<%BANNER%>

Observation of Defects Evolution in Electronic Materials

Permanent Link: http://ufdc.ufl.edu/UFE0024741/00001

Material Information

Title: Observation of Defects Evolution in Electronic Materials
Physical Description: 1 online resource (128 p.)
Language: english
Creator: Jang, Jung
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: diffraction, film, microstructure, thin
Materials Science and Engineering -- Dissertations, Academic -- UF
Genre: Materials Science and Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Advanced characterization techniques have been used to obtain a better understanding of the microstructure of electronic materials. The structural evolution, especially defects, has been investigated during the film growth and post-growth processes. Obtaining the relation between the defect evolution and growth/post-growth parameters is very important to obtain highly crystalline films. In this work, the growth and post-growth related defects in GaN, ZnO, strained-Si/SiGe films have been studied using several advanced characterization techniques. First of all, the growth of related defects in GaN and p-type ZnO films have been studied. The effect of growth parameters, such as growth temperature, gas flow rate, dopants used during the deposition, on the crystalline quality of the GaN and ZnO layers was investigated by high resolution X-ray diffraction (HRXRD) and transmission electron microscopy (TEM). In GaN films, it was found that the edge and mixed type threading dislocations were the dominant defects so that the only relevant figure of merit (FOM) for the crystalline quality should be the FWHM value of ?-RC of the surface perpendicular plane which could be determined by a grazing incidence x-ray diffraction (GIXD) technique as shown in this work. The understanding of the relationship between the defect evolution and growth parameters allowed for the growth of high crystalline GaN films. For ZnO films, it was found that the degree of texture and crystalline quality of P-doped ZnO films decreased with increasing the phosphorus atomic percent. In addition, the result from the x-ray diffraction line profile analysis showed that the 0.5 at % P-doped ZnO film showed much higher miscrostrain than the 1.0 at % P-doped ZnO film, which indicated that the phosphorus atoms were segregated with increasing P atomic percentage. Finally, post-growth related defects in strained-Si/SiGe films were investigated. Post-growth processes used in this work included high temperature N2 annealing, ion-implantation, and thermal oxidation. Advanced characterization techniques have been used to obtain information about strain, relaxation, layer thickness, elemental composition, defects, surface/interface morphology changes and so on. Based on the understanding of defects behavior during the strain relaxation after post thermal processes, a new manufacturing process to obtain highly-relaxed and thin Si1-xGex layers, which could be used as virtual substrates for strained-Si applications, was found.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Jung Jang.
Thesis: Thesis (Ph.D.)--University of Florida, 2009.
Local: Adviser: Craciun, Valentin.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0024741:00001

Permanent Link: http://ufdc.ufl.edu/UFE0024741/00001

Material Information

Title: Observation of Defects Evolution in Electronic Materials
Physical Description: 1 online resource (128 p.)
Language: english
Creator: Jang, Jung
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: diffraction, film, microstructure, thin
Materials Science and Engineering -- Dissertations, Academic -- UF
Genre: Materials Science and Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Advanced characterization techniques have been used to obtain a better understanding of the microstructure of electronic materials. The structural evolution, especially defects, has been investigated during the film growth and post-growth processes. Obtaining the relation between the defect evolution and growth/post-growth parameters is very important to obtain highly crystalline films. In this work, the growth and post-growth related defects in GaN, ZnO, strained-Si/SiGe films have been studied using several advanced characterization techniques. First of all, the growth of related defects in GaN and p-type ZnO films have been studied. The effect of growth parameters, such as growth temperature, gas flow rate, dopants used during the deposition, on the crystalline quality of the GaN and ZnO layers was investigated by high resolution X-ray diffraction (HRXRD) and transmission electron microscopy (TEM). In GaN films, it was found that the edge and mixed type threading dislocations were the dominant defects so that the only relevant figure of merit (FOM) for the crystalline quality should be the FWHM value of ?-RC of the surface perpendicular plane which could be determined by a grazing incidence x-ray diffraction (GIXD) technique as shown in this work. The understanding of the relationship between the defect evolution and growth parameters allowed for the growth of high crystalline GaN films. For ZnO films, it was found that the degree of texture and crystalline quality of P-doped ZnO films decreased with increasing the phosphorus atomic percent. In addition, the result from the x-ray diffraction line profile analysis showed that the 0.5 at % P-doped ZnO film showed much higher miscrostrain than the 1.0 at % P-doped ZnO film, which indicated that the phosphorus atoms were segregated with increasing P atomic percentage. Finally, post-growth related defects in strained-Si/SiGe films were investigated. Post-growth processes used in this work included high temperature N2 annealing, ion-implantation, and thermal oxidation. Advanced characterization techniques have been used to obtain information about strain, relaxation, layer thickness, elemental composition, defects, surface/interface morphology changes and so on. Based on the understanding of defects behavior during the strain relaxation after post thermal processes, a new manufacturing process to obtain highly-relaxed and thin Si1-xGex layers, which could be used as virtual substrates for strained-Si applications, was found.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Jung Jang.
Thesis: Thesis (Ph.D.)--University of Florida, 2009.
Local: Adviser: Craciun, Valentin.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0024741:00001


This item has the following downloads:


Full Text

PAGE 1

1 OBSERVATION OF DEFECTS EVOLUTION IN ELE CTRONIC MATERIALS By JUNG HUN JANG A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2009

PAGE 2

2 2009 Jung Hun Jang

PAGE 3

3 To my parent s Heeja and Dr. Craciun

PAGE 4

4 ACKNOWLEDGMENTS I would like to express my gratitude to Dr. Valentin Craciun for his great guidance a nd encouragement. His friendship encouraged me to a higher level of the success. I also thank my committee members, Dr. Stephen Pearton, Dr. Cammy Abernathy, Dr. Rajiv Singh, Dr. Timothy Anderson for their support. I would like to thank Dr. Luisa Ameria De mpere for giving me a great opportunity to work in the Major Analytical Instrumentation Center (MAIC). Also, I would like to thank Dr. Gerald Bourne, Kerry Siebein, Wayne Acree, Eric Lambers for training me to use the TEM, FIB, SEM, AES, and XPS instrument s I would like to thank Rosabel Ruiz for assisting with all the administrative matters in MAIC. I extend my appreciation to Dr. Brent Gila, Dr. Hyunsik Kim, Dr. Michelle Phen, Dr. Andrew Herrero, and Gerger Andrew for many helpful discussions and growing the samples that I worked with I would like to thank my friends, Seungyoung, Wantae, Junhan, Jaewon, Dohwon, Dongjo, Chanwoo, Jaeseok, Taegon, Jinsu, Byungwook, Myunghwan, Dongwoo, Doyoung, Mingi, Hyungjun, Takgun, Myunghwa and so on. I cannot forget eve rything happened with them in Gainesville, forever. Finally, I thank my family and Heeja for their love and encouragement. They always believe that I could accomplish anything.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................................... 4 LIST OF TABLES ................................................................................................................................ 8 LIST OF FIGURES .............................................................................................................................. 9 LIST OF ABBREVIATIONS ............................................................................................................ 13 ABSTRACT ........................................................................................................................................ 14 CHAPTER 1 INTRODUCTION ....................................................................................................................... 16 Motivation .................................................................................................................................... 16 Objective ...................................................................................................................................... 16 Dissertation Organization ........................................................................................................... 17 Background and Literature Review ........................................................................................... 18 Strained -Si/SiGe .................................................................................................................. 18 Strain t echnology.......................................................................................................... 18 Strain r elaxation and d efec ts ........................................................................................ 19 High Crystalline GaN .......................................................................................................... 20 GaN ............................................................................................................................... 20 Growth and d efects of GaN ......................................................................................... 21 P type ZnO ........................................................................................................................... 22 ZnO ............................................................................................................................... 22 Microstructure of p type ZnO ..................................................................................... 23 2 EXPERIMENTAL METHODS ................................................................................................. 29 Materials Processing ................................................................................................................... 29 M olecular Beam E pitaxy (MBE) ........................................................................................ 29 Chemical V apor D eposition (CVD) ................................................................................... 30 M etal O rganic C hemical V apor D eposition (MOCVD) ................................................... 31 P ulsed L aser D eposition (PLD) .......................................................................................... 31 Materials Characterization .......................................................................................................... 33 X ray Diffraction (XRD) ..................................................................................................... 33 Instrument: XPert s ystem for PANalytical ................................................................ 33 Measurement m ethod ................................................................................................... 34 Tran smission E lectron M icroscopy (TEM) ....................................................................... 41 X ray P hotoelectron S pectroscopy (XPS) .......................................................................... 42 Atomic F orce M icroscopy (AFM) ...................................................................................... 43 Hall M easurements .............................................................................................................. 43

PAGE 6

6 3 DEFECTS EVOLUTION IN P TYPE ZNO ............................................................................. 54 Phosphorus Doped ZnO Thin F ilms Grown on Sapphire Substrate s ....................................... 54 Experimental D esign ........................................................................................................... 54 Results and D iscussion ........................................................................................................ 55 Microstructure of p hosphorus d oped ZnO f ilms ........................................................ 55 Line p rofile a nalysis of p hosphorus d oped ZnO ........................................................ 57 Pho sphorous s egregation ............................................................................................. 58 Conclusion............................................................................................................................ 58 4 DEFECTS EVOLUTION IN G aN EPITAXIAL LAYERS ..................................................... 64 GaN G rown on S apphire S ubstrate ............................................................................................ 64 Experimental D esign ........................................................................................................... 64 Results and D iscussion ........................................................................................................ 64 Grazing i ncidence x -ray d iffraction (GIXD ) t echnique ............................................. 64 Relationship b etween g rowth c onditions and d efects ................................................ 68 Conclusion............................................................................................................................ 71 5 DEFECTS EVOLUTION IN STRAINED LAYERS DURING STRAIN RELAXATION ........................................................................................................................... 76 Strained -Si Layer Grown on Graded SiGe Substrate ................................................................ 76 Experimental Design ........................................................................................................... 76 Results and Discussion ........................................................................................................ 76 Strain r elaxation d uring t hermal a nnealing ................................................................ 76 Calculation of d islocation d e nsity from FWHM of o mega rocking c urve s ............. 79 Conclusion............................................................................................................................ 83 Strained SiGe G rown on Si ........................................................................................................ 83 Experimental D esign ........................................................................................................... 83 Results and D iscussion ........................................................................................................ 84 Determination of Ge c omposition, l ayer t hickness, and s train .................................. 84 Defects s tructure in s trained SiG e ............................................................................... 87 Conclusion............................................................................................................................ 90 Ion I mplantation of S trained L ayer s .......................................................................................... 91 Experi mental D esign ........................................................................................................... 91 Results and D iscussion ........................................................................................................ 92 Conclusion............................................................................................................................ 96 Oxidati on of S trained SiGe L ayers ............................................................................................ 96 Experiment D esign .............................................................................................................. 96 Results and D iscussion ........................................................................................................ 97 Conclusions ........................................................................................................................ 101

PAGE 7

7 6 SUMMARY ............................................................................................................................... 119 Strained -Si/SiGe ........................................................................................................................ 119 High Crystalline GaN F ilms ..................................................................................................... 1 19 P type ZnO ................................................................................................................................. 120 LIST OF REFERENCES ................................................................................................................. 121 BIOGRAPHICAL SKETCH ........................................................................................................... 128

PAGE 8

8 LIST OF TABLES Table page 1 1 Material p roperties of c ommon s emiconductor s [10]. ......................................................... 24 1 2 Lattice parameter and thermal expansion coefficient of substrates for ZnO crystalline films [29]. ............................................................................................................................... 25 3 1 Comparison of results from Williamson Hall plot and W arren -Averbach method applied to as -grown and phosphorus doped ZnO films ...................................................... 60 4 1 The growth conditions used during the deposition of GaN nucleation layer and 3D islands .................................................................................................................................... 72 4 2 FWHM values (arc sec) of in asymmetric and out of RCs. The FWHM value s of the ) 0 1 10 ( RC are much higher than one of the ) 0002 ( -RC. ................... 72 5 1 S imulation s of the XRR curves acquired from the as -grown and annealed strained silicon samples. A model consisting of r elaxed SiGe buffer, interfacial, strained silicon, and surface layers were used. ................................................................................. 102 5 2 L ine profile analysis and calculated dislocation density of as -grown and annealed strained silicon. The diffraction profiles are fitted by Voigt and Pseudo -Voigt functions and fitting parameters are estimated by the least square method .................... 102 5 3 XRR simulation results for as -grown Si1xGex samples. ................................................... 102 5 4 XRR simulation results for annealed Si1 xGex samples .................................................... 103 5 5 SiGe layer thickness and Ge content estimated from XRR and /2 RCs simulations and measured by XTEM ..................................................................................................... 103

PAGE 9

9 LIST OF FIGURES Figure page 1 1 The lattice and ellipsoids of constant energy in reciprocal space.. ..................................... 26 1 2 Energy band diagram for wide (Semiconductor 2) and narrow (Semiconductor 1) bandgap semiconductors [ 11]. ............................................................................................... 26 1 3 Crystal structure of GaN and polarization of AlGaN/GaN. ................................................ 27 1 4 The mismatch lattice parameter and expansion thermal coefficient for alternative substrates with respect to GaN layer[ 10, 15]. Most used substrates for GaN growth are sapphire and SiC. ............................................................................................................. 27 1 5 Low temperature scattering processes limiting electron mobility in AlGaN/GaN 2DEG system with three different dislocation densities [ 21]. ............................................. 28 1 6 ZnO crystal structure (Wurtzite) .......................................................................................... 28 2 1 A schematic of a conventional MBE system [35]. ............................................................... 45 2 2 Vapor pressure dependence with temperature for different silicon sources used in epitaxy [37]. ............................................................................................................................ 45 2 3 The schematic diagram of a pulsed laser deposition ( PLD) system [47]. .......................... 46 2 4 PANalytical XPert system. ................................................................................................... 46 2 5 Mirrors used in primary optics. ............................................................................................. 47 2 6 Secondary optics. ................................................................................................................... 47 2 7 Powder x ray diffraction pattern of Mo powder. ................................................................. 48 2 8 (102) reflectio n of the GaN grown on sapphire substrate and curve fitting using Gauss, Cauchy, Pearson, and Pseudo-Voigt functions ....................................................... 48 2 9 Experimental and fitted X ray reflectivity curves of a strained Si1xGex on silicon substrate. The simulation was performed by using Wingixa software from PANalytical. ........................................................................................................................... 49 2 10 Experimental (004) o mega 2 t heta rocking curve and simulation curve The simulation w as carried out by using Epitaxy software from PANalytical. ......................... 49 2 11 Omega 2theta rocking curve of asymmetric diffraction plane. ........................................... 50 2 12 Reciprocal space map (RSM). ............................................................................................... 50

PAGE 10

10 2 13 Reciprocal space map and real crystal lattice. ...................................................................... 51 2 14 Mosaicity. ............................................................................................................................... 51 2 15 A schematic of a transmission electron microscop e (TEM) [62]. ....................................... 52 2 16 A schematic of a XPS instrument and its electron excitation process [ 62]. ....................... 52 2 17 A schematic of a AFM instrument and its operation [6 4 ]. .................................................. 53 3 1 Crystal orientation of ZnO films. .......................................................................................... 61 3 2 Omega rocking curves of symmetric and asymmetric diffraction plane s of as -grown, 0.5, and 1.0 at % P -doped ZnO films ................................................................................... 61 3 3 Williamson Hal l plot of as -grown, 0.5, and 1.0 at % P -doped ZnO films ........................ 62 3 4 Mi crostrain of as grown, 0.5, and 1.0 at % P -doped ZnO films obtained from the Williamson Hall plot and Warren -Averbach method ......................................................... 63 3 5 XPS spectra acquired from 0.5 and 1.0 at % P -doped ZnO films. ...................................... 63 4 1 Transmission and edge geometries for the determinati on of the azimuthal spread. .......... 73 4 2 Grazing incidence x -ray diffraction (GIXD). ....................................................................... 73 4 3 Cross sectional bright field TEM images of GaN films. ..................................................... 74 4 4 RCs measured from different diffraction planes of sample A (t3D=10 min) and B (t3D=16.7 min) ................................................................................................. 74 4 5 The screw and edge type dislocation density of GaN films grown using various 3D growth mode times. ................................................................................................................ 75 4 6 Cross sectional bright field images of sample D .............................................................. 75 5 1 (113) Reciprocal space map of annealed s trained Si/Si0.7Ge0.3/graded -SiGe/Si substrate ............................................................................................................................... 104 5 2 X ray reflectivity (XRR) curves recorded from the as -grown and annealed strainedSi/Si0.7Ge0.3/Graded SiGe/Si .............................................................................................. 104 5 3 Raman spectra of strained -Si on SiGe. ............................................................................... 105 5 4 Curve fitting of (113) rocking curve of annealed strained silicon. ................................... 105 5 5 The coherence length distribution calculated for the as grown and annealed strained silicon from the second derivative of the Fourier size coefficient ................................... 106

PAGE 11

11 5 6 Atomic force microscopy (AFM) and transmission electron microscopy (TEM) images of strained silicon. ................................................................................................... 106 5 7 RBS spectrum and its fit using RUMP software [ 131] for a sample containing a 25% Ge fraction ........................................................................................................................... 107 5 8 (113) RSM from a Si0.85Ge0.15 layer on (001)Si ................................................................ 107 5 9 RC acquired from a sample with 15 % Ge and its simulation (displaced vertically for better view) ................................................................................................... 108 5 10 AFM micrographs of the as -grown SiGe layers surface ................................................... 108 5 11 (004) and (113) /2 RCs acquired from the as -grown Si1xGex on Si and after the thermal anneal ..................................................................................................................... 109 5 12 Plan -view TEM for strained SiGe samples as grown and annealed at 800 C for 30 min ........................................................................................................................................ 109 5 13 Omega 2theta rocking curves and TEM images. ............................................................... 110 5 14 (004) and (113) omega RCs) acquired from the as -grown and annealed strained Si80Ge20 layers ....................................................................................... 111 5 15 (113) reciprocal space maps of as -grown and annealed at 900 for 30 min 100 nm thick Si75Ge25 layers. ............................................................................................................ 112 5 16 -RCs acquired from the as grown and annealed at 900 for 30 min Si75Ge25 100 nm thick layers. ....................................................................................... 112 5 1 7 TEM images of annealed Si75Ge25 film. ............................................................................. 113 5 1 8 Bright field transmission electron microscopy image of implanted Si76Ge24 layer annealed at 500 for 30 min. The SiGe/Si and amorphou s -crystalline interfaces are clearly marked in this figure ............................................................................................... 113 5 1 9 (113) reciprocal space maps (RSMs) of Si76Ge24 layers. ................................................... 114 5 20 Surface undulation of SiGe layers. ...................................................................................... 114 5 2 1 Average peak to -peak distance (black) and r.m.s. roughness (blue) of annealed Si76Ge24 layers without/with ion implantation. .................................................................. 115 5 2 2 (113) RSMs of ionimplanted Si76Ge24 layers after thermal annealing. ........................... 115 5 2 3 Strain relaxation and full width at half maximum (FWHM) val RCs of ion implanted Si76Ge24 after thermal anneal. The black and blue lines indicate strain relaxation and FWHM values, respectively ...................................................................... 116

PAGE 12

12 5 2 4 Transmission electron microscopy images of an ionimplanted Si76Ge24 films ............. 116 5 2 5 / -grown strained -Si75Ge25 and after oxidation at 800 900 and 1000 for 1hr. Red dot lines indicate the SiGe peak p osition of as -grown film as a reference ....................................................................................................................... 117 5 2 6 Scanning transmission electron microscopy and reciprocal space map of Si75Ge25 oxidized in 800 ............................................................................................................... 117 5 2 7 (113) reciprocal space maps (RSMs) of strainedSi1xGex layers after the oxidation at 1000 for 1 hr ................................................................................................................... 118 5 2 8 STEM images of oxidized Si75Ge25 layers and their depth profile. .................................. 118

PAGE 13

13 L IST OF ABBREVIATIONS strain D crystallite size omega theta psi phi 2 / omega/2 theta x ray wave length thermal expansion coefficient inclination angle t3D 3D growth mode time FOM figure of merit GIXD grazing incidence x ray diffraction LT low temperature HT h igh temperature NL nucleation layer XRR x ray reflectivity RSM reciprocal space map RLP reciprocal lattice point g diffraction vector GRL germaniu m rich layer GDL germanium deficient layer

PAGE 14

14 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 OBSERVATION OF D EFECTS EVOLUTIONS IN ELECTRONIC MATERIALS By Jung Hun Jang August 2009 Chair: Valentin Craciun Major: Materials Science and Engineering A dvanced characterization technique s have been use d to obtain a better understanding of the microstructur e of electro nic materials. The structural evolution, especially defects, has been investigated during the film growth and post -growth processes. Obtaining the relation between the defect evolution and growth/post growth parameters is ver y important to obtain highly cr ystalline films In this work, the growth and post -growth related defects in GaN, ZnO, strainedSi/SiGe films have been studied using several advanced characterization techniques First of all, the growth of related defects in GaN and p type ZnO films have been studied. The effect of growth parameters, such as growth temperature, gas flow rate, dopants used during the deposition on the crystalline quality of the GaN and ZnO layers was investigated by high resolution X ray diffraction (HRXRD) and transmissi on electron microscopy (TEM). In GaN films, it was found that the edge and mixed type threading dislocations were the dominant defects so that the only relevant figure of merit (FOM) for the crystalline quality should be the RC of the surfa ce perpendicular plane which could be determined by a grazing incidence x ray diffraction (GIXD) technique as shown in this work The u nderstand ing of the relationship between the defect evolution and growth parameters allowed for the growth of high crysta lline GaN films. For ZnO films, i t was found that the degree of texture and crystalline

PAGE 15

15 quality of P doped ZnO films decreased with increasing the phosphorus atomic percent. In addition, the result from the xray diffraction line profile analysis showed th at the 0.5 at % P doped ZnO film showed much higher miscrostrain than the 1.0 at % P -doped ZnO film, which indicated that the phosphorus atoms were segregated with increasing P atomic percent age Finally post growth related defects in strained -Si/SiGe fil ms were investigated Post growth processes used in this work included high temperature N2 annealing, ionimplantation, and thermal oxidation. A dvanced characterization technique s have been used to obtain information about strain, relaxation, layer thickne ss, element al composition, defects, surface/interface morphology changes and so on. Based on the understanding of defects behavior during the strain relaxation after post thermal processes, a new manufacturing process to obtain highly relaxed and thin Si1 xGex layers, which could be used as virtual substrates for strained-Si applications, was found.

