<%BANNER%>

Growth of Gallium Nitride and Indium Nitride Films and Nanostructured Materials by Hydride-Metalorganic Vapor Phase Epitaxy


PAGE 1

GROWTH OF GALLIUM NITRIDE AN D INDIUM NITRIDE FILMS AND NANOSTRUCTURED MATERIALS BY HYDRIDE-METALORGANIC VAPOR PHASE EPITAXY By HYUN JONG PARK A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2006

PAGE 2

Copyright 2006 by Hyun Jong Park

PAGE 3

To Cathy, who endured and enjoyed our life in Gainesville, Florida

PAGE 4

iv ACKNOWLEDGMENTS I would like to thank Dr. Tim Anderson for being my advisor and chairman. Although the expectations placed on me were always challenging, it led me to a level of academic excellence that was unthinkable without his expectations. I want to sincerely thank Dr. Olga Kryli ouk for being a coadvisor and cochair. Dr. Kryliouk showed me new ways of thinking in our daily discussions. She was a mentor, friend, and sometime critic, who allowed me to a ttain a higher level of research. I am in debt for her the effort and creativity she provided throughout my time at UF. I thank Dr. Fan Ren, Dr. Cammy Abernat hy, and Dr. Jason Weaver for being my committee members. Their thoughtfu l advice is greatly appreciated. Without the senior students contributions, this work would not be possible. I am truly in debt to Dr. Mike Reed who greatly modified th e Hydride-Metalorganic Vapor Phase Epitaxy system in Microfabritech from RCA hydride for VPE of InP. The optimism and kindness he provided helped even after leaving the group were crucial to me in solving numerous technical problems. I would like to thank Dr. Mike Mastro fo r being my mentor and teaching me how to operate the H-MOVPE system. He gave me valuable advice during my first year in Dr. Andersons group. Dr. Sang Won Kang opened a new door fo r InN growth using the H-MOVPE system. His creative thinking, logical a pproach, experience, and kindness have influenced me. He is the be st colleague I have ever had.

PAGE 5

v I am thankful for the enormous help of Woo Kyoung Kim, from whom I learned to organize things and how to work effectivel y. He also helped to perform HT-XRD and analyze the data. He was a counselor, friend, and mentor who has permanently influenced my life. YongSun Won is the smartest person that I have known who understands matters in a scientific way. His way of thinking wa s always beyond my knowledge. I appreciate his help and valuable discussions. Josh Mangum embodies the right attitude to wards research and people. He is a good communicator and friend. His willingness to help was essential for me to complete this project. Dr. Jang Yeon Hwang and Young Seok Kim helped me with Raman spectroscopy and hydrodynamics simulations. Their th orough understanding of the physical world deeply influenced me. Dr. Seokhyun Yoon helped me with SEM and EDS when I did not know how to use them. He also shared with me a happiness of a loving family when I was alone. Dr. Byoung Sam Kang and Hung-Ta Wang in Dr. Fan Rens group fabricated InN nanorods based gas sensor. The completeness of my work would not be possible without their help. I would like to thank my colleagues, KeeChan Kim, Oh-Hyun Kim, and Do Jun Kim for providing helpful discussions and en couragement during my research. KeeChan Kim and Oh Hyun Kim especially helped me automate the H-MOVPE system. Their way of working effectivel y was very influential.

PAGE 6

vi Without the great staff at UF, this work w ould not have been possible. I would like to acknowledge James Hinnant and Dennis Vince in Chemical Engineering, Scott Gapinski, Di Badylak, and Chuck Rowland in Mi crofabritech. I sincerely thank Valentin Craciun for HR-XRD, Eric Lambers for AES and XPS, Maggie Puga-Lambers for SIMS, Kerry Siebein for TEM at the Major Analytical Instrument Center, UF. I also would like to acknowledge my co llaboration outside UF. The support and discussion with Dr. Chinho Park (Yeung Nam University) was greatly appreciated. He and his students, Seok-Ki Yeo, helped analyz e the Raman spectroscopy data and gave me valuable advice. Dr. Jianyun Shen taught me how to use Thermo-Calc and strain energy modeling. Drs. Albert Davydov and Igor Levin helped me verify the thermochemical data and did excellent characterization sa mples at NIST by FE-SEM, TEM, and CBED. Drs. Dmitry Khokhlov at Moscow State University and Timur Burbaev at Physical Institute, Moscow carried out the PL measur ements for the InN films and nanorods. Dr. Jaime Freitas at Naval Research Lab perf ormed the Raman spectroscopy and PL on the InN nanorod samples. Dr. Zuzanna Liliental-Weber at Lawrence Berkeley Lab characterized InN nanorods by HR-TEM. Dr Talmage Tyler at the International Technology Center measured the field emission properties of InN nanorods. Dr. Mee Yi Ryu at the Air Force Institute of Technol ogy performed CL and Hall measurements of InN films and nanorods. I would not have re ached this point wit hout their enormous help. I thank Drs. Seong-Geun Oh, Yeong Koo Yeo, and Hong-Woo Park in Chemical Engineering, Hanyang University for recommen ding me for studying in the University of Florida. I would not have this great oppor tunity without their kind and thoughtful help.

