. A. Polman, Journal of Applied Physics 82 (1997) 1-39. 45. L. Tsybeskov, S. P. Duttagupta, K. D. Hirschman, P. M. Fauchet, K. L. Moore, D. G. Hall, Applied Physics Letters 70 (1997) 1790-1792. 46. H. A. Lopez, P. M. Fauchet, Applied Physics Letters 77 (2000) 3704-3706. 47. M. Markmann, E. Neufeld, A. Sticht, K. Brunner, G. Abstreiter, C. Buchal, Applied Physics Letters 75 (1999) 2584-2586. 48. A. R. Zanatta, L. A. O. Nunes, L. R. Tessler, Applied Physics Letters 70 (1997) 511-513.

PAGE 226

205 49. W. Jantsch, S. Lanzerstorfer, L. Palmetshofer, M. Stepikhova, G. Kocher, H. Preier, Physica B-Condensed Matter 274 (1999) 330-333. 50. R. E. Hummel, "Photonic and Magnetic Properties of Spark-Processed Silicon", in: H. S. Nalwa (Ed.) Silicon-Based Materials and Devices, vol 1, Academic Press, 2001, pp. 237-266. 51. J. V. StJohn, J. L. Coffer, Y. G. Rho, R. F. Pinizzotto, Applied Physics Letters 68 (1996) 3416-3418. 52. N. Shepherd, R. E. Hummel, Journal of Physics and Chemistry of Solids 64 (2003) 967-974. 53. N. Shepherd, R. E. Hummel, Physica Status Solidi a-Applied Research 197 (2003) 222-227. 54. S. Ossicini, L. Pavesi, F. Priolo, Light emitting silicon for microphotonics, Springer, 2003, p. 282. 55. J. V. St John, J. L. Coffer, Y. G. Rho, P. Diehl, R. F. Pinizzotto, T. D. Culp, K. L. Bray, MRS Symposium Proceedings 486 (1998) 281. 56. G. Franzo, F. Priolo, S. Coffa, A. Polman, A. Carnera, Applied Physics Letters 64 (1994) 2235-2237. 57. H. Przybylinska, W. Jantsch, Y. SuprunBelevitch, M. Stepikhova, L. Palmetshofer, G. Hendorfer, A. Kozanecki, R. J. Wilson, B. J. Sealy, Physical Review B 54 (1996) 2532-2547. 58. F. Priolo, G. Franzo, S. Coffa, A. Carnera, Physical Review B 57 (1998) 4443-4455. 59. W. Jantsch, G. Kocher, L. Palmetshofer, H. Przybylinska, M. Stepikhova, H. Preier, Materials Science and Engineering B-Solid State Materials for Advanced Technology 81 (2001) 86-90. 60. M. J. A. de Dood, P. G. Kik, J. H. Shin, A. Polman, MRS Symposium Proceedings 422 (1996) 219-225. 61. P. G. Kik, M. J. A. de Dood, K. Kikoin, A. Polman, Applied Physics Letters 70 (1997) 1721-1723. 62. J. Stimmer, A. Reittinger, J. F. Nutzel, G. Abstreiter, H. Holzbrecher, C. Buchal, Applied Physics Letters 68 (1996) 3290-3292. 63. S. Coffa, G. Franzo, F. Priolo, Applied Physics Letters 69 (1996) 2077-2079. 64. P. G. Kik, A. Polman, MRS Bulletin 23 (1998) 48.