PAGE 16

16 CHAPTER 1 INTRODUCTION Motivation Many of the available technology are exploiting t he special physical and chemical properties of the materials, such as thickne ss, element al composition, interface roughness, structural defects and so on. These device properties of the bulk materials can be controlled or engineered by reducing the device dimension s Obtaining desirable material properties is one of the fundamental research es in the MSE field. For example, dislocation s play a key role in the strain hardening of the materials in which the dislocation motion will be impeded by the repulsive force between the dislocations [1 ]. In addition, the yield strength of most cr ystalline solids increases with decreasing grain size, which is well described by the Hall-Petch relationship [2, 3 ]. It seems intuitively correct that fine grain solids should have high yield strengths because the grain boundary can act as an obstacle to d islocation motion. Therefore, accurate characterization techniques for the micro structur e analysis are necessary for the desired materials performance. Many analytical methods are now available ; transmission electron microscopy (TEM), x ray diffraction (XRD), atomic force microscopy (AFM), ellipsometry, photoluminescence and so on. These analytical techniques play a specific role in the determination of the structural details and have been continuously developed because of the continuous challenges resultin g from the development of new materials and scaling of the material dimension This dissertation will endeavor to encompass the advances in the an a lysis of a wide spectrum of the electronic materials by using advanced characterization techniques Objective In this work advanced x ray based techniques have been employed to observe and understand the structural evolutions in thin films. The structural modifications occurred during

PAGE 17

17 the film growth as well as post -growth processes. When the films are grown, th e understand ing of the effect of the growth parameters, such as growth temperature and ambient pressure, gas flow rate, and dopants, on the microstructural evolutions is extremely important for the growth optimization process to obtain high crystalline fil ms. After the growth stage the film will undergo various post growth processes, such as high temperature anneal ing ion implant ation and thermal oxidation During these post -growth processes, other types of the structural evolutions are generated. Therefo re, o bservation and understand ing of these microstructural evolutions induced during the film growth and post -growth processes are necessary to obtain the desired material performances Dissertation O rganization The contents of this work are categorized i nto six chapters. Chapter 1 outlines the motivation and objective s as well as the introduction of the electronic materials used in this work. Chapter 2 provides the experimental method s including the materials processing and the advanced characterization methods. Chapter 3 and 4 introduce the defects evolution during film growth in GaN and ZnO films respectively In chapter 3 the influence of a p type dopant, phosphorous on the microstructure of ZnO films is investigated. In chapter 4 the effects of th e growth parameter s on the microstructure of GaN films as well as the case of advanced characterization techniques to assess their quality are presented Chapter 5 discusses the microstructural evolution in the strained -Si/SiGe layers during the post -growt h processes, such as high temperature nitrogen annealing, ion-implantation, and thermal oxidation. In addition, a new route for the fabrication of highly relaxed Si1 xGex layers is presented. Finally, chapter 6 summarizes the obtained results.

PAGE 18

18 Background a nd Literature R eview StrainedSi/SiGe Strain technology : A s trained layer is one in which the lattice dimensions are stretched or compressed beyond their equilibrium interatomic distance. This is accomplished by epitaxially growing the layer over another material with a larger or smaller lattice parameter The best known example is strained -Si grown on SiGe layer. Because the lattice parameter of germanium (Ge) is just 4% larger than that of silicon (Si) and both have the same lattice type SiGe layers are a useful substrate f or implementing the idea of strained silicon technology. Since Intel announced that strained silicon can speed up the flow of electrons through transistor s increasing their performance and decreasing power consumption, the strained-silicon technology has been intensively studied [ 4 ]. Basically, the carrier mobility is given by [ 5 ], *m q (1 1 ) w here is the carrier mobility, *m the effective mass of charge carrier, the carrier life time and q the carrier charge, respectively. Fig. 1 shows a strained -Si grown on SiGe layer In bulk -Si at room temperature the conduction band is comprised of six degenerate valleys, as shown in Fig. 1 1 (a). These valleys are of equal energy, as indicated 1 (a) and (c). Since the effective mass for any direction is the reciprocal value of the curvature of the electron energy function in that direction and the effective mass of each ellipsoid is anisotropic, the total electron conductivity m ass, m*, is obtained by adding the contributions of the six degenerate valleys and is given by [ 6 ], 1 *4 2 6 1 t lm m m (1 2)

PAGE 19

19 where, ml and mt are longitu dinal and transverse mass, respectively. On the other hand, the strain removes the degeneracy between the four in-plane 1 means that they are preferentially occupied by electrons. The electron mobility partial ly improves via a reduced in -plane and increased out -of plane m* due to the favorable mass of the plane transverse effective mass (mt=0.19m0) and out -of plane longitudinal mass (ml=0.98m0). In addition, the electron scattering is reduced due to the conduction valleys splitting into two different energy levels, which lowers enhances the mobility by decreasing the e ffective mass and intervalley scattering of charge carriers [ 6 ]. Using strained silicon, one can achieve a high er device performance without shrink ing the size of the transistor s Strain r elaxation and d efects : t here are several ways to induce a strain int o silicon for higher electron mobility. One of these is an Epitaxial Strain Inducing Template (ESIT) in which the silicon can be pseudomorphically grown on a relaxed Si1xGex buffer layer, resulting in the biaxial tensile strain, as shown in Fig. 1 1 This is the most used approach in HBTs (hetero bipolar transistors) In addition, the compressive strain in the MOS (metal -oxide semiconductor) channel can be induced by introducing a Si1xGex layer into source and drain regions [ 7 8 ]. During the production of strained -Si/ SiGe or post -growth processes the strain begins to relax by generating misfit defects The presence of the se defects, such as misfit/ threading dislocations and stacking faults, has an undesirable influence on the device performance, such as i ncreased leakage current in MOSFETs [ 9 ]. Therefore, i t is important to determine the ideal thickness of

PAGE 20

20 the strained layer in order to avoid strain relaxation, which could begin once the thickness reaches a critical value. High C rystalline GaN GaN : w ide ba ndgap semiconductors are extremely attractive for devices that work under harsh conditions, such as high temperature. GaN is the most promising material because of its intrinsic properties: wide bandgap energy, high saturation velocity, and high breakdown electric field. As summarized in Table 1 1 GaN is suitable for device operation in high frequency and temperature regions [ 10]. In order to maintain high transconductance in FET s the channel conductivity must be as high as possible. Basically, the conduc tivity can be increased by increasing the doping in the channel. However, the increased doping also causes increased scattering by the ionized impurities, which leads to a degradation of mobility. What is needed is a way of creating a high electron concent ration in the channel of a MESFET by other mean s than doping. A n ingenious approach to achieve this requirement is to use dissimilar heterostructure materials with different bandgap s [11]. T his can be achieved when the donor energy level in the wider -bandgap material lies above the conduction -band edge of the lower bandgap material, resulting in the occurrence of the band bending and presence of discontinuit ies in the conduction and valence band s The discontinuity in the conduction band creates a triangula r quantum well in which electrons are trapped forming a so called 2 -Dimensional Electron Gas (2 DEG), as shown in Fig. 1 2. Since the narrower bandgap semiconductor is usually undoped, the electrons trapped in this region could move without the ionized im purity scattering. Therefore, the electron mobility at lower temperature s where the impurity scattering is dominant, is significantly enhanced over the usual value s measured in epitaxial layer s of equivalent doping density. While it is possible to get a h igh charge density in the interfacial region by high doping of wide bandgap semiconductor in conventional III -V based HEMT, AlGaN/GaN heterostructure can

PAGE 21

21 have higher charge carrier density at the interface without intentionally doping. GaN has a wurtzite c rystal structure in which the unit cell is asymmetr ic along the c axis, as shown in Fig. 1 3 (a). This asymmetric arrangement of atoms in GaN can induce a spontaneous polarization which could have a great influence o n the device performance, as shown by s everal studied [12,1 3 ]. Also, the in -plane lattice constant of AlN (a=3.111 and c=4.978) is smaller than that of GaN (a=3.189 and c=5.185) so that AlGaN epitaxial layer experiences an in-plane biaxial tensile strain, resulting in a piezoelectric polari zation. Therefore, the net polarization in AlGaN/GaN heterojunction is the sum of the spontaneous polarization of GaN and the piezoelectric polarization of strained -AlGaN, which results in the formation of a positive polarization in the heterointerface, as shown in Fig. 1 3 (b) [ 14]. The electrons are attracted by this positive charge, and then they accumulate at the interface. This results in a higher charge density as well as a higher carrier mobility, even without intentionally doping. Growth and d efect s of GaN : s ap phire and SiC substrates are commonly used in GaN film growth due to the lack of GaN bulk single crystal. The growth of GaN on these substrates generates many misfit defects because of a large mismatch of both in -plane lattice constants and th ermal expansion coefficients between GaN and substrate. Fig. 1 4 displays the lattice mismatch and thermal expansion coefficient of several materials with respect to those of GaN [ 15]. To minimize the dislocation density and obtain a smooth surface, a two -step growth method has been proposed, where the first layer is grown at low temperature (500~600 ), known as low temperature (LT) layer or nucleation layer(NL), and then the second layer is grown at higher temperatures (over 1000), being known as high temperature (HT) layer [1618]. GaN nucleation layer grown at low temperature contains several type s of defects, such as dislocations, point defects, and stacking faults because a 3 dimensional island growth mode (3 D Volmer -

PAGE 22

22 Weber mode) is active here [ 19, 20]. These defects have a significant effect on the GaN based device performance Fig. 1 5 shows th at there are several scattering factors that affect the electron mobility [ 21, 22]. The carrier transport mechanism at low temperature s is affected by scattering due to the background unintentional donor (Ns), the alloy disorder of AlGaN barrier, the interf ace roughness at the AlGaN/GaN heterojunction, and the threading dislocations. The acoustic and optical phonon scattering can be ignored because their effects on the carrier mobility are not significant at low temperature s T he scattering effect of the thr eading dislocations on the electron transport is considered by the coulombic interaction from a charged core as well as the deformation potential interaction from the strain field surrounding dislocations. In Fig. 1 5, it is clear that the threading disloc ation density is the most important scattering factor limiting mobility. In addition to the mobility limiting factor, the threading dislocations act as a leakage current path. It was reported that the leakage current increased with increasing the dislocati on density in GaN based l ight e mitting d iodes (LEDs) [ 23]. As these structural defects in GaN have a great influence o n the device performance, th ey need to be studied in depth. P -type ZnO ZnO : ZnO has attracted a great deal of attention for many years as the most promising material for optoelectronics device applications due to its wide bandgap energy (3.37 eV) and large exciton binding energy (60 meV) [ 242 6 ]. Also, bulk ZnO substrate s are readily available as compared to GaN. However, the key issue is t o produc e p type ZnO films because of the deep level location of the candidate acceptors from ZnO valence band and their self -compensation by native point defects [ 2 7 2 8 ]. Therefore, it is critical to be able to produce reliable and reproducible p -type ZnO films.

PAGE 23

23 Microstructure of p -type ZnO : ZnO ha s a wurzite crystal structure, as shown in Fig. 1 6 In fact, there are three kinds of possible crystal structure for ZnO: wurzite, zinc blend, and rock salt structures [ 29]. The wurzite crystal structure is the most thermody namically stable phase It has been reported that high quality ZnO films could be grown on a number of substrates, such as silicon, sapphire, GaN, AlN, and GaAs [3033]. In order to reduce strain and dislocation density of films, the lattice m ismatch and thermal expansion coefficient between the substrate and film should be small. The lattice parameter and thermal expansion coefficient of several possible substrates are shown in Table. 1 2 [2 9 ]. A w idely used substrate for high quality ZnO film s is the sapphire due to its transparency and relatively low cost. However, the films grown on sapphire substrate show mosaicity, residual carrier concentration, and low mobility because the in -plane lattice mismatch is 18 % even for a 30 -plane rotation of the film lattice with respect to the substrate. In this study, the effect of phosphorus atoms as a ptype dopant on the microstructure and defects evolution during film growth is investigated in detail.

PAGE 24

24 Table 1 1 Materi al p roperties of c ommon s emiconductor s [10]. Attribute Si GaAs SiC GaN Energy Gap(eV) 1.11 1.43 3.2 3.4 Breakdown E -Field(V/cm) 6.0105 6.5105 3.5106 3.5106 Saturation Velocity(cm/s) 1.0107 2.0107 2.0107 2.5107 Electron Mobility(cm2/V -s) 1350 6000 800 1600 Thermal Conductivity (W/cm -K) 1.5 0.46 3.5 1.7 BFOM Ratio* 1.0 9.6 3.1 24.6 JFOM Ratio* 1.0 3.5 60 80 Heterostructures SiGe/Si AlGaAs/GaAs InGaP/GaAs AlGaAs/InGaAs None AlGaN/GaN InGaN/GaN *BFOM: Baligas figure of merit for power trainsis tor performance which is function of thermal conductivity and carrier mobility *JFOM: Johnsons figure of merit for power transistor performance which is function of breakdown field and carrier saturated velocity

PAGE 25

25 Table 1 2 Lattice parameter and thermal expansion coefficient of substrates for ZnO crystalline films [29]. Material Crystal S tructure Lattice parameter In -plane lattice mi smatch Thermal expansi on coefficient 1 ) a () a (10 6 ) c () (%) c (10 6 ) ZnO Hexagonal 3.252 2.9 5.213 4.75 GaN Hexagonal 3.189 1.8 5.17 5.185 4.55 AlN Hexagonal 3.112 4.5 5.3 4.980 4.2 -Al2O3 Hexagonal 4.757 (18.4 % after 30 in plane rotation) 7.3 12.983 8.1 6H SiC Hexagonal 3.080 3.5 4.2 15.117 4.68 Si Cubic 5.430 40.1 3.59 ScAlMgO4 Hexagonal 3.246 0.09 25.195 GaAs Cubic 5.652 42.4 6.0

PAGE 26

26 Figure 1 1. The lattice and ellipsoids of constant energy in reciprocal space (a) the equilib rium and (b) strained silicon (not to scale) (c) is the energy level at the bottom of the six conduction band valleys [6 ]. Figure 1 2 Energy band diagram for wide (Semiconductor 2) and narrow (Semiconductor 1) bandgap semiconductors [ 11].

PAGE 27

27 Figure 1 3 Crystal structure of GaN and polarization of AlGaN/GaN. (a) c rystal structure of Ga face GaN (b) net positive charge at the AlGaN/GaN interface caused by the sum of the net spontaneous and piezoelectric polarizations between AlGaN/GaN [ 10 ]. Figur e 1 4 The mismatch lattice parameter and expansion thermal coefficient for alternative substrates with respect to GaN layer[ 10, 1 5 ]. Most used substrates for GaN growth are sapphire and SiC.

PAGE 28

28 Figure 1 5 Low temperature scattering processes limiting electron mobility in AlGaN/GaN 2DEG system with three different dislocation densities [ 21]. Figure 1 6 ZnO crystal structure (Wurtzite)

PAGE 29

29 CHAPTER 2 EXPERIMENTAL METHODS Materials P rocessing M olecular Beam Epitaxy (MBE) Cho and Arthur, the pioneers of m olecular beam epitaxy (MBE) technique described that MBE was the epitaxial growth of semiconducting films by the reaction between the molecular beams of its constituents and a crystalline surface held at a certain temperature under ultra high vacuum (UHV) conditions [ 34]. Fig. 2 1 shows a schematic diagram of conventional molecular beam epitaxy system The molecular beam of some constituents such as Ga, Al, and As, are generated from the heated Knud s en effusion cells and the flux of each source can be rap idly controlled by a shutter that blocks the beam from hitting the substrate. The molecular beams generated from the effusion cells impinge on a rotating substrate mounted on a heated Mo block. The cell was fabricated from pyrolytic boron nitride (PBN) or high -purity graphite because of their chemical inert ness and endurance to a high temperature. The flux density emanating from the effusion cell is given by [35] 2 / 1 2) 2 ( cos T k m L p A JB (2 1) where A is the area of the aperture, p is the equilibrium vapor pressure in the cell, L is the distance between the cell and the substrate, m is the mass of the effusing species, Bk is the Boltzmann constant and T is the cell temperature. Note that the above equation is only valid when the cell aperture size is smaller than the mean free path of vapor molecules within the cell. After the species impinge on the substrate, they experience adsorption and migrat ion on to the surface until they combine to form the films. Th is whole process can be described by a

PAGE 30

30 sticking coefficient, S, which is defined as the fraction of the total number of impinging atoms or molecules that stick to the surface to that which are incorporated into the film. The growth step for the desired film by MBE can be monitored by reflection highenergy electron diffraction (RHEED) which allows for the arrangement of the top few monolayers and crystal structure to be observed As elemental s ources in MBE growth technique, gas or metal -organic sources such as PH3 and AsH3, can be use d due to easy control of the vapor pressure [36]. C hemical V apor D eposition (CVD) The chemical vapor deposition (CVD), which is sometimes also called by vapor pha se epitaxy (VPE), is a technique in which the species for the growth of epitaxial film are transported as vapors to the substrate using a carrier gas. As the substrate is kept at a high temperature, the species experience surface diffusion and then combine with each other or other species to form the crystalline films. Typically, the following steps are common to CVD/VPE processes [37]: 1) T ransport of the species to the substrate located in reaction chamber by the carrier gas 2) Adsorption of the species o n the substrate surface 3) Reaction between the adsorbed species to form an epitaxial film 4) Desorption of reaction products from the substrate surface 5) Transport of the by -products away from the reaction chamber Most of silicon and silicon -germanium f ilms ha ve been grown by the CVD technique because they have many available gas sources. Four sources of silicon have been widely used for epitaxial film: silicon tetrachloride (SiCl4), trichlorosilane (SiHCl3), dichl orosilane (SiH2Cl2), and silane (SiH4). At a given temperature, the vapor pressure of silane is greater than other sources and that of silicon tetrachloride is the lowest of them as shown in Fig. 2 2 This indicates that the silicon tetrachloride is the most stable silicon source, in turn the g rowth

PAGE 31

31 temperature (1100 ~1250 ) using by SiCl4 gas is the highest for the epitaxial film. On the other hand, the grow th temperature when using t he silane and /or dichl orosilane is around 500 ~600 because they pyrolyze easily. In this study, the dichlorosilane as a silicon sou rce was used because lower growth temperature was required for the strained epitaxial films. M etal O rganic C hemical V apor D eposition (MOCVD) Metal -organic chemical vapor deposition (MOCVD) is a variant of CVD technique. The o nly difference is the use of me tal -organic materials as source s For example, the mixture of trimethygallium ((CH3)3Ga) and arsine under the presence of hydrogen can be pyrolized at 600 to deposit layers of GaAs. There are two major advantages in MOCVD technique. First of all, the metal alkyls, which serve as sources of group III elements, are volatile liquids at room temperature so that they can be transported easily using a carrier ga s. Second, the pyrolysis temperature of metal alkyls is lower than that of the metal halides, resulting in lower growth temperature [ 35]. The MOCVD technique is widely used for a variety of materials, such as GaAs, GaAlAs, InGaAsP, GaAsSb, InAs N GaInP, Zn Se, ZnTe, AlN, GaN and so on [38 46]. In addition, the superlattices and quantum -well laser heterostructure can be easily deposited. In this study, MOCVD technique has been used for high crystalline GaN films. P ulsed L aser D eposition (PLD) Pulsed laser dep osition (PLD) is one of the most famous techniques in research laboratories due to its wide application to almost any materials, in particular for compounds that are difficult or impossible to produce in thin film form by other techniques. In addition, PLD has the ability to closely maintain the target composition in the deposited thin films because of the very short duration and high energy of the laser pulse. On the other hand, there are some

PAGE 32

32 disadvantages. The a blation plume cross section is generally sm all due to a limited laser spot size. In turn, this limits the sample size that can be prepared by PLD. Also, the plume of ablated material is highly forward directed, which causes poor conformal step coverage. Because of these disadvantages, PLD is mainly used for the investigation of new materials in a research environment. A schematic diagram of the basic PLD configuration is shown in Fig. 2 3 [47]. Lasers that are commonly used are ArF, KrF excimer layers, and Nd:YAG laser. It is generally recognized th at the shorter the wave length, the more effective the ablation process is [48]. In this study, a KrF excimer laser was used as the ablation source and p -type ZnO films were grown by PLD A focused laser pulse (pulse duration, 10~30 ns) strikes a target of the desired composition in a vacuum chamber. This incident high power pulsed laser beam (typically 2~5 J/cm2) heats up the target materials well beyond the evaporation temperature and produces an ejected plasma or plume of atoms, ions, and molecules. The materials dissociated from the target surface are deposited (velocities typically ~106 cm/s in vacuum) as a thin film on a substrate. The growth process is strongly dependent on several parameters, such as the laser fluence and wavelength, the structural a nd chemical composition of the target materials, the chamber pressure and the chemical composition of the background gas, and the substrate temperature and the distance between the target and the substrate. These variables have to be optimized to achieve h igh quality epitaxial films.