PAGE 7

vii I also would like to thank Andrew Wislocki for proofreading this dissertation. He corrected grammatical errors and made the contents clearer by helpful discussions. I cannot thank my family enough for thei r boundless love. My father (Jong Soo Park), mother (Sun Kyoung Kim), and sister (Soo Kyoung Park) have been an essential part of my life. Although they were far away in Korea, they were always with me in my heart throughout my life as a grad uate student in Gainesville. My sons, David and Justin Park, showed me what true love is. Although I needed more time and effort to balance family life and research, I would not trade their love for anything in the world. Finally, I would like to thank my wife, Ca therine Park, for her endless support and trust. I was always happy for her pr esence and I will feel this way forever.

PAGE 8

viii TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES............................................................................................................xii LIST OF FIGURES.........................................................................................................xiv ABSTRACT....................................................................................................................xxi i CHAPTER 1 INTRODUCTION........................................................................................................1 1.1 History of GaN and InN Development...................................................................1 1.1.1 GaN Development........................................................................................1 1.1.2 InN Development.........................................................................................3 1.2 Literature Review...................................................................................................9 1.2.1 Equilibrium Analysis of GaN and InN.........................................................9 1.2.1.1 Thermochemical Data for GaN and InN............................................9 1.2.1.2 Equilibrium Calculations of GaN Growth by HVPE.......................10 1.2.2 Stress in GaN Films....................................................................................14 1.2.2.1 In situ Stress Measurements.............................................................16 1.2.2.2 Ex situ Stress Measurements............................................................18 1.2.3 GaN Growth on Si......................................................................................21 1.2.4 Growth of InN Films and Nanostructured Materials..................................25 1.2.4.1 Growth of InN Films........................................................................25 1.2.4.2 Growth of InN Nanostructured Materials........................................29 2 CHEMICAL EQUILIBRIUM ANAL YSIS OF H-MOVPE SYSTEM.....................32 2.1 Introduction...........................................................................................................32 2.2 Thermochemical Data Collections.......................................................................33 2.3 Chemical Equilibrium Calculations......................................................................35 2.3.1 Thermochemical Data Verification............................................................35 2.3.2 Complex Chemical Equilibrium Calculations............................................38 2.3.2.1 Ga-C-H-Cl-Inert System..................................................................38 2.3.2.2 Ga-C-H-Cl-N System.......................................................................45 2.3.2.3 In-C-H-Cl-Inert System....................................................................47 2.3.2.4 In-C-H-Cl-N-Inert System...............................................................53