PAGE 227

206 65. L. H. Slooff, M. J. A. de Dood, A. van Blaaderen, A. Polman, Applied Physics Letters 76 (2000) 3682-3684. 66. E. Snoeks, A. Lagendijk, A. Polman, Physical Review Letters 74 (1995) 2459-2462. 67. U. Wahl, J. G. Correia, J. P. Araujo, A. Vantomme, G. Langouche, I. Collaboration, Physica B-Condensed Matter 273-274 (1999) 342. 68. C. K. Ryu, H. Choi, K. Kim, Applied Physics Letters 66 (1995) 2496-2498. 69. X. Orignac, D. Barbier, X. M. Du, R. M. Almeida, Applied Physics Letters 69 (1996) 895-897. 70. M. Benatsou, B. Capoen, M. Bouazaoui, W. Tchana, J. P. Vilcot, Applied Physics Letters 71 (1997) 428-430. 71. A. M. Dorofeev, N. V. Gaponenko, V. P. Bondarenko, E. E. Bachilo, N. M. Kazuchits, A. A. Leshok, G. N. Troyanova, N. N. Vorosov, V. E. Borisenko, H. Gnaser, W. Bock, P. Becker, H. Oechsner, Journal of Applied Physics 77 (1995) 2679-2683. 72. O. B. Gusev, A. N. Kuznetsov, E. I. Terukov, M. S. Bresler, V. K. Kudoyarova, I. N. Yassievich, B. P. Zakharchenya, W. Fuhs, Applied Physics Letters 70 (1997) 240-242. 73. M. S. Bresler, W. Fuhs, T. Gregorkiewicz, O. B. Gusev, P. E. Pak, E. I. Terukov, K. D. Tsendin, I. N. Yassievich, Physica B-Condensed Matter 274 (1999) 354-357. 74. I. N. Yassievich, M. S. Bresler, O. B. Gusev, Journal of Physics-Condensed Matter 9 (1997) 9415-9425. 75. M. Fujii, M. Yoshida, Y. Kanzawa, S. Hayashi, K. Yamamoto, Applied Physics Letters 71 (1997) 1198-1200. 76. P. G. Kik, A. Polman, Materials Science and Engineering B-Solid State Materials for Advanced Technology 81 (2001) 3-8. 77. M. Fujii, M. Yoshida, S. Hayashi, K. Yamamoto, Journal of Applied Physics 84 (1998) 4525-4531. 78. G. Franzo, V. Vinciguerra, F. Priolo, Applied Physics a-Materials Science & Processing 69 (1999) 3-12. 79. C. E. Chryssou, A. J. Kenyon, T. S. Iwayama, C. W. Pitt, D. E. Hole, Applied Physics Letters 75 (1999) 2011-2013.

PAGE 228

207 80. C. E. Chryssou, A. J. Kenyon, C. W. Pitt, Materials Science and Engineering B-Solid State Materials for Advanced Technology 81 (2001) 16-18. 81. A. J. Kenyon, C. E. Chryssou, C. W. Pitt, T. Shimizu-Iwayama, D. E. Hole, N. Sharma, C. J. Humphreys, Journal of Applied Physics 91 (2002) 367-374. 82. H. Takagi, H. Ogawa, Y. Yamazaki, A. Ishizaki, T. Nakagiri, Applied Physics Letters 56 (1990) 2379-2380. 83. M. L. Brongersma, A. Polman, K. S. Min, E. Boer, T. Tambo, H. A. Atwater, Applied Physics Letters 72 (1998) 2577-2579. 84. M. Fujii, S. Hayashi, K. Yamamoto, Applied Physics Letters 73 (1998) 3108-3110. 85. G. Z. Ran, Y. Chen, F. C. Yuan, Y. P. Qiao, J. S. Fu, Z. C. Ma, W. H. Zong, G. G. Qin, Solid State Communications 118 (2001) 599-602. 86. H. A. Lopez, P. M. Fauchet, Applied Physics Letters 75 (1999) 3989-3991. 87. S. Coffa, G. Franzo, F. Priolo, MRS Bulletin 23 (1998) 25. 88. H. A. Lopez, P. M. Fauchet, Materials Science and Engineering B-Solid State Materials for Advanced Technology 81 (2001) 91-96. 89. R. E. Hummel, "Differential Reflectance Spectroscopy in Analysis of Surfaces", in: R. A. Meyers (Ed.) Encyclopedia of Analytical Chemistry, Wiley, Chichester, 2000, pp. 9047-9071. 90. R. E. Hummel, Electronic Properties of Materials, Springer, New York, 2001, p. 438. 91. R. E. Hummel, W. Xi, P. H. Holloway, K. A. Jones, Journal of Applied Physics 63 (1988) 2591-2594. 92. R. Plugaru, J. Piqueras, E. Nogales, B. Mendez, J. A. Garcia, T. J. Tate, Journal of Optoelectronics and Advanced Materials 4 (2002) 883-892. 93. S. B. Aldabergenova, H. P. Strunk, P. C. Taylor, A. A. Andreev, Journal of Applied Physics 90 (2001) 2773-2780. 94. P. H. Citrin, P. A. Northrup, R. Birkhahn, A. J. Steckl, Applied Physics Letters 76 (2000) 2865-2867. 95. R. E. Hummel, M. H. Ludwig, J. Hack, S. S. Chang, Solid State Communications 96 (1995) 683-687. 96. M. Ludwig, Solid State and Materials Sciences 21 (1996) 265-351.