PAGE 33

33 Materials Characterization X -ray Diffraction (XRD) Instrument: XPert s ystem for PANalytical X ray source: The ceramic filament with Cu target is used as the x ray source in XPert system. The wavelength s of the characteristic X -ray lines for Cu target are 54056 11k 54439 12k and 396 1 k Goniometer: The goniometer for the XPert system has four axis for rotation 2 4 2 the diffracted x ray beam and the incident beam the plane rotation an gle. Primary optics (Incident x ray beam): Two kinds of primary optics have been used: x -ray mirror and hybrid mirror which are shown in Fig. 2 5. First, the x ray mirror gives a dichromatic beam of 1k and 2k elim inating k The incident x -ray beam is 1.2 mm wide and 20 mm high without a mask or divergence sli t and quasi -parallel to the samples surface with a horizontal divergence of 0.05. The x ray mirror is primarily used for 2theat / -ray reflectivity (XRR), and grazing incidence x ray diffraction (GIXD) scans. Second, the hybrid mirror gives the best resolution due to its monochromatic x-ray line in which 1k is only used. Secondary optics (diffraction beam): there are two kinds as the secondary optics: parallel plate collimator over detector and triple axis optics which are shown in Fig. 2 6 The parallel plate collimator is used for x ray ref lectivity (XRR ), 2theta / -ray diffraction (GIXD) scans and its divergence is 0.27 arm attachment for high resolution applications. One arm is the rocking curve attachment and detector with a 6 mm ape rture, which is corresponding to an acceptance angle of around 1

PAGE 34

34 Another arm is a germanium crystal channel cut analyzer in which the diffracted beam from the sample experiences (022) reflections before entering the detector. The acceptance angle of this analyzer is 12 a rc secs which is the highest resolution for all optic systems. Measurement m ethod X ray diffraction line profile analysis ( microstrain and crystallite size ): One of many characterization techniques for microstructure analysis is x ray d iff raction l ine p rofile a nalysis (LPA) in which the finite crystallite size and microstrain can be estimated by using the broadening of the diffraction peak. In 1918, Scherrer devised a famous formula explaining how the width of a Bragg reflection is affected by finite crystallite size [ 49]. BK D cos (2 2) where D is a size parameter, B Bragg angle, a full width half maximum (FWHM), a wavelength and K a constant which is near unit. The diffraction peak broadening is also affected by other microstructural features, such as microstrain, dislocation, twin planes and stacking faults. Especially, the strain field in the vicinity of the dislocations makes d -spacing variable, resulting in the line broadening. The relationship between the strain and line broadening can be obtained by differentiating the Bragg equation. cot 2 sin 4 cos ) 2 (0 2d d (2 3) tan 2rms D DK (2 4) where D is the integral breadth, rms root mean square strain and DK a scaling factor. Fig. 2 7 shows powder Xray diffraction patterns from high temperature annealed and ball mi lled Mo powder [ 50]. Annealed Mo has very sharp peaks which indicate free -strain and very large gr ain size. However, the width of the diffraction peak obtained from ball milled Mo (120

PAGE 35

35 hr) is very broad which means that this sample has both finite grain si ze and microstrain. Since these two effects are generally mixed, they should be separated in order to estimate strain and size. The total peak broadening can be expressed with the combination of Eq. (2 2 ) and ( 2 4 ), tan 2 cos2 rms D V sK D K (2 5) The first and second terms in the right side relate to the finite size and microstrain, respectively. It i s possible to separate these two effects using their different angle depen dence. Multiplying both sides of above equation by cos we get : sin 2 cos2 rms D V SK D K (2 6) The plot of cos vs sin 2 which is called Williamson Hall plot [ 51], should be a line, from which the average crystallite size and root mean square (RMS) strain can be determined from the intercept and slope, respectively. In addition, the separation of the effects of cryst allite size and microstrain on the line broadening is possible by using the Warren -Averbach method, which makes use of the Fourier coefficients of the diffraction lines [52]. In this method, the Fourier coefficient ( LA ) is the product of the size coefficient( S LA ) and the strain coefficient(D LA) [53], D L S L LA A A (2 -7) War ren has also shown that the Fourier coefficient s can be written as following, 2 2 2 2 2/ 2 ) ( ln ) ( ln a h L L A L AL S (2 8) where L is the Fourier length and a the lattice par ameter. If the ) ( ln L A values are plotted versus the square of reflection order, 2h the slope and intercept of the curves could be

PAGE 36

36 determined which are corresponding to the average microstrain and size coefficient, res pectively. In addition to the line broadening by the finite size and microstrain, another broadening source is the instrumental broadening which is dependent o n the used slit system and X ray, imperfect focusing and so on. In order to perform l ine p rofile a nalysis, the Bragg diffraction peak needs to be fitted by mathematical functions. Since LPA is strongly dependent on the used fitting function, this needs to be careful ly select ed In general, the measured diffraction profile is not well fitted by pure Ga ussian or Cauchy (or Lorentzian) function s so that several other mathematical functions, such as Pearson Voigt and Pseudo-Voigt functions [5456], are suggested for an accurate Line Profile Analysis Their shape is shown in Fig. 28 X ray reflectivity (thickness, density, and roughness of the thin films ): X ray r eflectivity (XRR) curves give us useful information about the thin film, such as the thickness, roughness, and density. The X ray reflectivity curve obtained from a strained -Si1 xGex/Si layer an d its simulation are shown in Fig. 2 9 First of all, we can see the appearances of thickness fringe which are so called Kiessig fringes [ 57]. Using th e periodic of the oscillation s from the XRR curve, we can estimate the thickness of the layer. The electr on density of the material can be also determined by measuring the critical angle at which the total reflection occurs. In addition, XRR curve can be affected by free surface and interfacial roughness. The kinematical and dynamical theories for the reflect ivity allow us to determine the roughness. In Fig. 2 9 the layer model, which consists of interfacial layer, Si1 xGex, and free surface, w as applied for the best simulation. The result of the simulation for the strained -Si1 xGex sample is also shown in Fi g. 2 9 Omega 2theat rocking curve (thickness, strain, lattice mismatch, alloy composition ): In an omega Si1xGex, or Al1 xGaxN, Ge and Al compositions can be estimated by a separation of different Bragg peaks of

PAGE 37

37 substrate and strained layer. In the case of symmetric diffraction, an angular separation ( i ) is given by [ 58] B i 0 (Eq.2 9) where 0 is the angular separation due to the different amount of the refraction of the X -ray beam at the air layer and layer -substrate interface, and B is the Bragg angle di fference between the layer and substrate. Fig. 2 10 shows a rocking curve recorded around (004) Bragg angle of strained -Si1 xGex and its simulat ion using Philips Epitaxy Program The different angular position of the diffraction peaks from strained layer a nd substrate depends on Ge composition and strainrelaxation in the epitaxial layer. It was also found that the thickness oscillation occurred around the Bragg peak. The oscillation s can be understood by the multiple scattering of diffracted wave s along Qzaxis due to the different boundar ies This is similar to X ray Reflectivity. But, while XRR is independent of the crystal form, the reflection near Bragg angle in the rocking curve is dependent o n the lattice periodicity. For asymmetric Bragg diffraction, the inclination angle, which is an angle of the plane with respect to the sample surface, should be considered in an angular separation ( i ). If it i s assumed that the strained layer will be tetragonally distorted, the inclination angle of the substrate and strained layer at given diffraction plane has a different value. If the refraction is independent of the diffraction geometry, the separation of Bragg and inclination angles can be determined by using two complementary geometries, w hich are the so called and Fig. 2 11 shows the two different geometries for asymmetric Bragg diffraction and 2 recorded around for (115) plane of strained Si1xGex layer, respectively. These different geometries for asymmetric Bragg diffraction allow us to separate B and

PAGE 38

38 B B ) ( 2 1 ) ( 2 1 B (2 10) If the separation of Bragg diffraction and inclination between the substrate and top layer is given, it gives rise to the lattice mismatch components [ 58]. B B L cot tan (2 11) B B L cot cot// (2 12) where L and // L are perpendicular and parallel mismatch, respectively. The cubic lattice mismatch ( L ) is given by 12 11 12 // 112 2 c c c c a a aL L S S L L (2 13) where 11c and 12c are the elastic constants at each crystallographic direction and La is a cubic lattice parameter at certain Ge composition [ 59, 60]. Reciprocal space map (relaxation, strain, lattice m ismatch, alloy composition ): The reciprocal s pace m ap (RSM) represents an intensity distribution of the diffracted x rays along QZ and QX at which QZ-axis is parallel to the normal of the sample surface in outward direction and QXaxis is parallel to the s ample surface. Fig. 2 12 (a) shows the position of two reciprocal lattice points belonging to the substrate and the layer for symmetric and asymmetric geometry. In the case of symmetric geometry, the perpendicular mismatch between the substrate and layer c an be measured from reciprocal space map, while there is no information about the parallel mismatch due to same inclination angle. Since the asymmetric space map could reveal the parallel mismatch as well as the perpendicular mismatch, it is more helpful t o provide more

PAGE 39

39 information about the strained layer. In Fig. 2 12 (a), if the reciprocal lattice positions of the substrate and layer are Q(XS,ZS) and Q(XL, ZL), the lattice mismatch can be given by L S LZ Z Z (2 14) L S LX X X // (2 15) As the perpendicular and par allel mismatches are determined, it is possible to estimate Ge composition of the strained layer. In Fig. 212 (b), a r eciprocal s pace m ap recorded around (115) diffraction point of strained -Si1 xGex is shown. The reciprocal lattice points of silicon subst rate and strained Si1 xGex layer are located to above and below region, respectively. It was found that the parallel and perpendicular lattice mismatches were 116 ppm and 10350 ppm and the estimated Ge content was 15.70%. This small parallel mismatch indic ates that the strained -Si1xGex layer is pseudomorphically grown on silicon substrate. In other way, the shape and position of each reciprocal lattice point also contain microstructural information. Around the reciprocal lattice point of Si substrate, the coherent crystal truncation rod and analyzer streak appear ed In contrast to Si substrate, it is shown that the reciprocal lattice point of the strained layer is relatively broadening along QZaxis because it has a finite vertical thickness compared to the infinite thickness of the silicon substrate. Another information obtained from this reciprocal space map is the presence of a tetragonal distortion of the strained Si1 xGex layer which is confirmed because the different inclination angle between two cryst als can be expressed by the angle difference between two reciprocal lattice directions. Mosaicity (tilt, twist, and lateral & vertical coherence length ): T he lattice parameter, thermal expansion coefficient, and crystal structure of the epitaxial layer wit h respect to the substrate should be considered for high quality film growth due to the lack of bulk substrate.

PAGE 40

40 These mismatches during the film growth generate the mosaic structure which means that the films have sub -grain slightly misoriented with respe ct to each other and the underlying substrate. Fig. 2 13 represents a reciprocal space map (RSM). Fig. 2 13 (a) shows how the RSM is recorded around a given lattice point by omega 2theta scan s with different offset omega values. Fig. 2 13 (b) represents a comparison between reciprocal and real lattices. If it is assumed that the substrate has a perfect crystal structure, there is not any broadening effect in the reciprocal lattice point of the substrate. However, the epitaxial layer should have some structu ral defects which are a source of the broadening of the reciprocal lattice point. In general, since the thickness of the epitaxial layer is much thinner compared to the substrate, the reciprocal lattice point is vertically broadened. I n this study, the rec orded reciprocal lattice point s due to heteroepitaxial mosaicity were broadened in both the lateral and the vertical directions. Generally, the mosaic structure can be described by four parameters; tilt, twist, lateral and vertical coherence length s [61]. Fig. 2 14 (a) shows a mosaic structure. The tilt and twist are out of plane and in -plane misorientation, respectively. T he coherence length can be defined as an average size extension of the crystal lattice regions which scatter coherently and are defects free. These mosaic domains cause reciprocal lattice point broadening which in turn means that they are related to crystal structural defects. Fig. 2 14 (b) represents the reciprocal lattice point broadening due to the mosaic spread. The finite vertical c oherence length broadens the reciprocal lattice point along QZ axis, whereas the finite lateral coherence length broadens it along QX or QY axis. These coherence lengths give information on average crystallite size in a mismatched epitaxial layer. In addition, the mosaic tilt & twist broaden the reciprocal lattice point along the lattice tilt direction. In general, since these mosaic parameters are mixed, they need to be separated. Since the separation of these

PAGE 41

41 effects is still very complex using x ray diff raction, it is an active research topic The most interesting result in this study is that these mosaic parameters are strongly dependent of the specific type of the dislocations. This will be presented in chapter four Transmission Electron M icroscopy (TEM) To obtain highly magnified images of the samples structure transmission electron microscopy (TEM) technique has been widely used. Basically, TEM is very similar to the optical microscopy (OM) because both have a series of lenses system s to magnify a n i mage. However, TEM has a great strength in the resolution issue, approaching around 0.08 nm. The following is a resolution equation, A N s 61 0 (2 -16) where s is the resolution (the minimum distance between points or parts of an object), is the beam wavelength, A N is the numerical aperture (resolving power of the lens and the brightnes s of the image). In the optical microscopy, NA is around 1 and =500 nm, giving s= 300 nm. In TEM, while NA is around 0.01 due to large imperfection of the electro-magnetic lens system, the electron wavelength conventionally used in TE M under 100 kV is around 0.004 nm, giving s=0.25 nm which is much better than OM. A schematic of a TEM instrument is shown Fig. 2 15 [62]. The high energy electron beam with 100 to 400 kV voltages are generated and accelerated from the electron gun. There are two kinds of the electron gun: thermal and field emission electron gun. The electron beam can be focused on the thin sample (around 50~200 nm thick) by the electro-magnetic condenser lenses. The transmitted and scattered electron beams from the thin s ample give diffraction pattern and magnified image in the back focal and image planes of the objective lens respectively. There are

PAGE 42

42 several imaging methods using TEM: bright -field, centered beam dark -field, weak beam dark field and high resolution lattice imaging techniques. In this study, both the TEM 200 CX and TEM 2010F instruments located in MAIC, University of Florida, were used. X -ray P hotoelectron S pectroscopy (XPS) X ray photoelectron spectroscopy (XPS), which is sometime also known as electron spectroscopy for chemical analysis (ESCA), is a technique using the photoelectric effect. This technique is used for identifying chemical elements and bonding information for all elements except hydrogen and helium atoms. Mg K (1253.6 eV), Al K (1486.6 eV), or monochromatic Al K (1486. 7eV) xrays are usually used [ 63]. XPS is a surface sensitive technique because only egenerated in the upper 0.5 5 nm depth of the sample could exit the material without losing energy. Fig. 2 16 shows a schematic of a XPS instrument and its electron excitation process [62]. A high energy x ray beam is incident into the sample, and then the electrons are emitted from specific orbitals when x ray energy exceeds their binding energy. The conserv ation of energy for the ejected electrons taking into account the binding energy can be described as following, q E hv Edectector b (2 17) where bE is a binding energy, hv is the incident xray energy, dectectorE is the measured energy in the electron detector, and is the work function. The binding energy may be regarded as the e nergy difference between the initial and final states after the photoelectron has left the atom. Note that since the binding energy is dependent o n the incident x -ray energy, it should be monochromatic. Each element has a specific binding energy and it is possible to identify the element information using tabulated values and other references. In addition, since the electron

PAGE 43

43 binding energy is dependent of the chemical ambience, XPS allows for understanding of t he chemical bonding state. A tomic F orce M i croscopy (AFM) Atomic force microscopy (AFM) or scanning probe microscopy (SPM) is a high resolution surface imaging technique. The AFM consists of a microscale cantilever with a sharp probe, laser source, photodiode and detector & feedback electronics, as shown in Fig. 2 17 [6 4 ]. The principle of AFM is based on the effect of interaction force between the sharp tip and the sample surface that bends the cantilever When the tip is brought to the sample surface, the cantilever bends by responding to the inte raction force according to the topography of the surface, resulting in the dependent position of the incident laser beam on the detector. Finally the photodiode, which is sensitive to the cantilever deflection, generates topographic data. There are two different imaging methods in AFM: tapping and contact modes. In the tapping mode, the image can be produced by imaging the interaction force of the oscillating contacts of the tip with the sample surface. However, in the contact mode, the cantilever drags directly across the sample surface at constant force, resulting in possible surface damage. Hall M easurements Hall effects allow us to determine the carrier type, carrier density, carrier mobility and film resistivity. When the magnetic field is applied normal to the sample surface, the electric field is set up across the sample due to the charge separation This potential difference by the induced electric field is so called Hall voltage. Hall theory gives the Hall coefficient ( HR ) as following [62], ) ( ) ( ) ( ) (2 2type n qp r type p qp r bn p q n b p r RH (2 18)

PAGE 44

44 where HR is the Hall coefficient, r is the scattering factor, p is the hole density, n is the electron density, and ) ( ) ( mobility hole mobility electron bp n Typically, i t is assumed that r is of the order of unit. The Hall coefficient allows us to determine the carrier type as well as the carrier concentration.

PAGE 45

45 Figure 2 1. A schematic of a conventional MBE system [35]. 100 200 300 400 500 800 1000 2000 -20 0 20 40 60Temperature ( )Vapor pressure ( torr ) SiH2Cl2SiHCl3SiCl4 Figure 2 2. Vapor pressure dependence with temperature for different silicon sources used in epitaxy [37].

PAGE 46

46 Figure 2 3. The schematic diagram of a pulsed laser deposition (PLD ) system [47]. 2(a) (b)Diffracted beam Incident beam Figure 2 4. PANalytical X Pert system. (a) s chematic of XPert system and (b) four axi s variables.

PAGE 47

47 Figure 2 5. Mirrors used in primary optics. (a) x ray mirror and (b) hybrid mirror (a) (b) Figure 2 6. Secondary optics. (a) p arallel plate collimator and (b) triple axis optics

PAGE 48

48 Fig ure 2 7. Powder x ray diffraction pattern of Mo powder. (a) high temperature annealed Mo powder and (b) ball milled (120hr) Mo powder [ 50]. Fig ure 2 8 (102) reflection of the GaN grown on sapphire substrate and curve fitting using Gauss, Cauchy, Pearson, and Pseudo-Voigt functions

PAGE 49

49 Fig ure 2 9 Experimental and fitted X -ray reflectivity curves of a strained Si1xGex on silicon substrate. The simulation was performed by using Wingixa software from PANalytical. Fig ure 2 10. Experimental (004) o mega 2 t heta rocking curve and simulation curve Th e simulation was carried out by using Epitaxy software from PANalytical.