PAGE 9

ix 2.3.2.5 NH3 Partial Decomposition..............................................................60 2.4 Conclusions...........................................................................................................62 3 STRESS DETERMINATION STUDIES OF GALLIUM NITRIDE FILM ON SAPPHIRE.................................................................................................................64 3.1 Introduction...........................................................................................................64 3.2 Effects of Lattice and Thermal Expansion Mismatches.......................................67 3.2.1 Lattice Mismatch at Growth Temperature.................................................67 3.2.2 Thermal Expansion Mismatch....................................................................69 3.2.3 Combination of Lattice and Th ermal Expansion Mismatches...................70 3.3 Stress Measurements of GaN Films on Sapphire.................................................71 3.3.1 GaN Structural and Compositional Studies................................................72 3.3.2 Stress Measurements by Raman Spectroscopy..........................................77 3.3.2.1 Curvature Calculations.....................................................................79 3.3.2.2 Lattice Parameter Calculations.........................................................80 3.3.3 Lattice Parameter Measurements by XRD Reciprocal Space Mapping..........................................................................................................81 3.3.4 Curvature Measurements by XRD Rocking Curve....................................84 3.4 Stress Modeling of GaN on Sapphire...................................................................89 3.4.1 Frank-van der Merwes Semi-infinite Model.............................................89 3.4.2 Layer-by-Layer Growth Model..................................................................91 3.4.2.1 Strain Energy (Homogeneous Stress) Calculations.........................92 3.4.2.2 Dislocation Energy (Periodic Strain Energy) Calculation...............94 3.4.2.3. Total Strain Energy Calculations....................................................95 3.5 Conclusions...........................................................................................................97 4 GROWTH OF GALLIUM NITRID E ON SILICON BY H-MOVPE.....................100 4.1 Introduction.........................................................................................................100 4.2 H-MOVPE Growth Technique...........................................................................101 4.2.1 Chemical Reactions of H-MOVPE Technique.........................................101 4.2.2 H-MOVPE Reactor Schematics...............................................................102 4.3 Thick GaN Growth on Al2O3..............................................................................106 4.4 GaN Growth on Si..............................................................................................111 4.4.1 Growth of Thin and Thick GaN on Si......................................................111 4.4.1.1 Nitronex GaN/Si template..............................................................111 4.4.1.2 Growth of Thin GaN Using Nitronex Template............................114 4.4.1.3 Growth of Thick GaN Using the Nitronex Template.....................115 4.4.2 Growth of Thick GaN on Si Using InN Interlayers.................................118 4.4.2.1 InN Growth on Si(111)..................................................................119 4.4.2.2 LT-GaN Growth on InN Buffer Materials/Si(111)........................120 4.4.2.3 Thick HT-GaN Growth on LT-GaN/InN/Si(111)..........................123 4.5 Conclusions.........................................................................................................126 5 GROWTH OF INDIUM NITR IDE NANORODS BY H-MOVPE.........................128

PAGE 10

x 5.1 Introduction.........................................................................................................128 5.2 Chemical Reactions for InN Growth by H-MOVPE..........................................130 5.3 InN Nanorods Growth Optimization..................................................................130 5.3.1 Experimental Procedure...........................................................................131 5.3.2 Results......................................................................................................131 5.3.2.1 Effects of Growth Temperat ure and HCl/TMI Molar Ratio..........131 5.3.2.2 Effects of NH3/TMI Molar Ratio and Substrate Material..............134 5.3.2.3 Equilibrium Analysis......................................................................135 5.4 Properties of InN Nanorods................................................................................138 5.4.1 Morphology..............................................................................................138 5.4.2 Crystallinity..............................................................................................139 5.4.3 Self-alignment..........................................................................................141 5.4.4 Growth Axis and Structural Properties.....................................................145 5.4.5 Polarity.....................................................................................................150 5.4.6 Chemical Composition.............................................................................151 5.4.7 Photoluminescence...................................................................................153 5.4.8 Cathodoluminescence...............................................................................156 5.4.9 Raman Spectroscopy................................................................................157 5.4.10 Electrical Properties................................................................................158 5.4.11 Field Emission Properties.......................................................................159 5.5 Pt-dispersed InN Nanorods for Select ive Detection of Hydrogen at Room Temperature.........................................................................................................163 5.5.1 Experimental Procedure...........................................................................163 5.5.2 Results......................................................................................................164 5.6 Conclusions.........................................................................................................167 6 EXPLORATORY STUDY OF INDIUM NITRIDE FILMS GROWTH BY HMOVPE....................................................................................................................169 6.1 Experimental Procedure......................................................................................169 6.2 Results.................................................................................................................170 6.2.1 Effect of HCl/TMIn Molar Ratio.............................................................170 6.2.2 Effect of Growth Temperature.................................................................172 6.2.3 Effect of NH3/TMIn Molar Ratio.............................................................174 6.2.4 Effect of Buffer Layer..............................................................................177 6.2.4.1 Surface Morphology of InN Films without Buffer Layer..............177 6.2.4.2 Growth of InN Films with Low Temperature Buffer Layer..........177 6.2.5 Growth of InN Films on Al2O3 and Si......................................................179 6.3 Conclusions.........................................................................................................181 7 EXPLORATORY STUDY OF GALLIUM NITRIDE NANORODS GROWTH BY H-MOVPE..........................................................................................................183 7.1 Experimental Procedure......................................................................................183 7.2 Results.................................................................................................................183 7.3 Conclusions.........................................................................................................189