PAGE 229

208 97. S.-S. Chang, Production and Characterization of Photoluminescing Spark-Processed Si, Department of Materials Science and Engineering, Ph.D. dissertation, University of Florida, Gainesville, 1994, p. 154. 98. M. H. Ludwig, A. Augustin, R. E. Hummel, Semiconductor Science and Technology 12 (1997) 981-986. 99. S. Ossicini, Physica Status Solidi a-Applied Research 170 (1998) 377-390. 100. A. G. Cullis, L. T. Canham, P. D. J. Calcott, Journal of Applied Physics 82 (1997) 909-965. 101. F. Priolo, G. Franzo, D. Pacifici, V. Vinciguerra, F. Iacona, A. Irrera, Journal of Applied Physics 89 (2001) 264-272. 102. X. W. Zhao, S. Komuro, H. Isshiki, S. Maruyama, Y. Aoyagi, T. Sugano, Applied Surface Science 114 (1997) 121-125. 103. X. W. Zhao, S. Komuro, H. Isshiki, Y. Aoyagi, T. Sugano, Applied Physics Letters 74 (1999) 120-122. 104. T. Kimura, A. Yokoi, H. Horiguchi, R. Saito, T. Ikoma, A. Sato, Applied Physics Letters 65 (1994) 983-985. 105. J. H. Shin, R. Serna, G. N. vandenHoven, A. Polman, W. G. J. H. M. vanSark, A. M. Vredenberg, Applied Physics Letters 68 (1996) 46-48. 106. M. S. Bresler, O. B. Gusev, V. K. Kudoyarova, A. N. Kuznetsov, P. E. Pak, E. I. Terukov, I. N. Yassievich, B. P. Zakharchenya, W. Fuhs, A. Sturn, Applied Physics Letters 67 (1995) 3599-3601. 107. A. Polman, G. N. Vandenhoven, J. S. Custer, J. H. Shin, R. Serna, P. F. A. Alkemade, Journal of Applied Physics 77 (1995) 1256-1262. 108. F. Priolo, S. Coffa, G. Franzo, C. Spinella, A. Carnera, V. Bellani, Journal of Applied Physics 74 (1993) 4936-4942. 109. G. N. Vandenhoven, J. H. Shin, A. Polman, S. Lombardo, S. U. Campisano, Journal of Applied Physics 78 (1995) 2642-2650. 110. U. Hommerich, F. Namavar, A. Cremins, K. L. Bray, Applied Physics Letters 68 (1996) 1951-1953. 111. J. S. Custer, A. Polman, H. M. Vanpinxteren, Journal of Applied Physics 75 (1994) 2809-2817. 112. S. Libertino, S. Coffa, G. Franzo, F. Priolo, Journal of Applied Physics 78 (1995) 3867-3873.

PAGE 230

209 113. A. Cavallini, B. Fraboni, S. Pizzini, Applied Physics Letters 72 (1998) 468-470. 114. J. L. Benton, J. Michel, L. C. Kimerling, D. C. Jacobson, Y. H. Xie, D. J. Eaglesham, E. A. Fitzgerald, J. M. Poate, Journal of Applied Physics 70 (1991) 2667-2671. 115. D. L. Adler, D. C. Jacobson, D. J. Eaglesham, M. A. Marcus, J. L. Benton, J. M. Poate, P. H. Citrin, Applied Physics Letters 61 (1992) 2181-2183. 116. H. Isshiki, A. Polman, T. Kimura, Journal of Luminescence 102 (2003) 819-824. 117. F. Lucarz, A. J. Kenyon, Proceedings of the London Communications Symposium 2003 (2003) 125-128. 118. H. A. Lopez, P. M. Fauchet, Physica Status Solidi a-Applied Research 182 (2000) 413-418. 119. S. Schmitt–Rink, C. M. Varma, A. F. J. Levi, Physical Review Letters 66 (1991) 2782-2785. 120. S. Coffa, F. Priolo, G. Franzo, V. Bellani, A. Carnera, C. Spinella, Physical Review B 48 (1993) 11782-11788. 121. G. Franzo, S. Coffa, F. Priolo, C. Spinella, Journal of Applied Physics 81 (1997) 2784-2793. 122. M. Matsuoka, S. Tohno, Applied Physics Letters 71 (1997) 96-98.

PAGE 231

BIOGRAPHICAL SKETCH Kwanghoon Kim was born in Busan, South Korea, where he attended the Haewoondae High school graduating with honors in 1993. He attended the Seoul National University at Seoul, South Korea, and graduated in 1998 with a Bachelor of Science in mineral and petroleum engineering. He continued studying mineral and petroleum engineering with a specialization in geochemistry at the Seoul National University and received a Master of Science degree in 2000. In 2000, he moved to the University of Florida to pursue his Ph.D. study in the Department of Materials Science and Engineering with a specialty of electronic materials. He was awarded his second Master of Science in 2002. His main research area under the supervision of Professor Rolf E. Hummel was the development of visible and infrared light emitting devices based on spark-processed silicon doped with rare-earth elements. He is a member of the American Vacuum Society and Alpha Sigma Mu. 210