PAGE 50

50 Fig ure 2 11. Omega 2theta rocking curve of asymmetric diffraction plane. (a) and + geometry and (b) o mega 2theta rocking curve obtained from and + geometry in strained Si1xGex on silicon substrate [58]. Perpendicular mismatch Parallel mismatch Perpendicular mismatch Substrate Reciprocal Point Layer Reciprocal Point QXQZ (a) (b)Silicon substrate Strained Si1 xGexlayer Q(XL, ZL) Q(XS, ZS) Perpendicular mismatch Parallel mismatch Perpendicular mismatch Substrate Reciprocal Point Layer Reciprocal Point QXQZ (a) (b)Silicon substrate Strained Si1 xGexlayer Q(XL, ZL) Q(XS, ZS) Fig ure 2 12. Reciprocal space map (RSM). (a) reciprocal lattice point s of substrate and layer in symmetric and asymme tric Bragg diffraction and (b) reciprocal space ma p recorded around (115) plane

PAGE 51

51 Fig ure 2 13. Reciprocal space map and real crystal lattice. (a) r eciprocal s pace m ap and (b) c omparison between reciprocal lattice point and real space lattice Fig ur e 2 14. Mosaicity. (a) m osaic structure of epitaxial layer and (b) reciprocal lattice point of (002) and (102) [61].

PAGE 52

52 Figure 2 15. A schematic of a transmission electron microscop e (TEM) [62]. EKEL1EL2,3EVECEVac Incident x ray Figure 2 16. A schematic of a XPS instrument and its ele ctron excitation process [62].

PAGE 53

53 Detector and Feedback Electronics PZT scanner Sample surface Cantilever & tip Laser Photodiode Detector and Feedback Electronics PZT scanner Sample surface Cantilever & tip Laser Photodiode Figure 2 17. A schematic of a AFM instrument and its operation [6 4 ].

PAGE 54

54 CHAPTER 3 DEFECTS EVOLUTION IN P TYPE ZNO Phosphorus D oped ZnO Thin F ilms G rown on S apphire S ubstrate s Experimental D esign The ZnO:P la yers were grown on c -plane sapphire substrate s by the pulsed laser deposition (PLD) technique at 700 in an oxygen partial pressure of 150 mTorr. Prior to ZnO:P deposition, an undoped ZnO buffer layer was deposited on the sapphire at 400 in 20 mTorr. The buffer layer was annealed at 650 in an oxygen ambient prior to ZnO:P deposition. This served as a nucleation layer for ZnO:P growth, greatly improving surface morphology and reproducibility in the transport properties. Phosphorus -doped ZnO targets were fabricated using high -purity ZnO powder (99.9995 %) mixed with P2O5 (99.998 %) powder as the doping agent and sintered at 1000 for 12 h in air. The phosphorus doping levels chosen for this study were 0.5 and 1 at. %. The target was ablated by a KrF excimer laser with a laser frequency of 1 Hz and energy density of approximately 1.5 J/cm2. The film thickness was approximately 400 nm for the ZnO:P layer and 100 nm for the undoped ZnO buffer layer. The microstructure of ZnO:P was examined by Philips MRD X Pert system for 2theta/theta (2 / ) scan and omega rocking curves ( RCs) The room temperature electrical properties were invest igated by Hall effect measurements using the four -point van der Pauw geometry with a commercial LakeShore Hall measurement system. X -ray photoelectron spectroscopy (XPS) was used to investigate the chemical bonding state of atoms in ZnO:P films with a Perkin Elmer PHI 5100 ESCA System. In XPS analysis, the position of the adventitious C 1s peak was considered as a standard reference with a binding energy at 284.6 eV.

PAGE 55

55 Results and D iscussion Microstructure of p hosphorus d oped ZnO f ilms : h igh quality ZnO films have been grown on several substrates, such as sapphire (Al2O3), silicon carbide (SiC), gallium nitride (GaN), silicon (Si), and so forth [30 33]. Of these substrates, c -plane sapphire substrates are commonly used for ZnO film growth due to its hexagonal crystal structure, transparency and relatively low cost. However, the large mismatch of the in plane lattice parameter and rather different thermal expansion coefficient between the sapphire (ain plane=4.757, Ka/ 10 3 76 ) and zinc oxide (ain plane=3.252 Ka/ 10 9 26 ) are a challenge for the high quality film growth [ 29]. Fig. 3 1 (a) shows the } 1 1 10 { pole figure of as -grown, 0.5, and 1.0 at. % P -doped ZnO films grown on the c -plane sapphire substrate. It is shown that the ZnO films are textured along the c axis and the degree of texture is degraded with increasing phosphorus atomic percent in ZnO. Interestingly, it is also found from the pole figure that there are two columnar structures with different in -plane or ientations in hexagonal crystal structure in which one possesses higher crystallinity than the other. In order to identify the crystal orientation, the } 1 1 10 { and } 4 1 10 { phi ( )-scans of ZnO film and sapphire substrate, respectively, w ere performed, as shown in Fig. 3 1 (b). While the broad periodic peaks are exactly matched to the sapphire peaks, the sharp ones are deviated from the substrate peaks. The measured deviation angle between the sharp and broad peaks was exactly 30. The growth of ZnO films on c plane sapphire substrate resulted in the in -plane epitaxial relationship of ] 0 2 11 [ // ] 0 1 10 [3 2O Al ZnO and ] 0 1 10 [ // ] 0 1 10 [3 2O Al ZnO for two columnar structures with sharp and broad peaks, respectively. T he reason for smaller peak width of ZnO films with 30 in plane rotation with respect to the sapphire substrate is a decrease in the lattice mismatch from 31.6 % to 18.4 %. Conventionally, the two step growth method has been used for high quality

PAGE 56

56 ZnO films in which the nucleation layer was grown at low temperature or low oxygen pressure, followed by subsequent growth of the main ZnO at high temperature and/ or high oxygen pressure [6 5 6 6 ]. In our study, the growth temperature used during the deposition of the nucleation layer was 400, which is lower than that conventionally used for high crystalline quality n type ZnO film growth. Relatively lower nucleation temperature causes a reduction of adatom surface mobility/diffusion length during the film growth, leading to poor er crystallin e quality. RCs) were recorded in order to investigate the crystalline quality, as shown Fig. 3 2 (0002) and ) 1 1 10 ( RCs were measured in symmetric and asymmetric geometr ies For all samples the RC s show ed higher intensity and smaller full width at half maximum (FWHM) values than the ) 1 1 10 ( RC s, which indicates that the in plane columnar domains are twisted with a certain angle due to the presence of threading dislocations, resulting in the formation of the low angle grain boundary. It was reported that the low angle grain boundary was due to the edge type dislocations generated from the high defective initial nucleation layer [ 6 7 ]. Hence, the ZnO samples consisted of highly textured colum nar grains along c axis with certain width. This width can be described by the lateral coherence length, which can be defined by the average extension of the crystal lattice regions which scatter coherently. [ 61, 6 8 ]. The lateral coherence length calculated RC gives under estimated value because an influence of the edge type dislocations in the low angle boundary on the change in the d -spacing of out of plane is very slight. Therefore, the coherence length should be measured from the FWHM of th RC of the diffraction plane with high inclination angle ( ), which is the angle between sample surface normal direction and diffraction plane normal direction. From the Scherrer equation and FWHM values of ) 1 1 10 ( RC ( =60)

PAGE 57

57 [49], the calculated lateral coherence length is 67 40 and 15 for as -grown, 0.5, and 1.0 at % P doped ZnO films, respectively. It is clear from Fig. 3 2 that the crystalline quality degrades and the columnar grain width decreases with increasing phosphorus atomic percent in ZnO films. Line p rofile a nalysis of p hosphorus d oped ZnO : for further analysis of the microstructure of P -doped ZnO films, an x ray diffraction line profile analysis (LPA) was performed. The width of Bragg reflection s is caused by the microstructural features, in which the finite crystallite size and microstrain play a key role in the diffraction line broadening. Several {0001} diffraction planes were recorded in 2 / scan mode. Using a Williamson Hall plot, the strain a nd size broadening effects can be separated, as shown in Fig. 3 3 [ 51]. The plot of cos versus sin 2 in which is the integral breadth, is the Bragg angle, and is the x ray wavelength gives a straight line which allows for the determination of average crystallite size (Vrms) from the intercept and slope, respectively. It is clearly shown from the Williamson Hall plot th at while the crystallite size is almost identical at around 500~600 in all ZnO samples, the 0.5 at % P -doped ZnO film shows the highest microstrain of around 0.119 %. For an accurate separation of these two contributions, the data were analyzed by the Wa rren -Averbach method which makes use of the Fourier coefficients from at least two harmonic reflections [ 53]. In this study, the Warren -Averbach method was performed by Philips Line Profile Analysis program using the (0002) and (0004) diffraction planes. T he result i s shown in Fig. 3 4 and table 3 1 Although two methods give different strain values about ZnO films, the 0.5 at % P -doped ZnO film shows the highest strain by both methods. The behavior of the P atoms in ZnO films is still under debate. It was reported that the P atoms are substitutionally incorporated on the oxygen site s resulting in the formation of a deep level acceptor from the valence band of ZnO [ 6 9 ]. Another report showed that the phosphorus doping

PAGE 58

58 in ZnO formed the antisite complex (PZn2VZn) due to the large mismatch between the phosphorus and oxygen ionic radii [70]. Regardless of the different behavior of the phosphorus as mentioned above, we believe that its incorporation into the ZnO films must generate an internal strain. From the x ray diffraction line profile analysis, the phosphorus in 0.5 at % P doped ZnO was effectively incorporated into ZnO films, resulting in the highest strain. From the Hall measurement, the 1.0 at % P -doped ZnO film ( 3 1710 8 1 cm ) showed lower ca rrier density than the 0.5 at % P -doped ZnO film ( 3 1710 2 6 cm ) which is consistent with lower substitutional incorporation confirming the results of the x -ray diffraction line profile analysis. Phosphorous s egregation : for the chemical analysi s of the films x -ray photoelectron spectroscopy (XPS) investigation s were performed. Fig. 3 5 shows high resolution scan s of the P 2p3/2 region for the 0.5 and 1.0 at % P -doped ZnO films. The acquired spectra can be deconvoluted into two different peaks c entered at binding energ ies of 130.0 and 133.5 eV [7 1 ]. The lower binding energy, denoted by P1, corresponds to P -P bonding state, while the higher binding energy could be regarded as P O bonding state. Hence, P incorporated into ZnO films exists as P -P an d PZn. In Fig. 3 5 the ratio of AP1/AP2 was determined, indicating that P in 0.5 at % P doped ZnO film has been effectively incorporated into Zn site while P in 1.0 at % P -doped ZnO film tended to segregat e From the analysis of XRD and XPS results, it ap pears that 0.5 at % P doped ZnO film show s higher strain due to the occupancy of P into Zn site while 1.0 at % P doped ZnO film relaxed a strain with the P segregation, resulting in the degradation of the crystalline quality and electrical properties Conc lusion The microstructure of P -doped ZnO films with different phosphorus atomic percent was investigated. From XRD and XPS data, a significant fraction of phosphorus atoms in 1.0 at % P -

PAGE 59

59 doped ZnO film are segregated while in 0.5 at % P doped ZnO film they are effectively incorporated into ZnO film, resulting in high internal strain. The degree of texture and crystalline quality of ZnO films degraded with increasing phosphorus atomic percent.

PAGE 60

60 Table 3 1 Comparison of results from William son Hall plot and Warren -Averbach method applied to as -grown and phosphorus doped ZnO films Williamson Hall plot Warren Averbach method Strain (%) Crystallite size ( ) Strain (%) Crystallite size ( ) As grown 0.002 540 0.038 550 0.5 at. % P doped 0 .119 680 0.072 618 1.0 at. % P doped 0.035 558 0.059 535

PAGE 61

61 Figure 3 1. Crystal orientation of ZnO films. (a) p ole figure s of as -grown, 0.5 and 1.0 at % P doped ZnO films and (b) phi scan s of } 4 1 10 { sapphire substrate and } 1 1 10 { ZnO film (0002) (1011)19.148 3.235 1.0 at. % 7.082 1.683 0.5 at. % 4.209 0.779 As grown (1011) (0002) FWHM ( ) 19.148 3.235 1.0 at. % 7.082 1.683 0.5 at. % 4.209 0.779 As grown (1011) (0002) FWHM ( ) 5 10 15 20 25 30 100 1000 10000 100000 0 5 10 15 20 25 30 35 100 1000 Counts Fig ure 3 2. Omega rocking curves of symmetric and asymmetric diffraction plane s of as -grown, 0.5, and 1.0 at % P -doped ZnO films

PAGE 62

62 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 0.0018 0.0020 0.0022 0.0024 0.0026 0.0028 0.0030 sin 2 cos As-grown ZnO 0.5% at p-doped ZnO 1.0% at p-doped ZnO 0002 0004 0006 Figure 3 3. Williamson Hall plot of as -grown, 0.5, and 1.0 at % P -doped ZnO films As-grown 0.5% P-doped 1.0% P-doped 0.00 0.02 0.04 0.06 0.08 0.10 0.12 Samplesstrain (%) Williamson-Hall plot Warren-Averbach method

PAGE 63

63 Figure 3 4. Mi crostrain of as -grown, 0.5, and 1.0 at % P -doped ZnO films obtained from the Williamson Hall plot and Warren -Averbach method Binding energy ( eV ) Binding energy ( eV )Intensity ( a.u .) Intensity ( a.u .)P1 P2 P1 P2 AP1/AP2=0.47 AP1/AP2=1.04(a) (b) Figure 3 5 XPS spectra acquired from 0.5 and 1.0 at % P -doped ZnO films. P 2p3/2 of (a) and 0.5 (b) 1.0 at % P -doped ZnO films.

PAGE 64

64 CHAPTER 4 D EFECTS EVOLUTION IN G AN EPITAXIAL LAYERS GaN G rown on S apphire S ubstrate Experimental D esign The GaN films were grown on c -plane sapphire substrates using a Veeco P75 vertical MOCVD reactor via a conventional two -step growth method [ 1 6 1 8 ]. Trimethylgalliu m (TMGa), ammonia (NH3), and hydrogen (H2) were used as the Ga and N precursors, and carrier gas, respectively. The growth process was monitored by the reflectance transient method [ 7 2 ]. A 20 nm thick GaN nucleation layer (NL) was grown initially at 532~550 Then, the temperature was ramped up to 1028 1096 for high temperature three dimensional (3D) islands growth. The growth conditions for the high temperature islands and low temperature nucleation layer which are shown in table 4 1, were controlled by changing TMGa flow rate with the same V/III RCs and -scans in skew and grazing incidence geometries. For the cross sectional TEM imag ing of the GaN layers, a JEOL TEM 200CX was used and the sample preparation was carried out by using the focused ion beam (FIB) technique. Results and D iscussion Grazing i ncidence x -ray d iffraction (GIXD) technique : m any investigations have been performed using high resolution xray diffraction (HRXRD) for the estimation of threading dislocation density of high quality GaN films because of their specific dislocation structure i n which the line direction and Burgers vectors are ] 0001 [ l ] 0001 [ screwb ] 0 2 11 [ 3 1 edgeb

PAGE 65

65 respectively [7 3 7 4 ]. The structural defects of GaN with a high dislocation density can be described by mosaic parameters: tilt, t wist, lateral and vertical coherence lengths [61, 6 8 ]. Tilt and twist are out -of (polar spread) and in -plane (azimuthal spread) misorientations, respectively, and the lateral and vertical coherence lengths are an average crystallite size that is free of def ects. The reciprocal lattice point s are broadened along specific direction s by these mosaic parameters. The strain fields near the screw and edge dislocations, which distort the d -spacing of the planes parallel and perpendicular to the surface, respectivel y, are associated with the polar and azimuthal spreads [75]. If the widths of the omega rocking RCs) measured from parallel and perpendicular planes are obtained, then the screw and edge dislocation densities can be estimated. Measurement of the polar spread is very straightforward and directly obtained by a symmetric high resolution X ray diffraction setting. The determination of the azimuthal spread requires diffraction from planes normal to the surface and it can be achieved by transmission or edge geometry diffraction methods as shown in Fig. 4 1 [ 7 6 7 7 ]. However, the diffracted beam fro m the film is too heavily attenuated by the thick substrate for the transmission geometry and a micro -focusing X ray beam is required for the edge geometry [78]. On the other hand, Srikant et al. indirectly estimated the twist angle by using the plot of the full width at half maximum RCs vs the inclination angles ( ) and Heinke et al. reported that RC of ) 2 3 30 ( plane with a large inclination angle ( 72 ) could be used as a figure of merit (FOM) for the edge dislocation density [7 9 80]. Also, the theoretical model for the determination of the mosaic parameters developed by Lee et al. showed much improved agreement between the XRD and TEM results [81]. Here, this stu dy presents the results of the grazing incidence phi ( )-scans to directly obtain the twist parameter without any complicated computation.

PAGE 66

66 RCs of onaxis reflections, such as (0002), (0004), and (0006), are measured, the tilt angle can be estimated by using the Williamson Hall plot in which the slope and y axis interception correspond to the tilt angle and lateral coherence length, r espectively [73]. On the other hand, when diffraction from the surface normal planes ( 90 ), such as ) 0 1 10 ( or ) 0 2 11 ( is required for the determination of the twist angle, the grazing incidence -scan is performed. In this study, the broadening of the RCs due to the finite vertical and lateral coherence length is neglected because its effect is minor. Other broadening effects, such as X ray extinction, film curvature, and instrument al limits, are also negligible because they contribute to only a few arcsec [82]. -RCs of (0002) and -scan of ) 0 1 10 ( are considered as a FOM for screw and edge type dislocation densitie s, respectively. Fig. 4 2 (a) shows a -scan of the } 0 1 10 { plane recorded by grazing incidence XRD technique. The six diffraction peaks are well defined, which represent the six -fold symmetry of the wurtzite GaN. Sinc axis is co axial with the axis for 90, the diffraction profile of the ) 0 1 10 ( RC measured under grazing incidence angle is shown in Fig. 4 2 defined as the angle between the X -ray incident beam and the diffraction plane parallel direction and The in RC is 0.0001, which is much finer than -scan b ecause the maximum resolution step of the axis in our XPert system is just 0.01. Therefore, all grazing incidence RC technique measured in In addition, the line profiles were fitted by using a Pseudo-Voigt function, because a pure Gaussian function was not suitable due to the broadening near the tail region of the diffraction profiles [56].