PAGE 11

xi 8 FUTURE WORK AND RECOMMENDATIONS..................................................190 8.1 Nucleation and Growth Mechan ism Studies of InN Nanorods..........................190 8.2 Self-aligned InN Nanorods as Buffer Material for GaN on Si...........................190 8.3 Buffer Layer Optimization for GaN growth on Si Substrate..............................191 8.4 Use of Negative Thermal Expansion Materials..................................................192 8.5 Double-sided Growth of GaN on Si and Sapphire.............................................195 APPENDIX: THERMOCHEMICAL DATA FOR AL-IN-GA-C-H-CL-N SYSTEM..................................................................................................................196 LIST OF REFERENCES.................................................................................................222 BIOGRAPHICAL SKETCH...........................................................................................243

PAGE 12

xii LIST OF TABLES Table page 1-1 Reported values of thermochemical data for solid GaN and InN............................121-2 Reported values of the Raman E2 (high) peak positions..............................................201-3 Techniques used for growth of crack-free GaN on Si..............................................232-1 Commonly considered species in Ga-In-H-C-Cl-N system.....................................342-2 Additionally included gas phase species..................................................................352-3 Base inlet conditions for sources for Ga N growth and atomic mole fractions for calculation................................................................................................................392-4 Typical growth conditions for InN and atomic mole fractions for calculation........483-1 Lattice parameters of GaN and widely used substrates at 300 and 1300 K.............683-2 Lattice constants at 1300 a nd 300 K and thermal strain ( T = 1000 K) of GaN and widely used substrates.......................................................................................703-3 The growth conditions of the two samples for stress measurements.......................723-4 GaN a and c lattice parameters measured at 300 K ()...........................................843-5 Stress, lattice parameter, and curv ature of the H-MOVPE GaN with depth............863-6 Calculated misfit dislocation density of GaN on widely used substrates with Vernier period..........................................................................................................903-7 Lattice constants () of GaN, AlN, and Al2O3........................................................933-8 Elastic stiffness coefficients cij (GPa) and compliances sij (1/GPa) of GaN, AlN, and Al2O3..................................................................................................................933-9 Shear moduli, Poisson's ratios, and la ttice parameters of GaN, AlN, and Al2O3.....943-10 Calculated lattice constants and total energy...........................................................964-1 Electrical properties of GaN film grown on GaN/Al2O3 template.........................111

PAGE 13

xiii 4-2 Sheet resistance and resistivit y measured by four-point probe..............................1114-3 Growth conditions for InN buffer interlayer on Si.................................................1195-1 Base conditions of InN nanorods growth...............................................................1315-2 Intensity ratio comparison with XRD powder pattern with different substrates...............................................................................................................1405-3 Electrical properties of InN films and nanorods....................................................1596-1 Base conditions of InN film growth.......................................................................1697-1 Base conditions for Ga N micro/nanorods growth..................................................1838-1 Lists of NTE materials...........................................................................................192A-1 Gas Phase...............................................................................................................197A-2 Gas Phase (Metalorganics Adducts, Oligomers, etc)............................................217A-3 Liquid Phase...........................................................................................................218A-4 Solid Phase.............................................................................................................220A-5 Carbon (Graphite)..................................................................................................221A-6 GaN and InN..........................................................................................................221