PAGE 67

67 T he measured FWHM values of the in -plane, asymmetric pl ane and out of RCs of the GaN films used in this study are shown in T able 4 2 In this study, the instrumental broadening was neglected because it contributed to only a few arc sec. The tilt and twist angles of the high defective GaN layers are associated with the screw and edge type dislocations ( ]0001 [ l ] 0001 [ screwb ] 0 2 11 [ 3 1 edgeb ), respectively [ 73, 7 4 ]. Heying et. al. reported that FWHM of ) 2 1 10 ( -RC could be a relevant figure of merit (FOM) for the crystal quality because this can be broadened by both type disloc ations [ 83]. As shown in T able 4 2 the FWHM value s of RC of the surface normal plane are much larger than one of the surface parallel plane, which is indicative that the edge type threading dislocations are dominant. Therefore, the crystalline quality of GaN layer should be determined only by the edge dislocation density. Typical bright field images from sample s A and E in Table 4 1 are shown in Fig. 4 3 Under 0002 g the type of the threading dislocations that can be observed a re pure screw or mixed ( ] 0001 [ l ] 3 2 11 [ 3 1 mixedb ), while the edge or mixed dislocations are visible under 0 2 11 g By analyzing Fig. 4 3, it was found that the edge and mixed type dislocations were the dominant defects in the GaN films, w hich is consistent with XRD data mentioned above. Heinke. et al. suggested that t he FWHM of ) 2 3 30 ( RC could be a figure of merit for the edge dislocation density because this plane has a high inclination angle (~71) with respect to the surface plane. They reported that the twist angle was 1.14 0.04 times the FWHM value of the ) 2 3 30 ( RC [84]. In our case, the sample D shows the minimum FWHM value of the ) 2 3 30 ( RC, but the sample E shows the minimum FWHM value of the ) 0 1 10 ( -RC and the best crys talline quality of all GaN films. This discrepancy might be due to the dependent distribution of the tilt and twist angles. Srikant et al. used a mathematical formulation to obtain a relationship between

PAGE 68

68 the tilt and twist distributions [ 79]. Therefore, a FOM for the crystalline quality should be RC of the surface perpendicular plane. In this aspect, the grazing incidence X -ray diffraction is a more suitable technique than above mentioned theoretical method for the stud y of the growth optimization since it requires only one omega rocking curve and the twist angle can be directly measured without any theoretical computation. Relationship b etween g rowth c onditions and d efects : Fig. 4 4 shows a plot of the RC of sample A (t3D=10.0 min) and B (t3D=16.7 min) and inclination angles ( RC increase with increasing the inclination angle. While Zheng et al. determined the mean twist angle by taking the average values of FWH M RC and -scan of ) 1 3 12 ( plane with the inclination angle 78.6, this study uses a FWHM value of ) 0 1 10 ( RC as a twist angle, which is directly obtained from the grazing incidence ge ometry [85]. Now, the screw and edge -type dislocation densities can be estimated RC and ) 0 1 10 ( -scan into the following equations, 2 2 00029screw screwb FWHM D (4 1) 2 2 0 1 109edge edgyb FWHM D (4 2) where screwb (=0.5185nm) and edgeb (=0.3189nm) are Burgers vectors [86]. The dislocation density of GaN layers grown using different 3D growth mode times is shown in Fig. 4 5. The 3D growth mode time during the HT island growth increases with dec reasing growth temperature and TMGa flow rate, as shown in Table 4 1 All GaN layers grown under same growth condition for the nucleation layer possessed almost the same screw dislocation densities, around

PAGE 69

69 7107~8107/cm2, which was independent of the 3D H T island growth. However, the edge threading dislocations decreased with increasing 3D growth mode time, as shown in Fig. 4 5. Mixed type dislocations with ] 3 2 11 [ 3 1 mixedb contributed to the broadening of the RCs of both parallel and perpendicular surface planes. However, since the FWHM values of (0002) rocking curves of all samples used in this study were almost the same, it could be inferred that the defects generated during the 3D HT island growth ar e pure edge -type dislocations. For further inve stigation of the defects structure of GaN films TEM investigations were performed. Typical bright field images obtained from sample D (t3D=21 min) are shown in Fig. 4 6. The Burgers vector of dislocations in GaN films was determined using the method, kn own as the g b=0 invisibility criterion [87]. Bright field images were taken near the ] 00 1 1 [ zone axis (Fig. 4 6 (a)) using a two -beam condition. The results, displayed in Fig. 4 6 (b) and (c), were obtained using different diffraction vectors in order to figure out the type of the threading dislocations. Under 0002 g the types of the threading dislocations that can be observed are pure screw or mixed, while the edge or mixed dislocations are present under 0 2 11 g It was found from TEM results that the threading dislocations of the GaN layers are either pure edge or mixed types. Very few pure screw dislocations were found in these GaN samples. An increase of the number of defects in the interfacial region could be seen, which corresponds to the initial nucleation layer. The data shown above provides us several information including the generation and reduction of the threading dislocations of GaN films. It is clear from the TEM images that the threading dislocations are mixed and pure edge -types in these GaN films. Therefore, the XRD data indicates that the broadening of (0002) and ) 0 1 10 ( -RCs is due to mixed and pure edge -

PAGE 70

70 types threading dislocations, respectively. For sample A and B the ramping rate and HT island growth temperature were the same, only the TMGa flow rate w as varied, resulting in different lateral growth rate (t3D(A): 10 min, t3D(B): 16.7min). A low island growth temperature in sample C gives a reduced number of HT island density and slow the lateral growth rate (t3D(A): 18.5 min), compared to sample A and B. For longer 3D growth mode time with respect to other samples, sample D (t3D(A): 21 min) was grown at a relatively low temperature and TMGa flow rate. The mixed type threading dislocation density of A, B, C, and D samples described in this study is almost the same, as shown in Fig. 4 5. Hence, the mixed -type dislocations are independent of the HT island density and lateral growth rate. The edge type dislocation density decreased with decreasing the lateral growth rate as well as the HT island density. The reason is that the longer the 3D growth mode time was, the more likely it was for the edge type dislocations to annihilate each other. It is clearl y seen in Fig. 4 6 (c) that edge dislocations near the interfacial region between GaN film and sapphire substrate are bent, which indicates concrete evidence for the dislocation annihilation by their interaction. Next, the nucleation layer would be thought as another source of defects generation because it acts as a template for high temperature main GaN films growth As shown above, since the mixed threading dislocation density was not changed even though using different growth conditions used during the high temperature island growth, we believe that the mixed dislocations could be controlled by changing the growth condition for the nucleation layer. In addition, the initial GaN layers grown at low temperature were found to have a zinc blend crystal struct ure with high density of stacking faults in one set of {111} planes, and then they were transformed into the wurtzite crystal structure by introducing Shockley partial dislocations during the high temperature ramping [ 19, 88]. In this time, the edge type di slocation can be also

PAGE 71

71 generated from the reaction between the sub -boundary dislocations and these Shockley partial dislocations. Since this large stacking disorder of nucleation layers was caused by a low growth temperature, it has been thought that the gr owth temperature must be an important parameter having a great influence on the crystalline quality of the main GaN layers. Sample s E and F have a nucleation layer grown at a relatively lower temperature than sample s A, B, C, and D, as shown i n Table 4 1 Sample F which have the nucleation layer grown at 542 and the 3D growth mode time of 19.3 min showed the highest crystalline quality among our samples. Conclusion -RC mea sured by grazing incidence X -ray diffraction. The threading dislocation density could be controlled by using different growth conditions during the deposition of the nucleation layer and high temperature islands growth. The results from XRD and TEM investi gations showed that the pure edge and mixed type dislocations were the dominant defects in our GaN films. Therefore, the relevant figure of merit (FOM) for the crystalline quality should be determined by the FWHM RC of the surface perpendicular planes. In our study, a GaN film grown at 542 for the nucleation layer and with the 3D growth mode time of 19.3 min showed the best crystalline quality so far.

PAGE 72

72 Table 4 1. The growth conditions used d uring the deposition of GaN nucleation layer and 3D islands Growth conditions Samples A B C D E F Nucleation layer Temperature ( ) 550 550 550 550 532 542 V/III 19880 19880 19880 19880 19880 19880 3D island growth Temperature ( ) 1094 1095 1028 1062 1085 1096 TMGa (sccm) 10 8 10 9 9 9 V/III 5964 6088 5964 6012 6012 6012 Time (min) 10.0 16.7 18.5 21.0 16.6 19.3 Table 4 2. FWHM values (arc sec) of in asymmetric and out of RCs. The FWHM value s of the ) 0 1 10 ( RC are much higher than one of the ) 0002 ( -RC. Plane Samples Si reference samples A B C D E F FWHM ) 0002 ( 0.0 195 199 204 195 215 194 0.0 11 ) 2 1 10 ( 42.9 419 346 330 310 313 333 35.6 11 ) 2 3 30 ( 70.2 537 444 399 3 67 387 414 72.1 8 ) 0 1 10 ( 90.0 593 490 477 451 443 423 90.0 13

PAGE 73

73 Figure 4 1. Transmission and edge geometries for the determination of the azimuthal spread Figure 4 2 Grazing incidence x ray diffraction (GIX D). (a) g razing incidence } 0 1 10 { scan (b) (10 10) RC of sample A (t3D=10 min). The acquired phi scan corresponds to the six -fold symmetry of wurtzite structure of GaN

PAGE 74

74 Figure 4 3 Cross sectional bright field TEM images of GaN films. (a, b) and (c, d) were taken from sampl e A and E respectively Figure 4 4 -RCs measured from different diffraction planes of sample A (t3D=10 min) and B (t3D=16.7 min)

PAGE 75

75 Figure 4 5 The screw and edge type dislocation density of GaN films grown using various 3D growth m ode times. Figure 4 6 Cross sectional bright field images of sample D The images were taken near (a) ] 00 1 1 [ zone axis using two -beam condition ((b) 0002 g and (c) 0 2 11 g ). The edge and mixed type d islocations are marked as e and m, respectively

PAGE 76

76 CHAPTER 5 DEFECTS EVOLUTION IN STRAINED LAYERS DURING STRAIN RELAXATION StrainedSi Layer G rown on G raded SiGe Substrate Experimental D esign A compositionally graded layer to confine threading dislocatio ns was grown by gradually accommodating the lattice mismatch between Si1 xGex layers [ 89, 90] on (100) silicon substrate via Molecular Beam Epitaxy (MBE) technique. T he growth condition s w ere shown in ref erences [91, 9 2 ]. A 630 nm thick layer of the fully re laxed Si0.7Ge0.3 was grown on top of the graded layer, followed by a 50 nm strained Si capping layer. After growth, some of the strained samples were annealed at 800 for 30 min in a furnace under N2 atmosphere to study how temperature affects the microstructure of the strained layer. A PANanalytical MRD XPert system equipped with a mirror and a Ge (220) monochromator on the primary optics and a Ge (220) analyzer on the secondary optics was used to collect high resolution X ray diffraction rocking curves (RCs) and reciprocal space maps (RSM). For X ray reflectivity (XRR) spectra, X ray mirror and parallel plate collimator were used as the primary and secondary optics, respectively. The scan conditions for all the samples were 0.002 step size -RCs and 0.005 and 3 seconds for XRR. In order to confirm the strain relaxation, the presence of misfit dislocations and investigate the surface morphology of the strained silicon layer, transmission electron microsc opy (TEM) and atomic force microscopy (AFM) investigations were carried out with the aid of a JEOL TEM 200CX and a Digital Instruments Nanoscope AFM set up under tapping mode, respectively. Results and D iscussion Strain r elaxation d uring thermal a nnealing : Fig. 5 1 shows a typical high resolution reciprocal space map (RSM) recorded around the asymmetric (113) reflection from an annealed

PAGE 77

77 strained Si/Si0.7Ge0.3/Si1xGex/Si -substrate sample. This diffraction pattern is displayed as a function of scattering vec tor XQ parallel and YQ perpendicular to the surface. From this RSM, several information regarding the sample structure can be extracted. First of all, it is found that the Si0.7Ge0.3 buffer layer is fully relaxed. If a buffer layer is relaxed and the thickness of strained silicon is less than the critical thickness [93, 94], there is a small driving force for the movement of the dislocations from the interface between strained layer and relaxed buffer layer to the su rface layer which indicates that misfit dislocations are not present As suggested by Loo et al. [95], although the thickness of the strained silicon layer is below the critical thickness, thermal treatment can result in the formation of misfit dislocation originating at the heteroepitaxial interface between strained and buffer layers if the latter is not fully relaxed. The presence of the graded layer (Si1xGex, with x from 0 to 0.3) between the Si -substrate and Si0.7Ge0.3 in which Ge was incorporated at a rate of 10 at.% per micrometer up to a composition of 30 at.%Ge is clearly seen in Fig. 5 1. This continuous graded layer was employed for an effective reduction of threading dislocations density. Since the misfit dislocations in the graded layer are dist ributed homogeneously, the movement of the rather mobile threading dis locations will be blocked [ 96]. Finally, the topmost strained silicon peak appears to the upper left region of the silicon substrate peak. Since the peak width of the strained layer is very broad compared to Si -substrate or fully relaxed Si0.7Ge0.3, it can be expected that this broadening is due to defects, such as dislocations, and finite crystallite size. Also, the measured parallel and perpendicular mismatches between strained silicon and relaxed buffer layers are 310 01 1 % and 25 2 %, respectively. The horizontal shift of the strained silicon with respect to the relaxed buffer layer is very small, suggesting that there are two possibilities for the shif t of th is magnitude: (I) small strain relaxation due to post -growth annealing process and (II) presence of surface steps due to a small

PAGE 78

78 wafer misorientation [97]. Therefore it is difficult to make an assertion whether the observed peak shift of the strain ed silicon layer with respect to the SiGe buffer layer is due to strain relaxation. Fig. 5 2 displays the x ray reflectivity (XRR) curves acquired from as -grown and annealed strained samples. Since x ray reflectivity curves were obtained using the symmetri c /2 configuration with very small angles, this technique is suitable for the investigation and characterization of thin film s c), which is affected by an experimental deviation of around % 3 in our study. The non -spec ular diffuse scattering as well as the specular scattering was considered in order to obtain the surface and interfacial roughness values. Fig. 5 2 shows different reflectivity curves for the as -grown and annealed s trained Si deposited on a SiGe buffer layer, resulting from the changes of the strained layers roughness and interfacial layers density. Using the Wingixa software from PANanalytical to simulate the XRR curves one can get layer roughness, density, and thickness values, which are listed in Table 5 1 In the simulation of XRR curves a multi layer model, which consists of the Si1xGex buffer, interfacial, strained silicon and surface layers, was used to obtain the best simulation. Estimated Ge composition of relaxed SiGe buffer layer was about 31~32% [ 59]. It is found in Table 5 1 that the density of the interfacial layer and roughness of the strained layer in the annealed sample are higher than those in the as grown sample. It appears that Ge atoms in the relaxed Si0.7Ge0.3 layer slightly diffused into the strained silicon layer and strain was relaxed during the p ost -growth annealing process [ 98]. T he Ge interdiffusion near the interfacial layer between strained -Si and fully relaxed SiGe buffer layer was a lso confirmed by using Raman spectroscopy, as shown in Fig. 5 3. The curve shown in Fig. 5 -3 (a) was fitted using a Lorentzian function in order to determine the peak position and intensity of the Raman

PAGE 79

79 scattered light arising from the Si -Si vibrational modes in the SiGe buffer layer, strained Si, and Si substrate, respectively. Fig. 5 3 (b) shows Si -Si vibrational modes in the strained Si of as grown and annealed samples, in which the intensity of the strainedSi peak for the annealed sample is smaller tha n that recorded from the as grown sample probably due to the thinning of the strained Si layer. Therefore, this decrease of strained -Si peak intensity of annealed sample in Raman spectroscopy is attributed to interdiffusion at the Si/SiGe interface. Calcu lation of d islocation d e nsity from FWHM of o mega r ocking c urve s : t he RC) acquired from a sample results from several factors: (I) finite crystallite size, (II) film curvature, (III) strain of the specimen, and (IV) i nstrumental limit [ 82]. Of these effects, the size and strain effects are the two main causes of line broadening. The former depends on the size of the coherent domain which is limited by planar defects, such as stacking faults, twin faults and subgrain boundaries while the latter is caused by dislocations and point defects. Analyzing these line broadening effects, one can estimate the dislocation density and coherence length. In the line profile analysis (LPA) of x ray diffraction peaks the primarily us RCs are Gaussian, Lorentzian, Voigt and Pseudo -Voigt [ 55, 56, 99, 100]. In this study, the Hordon -Averbach method was used for the -RCs to estimate the dislocation density [ 101]. In conventional HordonAverbach meth -RC has a Gaussian shape. Therefore, a full width at half maximum (FWHM, m ) of the measured rocking curve can be expressed as follows [ 82], 2 2 2 2 2 2 0 2)] ( [ )] ( [ )] ( [ )] ( [ )] ( [ )] ( [ )] ( [ hkl hkl hkl hkl hkl hkl hklInst r L m (5 1) where, 0 represents X ray extinction broadening, L a finite crystallite size broadening, r a specimen curvature broadening, an angular broadening nearby the dislocations, a strain broadening, and Inst an instrumental broadening. Since the FWHM of the RC of

PAGE 80

80 commercially available silicon single crystal with a dislocation density of 3000/cm2 is just 8 arcsec, the x ray extinction, curvature, and instrum ental broadening effects could be safely ignored. The broadening of the RC due to the dislocations and finite crystallite size is given by [102], 2 2 2 2 2 2) 2 sin ( tan ) ( ) ( ) ( ) (B LB L mK K K (5 2) where, K K and LK are constants, B the Bragg angle, and the wavelength. If the -RCs are measured at different Bragg angles and Eq. ( 5 2) is fitted by the least square method, K, K and LK can be estimated. If the dislocations are randomly dis tributed, the dislocation density can be given as [ 86], ) 2 ln( 22b K Ndis (5 3 ) where, b is the Burgers vector and disN is the dislocation density. However, a pure Gaussian function used for fitting the rocking curves of the strained silicon was not suitable because of the line broadening in the tail region of the diffraction profile. In this study, the X -ray rocking curves were fitted by a Pseudo-Voigt function, which is described by the following equation [ 56], )] ( ) 1 ( ) ( [ ) (0x G x L I x PV (5 4) where, ) ( x L and ) ( x G are Lorentzian and Gaussian functions, and ( 1 0 ) represents the fraction of the Lorentzian component in Pseudo -Voigt function. Now, since th e -RCs measured from the strained silicon layer is not a pure Gaussian function, but a Pseudo -Voigt function, Eq. ( 5 2) needs to be modified as following, n L n n n mhkl hkl hkl hkl )] ( [ )] ( [ )] ( [ )] ( [ (5 5)

PAGE 81

81 wh ere, n is a constant that allows for the best fit. If Eq. ( 5 5) is also fitted by the least square method and then K is estimated, the dislocation density of the strained silicon layer can be calculated by Eq. ( 5 3 ). Fig. 5 4 shows a (113) omega rocking curve acquired from annealed strainedSi, which is fitted with Pseudo-Voigt and Gaussian functions. It is clear that a pure Gaussian function does not fit well in the tail region of the diffraction profile because of the absence of a Lo rentzian component. Therefore, a Pseudo -Voigt function is more suitable to fit the RCs, as shown previously [ 100 -RCs of (113), (004), and (224) planes were analyzed for the estimation of the dislocation density by using Pseudo-Voigt func tion fitting and the modified Hordon-Averbach method. Notably, the (113) and (224) planes were measured because they require the case of grazing incidence angles, around 2.8 and 8.8, respectively. Such a small incidence angle results in a large surface i nteraction volume, therefore being more sensitive to the surface layer. The RCs of the strained layers were fitted with both Voigt and Pseudo -Voigt functions in addition to Gaussian function, in order to find out the best diffraction line profile shape, a s shown in Table 5 2 It was found that while the RCs of as -grown strained silicon have almost a pure Gaussian shape with very small Lorentzian fractions, those of annealed strained silicon have Lorentzian as well as Gaussian shapes due to a higher Lorentz ian fraction. It means that the annealing has a strong effect on the line broadening in the tail region of the RCs. In order to estimate the dislocation density, the FWHM values of several RCs were used in the modified Hordon-Averbach method. By optimizing several parameters shown in Eq. ( 5 5) obtained from the least square fitting, the dislocation density of the strained silicon was calculated and is displayed in Table 5 2 It was found that the dislocation density of the annealed strained silicon is large r than that of the as grown one, but the fitting parameter (n) of the former is smaller than

PAGE 82

82 that of the latter. In the as -grown sample, n wa s 1.92 which is quite close to 2 the value used by the conventional Hordon-Averbach method, as shown in Eq. ( 5 2). In the annealed strained silicon, however, n wa s found to be 1.30, which is considerably smaller than 2 because the shape of RCs is no longer pure Gaussian. This result raised the question on what is the physical meaning of the tail part of the x -ray ome ga -RC is related to the finite coherence length which can be defined as an average crystallite size which scatters coherently and therefore is free of defects [61]. The coherence length can be statistically estimated by using Fourier analysis [ 52]. First of all, the Fourier coefficients of the diffraction line profile can be determined from the Gaussian and Lorentzian component of the Voigt function and then separated into size and mean squared strain coefficients which are expressed by the following equations [ 103], ) 2 exp( ) (2 2 G LW L LW L A (5 6 ) ) 2 exp( ) ( ) (2 2 2 L Q L A L AS (5 7 ) where, ) ( L A and ) ( L AS are the Fourier and size coefficients, L W and G W are the FWHM of Lorentzian and Gaussian functions, L Q and are the coherence length, reciprocal lattice spacing and strain, respectively. Finally, the coherence length distribution (P(L)) can be obtained from the second derivative of the size coefficient [ 104]. Fig 5 5 shows the coherence length distribution of annealed and as grown strained silicon. It is found that the mean coherence length of annealed strained silicon is smaller than that of as -grown one due to the generation of misfit defects. Fig. 5 6 shows th e surface morphologies and plan view images of the as grown and annealed strained silicon obtained from AFM and TEM investigations The large -scale cross hatching characteristic [105], which is the presence of misfit defects are clearly shown in AFM and T EM images, respectively. We clearly see that the annealed strained silicon has a more

PAGE 83

83 cross -hatching patterns than the as -grown caused by the strain relaxation, consistent with the results obtained from X ray techniques. Conclusion In summary, annealed a nd as -grown strained silicon layers were characterized by using high resolution X ray diffraction. Although the parallel mismatch between the annealed strain silicon and SiGe buffer layer measured from RSM was very small, the evidence obtained from XRR and -RC measurement clearly showed the occurrence of t he strain relaxation during annealing process at 800 for 30 min. T he dislocation density of the strained silicon was -RCs by using the modified Hordon -Averbach method. Conven tional Hordon-Averbach method in which a rocking curve is fitted by only a Gaussian function wa s not appropriate for this experiment due to a significant peak broadening in the tail region. The tail of the rocking curve wa s fitted well with Voigt or Pseu do -Voigt functions that have Lorentzian as well as Gaussian components. It was found that the dislocation density of annealed strained silicon is larger than that of the as -grown one as a result of strain relaxation. In addition, it was found that the cohe rence length of the strained silicon decreased after high temperature annealing, which induced a line broadening in the tail RCs. The results -RC are more suitable techniques than RSM to a ccess the strain relaxation when its value is rather small. Strained SiGe G rown on Si Experimental D esign Si1xGex layers with a nominal Ge fraction x=15, 20, and 25 % and thickness of 500 and 1000 were grown on (100) Si by reduced pressure chemical vapor deposition (RPCVD) at a pressu re of 10 Torr and growth temperature of 700 using dichlorosilane (DCS, SiH2Cl2) and

PAGE 84

84 germane (GeH4) as silicon and germanium sources, respectively. This thickness is significantly larger than the critical thickness for SiGe layers with those Ge concentrations [ 106]. After growth, the layers were an nealed in a quartz tube furnace at 800 and 900 oC for 30 min under a N2 ambient to study the effect of thermal processing on the strain state of the strained SiGe layer. A PANalytical MRD XPert system equipped with a 1/2o slit, a mirror and a Ge(220) mono chromator on the primary optics and a channel cut Ge (220) analyzer on the second ary optics was employed to collect high resolution omega reciprocal space maps (RSM). For x -ray reflectivity (XRR) spectra acquisition a 1/32o slit and a mirror were used on the primary optics while a parallel plate co llimator and a 0.1 mm slit were used on the secondary optics. To confirm the degree of strain relaxation and Ge composition, and investigate the created defects, transmission electron microscopy (TEM) investigations were performed with the aid of a JEOL 20 0CX. The thickness and composition of the 25% Ge sample was also investigated using Rutherford backscattering spectroscopy (RBS). The effect of the thermal treatment on surface morphology was also investigated by atomic force microscopy (AFM) with a Digita l Instrument Nanoscope under tapping mode. Results and D iscussion Determination of Ge c omposition, l ayer t hickness, and s train : XRR curves acquired from strained Si1xGex small angles so that this technique is suitable for the investigation and characterization of thin films. The film density was determined from the position in the XRR curves of the critical angle C), which is proportional to the electron and mass density [ 107, 108 ]; the experimental error was estimated to be around 4 % in our study. The diffuse as well as the specular scattering were considered in order to obtain the surface and interface roughness values. The simulation of the XRR curves was performed by using the comme rcially available Wingixa software package

PAGE 85

85 from PANanlytical. For the simulation we used a four -layers model: substrate (Si), interfacial layer (transition layer), Si1xGex layer, and surface contamination layer. The information of the interfacial and str ained layers obtained from the simulations for as -grown and annealed samples is presented in Table 5 3 and Table 5 4 respectively. It is clearly shown that both density and thickness of the interfacial layer increase with the increase of Ge fraction. As t he interface could not be infinitely sharp, especially when the growing temperature was 700 the higher the change in Ge concentration, the thicker this transition layer is, as expected. The t hickness of the grown Si1 xGex layers are different from the nominal values, more so for higher Ge fractions. The rather large thickness deviation observe d for the sample containing 25% Ge concentration might be caused by an error in the measurement due to surface roughness as a cross section TEM found a (local) thickness value of 510 much closer to the 500 nominal value. Since the sample with x=25% Ge had such a large variation of the thickness with respect to the nominal value, we p erformed a Rutherford backscattering spectroscopy measurement. The result, shown in Fig. 5 7 was modeled using the RUMP program and indicated a composition of Si0.76Ge0.24 and a thickness of 496 very close to the nominal value. After thermal anneal, it was found that the thickness of the SiGe layers decreased due to unwanted oxidation. However, the increase of the interfacial layer thickness and roughness coupled with th e decrease of the Si1xGex layer density points toward a significant Ge interdiffusion [ 109]. A reciprocal space map (RSM) acquired for the (113) peak region from as -grown Si0.85Ge0.15 sample is displayed in Fig. 5 8 [58, 110]. This diffraction pattern i s d isplayed as a function of scattering vector QX parallel and QY perpendicular to the surface. As one can see, there is no measurable relaxation, the SiGe peak being located exactly below the Si peak, at the same QX

PAGE 86

86 performed with zero relaxation, as found by RSM results, are shown in Fig. 5 9 The SiGe layer the values extracted from XRR simulations are presented in Table 5 5 One can note that there is a rather good agreement between these values measured by both xray based techniques. A second observation is that for the lowest Ge fraction used in this stud y, the measured values are closer to the nominal ones, while for the highest value the difference is the largest. The AFM results regarding the samples surface morphology are presented in Fig. 5 10. One can note that for lower Ge concentration the surface is quite smooth, its rms (root mean square) values being around 3 However, the sample contained 25% Ge exhibited a much rougher surface, with a RMS value of 9 The surface also showed an undulation, which might explain the over evaluation of the layer previously reported for SiGe and are indications of the relaxation process [ 111]. from as -grown sampl es, as shown in Fig. 5 11 from annealed Si1xGex layers are different as well as SiGe layer peaks are broader compared to as -grown samples, resulting in the strain relaxation. Especially, the peak shift of annealed Si1 xGex toward Si The misfit dislocations generated during strain relaxation in strained Si1xGex are a perfect 60 type, } 111 { 110 2 / 1 Burgers vector and slip plane, respectively [ 106, 1 12]. These dislocations are affecting the x -ray diffraction from asymmetric planes but not from symmetric ones, as one can see in Fig. 5 11. In addition, the amplitude of the oscillations and integrated intensity of Si1xGex peak shown in the annealed samples, especially for higher Ge compositions, were greatly dampened and reduced, an indication of strong Ge interdiffusion and degradation of the

PAGE 87

87 sharpness of the interfaces. Si and Ge atoms diffuse near the interface and this inte rdiffusion affects the structural factor of x ray diffracted beam. Zheng et al. showed that the interdiffusivity could be calculated from the decay of integrated intensity of Si1xGex layers with annealing time [113]. Fig. 5 12 shows plan -view TEM images o f as -grown and annealed Si1xGex layers. After thermal annealing, the misfit dislocations increased in density indicating that strain relaxation has taken place which agrees with the observation from the x ray based techniques. For the highest germanium c oncentration, 25% Ge, misfit dislocations are already present in the as grown state, the layer thickness being much larger than the critical thickness [ 106]. The presence of these defects in the 25% Ge sample in the as -grown state could be another cause of the thickness measurement discrepancy between TEM and x ray based techniques. It also appears that the simulation of the RCs could be carried out more accurately if the strain relaxation is known from a measurement using a different analysis technique. De fects s tructure in s trained SiGe : Fig. 5 13 (a) shows (004) omega 2theta rocking -grown and annealed at 800 for 30 min Si80Ge20 films 50 nm thick. The RC of as -grown sample shows the well -defined thickness fringes which indicate the formation of an abrupt interface between Si substrate and strained Si80Ge20 layer. From the 1xGex layer were found to be 46.8 nm and 21 %, respectively. However, the RC of Si80Ge20 layer annealed at 800 for 30 min is different from that of as -grown sample. The intensity of annealed Si1xGex layer was reduced and the thickness oscillations were dampened, compared to as -grown sample. In addition, the Bragg peak positi on of strained layer slightly shifts towards Si substrate position. These changes indicate a strain relaxation accompanied by misfit defects generation and Ge interdiffusion near the interface [ 113]. Fig. 5 13 (b) shows plan view

PAGE 88

88 transmission electron micr oscopy (PV TEM) images of as -grown and annealed Si80Ge20 samples. The a s -grown sample does not show any misfit dislocations, indicating that the strained layer was pseudomorphically grown and the interface was abrupt, which is consistent with the results o btained from investigations However, the annealed sample shows many misfit dislocations with a cross -hatching pattern which consist of perfect 60 dislocations, 111 110 2 1 Burgers vector and slip plane [ 112, 114]. An omega rocking cur RC) investigation was further performed to study the effect of these misfit dislocations on the peak broadening. The full width at half maximum (FWHM) value of the RC has been used as a figure of merit (FOM) for the crystalline quality of the layer s Very small FWHM values, similar to the instrument resolution limit of 0.0021 measured RCs that were acquired from the as -grown Si80Ge20 layer and are displayed in Fig. 5 14 (a) and (b) indicate that -RCs acquired from the annealed sample and shown in Fig. 5 14 (c) and (d) were also similar to those of as -grown sample. However, it needs to be noted that the s -RC of annealed sample is RC is still RC is not sensitive to the presence of 60 misfit dislocations within the annealed strained layer as shown by TEM results, th -RC is. For further study on the defects evolution during strain relaxation, 100 nm thick strained Si75Ge25 layers with higher strain energy areal density (ES) than 50 nm thick strained Si80Ge20 samples were grown and then anneale d at 900 for 30 min. This strain energy areal density can be given as Bh ES 2 where is the in plane strain, B is the Burgers vector and h is the strained layer thickness [ 115]. The (113) reciprocal space map acqui red from the as -grown 100

PAGE 89

89 nm thick Si75Ge25 sample in Fig. 5 15 indicated a slight strain relaxation, less than 1.0 %, while strain relaxation of annealed sample was 19.1 %. The defect structure of as grown and annealed strained Si75Ge25 samples has been a lso studied by using RC as shown in Fig 5 1 6 It is found -RCs of as -grown Si75Ge25 are very similar to those recorded from annealed Si80Ge20 layer, as shown in Fig. 5 14 (c) and (d) because the layer thickness of strained Si75Ge25 is much thicker than the critica l thickness for the pseudomorphic growth indicating the generation of 60 -RCs of annealed Si75Ge25 sample show a different behavior compared to slightly strain relaxed sample with 60 misfit dislocations. First of all, it is RCs of 100 nm thick Si75Ge25 annealed at 900 for 30 min are much larger than those of slightly strain-relaxed samples. The -RC was not sensitive t o 60 misfit dislocations. It is suggested that the misfit defects in annealed strained Si75Ge25 layer with higher strain energy areal density should be different from 60 misfit dislocations. Second, the shape of both (004) RCs of annealed sam ple is almost symmetric. Therefore, it is inferred that the defect structure of highly strain relaxed SiGe sample is different from that of slightly strain relaxed sample. Fig. 5 1 7 shows cross -sectional transmission electron microscopy (TEM) images of 100 nm thick Si75Ge25 annealed at 900 for 30 min. The strained-SiGe/Si interface is clearly visible and the misfit dislocations lying at the interface and stacking faults in the strained SiGe layer are also shown in Fig. 5 1 7 (a). Fig. 5 1 7 (b) shows a highresolution TEM lattice image of th e annealed sample. Typical 60 dislocations present in FCC crystal structure have Burgers vectors 110 2 / 160 b gliding on the {111} closed packed plane s It was experimentally found and theoretically predicted that 60 dislocations w ill dissocia te into two partial dislocations in order

PAGE 90

90 to lower the strain energy [ 116, 117 ]. The Burgers vectors of these partial dislocations are 1 1 2 6 / 1b and 2 11 6 / 1b between the dislocation line and its Burgers vector. In addition, it was reported that the dissociation of 60 dislocations leaded to a different geometry in strain relaxation for the tensile and compres sive strained films [ 118]. Fig. 5 1 7 (b) shows stacking faults on {111} planes near RCs of 100 nm thick strained Si75Ge25 annealed at 900 for 30 min, as shown in Fig 5 1 6 (c) and (d), is due to these partial dislocations associated with the stacking faults. Conclusion SiGe layers that were grown by RPCVD underwent an anneal treatment at 800 oC for 30 min. The results obtained from XRR and RCs investigations showe d that the most important modification after the thermal treatment was the Ge interdiffusion and misfit dislocation s generation, conducting to wider and rougher interfaces. The strain relaxation was confirmed by presence of misfit dislocations in strained layer s At the onset of relaxation the surface morphology exhibited a waviness followed by the appearance of a cross hatched pattern. These results ind icate that x -ray based techniques can be used to accurately determine Ge content, thickness, and roughness of SiGe layers when the strain state of the material is known as seen with the 15 and 20% Ge samples. Once relaxation has taken place, it is imperative to know the degree of relaxation in order to correctly simulate the structure of the film. The result showed that the 60 dislocations generated from slightly strainrelaxed SiGe layers had an influenced on the profile shape of the RC while they did not contribute to the peak broadening of the RC. On the other hand, stacking faults bounded by two

PAGE 91

91 RCs symmetrically broad. Since the presence of the stacking faults in compressively stressed films is unusual this observation of the stacking faults RC contributes to the understanding of defects evolution and process design for high quality strained layer s Ion I mplantation of S trained Layer s Experimental D esign Si1xGex layers with a nominal Ge fraction x=0.25 and a thickness of 50 nm were grown on (100) Si -substrate by reduced pressure chemical vapor deposition (RPCVD) at a pressure of 10 Torr and growth temperature of 700 using dicholorosilane (DCS, SiH2Cl2) and germ ane (GeH4) as silicon and germanium sources, respectively. The Ge composition and layer thickness of Si1xGex measured from Rutherford Backscattering Spectroscopy (RBS, as shown in Fig. 5 7 ) were 24 % and 49.6 nm, respectively. After growth, Si+ ion implantation with an energy of 12 keV at a fluence of 1510 1 atoms/cm2 was carried out, followed by thermal annealing in N2 ambient using a quartz tube furnace. Fig. 5 1 8 shows bright field transmission electron image of implanted SiGe layer ann ealed at 500 for 30 min. The initial amorphous layer thickness after the ion implantation was estimated to be around 28 nm T he amorphous layer began to crystallize during thermal annealing, a process which is called solid phase epitaxy re -growth (SPER) [119]. One can note that the amorphous -crystalline (a -c) interface was not planar but rough, as shown in Fig. 5 1 8 For comparison reason, the as -grown SiGe layers without ion implantation were also annealed under the same conditions, as those employed for the implanted samples. A PANalytical MRD XPert system equipped with 1/2 slit, mirror and Ge (220) monochromator on the primary optics and a channel cut Ge (220) analyzer on the secondary optics was employed to collect high -resolution omega rocking curves RCs) and reciprocal space maps (RSMs). For

PAGE 92

92 the investigation of the surface morphology, a Digital Instruments Nanoscope atomic force microscopy (AFM) instrument was employed under tapping mode. A JEOL TEM 200CX was used for the cross sectional transmis sion electron microscopy (TEM) images of SiGe layers and the sample preparation was carried out by using the focus ion beam (FIB) technique. Results and D iscussion First of all, the strain relaxation of annealed SiGe layers without ion implantation was ana lyzed to evaluate various possible sources for the dislocations nucleation. Fig. 5 1 9 shows (113) reciprocal space maps (RSMs) of as -grown and annealed at 800 for 30 min Si76Ge24 layers on Si substrate. The reciprocal lattice points (RLPs) are displayed as a function of scattering vector QX parallel and QY perpendicular to the surface. It is clearly seen from Fig. 5 1 9 (a) that the QX values of SiGe layer and Si substrate are exactly the same and the shape of SiGe RLP is very sharp and symmetric, indicating that the as grown layer was pseudomorphically (relaxation=0 %) grown and the interface between the strained layer and substrate was abrupt. During post -grow th thermal treatment, the SiGe layer begins to relax due to misfit dislocations generation caused by the thermal budget. In Fig. 5 1 9 (b), one can note that the SiGe RLP slightly shifts along the relaxation line and seems to be asymmetric. In previous pape r, it was reported that the asymmetric RLP of the strained layer was due to the presence of 60 misfit dislocations at the interface [ 120]. It was also found from RSM results that the strain relaxation of the SiGe layer annealed at 800 for 30 min was les s than 2 %. This insignificant strain relaxation even after such a high temperature anneal indicates that there exists a kinetic barrier for the generation of the misfit dislocations at the interface because of few sources for the dislocations nucleation. These sources include pre -existing threading dislocations from the Si substrate and growthrelated defects induced during the growth of the strained epilayer. The

PAGE 93

93 threading dislocation density of the commercially available Si substrate is less than 104/cm3, which is not enough for an efficient strain relaxation. In addition, one of the growth -related defects in strained SiGe is the surface undulation, which is the formation of a wavy surface characteristic, as shown in Fig. 5 20. It was theoretically predic ted and experimentally reported that the surface of strained layers with relatively high Ge concentration becomes undulated to reduce the system energy because the film stress of a wavy surface is smaller than that of the planar surface [ 121]. In addition, Ge composition near the cusps was larger than that near the troughs, resulting in the lateral modulation of the lattice constant in the SiGe surface [ 122]. These facts indicate that the cusps can be a preferential site for the dislocations nucleation due to high stress. Fig. 5 2 1 shows average peakto -peak distance and r.m.s. roughness of annealed Si76Ge24 layers without/with Si+ implantation. As expected, it is shown that the average peak to peak distance in t hese samples decreases with increasing anneal ing temperature. Note that a change in the peak to -peak distance with a degree of the thermal budget is almost the same in the Si76Ge24 films without/with ion implantation, as shown in Fig. 5 2 1 It is also found that while the r.m.s. roughness of the samp les without ion implantation increases with the annealing temperature, that of the samples with ionimplantation is almost independent of the temperature even though annealed layers are smoother than as implanted layer. This difference of the r.m.s. roughness values between the two sets of samples is due to different surface morphology. In Fig. 5 20 (b), the cross hatch pattern, which is a general characteristic of the strain -relaxed films [ 123], is clearly seen in the annealed SiGe layers without ion-impla ntation. However, the AFM image of the annealed layer with ion implantation does not show any cross -hatch pattern, as one can see in Fig. 5 20 (c).

PAGE 94

94 So far, two possible sources of dislocations nucleation for strain relaxation have been presented : pre -exist ed threading dislocations from the substrate and surface undulation of the strained layer. However, these sources did not play an efficient role to obtain highly strain relaxed SiGe layers. Another source for the dislocations nucleation in these samples co uld be the point defects induced by ion implantation. Fig. 5 22 shows (113) RSMs of ion-implanted Si76Ge24 annealed at three different temperatures. The (113) RSM of ion -implanted Si76Ge24 annealed at 700 shows that two different layers, a fully -strained layer (defect -free layer) and a relaxed layer (defective layer), co -exist. After ion implantation, a top SiGe layer of around 28 nm thick was amorphized and there was an amorphous -crystalline interface (a -c interface), as shown in Fig. 5 1 8 When the top amorphous layer began to crystallize during the SPER process, we believe that most of defects would be nucleated. Especially, dislocation loops can be rather easily nucleated due to a high concentration of point defects. These generated defects propagated into the interface region and the relaxation process was efficiently enhanced, as compared to the samples without ion -implantation. The thickness of the defect -free layer in ion -implanted samples decreased with increasing annealing temperature and finally disappeared after the 900 anneal, as shown in Fig 5 2 2 This fact indicates that the propagation of induced defects to the interface is a thermally activated process. Fig. 5 2 3 shows the degree of strain relaxation and full -RCs) recorded from ion implanted Si76Ge24 layers. It is evident that the strain relaxation of ion implanted Si76Ge24 layers increases with increasing annealing temper ature. The strain relaxation is dramatically increased over 700 which again demonstrates that there exists an energy barrier for the RCs after different thermal treatments needs to be noted. For low annealing temperatures, at which the

PAGE 95

95 layer is a stack of a defect -RC increase with increasing temperature. However, FWHM values begin to decrease with increasing temperature over 700 simultaneously with a strain relaxation. Fig. 5 2 4 shows cross sectional transmission electron diffraction (X TEM) bright field images of ion -implanted Si76Ge24 layer after thermal annealing. One can see that almost all defects began to nucleate from the a -c interface during SPER process. The dislocation loops were easily nucleated due to a high concentration of point defects induced by the ion implantation. It is clearly shown in Fig. 5 2 4 (a) that there exists a defect -free region between the initial a -c interface and SiGe/Si interface in the implanted layer annealed at 700 even though there were few defects observed at the SiGe/Si interface. However, the implanted Si76Ge24 layer annealed at 800 was much more defective as shown in Fig. 5 2 4 (b), where the dislocation loops as well as the misfit dislocations are clearly visible at the SiGe/Si interface. These results are consistent with the previous (113) RSM results of ionimplanted Si76Ge24 layer after thermal treatment, shown in Fig. 5 2 2 Evaluating the possible sources for the nucleation of the misfit segments at the SiGe/Si interface, it has been concluded that the pre -existing threading dislocations and surface undulation are not enough to nucleate large concentration of misfit dislocations. In pr evious work, where 100 nm thick Si76Ge24 layers were annealed at 900 for 30 min without ionimplantation, we observed relatively high strain relaxation, which was caused by the stacking faults associated with partial dislocations dissociated from previou sly generated misfit dislocations [ 120]. On the other hand, based on the results from this study, the relaxation mechanism of the ion implanted strained SiGe layers after thermal annealing is different: the glissile dislocation loops can be easily nucleate d due to a high concentration of point defects