PAGE 14

xiv LIST OF FIGURES Figure page 1-1 Reported values of electric properites of InN (a) Hall mobility and (b) Carrier concentration with calendar year................................................................................61-2 Reported bandgap energy of InN as a function of carrier concentration...................71-3 Examples of equilibrium calculations. (a) Driving force for the deposition as a function of growth temperature with va rious parameters for F (b) Comparison between calculated growth ra tes and experimental data..........................................111-4 Driving force for the deposition of InN ( PIn) using InCl (HVPE) and InCl3 (THVPE) as a function of growth temperature........................................................131-5 Schematic of GaN growth on SiC (a ) GaN/SiC and (b) Ga N/AlN/SiC growth......161-6 In situ curvature measurements during the growth of a ~ 6 m thick GaN layer showing the influence of AlN interlay ers on the curvature. The sample was crack-free after growth.............................................................................................181-7 Raman peak shifts of 2 m thick GaN with different LT-GaN buffer layer thickness; LT-GaN buffer layer thickne ss (a) 10 nm, (b) 50 nm, (c) 75 nm, and (d) 85 nm..................................................................................................................212-1 Phase diagram of In-N systems at P = 0.1 MPa and experimental InN decomposition data...................................................................................................362-2 Calculated Grxn for GaN(s) and (GaN)3(g) formation reactions............................372-3 Phase diagram of Ga-N systems at P = 0.1 MPa......................................................382-4 Schematic of the inlet of HMOVPE technique for GaN growth............................382-5 Equilibrium partial pressures of gas phase molecules and mole fractions of condensed phases with respect to the temperature in Ga-C-H-Cl-Inert system.......402-6 Equilibrium partial pressures of gas phase molecules and mole fractions of condensed phases with respect to X(H) /{X(Inert) + X(H)} in Ga-C-H-Cl-Inert system.......................................................................................................................42

PAGE 15

xv 2-7 Equilibrium partial pressures of gas phase molecules and mole fractions of condensed phases with respect to X( Cl)/X(Ga) ratio in Ga-C-H-Cl-Inert system......................................................................................................................442-8 Schematic of H-MOVPE for GaN growth: Ga-C-H-Cl-N system. N is present and H2 is used as a carrier gas in the system compared to Figure 2-4.....................452-9 Calculated growth-etch transition temperature as a function of Cl/Ga ratio............462-10 Calculated transition of growth-etc h and Ga(l) formation as a function of temperature and Cl/Ga ratio.....................................................................................472-11 Schematic of the inlet of HMOVPE technique for InN growth..............................482-12 Equilibrium partial pressures of gas phase molecules and mole fractions of condensed phases with respect to the temperature in In-C-H-Cl-Inert system........502-13 Equilibrium partial pressures of gas phase molecules and mole fractions of condensed phases with respect to X(H) /{X(Inert) + X(H)} in In-C-H-Cl-Inert system.......................................................................................................................512-14 Equilibrium partial pressures of gas phase molecules and mole fractions of condensed phases with respect to X(Cl)/ X(In) ratio in In-C -H-Cl-Inert system.....522-15 Schematic of H-MOVPE inlet wi th In-C-H-Cl-N-In ert system for thermodynamic calculations. N is present in the system compared to Figure 211............................................................................................................................. .542-16 Calculated growth and etch regime with respect to temperature and Cl/In ratio. N/In ratio was varied from 100 to 7000...................................................................552-17 Calculated growth and etch regime with respect to temperature and Cl/In ratio. Cl/In ratio was varied from 1.1 to 10.......................................................................562-18 Calculated results of the growth, no gr owth (etch), and In droplet regimes (a) Cl/In = 0 to 0.5, N/In = 4.0 x 104 to 7.0 x 104, (b) Cl/In = 0 to 1, N/In = 1.0 x 103 to 7.8 x 104, and (c) In droplet etching c onditions at T = 1153 K, P = 105 Pa.........572-19 Calculated growth-etch transition temperatures (a) In droplet etch conditions when no HCl was present (b) growth-etc h transition temperature at high N/In ratios: 105, 106, and 107............................................................................................592-20 Equilibrium partial pressures for decomposition of NH3 at 1 atm total pressures calculated by Thermo-Calc......................................................................................602-21 Magnitude of the Gibbs energy correcti on required to achieve a value for partial decomposition ( ) at 950 C....................................................................................61