PAGE 96

96 during solid phase epitaxy re -growth process and they propagate toward the SiGe/Si interface, where they can efficiently lower the kinetic barrier energy for the nucleation of the misfit dislocations. Also, th e threading dislocations accompanied with the misfit segments can be easily annihilated after subsequent thermal annealing, as shown in Fig. 5 2 3 Conclusion In summary, the defects behavior and relaxation mechanism in strained Si76Ge24 with and witho ut ionimplantation after thermal annealing were analyzed by considering possible sources for the dislocations nucleation. The strain of SiGe layer without ionimplantation was only slightly relaxed after even high temperature annealing because the pre -exi sting threading dislocations in the substrate and surface undulation could not act as efficient nucleation sources for the misfit dislocations at SiGe/Si interface. On the other hand, the results showed that the high concentration of point defects in impla nted SiGe films helped generating defects during SPER process. These defects propagated to the SiGe/Si interface, where they induce d the formation of misfit dislocations, resulting in the enhancement of the strain relaxation. It was also found that the pro pagation of these dislocations induced during SPER is a thermally activated process. Finally, the dislocations can be easily annihilated after subsequent thermal treatment even though the strain was efficiently relaxed. These results clearly indicate that ion implantation is a very promising technique for the fabrication of high crystalline relaxed SiGe films. Oxidation of S trained SiGe L ayers Experiment D esign Si1xGex layers with a nominal Ge fraction x=15, 20 and 25 %, and a thickness of 50 nm were grown on (100) Si -substrate s by reduced pressure chemical vapor deposition (RPCVD) at a pressure of 10 Torr and growth temperature of 700 using dicholorosilane (DCS, SiH2Cl2) and

PAGE 97

97 germane (GeH4) as silicon and germanium sources, respectively. The Ge compositio n and layer thickness of Si1 xGex measured from Rutherford Backscattering Spectroscopy (RBS, shown previously) were very similar to the nominal thickness and composition values As -grown Si1xGex layers were oxidized in a furnace at 800 900 and 1000 for 1 hr under a flow rate of 0.5 L/min of high purity oxygen gas. A PANalytical MRD XPert system equipped with a 1/2 slit, mirror and Ge (220) monochromator on the primary optics and a channel cut Ge (220) analyzer on the secondary optics was employe d to collect high resolution omega reciprocal space maps (RSMs). A JEOL TEM 2010F was used for the cross -sectional scanning transmission electron microscopy (STEM) images and energy dispersive spectroscopy (EDS) for Ge depth profile The sample preparation was carried out by using the focus ion beam (FIB) technique. Results and D iscussion Fig. 5 2 5 shows (113) omega Si75Ge25 layers after thermal oxidation at 800 900 and 1000 for 1 h ou r. The as grown film shows intensity oscillation fringes due to the multiple scattering of diffracted x ray beam along sample normal direction. Th ese oscillation s observed from as -grown layers indicate that the interface between the substra te and film is extremely abrupt, so that the as grown layer was pseudomophically grown on Si substrate. After the thermal oxidation, the shape of the significantly changes. It needs to be noted that the change s 1xGex (x=15, 2 0, and 25 %) layers with different oxidation temperatures are very similar, even though those of Si75Ge25 layer s are only shown in Fig. 5 2 5 First of all, from the shape of the (113) strained Si75Ge25 layer oxidized in 800 it can be inferred that two different SiGe layers co -

PAGE 98

98 exist: a Ge rich layer (GRL) and a Ge deficient layer (GDL), as indicated in Fig. 5 2 5 Since SiO2, rather than GeO2, was preferentially formed during the oxidation due to the large difference of the heat formation of SiO2 and GeO2 [124126], Ge atoms were rejected from the SiO2, resulting in a Ge pile up below the oxide. At the same time, Ge diffused into the unoxidized pre -existing SiGe layer, inducing strain relaxation occurred during the oxidation. The strain can be ra ther easily relaxed by thermal oxidation, compared to N2 thermal annealing because a large concentration of point defects is generated during the oxidation [127], which lower s the nucleation energy for the generation of misfit dislocations as shown previo usly by us As a matter of fact, the Ge diffusion into the Si substrate at 800 is insignificant. It was reported that the activation energy for Ge diffusion in Si1xGex (x=2 0 ~3 0 % ) alloy was around 3.8~4.3 eV [ 128, 129 ], so that 800 is not enough for Ge atoms to efficiently diffuse into the substrate. However, the existence of GDL indicates that the activation energy for the diffusion of Ge atoms into Si during thermal oxidation is smaller than that under N2 annealing. 1xGex layer after the oxidation at 900 show ed a very broad SiGe layer peak in dicating that Ge atoms diffuse d deeper into the Si Ge compared to the oxidation in 800 This clearly show s that a G RL still exists in the oxidation at 900 as well as 800 Finally, t -Si75Ge25 layer after o xidation at 1000 show ed a different Ge behavior. One can see that the GRL is no longer present a nd SiGe peak located in the right of a red dot line ranges over Si substrate. The Ge atoms rejected from the oxid e are no longer accumulating because they could efficiently diffuse into Si Ge at very high temperature. While the Ge segregation dominantly occurred in the oxidation at 800 the Ge diffusion preferentially occurred at 1000 RCs of strained -SiGe layers

PAGE 99

99 suggest that the competition between Ge accumulation and Ge diffusion into Si Ge is strongly dependent on the oxidation temperature Fig. 5 2 6 shows scanning transmission electron microscopy (STEM) image and Ge concentration meas ured by energy dispersive spectroscopy (EDS) and (113) reciprocal space map (RSM) of strained -Si75Ge25 oxidized in 800 The TEM image c learly shows the presence of two layers distinguished by the mass contrast in which dark and bright region s in the SiG e film are associated with GRL and GDL, respectively While initial Ge composition of as -grown SiGe layer was around 25 %, the Ge composition in GRL and GDL as measured by EDS was 31.65 % and 19.51 %, respectively. A typical high resolution reciprocal sp ace map (RSM) recorded around the asymmetric (113) reflection from SiGe layer after 800 oxidation is shown in Fig. 5 2 6 (b). In Fig. 5 2 6 (b), GRL and GDL are located at bottom and top, respectively, with respect to a red dot that indicates the position of the RLP of the as -grown SiGe layer. The presence of both GRL and GDL in the (113) RSM is consistent with the STEM image. On the other hand, it wa s found from Fig. 5 2 6 (b) that a large degree of strain relaxation of both GRL and GDL occurred after the t hermal oxidation in 800 In a previous study, it was reported that the strain relaxation after N2 annealing in 800 wa s very s mall and only a few perfect misfit dislocations were generated [ 120]. This fact indicates that the thermal oxidation process is more efficient than N2 annealing to obtain highly relaxed SiGe layers. While the Ge pile up in the strained Si1xGex layers occurred during thermal oxidation for both 800 and 900 the SiGe layers oxidized at 1000 did not show any Ge pile up, as in ferred from the shape of the RCs shown Fig. 5 2 5 For further investigation of Ge behavior and strain relaxation of SiGe layers after the thermal oxidation at 1000 the (113) RSMs of strained Si1xGex layers with different Ge compositions were acquired In Fig. 5 2 7 the SiGe

PAGE 100

100 RLP of the oxidized film extends over that of Si one confirming the formation of the Ge graded layer (GGL) for all of Si1xGex samples because the diffusion, rather than the accumulation of Ge atoms into Si substrate is much more d ominant. A w eak intensity and broadening of the SiGe RLP indicate that the layer has a mosaic structure in which the layer is tilt or twist along certain direction, causing the modulation of the lattice parameter. What wa s also noted after the oxidation of strained Si1xGex layers is a significant degree of the strain relaxation. In the RSM of Si85Ge15, QX values of Si and SiGe RLPs are the same indicating that their in plane lattice parameters are still well matched even though the SiGe RLP is very broad. The RLP of SiGe layer shifts to leftupper direction with increasing Ge composition during t hermal oxidation. Finally, the strain of the Si75Ge25 layer was almost completely relaxed after the oxidation in 1000 with a formation of GGL Therefore, this fa ct indicates that the strain relaxation induced by thermal oxidation is strongly dependent of the initial Ge composition in as -grown layer because a large concentration of the point defects lowers the activation energy for the dislocation nucleation. Fig. 5 2 8 shows scanning transmission electron microscopy (STEM) images of Si85Ge15 and Si75Ge25 layers, and Ge depth profile from EDS after thermal oxidation at 1000 In contrast to Fig. 5 2 6 (a), in which the GRL and GDL co -exist, the dark contrast in the r egion of SiGe layer after the thermal oxidation at 1000 was gradually getting bright along the Si substrate, indicating that Ge composition in SiGe layer gradually decreases from the Si1xGex/SiO2 interface to Si substrate, as shown in STEM images and co nfirmed by the EDS results in Fig. 5 2 8 The existence of GGL after the o xidation at 1000 confirmed by S TEM is consistent with the results from (113) RSMs displayed in Fig 5 2 7 Ge atoms diffuse deeper into the Si substrate with increasing Ge composition because the activation energy for Ge diffusion in

PAGE 101

101 SiGe layers decrease s with increasing Ge composition [ 129], as one can see Fig. 5 2 8 In addition, Ge composition of Si85Ge15 and Si75Ge25 layers near the SiO2/Si1 xGex interface after the thermal oxidatio n at 1000 for 1 hr is around 12 % and 18 %, respectively. Therefore, i t is suggested from Fig. 5 2 8 (c) that the thermal oxidation of strained -SiGe layer with a high Ge composition at high temperature be a promising technique to fabricate relaxed and lin ear -graded Si1xGex as virtual substrate s for strained Si applications Conclusions A Ge graded layer (GGL) was fabricated by using thermal oxidation at high temperatures. It was found that th e competition between the Ge accumulation below the oxide interf ace and diffusion into the remaining SiGe layer was strongly dependent of the oxidation temperature. Ge accumulation occurred during oxidation at 800 and 900 but it was no longer found after the oxidation at 1000 for which Ge diffusion preferentially occurred In addition, it was found that the strain was efficiently relaxed during the thermal oxidation and the initial Ge composition in SiGe layer had a great influence on the degree of strain relaxation. The Ge graded layer fabricated by using therm al oxidation of strained Si75Ge25 layer at a high temperature can be used as a virtual substrate for strained Si on fully -relaxed thick SiGe applications.

PAGE 102

102 Table 5 1. S imulation s of the XRR curves acquired from the as -grown and annealed strained silic on samples. A model consisting of r elaxed SiGe buffer, interfacial, strained silicon, and surface layers were used. Layer As grown strained silicon Annealed strained silicon Thickness () Density (g/cm 3 ) Roughness () Thickness () Density (g/cm 3 ) Roughn ess () Buffer layer 3.40 3 3.40 2 Interfacial layer 15 2.69 23 15 2.91 26 Strained layer 530 2.33 7 533 2.31 15 Surface layer 5 2.13 44 6 2.14 47 Table 5 2. L ine profile analysis and calculated dislocation density of as -grown and annealed strai ned silicon. The diffraction profiles are fitted by Voigt and Pseudo -Voigt functions and fitting parameters are estim ated by the least square method Voigt Pseudo Voigt n K Dislocation de nsity (#/cm2) W G 1 W L 2 W Psd 3 4 As grow n strained Si/Si0.7Ge 0.3 (113) 0.151 0.006 0.154 0.05 1.92 2.50106 1.2109 (004) 0.149 ~0 0.150 ~0 (224) 0.155 ~0 0.155 ~0 Annealed strained Si/Si0.7 Ge 0.3 (113) 0.138 0.066 0.176 0.46 1.30 8.60106 4.0109 (004) 0.131 0.078 0. 177 0.52 (224) 0.150 0.053 0.179 0.37 WG 1: FWHM of Gaussian function WL 2: FWHM of Lorentzian function WPsd 3: FWHM of Pseudo-Voigt function 4: Fraction of the Lorentzian component (0 Table 5 3. XRR simulation results for as grown Si1xG ex samples Thickness ( ) Roughness ( ) Density (g/cm 3 ) Ge concentration (%) 15 20 25 15 20 25 15 20 25 Interfacial layer 4 6 13 2 3 9 2.56 2.98 3.02 Strained layer 497 470 579 10 10 10 2.84 3.04 3.24 Nominal value 1 500 500 500 2.84 3.01 3.17 1 Fr om the reference with [ 130]

PAGE 103

103 Table 5 4. XRR simulation results for annealed Si1 xGex samples Thickness ( ) Roughness ( ) Density (g/cm 3 ) Ge concentration (%) 15 20 25 15 20 25 15 20 25 Interfacial layer 11 20 36 2 5 12 2.42 2.85 2.92 Strained layer 487 452 530 3 3 8 2.83 3.03 3.21 Table 5 5. SiGe layer thickness and Ge content estimated from XRR and /2 RCs simul ations and measured by XTEM Samples XRR XTEM Thickness () Ge content (%) Thickness () Ge content (%) Thickness () 15% G e 497 15.0 490 15.0 500 20% Ge 470 21.0 468 22.0 470 25% Ge 579 27.0 580 26.0 510

PAGE 104

104 Relaxation Strained silicon Silicon substrate Relaxed SiGe buffer layer Graded Si1 xGexlayer Si -sub. Graded Si1 xGex Relaxed Si0.7Ge0.3 50nm thick strained -Si Figure 5 1. (113) Reciprocal space map of annealed strained -Si/Si0.7Ge0.3/graded -SiGe/Si substrate 1 2 3 4 10 100 1000 10000 100000 1000000 10000000 Counts2 Theta As grown strained silicon on Si0.7Ge0.3 Annealed strained silicon on Si0.7Ge0.3 Figure 5 2. X -ray reflectivity (XRR) curves record ed from the as -grown and annealed strained Si/Si0.7Ge0.3/Graded SiGe/Si

PAGE 105

105 480 500 520 540 0 10000 20000 30000 40000 50000 60000 Si Si ( SiGe ) peak Si Si (strained Si ) peak Si Si (silicon sub.) peak 460 480 500 520 540 560 0 2000 4000 6000 8000 10000 Si Si (as grown strained Si ) peak Si Si (annealed strained Si ) peak (a) (b) Wave number Wave number Figure 5 3. Raman spectra of strained-Si on SiGe. (a) Raman spectra from the as grown strained Si and (b) Si Si vibrational modes in the strainedSi o f as -grown and annealed sam ples Figure 5 4 Curve fitting of (113) rocking curve of annealed strained silicon (a) Pseudo -Voigt and (b) Gaussian functions were used R2 is a goodness of fit parameter.

PAGE 106

106 Figure 5 5. The coherence length distribution calculated for the as -grow n and annealed strained silicon from the second derivative of the Fourier size coefficient Roughness: 45 Roughness: 48 (a) (c) (b) (d) Figure 5 6. Atomic force microscopy (AFM) and transmission electron microscopy (TEM) images of strained silicon. (a b) and (c d) were obtained from as -grown and annealed strained Si, respectively.

PAGE 107

107 Figure 5 7 RBS spectrum and its fit using RUMP software [ 131] for a sample containing a 25% Ge fraction Figure 5 8. (113) RSM from a Si0.85Ge0.15 layer on (001)Si

PAGE 108

108 Figure 5 RC acquired from a sample with 15 % Ge and its simulation (displaced vertically for better view) Figure 5 10. AFM micrographs of t he as grown SiGe layers surface

PAGE 109

109 Figure 5 11. (004) and (113) /2 RCs acquired from the as grown Si1xGex on Si and after the thermal annea l Figure 5 12. Planview TEM for strained SiGe samples as -grown and annealed at 800 C for 30 min

PAGE 110

110 Figure 5 13. Omega 2theta rocking curves and TEM images. / -view TEM images of as -grown and annealed at 800 for 30 min Si80Ge20 films 50 nm thick

PAGE 111

111 Figure 5 14 -RCs) acquired from the as grown and annealed strained Si80Ge20 layers

PAGE 112

112 Figure 5 15. (113) reciprocal space maps of as -grown and annealed at 900 for 30 min 100 nm th ick Si75Ge25 layers. Figure 5 1 6 RCs acquired from the as grown and annealed at 900 for 30 min Si75Ge25 100 nm thick layers.

PAGE 113

113 Figure 5 1 7 TEM images of annealed Si75Ge25 film. (a) Cross -sectional and (b) high resolution TEM images in <110> projection of a 100 nm thick stained Si75Ge25 annealed at 900 for 30 min. Figure 5 1 8 Bright field transmission electron microscopy image of implanted Si76Ge24 layer annealed at 500 for 30 min. The SiGe/Si and amorphous -crystalline inter faces ar e clearly marked in this figure

PAGE 114

114 Figure 5 1 9 (113) reciprocal space maps (RSMs) of Si76Ge24 layers. (a) and (b) were acquired from as -grown and annealed Si76Ge24 layers. The relaxation line is located between blue (relaxation=0 %) and red (re laxation=100 %) points Figure 5 20. Surface undulation of SiGe layers. (a) Schematic of the surface undulation of strained SiGe layer and AFM images of Si76Ge24 samples annealed at 800 for 30 min (b) without and (c) with ion implantation.

PAGE 115

115 Figure 5 2 1 Average peak -to -peak distance (black) and r.m.s. roughness (blue) of annealed Si76Ge24 layers without/with ion implantation. Figure 5 2 2 (113) RSMs of ionimplanted Si76Ge24 layers after thermal annealing.

PAGE 116

116 Figure 5 2 3 Strain relaxation and -RCs of ionimplanted Si76Ge24 after thermal anneal. The black and blue lines indicate strain relaxatio n and FWHM values, respectively Figure 5 2 4 Transmission electron microscopy images of an ioni mplanted Si76Ge24 films The layers were annealed at (a) 700 and (b) 800 The images were taken by using g=220 two beam condition

PAGE 117

117 1.6 2.0 2.4 2.8 101102103104105106 IntensityOmega 1.2 1.6 2.0 2.4 2.8 101102103104105106107 IntensityOmega 1.2 1.6 2.0 2.4 2.8 101102103104105106107 IntensityOmega 1.6 2.0 2.4 2.8 101102103104105106107 IntensityOmega Si SiGe As -grown 800 1000 900 Thickness oscillation Ge rich layer Ge deficient layer Figure 5 2 5 (113) / of as -grown strained -Si75Ge25 and after oxidation at 800 900 and 1000 for 1hr. Red dot lines indicate the SiGe peak position of as -grown film as a reference (a) (b)Si GDL GRL 50 nm 50 nm Spectrum 1 Spectrum 219.51 80.49 Spectrum 2 31.65 68.35 Spectrum 1 Ge Si Atomic % 19.51 80.49 Spectrum 2 31.65 68.35 Spectrum 1 Ge Si Atomic % GRL GDL Si SiO2Carbon Figure 5 2 6 Scanning transmission electron microscopy and reciprocal space map of Si75Ge25 oxidized in 800 (a) STEM image and Ge concentration measured by EDS and (b) (1 13) reciprocal space map (the red dot indicates the reciprocal lattice point (RLP) of as -grown Si75Ge25 layer )

PAGE 118

118 Si85Ge15Si80Ge20Si75Ge25 Figure 5 2 7 (113) reciprocal space maps (RSMs) of strained Si1xGex layers after the oxidation at 1000 for 1 hr Figure 5 2 8 STEM imag es of oxidized Si75Ge25 layers and their depth profile. (a) Si85Ge15 and (b) Si75Ge25 layers, and (c) Ge depth profile from EDS after the oxidation in 1000

PAGE 119

119 CHAPTER 6 SUMMARY StrainedSi/SiGe The strain relaxation and observation of the structural evolutions in strained-Si/SiGe layers after post thermal processes were investigated using advanced xray techniques. These techniques provided a wealth of information about strained layers, such as degree of strain / relaxation, element al composition, layer thickness, interface roughness, dislocation density, and dislocation behavior. Based on the experimental information, this work showed that the defect behavior in such strained layers was strongly dependent on the thermal processes. For the strain relaxat ion, the misfit defects should be nucleated at the interface between the strained layer and substrate. The activation energy for the nucleation of the misfit defects depended on the several nucleation source, such as pre existing threading dislocations from the substrate, surface of the strained layer, point defects induced by ionimplantation and oxidation processes. It was found from this work that the effects of pre -existing threading dislocations and surface step/undulation on the relaxation w ere not ve ry important so that high temperature N2 annealing induced only a slight relax ation However, the strain after the implantation or oxidation processes was highly relaxed due to the role of point defects assisted strain relaxation. The generation of the dis location loops by a large concentration of point defects lowered the activation energy of the misfit dislocation at the interface. The understand ing of the relaxation mechanism and defect behavior, highly relaxed and thin Si1xGex layers have been fabricat ed. High C rystalline GaN F ilms High resolution x ray techniques were us ed to acquire detailed information about optimiz ation of the growth conditions for high crystalline GaN films. Especially, a grazing incident x -ray diffraction (GIXD) method was used fo r the determination of the twist angle

PAGE 120

120 because most defects in GaN films were edge type threading dislocations It was found that the crystallinity of GaN films grown by the two -step growth method was strongly dependent on the growth conditions, such as gr owth temperature and gas flow rate in low temperature and high temperature growth steps. The edge type threading dislocations can be reduced by optimizing the 3 dimenstional growth mode time (t3D) in the high temperature island growth step. On the other ha nd, the screw -type threading dislocation density could be reduced by optimizing the growth temperature in the low temperature nucleation growth. P -type ZnO Phosphorus doped ZnO films were characterized using high resolution xray diffraction techniques. By analyzing the omega rocking curves, it was found that ZnO films grown on the low temperature buffer layer showed two columnar structures with different in-plane orientation which were surrounded by threading dislocations. For the strain evaluation a line profile analysis (LPA) using a William Hall plot and Warren -Averbach method was carried out. The results showed that while the crystallite size was almost the same in all ZnO films, 0.5 at. % P doped ZnO film showed the highest strain value, compared to 1 .0 at. % P -doped ZnO film. This suggested that the internal strain was increased with increasing incorporation of phosphorus atom s due to a mismatch between the atomic radii, resulting in the phosphorus segregation for the strain relaxation. This fact was found to be confirmed by XPS results.

PAGE 121

121 LIST OF REFERENCES 1 W D. Callister, Materials science and engineering: An introduction (John Wiley & Sons, New York, NY, 2006). 2 E. O. Hall, Proc. Phys. Soc. Ser. B 64, 7 47 (1951) 3 N. J. Petch, J. Iron and Steel Insti tute 25 (1953). 4 S. E. Thompson IEEE Transaction on Electron Devices 51, 1790 (2004) 5 J. P. McKelvey, S olid State Physics for Engineering and Materials Science (Krieger Publishing Company, Malabar, FL, 2003) 6 N Mohta and S E. Thompson, IEEE Circuit & Dev ices Magazine 18 (2005) 7 Z. Shi D. Onsongo, K. Onishi, J. C. Lee, S. K. Banerjee, IEEE Electron Device Lett. 24, 34 (2003). 8 O. Weber F. Ducroquet, T. Ernst, F. Andrieu, J. F. Damlencourt, J. M. Hartmann, B. Guillaumot, A. M. Papon, H. Dansas, L. Breva rd, A. Toffoli, P. Besson, F. Martin, Y. Morand, S. Deleonibus, in Symp. VLSI Technology Dig. Tech. Papers 42 (2004). 9 S. L. Wu, Y. P. Wang, S. J. Chang, Semicond. Sci. Technol. 21, 44 (2006) 10. U K. Mishra, P Parikh Y Wu, Processing of The IEEE 90 (2002) 11. L. Hsu W. Walukiewicz, Physical Review B 56, 1520 (1997) 12. J. Neugebauer, P hys. Stat. S ol. (b) 227, 93 (2001) 13. A. Smith, R. Feenstra, D. Greve, M. Shin, M. Skowronski, J. Neugebauer, J. Northrup, Surf. Sci. 423, 70 (1999) 14. O. Ambacher, J. Smart, J. R. Sh ealy, N. G. Weimann, K. Chu, M. Murphy, W. J. Schaff, L. F. Eastman, R. Dimitrov, J. Wittmer, M. Stutzman, W. Rieger, J. Hilsenbeck, J. Appl. Phys. 85, 3222 (1999). 15. R. Korbutowicz, J. Kozlowski, E. Dumiszewska J. Serafinczuk, Cryst. Res. Technol. 40, 503 (2005) 16. I. Akasaki H. Amano, Tech. Dig. Int. Electron Devices Meet 96, 231 (1996) 17. X. H. Wu, L. M. Brown, D. Kapolnek, S. Keller, S. P. Denbaars, J. S. Speck, J. Appl. Phys. 80, 3228 (1996) 18. K. Lorenz, M. Gonsalves, W. Kim, W. Narayanan, S. Mahajan, Appl. Phys. Lett. 77, 3391 (2000)

PAGE 122

122 19. X. H. Wu, P. Fini, E. J. Tarsa, B. Heying, S. Keller, U. K. Mishra, S. P. DenBaars, J. S. Speck, J. Crystal Growth 189/190, 231 (1998) 20. F. Degave, P. Ruterana, G. Nouet, J. H. Je, C. C. Kim, J. Phys. Condens. Matter 14, 13019 (2002) 21. D. Jena, I. Smorchkova, A. C. Gossard U. K. Mishra, Phys. Stat. Sol. B. 228, 617 (2001). 22. N. G. Weimann, L. F. Eastman, D. Doppalapudi, H. M. Ng, T. D. Moustakas, J. Appl. Phys. 83, 3656 (1998) 23. D. S. L i, J. Chen, H. B. Yu, H. Q. Jia, Q. Huang, J. M Zhou, J. Appl. Phys. 96, 1111 (2004) 24. Z. K. Tang, G. K. L. Wong, P. Yu, Appl. Phys. Lett. 72, 3270 (1998) 25. H. Ohota, K. Kawamura, M. Orita, M. Hirano, N. Sarukura, H. Hosono, Appl. Phys. Lett. 77, 475 (2000) 26. D. M. Bagnall, Y. F. Chen, Z. Zhu, T. Yao, S. Koyama, M. Y. Shen, T. Goto, Appl. Phys. Lett. 70, 2230 (1997) 27. D. C. Look, D. C. Reynolds, C. W. Litton, R. L. Jones, D. B. Eason, G. Cantwell, Appl. Phys. Lett. 81, 1830 (2002) 28. C. G. Van de Walle, Phys. Rev. Lett. 85, 1012 (2000) 29. U. Ozgur, Ya. I. Aliv ov, C. Liu, A. Teke, M. A. Reshchikov, S. Dogan, V. Avrutin, S. J. Cho, H. Morkoc, J. Appl. Phys. 98, 0411301 (2005) 30. R. D. Vispute, V. Talyansky, S. Choopun, R. P. Sharma, T. Venkatesan, M. He, X. Tang, J. B. Halpern, M. G. Spencer, Y. X. Li, L. G. Salam anca Riba, A. A. Iliadis, K. A. Jones, Appl. Phys. Lett. 73, 348 (1998) 31. V. Craciun, J. Elders, J. G. E. Gardeniers, J. Geretovsky, Ian W. Boyd, Thi n Solid Films 259, 1 (1995) 32. V. Craciun, R. K. Singh, J. Perriere, J. Spear, D. Craciun, J. Electrochemical Society 147, 1077 (2000) 33. S. Amirhaghi, V. Craciun, D. Craciun, J. Elders, I. W. Boyd, Microelectronic Engineering 25, 321 (1994) 34. A. Y. Cho, J. R. Arthur, Prog. Solid State Chem. 10, 157 (1975) 35. S. Mahajan, K. S. Sree Harsha, P rinciple of growth and proc essing of semiconductors (McGraw Hill, New York, NY, 1999). 36. M. B. Panish, H. Temkin, Ann. Rev. Mat. Sci. 19, 209 (1989)

PAGE 123

123 37. A. S. Grove, Ind. & Eng. Chem. 58, 48 (1966) 38. K. Mizuguchi, N. Hayafuji, S. Ochi, T. Murotani K. Fujikawa, J. Crystal Growth 77, 509 (1986) 39. F. Koyama, H. Uenohara, T. Sakaguchi K. Iga, Jpn. J. Appl. Phys. 26, 1077 (1987) 40. G. A. Fish, B. Mason, S. P. Denbaars L. A. Coldren, J. Crystal Growth 186, 1 (1998) 41. B. T. McDermott, E. R. Gertner, S. Pittman, C. W. Seabury M. F. Chang, J. Appl. Ph ys. 68, 1386 (1996) 42. H. Naoi, Y. Naoi S. Sakai, Solid State Electronics 41, 319 (1997) 43. S. L. Delage, M. a. di Forte Poisson, H. Blanck, C. Brylinski, E. Chartier, P. Collot, IEEE Electronics Letters 27, 253 (1991) 44. X. T. Zhang, Z. Liu, Y. P. Leung, Q. Li S K. Hark, Appl. Phys. Lett. 83, 5533 (2003) 45. S. J. C. Irvine, J. Bajaj H. O. Sankur, J. Crystal Growth 124, 654 (1992) 46. P. Chen, R. Zhang, Z. M. Zhao, D. J. Xi, B. Shen, Z. Z. Chen, Y. G. Zhou, S. Y. Xie, W. F. Lu, Y. D. Zheng, J. Crystal Growth 225, 150 ( 2001) 47. Douglas B. Chrisey, Graham K. Hubler, Pulsed Laser deposition of thin films (John Wiley & Sons, New York, NY 1994). 48. T. J. Jackson, S. B. Palmer, J. Phys. D: Appl. Phys. 27, 1581 (1994) 49. P. Scherrer, Nachr. Konigl. Gessel. Wiss. Gottingen 98 (1918) 50. I. Lucks, P. Lamparter, E. J. Mittemeijer, J. Appl. Cryst. 37, 300 (2004) 51. G. K. Williamson W. H. Hall, Acta Met. 1 22 (1953) 52. B. E. Warren B. L. Averbach, J. Appl. Phys. 21, 595 (1950) 53. B. E. Warren, X -ray Diffraction (Addison -Wesley New York, 1970). 54. M M. Hall, V. G. Veeraraghavan, H. Rubin, P. G. Winchell, J. Appl. Cryst. 10, 66 (1977) 55. J. I. Langford, J. Appl. Cryst. 11, 10 (1978) 56. G. K. Wertheim, M. A. Butler, K. W. West, D. N. E. Buchanan, Rev. Sci. Instrum. 45, 1369 (1974) 57. H. Kiessig, Ann. Phys. Leipzig 10, 769 (1931)

PAGE 124

124 58. U. Pietsch, V. Holy, T. Baumbach, High -Resolution X-ray Scattering from Thin Films to Lateral Nanostructure (Springer Pietsch, Ullrich 2004) 59. F. Schaffler, Semicond. Sci. Technol. 12, 1515 ( 1997). 60. J. P. Dismukes, L. Ekstrom, E. F. Steigmeier, I. Kudman, D. S. Beers, J. Appl. Phys. 35, 2899 (b) ( 1964). 61. R. Chierchia, T. Bottcher, H. Heinke, S. Einfeldt, S. Figge, D. Hommel, J. Appl. Phys. 93, 8918 (2003) 62. Dieter K. Schroder, Semicondcutor m aterial and device characterization (John Wiley & Sons, New Jersey, NY, 2006). 63. J. F. Moulder, W. F. Stickle, P. E. School, K. D. Bomden, Handbook of xray photoelectron spectroscopy (Physical Electronics Inc., Eden Prairie, Minnesota) 64. D. Sarid, Scanning force microscopy (Oxford University Press, N ew York NY, 1991) 65. T. Nakamura, Y. Yamada, T. Kusumori, H. Minoura, H. Muto, Thin Solid Films 411, 60 (2002) 66. S. Choopun, R. D. Vispute, W. Noch, A. Balsamo, R. P. Sharma, T. Venkatesan, A. Iliadis, D. C. Look, Appl. Phys. Lett. 75, 3947 (1999) 67. C. Liu, S H. Chang, T. W. Noh, M. Abouzaid, P. Ruterana, H. H. Lee, D. W. Kim, J. S. Chung, Appl. Phys. Lett. 90, 011906 (2007) 68. R. Chierchia, T. Bottcher, S. Figge, M. Diesselberg, H. Heinke, D. Hommel, Phys. Stat. Sol. (b) 228, 403 (2001) 69. D. C. Look, B. Clafi n, Phys. Status Solidi B 241, 624 (2004) 70. S. Limpijumnong, S. B. Zhang, S. H. Wei, C. H. Park, Phys. Rev. Lett. 92, 155504 (2004) 71. Kyoung -Kook Kim, HyunSik Kim, Dae -Kue Hwang, Jae Hong Lim, SeongJu Park, Appl. Phys. Lett. 83, 63 (2003) 72. S. M. Hubbard, G Zhao, D. Pavlidis, W. Sutton, E. Cho, J. Crys. Growth 284, 297 (2005) 73. T. Metzger, R. Hopler, E. Born, O. Ambacher, M. Stutzmann, R. Stommer, M. Schuster, H. Gobel, S. Christiansen, M. Albrecht, H. P. Strunk, Philosophical Magazine A 77, 1013 (1998) 74. D. Kapolnek, X. H. Wu, B. Heying, S. Keller, B. P. Keller, U. K. Mishra, S. P. DenBaars, J. S. Speck, Appl. Phys. Lett. 67, 1541 (1995) 75. T. A. Lafford, P. J. Parbrook, B. K. Tanner, Phys. Stat. Sol. (c) 0 542 (2002)

PAGE 125

125 76. V. V. Ratnikov, R. Kjutt, T. Shubina, J Appl. Phys. 88, 6252 (1998) 77. H. Amano, T. Takeuchi, H. Sakai, S. Yamaguchi, C. Wetzel, I. Akasaki, Mater. Sci. Forum 264-268, 1115 (1998) 78. J. Kozlowski, R. Paszkiewicz, M. Tlaczala, Phys. Stat. Sol. (b) 228, 415 (2001) 79. V. Srikant, J. S. Speck, D. R. Cla rke, J. Appl. Phys. 82, 4286 (1997) 80. H. Heinke, V. Kirchner, S. Einfeldt, D. Hommel, Phys. Stat. Sol. (a) 176, 391 (1999) 81. S. R. Lee, A. M. West, A. A. Allerman, K. E. Waldrip, D. M. Follstaedt, P. P. Provencio, D. D. Koleske, C. R. Abernathy, Appl. Phys. Lett. 86, 241904 (2005) 82. P. D. Healey, K. Bao, M. Gokhale, J. E. Ayers, F. C. Jain, Acta Cryst. A51 498 (1995) 83. B. Heying, X. H. Wu, S. Keller, Y. Li, D. Kapolnek, B. P. Keller, S. P. DenBaars, J. S. Speck, Appl. Phys. Lett. 68, 643 (1996) 84. H. Heinke, V. Kirchner, S. Einfeldt, D. Hommel, Appl. Phys. Lett. 77, 2145 (2000) 85. X. H. Zheng, H. Chen, Z. B. Yan, Y. J. Han, H. B. Yu, D. S. Li, Q. Huang, J. M. Zhou, J. Crys. Growth 255, 63 (2003) 86. R. Gay, P. B. Hirsch, A. Kelly, Acta Metall. 1 315 (1953) 87. J. W. Ed ington, Interpretation of Transmission Electron Micrographs: Monographs in practical electron microscopy in materials science (Macmillan, Philips Technical Library, 1975) 88. J. Narayan, Punam Pant, A. Chugh, H. Choi, J. C. C. Fan, J. Appl. Phys. 99, 054313 (2006) 89. S. H. Olsen, A. G. ONeill, S. Chattopadhyay, K. S. Kwa, L. S. Driscoll, D. J. Norris, A. G. Cullis, D. J. Robbins, J. Zhang, Semicond. Sci. Technol. 19, 707 (2004) 90. X. L. Yuan, T. Sekiguchi, J. Niitsuma, Y. Sakuma, S. Ito, S. G. Ri, Appl. Phys. Let t 86, 162102 (2005) 91. E. A. Fitzgerald, Y. H. Xie, M. L. Green, D. Brasen, A. R. Kortan, J. Michel, Y. J. Mii, B. E. Weir, Appl. Phys. Lett 59, 811 (1991) 92. M. S. Phen, R. T. Crosby, K. S. Jones, M. E. Law, J. L. Hansen, A. N. Larsen, Mater. Res. Soc. Sy mp. Proc. 864 (2005) 93. S. C. Jain, J. R. Willis, R. Bullough, Advances In Physics 39, 127 (1990) 94. B. W. Dodson, J. Y. Tsao, Annu. Rev. Mater. Sci. 19, 419 (1989)

PAGE 126

126 95. R. Loo, R. Delhougne, M. Caymax, M. Ries, Appl, Phys. Lett 87, 182108 (2005) 96. A. E. Romanov, W. Pompe, S. Mathis, G. E. Beltz, J. S. Speck, J. Appl. Phys 85, 182 (1999) 97. F. Fournel, H. Moriceau, B. Aspar, K. Rousseau, J. Eymery, J. Rouviere, N. Magnea, Appl. Phys. Lett. 80, 793 (2002) 98. L. H. Wong, C. C. Wong, K. K. Ong, J. P. Liu, L. Chan, R. Rao, K L. Pey, L. Liu, Z. X. Shen, Thin Solid Films 462-463, 76 (2004) 99. H. P. Klug L. E. Alexander, X -ray Diffraction Procedures (John Wiley and Sons., New York NY 1974) 100. R. A. Young D. B. Wiles, J. Appl. Cryst. 15, 430 (1982) 101. M. J. Hordon B. L. Averbach, Acta. Metallurgica. 9 237 (1961) 102. J. E. Ayers, J. Cryst. Growth 135, 71 (1994) 103. A. Boulle, C. Legrand, R. Guinebretiere, J. P. Mercurio, A. Dauger, Thin Solid Films 391, 42 (2001) 104. E. F. Bertaut, Acta Cryst 3 14 (1950) 105. K. Sawano, N. Usami, K. Arimot o, K. Nakagawa, Y. Shiraki, Thin Solid Films 508, 117 (2006) 106. T. B. Chen et al., Solid -State Electr. 50, 1194 (2006) 107. S. Zheng, J. Microelectr. 39, 53 (2008) 108. V. Ligatchev, T. K. S. Wong, S. F. Yoon, J. Appl. Phys. 95, 7681 (2004) 109. S. J. Koester, K. Rim, J. O. Chu, P. M. Mooney, J. A. Ott, M. A. Hargrove, Appl. Phys. Lett. 79, 2148 (2001) 110. S. R. Sheng, M. Dion, S. P. McAlister, M. L. Rowell, J. Crystal Growth 253, 77 (2003) 111. K. Grimm, L. Vescan, C. C. G. Visser, L. K. Nanver H. Luth, Materials Science and E ngineering B 69-70, 261 (2000) 112. L. B. Hansen, K. Sokbro, B. I. Lundqvist, K. W. Jacobsen, D. M. Deaven, Phys. Rev. Lett. 75, 4444 (1995) 113. S. Zheng, M. Kawashima, M. Mori, T. Tambo, C. Tatsuyama, Thin Solid Films 508, 156 (2006) 114. J. W. Matthews, S. Mader, T B. Light. J Appl Phys 41, 3800 (1970)

PAGE 127

127 115. S. W. Bedell, K. Fogel, D. K. Sadana, H. Chen, A. Domenicucci, Appl Phys Lett 85, 2493 (2004) 116. J. P. Hirth, J. Lothe. Theory of Dislocations (Wiley, New York NY, 1982). 117. J. Zou, D. J. H. Cockayne. J Appl Phy s 77, 2448 (1995) 118. P. M. J. Maree, J. C. Barbour, J. F. van der Veen, K. L. Kavanagh, C. W. T. Bulle Lieuwma, M. P. A. Viegers, J Appl Phys 62, 4413 (1987) 119. G. L. Olson, J. A. Roth, Mat. Sci. Reports 3 1 (1998) 120. J. H. Jang, M. S. Phen, K. Seibein, K. S. Jones, V. Craciun, Materials Letters 63, 289 (2009) 121. J. M. Baribeau, J. Vac. Sci. Technol. B 16, 1568 (1998) 122. C. Wu, R. Hull, J. Appl. Phys. 100, 083510 (2006) 123. J. H. Jang, M. S. Phen, A. Gerger, K. S. Jones, J. L. Hansen, A. N. Larsen, V. Craciun, Semi cond. Sci. Technol. 23, 035012 (2008) 124. B. G. Min, Y. H. Pae, K. S. Jun, D. H. Ko, H. Kim, M. H. Cho, T. W. Lee, J. Appl. Phys. 100, 016102 (2006) 125. D. K. Nayak, K. Kamjoo, J. C. S. Woo, J. S. Park, K. L. Wang, Appl. Phys. Lett. 56, 66 (1990) 126. L. P. Chen, Y. C. Chan, S. J. Chang, G. W. Huang, C. Y. Chang, Jpn. J. Appl. Phys. part 2 37, L122 (1998) 127. K. K. Linder, F. C. Zhang, J. S. Rieh, P. Bhattacharya, D. Houghton, Appl. Phys. Lett. 70, 3224 (1997) 128. N. R. Zangenberg, J. Lundsgaard Hansen, J. Fage Pedersen, A. Nylandsted Larsen, Physical Review Letters 87, 125901 (2001) 129. G. L. McVay, A. R. DuCharme, Physical Review B 9 627 (1974) 130. J. P. Dismukes, L. Ekstrom, R. J. Paff, J. Physical Chemistry 68, 3021 (1964) 131. L. N. Doolittle, Nucl. Instrum. Meth. B9 344 (1985)

PAGE 128

128 BIOGRAPHICAL SKETCH Jung Hun Jang was born in Wonju, Gangwon Do, South Korea. He earned a bachelor s degree in Department of Materials Science and Engineering, Hanyang University, Seoul, South Korea, in Feburary 2005. In August 2005, he enrolled in the Department of Materials Science and Engineering University of Florida, to pursue Ph.D under the guidance of Dr. Valentin Craciun. His main research involved growth and characterization of electronic materials, such as Si, SiGe, ZnO, and GaN. Especially, h e focused on the advanced characterization by using xray based techniques. During his Ph.D study, he worked in Major Analytical Instrumentation Center (MAIC), University of Florida. He received Korean Graduate Student Research Award from University of F lorida in 2008. He is author/co author of 12 journal and conference papers.