PAGE 16

xvi 2-22 NH3 mole fraction with respect to te mperature at different values of ..................623-1 Substrates for GaN plotted with thermal strain versus lattice parameters at room temperature...............................................................................................................703-2 XRD and -2 rocking curve of MOCVD and H-MOVPE grown GaN on sapphire, XRD FWHM; -rocking curve (MOCVD: 1261 arcsec, H-MOVPE: 1790 arcsec); -2 rocking curve (MOCVD: 92 arcsec, H-MOVPE: 122 arcsec)..........................................................................................723-3 Pole figures for the (116) sapphire substrate (2 theta = 57.490 ) and (112) GaN by MOCVD (2 theta = 69.185 ). The GaN in-plane axis is rotated by 30 with respect to the sapphire axis.......................................................................................733-4 Pole figures for the (116) sapphire substrate (2 theta = 57.490 ) and (112) GaN by H-MOVPE (2 theta = 69.185 ). The GaN in-plane axis is rotated by 30 with respect to the sapphire axis..............................................................................743-5 AES surface scan and depth profile of GaN on sapphire grown by MOCVD.........753-6 AES surface scan and depth profile of GaN on sapphire grown by H-MOVPE.....763-7 Raman E2 peak shifts at the surface (a) MO CVD, (b) H-MOVPE, and (c) depth profile.......................................................................................................................7 83-8 XRD reciprocal space mapping of H-MOVPE grown GaN (002) and (114) peaks.........................................................................................................................8 23-9 XRD reciprocal space mapping of MOCVD grown GaN( 002) and (114) peaks........................................................................................................................8 33-10 The procedure of determining curv ature by measurements of XRD rocking curves.......................................................................................................................853-11 Curvature measurements by XRD rock ing curve by displacing the H-MOVPE grown GaN/c-Al2O3 sample in x-direction by 5 mm...............................................863-12 SIMS oxygen depth profiles of MO CVD and H-MOVPE grown GaN films.........883-13 Frank-van der Merwes semi -infinite overgrowth model........................................903-14 Shen-Danns layer-by-layer model..........................................................................913-15 Minimum energy calculation of stress + dislocation energy....................................963-16 In-plane lattice consta nts of GaN on AlN/sapphire.................................................97

PAGE 17

xvii 4-1 Schematics of H-MOVPE (a) bird eye view of the entire reactor, (b) the source and growth zones with temperatur e profile in the source zone..............................1034-2 Process Flow Diagram (PFD) of H-MOVPE system.............................................1054-3 XRD -2 scan and -rocking curve of as received GaN/Al2O3 template from Uniroyal Optoelectronics. denotes the secondary peak due to Cu K radiation1074-4 SEM images of GaN film (a) Cross-sectional SEM of 125 m thick H-MOVPE GaN film on GaN/Al2O3 template, (b) SEM plan-view of the same sample. (c) Plan-view of LT-HVPE smoothing layer...............................................................1084-5 XRD of GaN film (a) XRD -2 scan of thick (45 m) GaN on GaN/Al2O3, (b) XRD -rocking curve of GaN (002) peak; FWHM = 780 arcsec.........................1094-6 Auger Electron Spectroscopy of 45 m thick GaN film (a) surface scan (b) sputtering depth profile (c) su rface scan after sputtering.......................................1104-7 X-SEM image of GaN on Si using AlGaN graded layer.......................................1124-8 XRD -2 scan and HR-XRD -rocking curve of as received GaN/Si from Nitronex, FWHM = 0.222 (800 arcsec)...............................................................1134-9 SEM plan and cross sectional views of 2 m H-MOVPE + 1 m MOVPE crackfree GaN grown on AlGaN/Si graded layer...........................................................1144-10 XRD -2 scan of 2 m H-MOVPE + 1 m MOVPE crack-free GaN grown on AlGaN/Si graded layer...........................................................................................1144-11 SEM plan-views of thick (55 m) GaN films grown on GaN/Si templates (4 hr) with (a) fast cooling and (b) slow co oling. Note two micrographs are at different magnification...........................................................................................1154-12 Schematic of crack generation of Ga N on Si and GaN on Nitronex template. Cracking penetration to Si we re observed in both cases........................................1164-13 XRD -2 scan of thick (55 m) GaN film on GaN/Si template for 4 hr..............1174-14 XRD -rocking curve of GaN (002) peak of 55 m thick GaN. FWHM = 0.248 (892 arcsec)..........................................................................................................1174-15 The schematic to obtain thick GaN on Si(111) using InN buffer materials...........1194-16 SEM plan-view of InN columnar f ilm, small nanorods (d = 250 nm), large nanorods (d = 500 nm), and microrods grown on Si(111).....................................120

PAGE 18

xviii 4-17 SEM plan-view of 10, 20, and 30 min LTGaN grown on (a) InN columnar film, (b) smaller nanorods (d = 250 nm), (c) la rger nanorods (d = 500 nm), and (d) microrods................................................................................................................1214-18 LT-GaN deposited on InN nanorods (d = 500 nm) on Si(111) for 30 min (~ 4 m) (a) plan-view of LT-GaN on InN na norods, (b) cross-sectional view (c) XRD -2 scan.......................................................................................................1224-19 HT-GaN/LT-GaN growth on various InN structures (a) InN columnar film, (b) small nanorods (d = 250 nm), (c) large nanorods (d = 500 nm), and (d) microrods. The right figures show expa nded views of the surface of HT-GaN films........................................................................................................................12 34-20 Thick (28 m) and freestanding (selfseparated) GaN grown on InN nanorods/Si (a) SEM plan-view of the GaN film, (b) XRD -2 scan......................................1244-21 Cross-sectional view of thick GaN grown on InN nanorods/Si(111). The surface was covered by LT-HVPE layer grown after thick HT-HVPE..............................1254-22 SEM plan-view of thick GaN grown on InN nanorods (d=500 nm)/Si(111). (a) as grown HT-GaN surface, (b) covered LT-GaN surface......................................1265-1 The growth map of InN at different HCl/TMI ratio with growth temperature; N/In = 250; Growth time = 1 hr; Substrate = c-Al2O3...........................................1325-2 Scanning electron micrographs of InN film and nanorods. NH3/TMIn were varied from 100 to 7000; Other growth conditions are HCl/TMIn = 4, T = 600 C; Growth time = 1 hr...........................................................................................1345-3 The calculated boundary of the growth and etch regimes and experimental observations. (a) deposition temperature vs HCl/TMIn ratio for selected N/In atom ratios; (b) depositio n temperature vs. N/In atomic ratio for selected HCl/TMIn ratios.....................................................................................................1365-4 Scanning electron micrographs of In N nanorods grown at optimal conditions; Cl/In = 4, N/In = 250, T = 600 C, for 1 hr growth on Si(111).............................1385-5 XRD -2 scan of InN nanorods grown on various substrates (a) a-Al2O3, (b) cAl2O3, (c) r-Al2O3, (d) Si(100), (e) Si (111), (f) GaN/c-Al2O3, and (g) PDF#501239 (powder diffraction file)................................................................................1395-6 Scanning Electron Micrographs of aligned InN nanorods grown on GaN/Al2O3 substrates. (a) sparse nanorods with sharp tips (b) dense nanorods with sharp tips (c) dense nanorods with flat tips............................................................................1425-7 XRD scan of aligned InN nanorods on GaN/c-Al2O3 substrates...........................1435-8 XRD -rocking curve of aligned InN nanorods on GaN/Al2O3............................143

PAGE 19

xix 5-9 XRD pole figure of self-aligned InN nanorods (Substrate: GaN/c-Al2O3)............1445-10 InN (103) plane and -axis for pole figure measurement......................................1445-11 Transmission electron micrographs of an InN nanorod (a) splitting of reflections and (BF) bright field image at the bottom, (b) sparse planar defects by darkand bright-field image...................................................................................................1465-12 TEM and SAD of InN nanorods (WZ) w ith growth axes (a) [0002], (b) [1010], and (c) [1124].........................................................................................................1475-13 SEM image of wurtzite InN nanorod grown on the side wall of a nanorod..........1485-14 TEM and SAD of InN and In2O3 nanorods (a) InN (ZB) growth axis = a few degrees off [220] (b) In2O3 (BCC) growth axis = [402]........................................1495-15 CBED of the InN nanorod......................................................................................1515-16 AES spectrum of surface scan of as grown InN nanorods at T = 600 C, HCl/TMIn = 4, NH3/TMIn = 250, 1 hr growth on c-sapphire...............................1525-17 Scanning electron micrographs with AE S of InN nanorods before and after AES sputtering damage..................................................................................................1525-18 EDS line-scan of an InN nanorod grown on Si substrate.......................................1535-19 Room temperature (300 K) PL of In N nanorods grown on Si substrate and InN film on GaN substrate, maximum intensity observed around 1.08 eV..................1545-20 Low Temperature (6 K to 300 K) PL of InN nanorods on Si substrate (a) whole detection range, (b) magnified range be tween 730 to 850 meV from 23 to 100 K region......................................................................................................................1555-21 CL spectra of InN nanorods grown on sapphire....................................................1565-22 Room-temperature Raman scattering of InN nanorods deposited on Si and GaN/Si substrates...................................................................................................1585-23 Scanning Electron Micrograph of sharp edge InN nanorods grown on Si (single NR on the left) and on GaN/Al2O3 substrates........................................................1605-24 I-V curves and Fowler-Nordhe im plots of InN nanorods on Si.............................1615-25 SEM images and schematic of hydrogen gas sensor made of Pt nanoparticle dispersed InN nanorods..........................................................................................1635-26 I-V characteristics from unc oated and Pt-coated InN nanorods............................164

PAGE 20

xx 5-27 I-Time plot of 10 to 250 ppm H2 test by Pt-InN nanor ods (left) and | R|/R(%)Time plot of 10 to 250 ppm H2 test by Pt-InN nanorods (right)............................1655-28 N2, N2O, ND3, and O2 test for InN nanorods.........................................................1666-1 Scanning electron micrographs of InN films (a) HCl/TMIn = 0, (b) HCl/TMIn = 0.3, (c) HCl/TMIn = 1, and (d) HCl/TMIn = 4. Other growth conditions: T = 560 C; P = 760 Torr; N/In ratio = 2500; substrate: GaN/Al2O3; growth time = 1 hr..................................................................................................1706-2 Growth rate and XRD of InN (a) InN gr owth rate and (b) XRD scan with respect to HCl/TMIn ratio..................................................................................................1716-3 SEM plan-views of InN films grown at different temper atures (a) 500, (b) 525, (c) 560, and (d) 600 C; Other growth conditi ons Cl/III ratio = 1; P = 760 Torr; N/In ratio = 2500; Substrate = GaN/c-Al2O3; growth time = 1 hr.........................1726-4 Growth rate and XRD scans with resp ect to the growth temperature (a) Loge (Normalized growth rate) vs 1000/T (K), (b) XRD scans in 2 theta ranges from 30 to 32; growth temperature 300 to 700 C; growth time = 1 hr.......................1736-5 Scanning electron micrographs of InN films grown in different NH3/TMIn ratios; other growth parameters Cl/In ratio = 1; P = 760 Torr; Growth T = 560 C; growth time = 1 hr............................................................................................1756-6 Growth rate and XRD scans with respect to NH3/TMIn ratio (a) Growth rate vs. N/In (NH3/TMIn) ratio, (b) XRD scans of 2 theta range from 30 to 32 when N/In was 100 to 50000, (c) enlarged view of XRD for N/In range 7000 to 50000; Cl/In ratio = 1; P=760 Torr; Growth T = 560 C; growth time = 1 hr...................1766-7 AFM images of the InN films. NH3/TMI = 10000, HCl/TMI = 0.3, T=560 C; Substrate = GaN/c-Al2O3; growth time = 1 hr.......................................................1776-8 SEM plan-view and cross-se ctional of InN films on GaN/Al2O3 substrate; N/In = 5000, Cl/In = 1, T = 560 C, and growth time = 1 hr with 450 C buffer layer growth for 30 min...................................................................................................1786-9 XRD -2 scan of InN grown on GaN/Al2O3 substrate; N/In = 5000, Cl/In = 1, T = 560 C, and growth time = 1 hr with 450 C buffer layer growth for 30 min.1786-10 XRD -rocking curve of InN grown on GaN/Al2O3 substrate; N/In = 5000, Cl/In = 1, T = 560 C, and growth time = 1 hr with 450 C buffer layer growth for 30 min...............................................................................................................1796-11 XRD -2 scan of InN film grown on c-Al2O3......................................................1806-12 XRD -2 scan of InN film grown on Si(111)......................................................180

PAGE 21

xxi 6-13 XRD -2 scan comparison of InN film grown on different substrates................1817-1 SEM plan-view micrographs of GaN grown at Cl/Ga ratio 2 and 3 on c-Al2O3, GaN/c-Al2O3, and Si at N/Ga = 250 w ithout nitridation in N2..............................1847-2 SEM micrographs of GaN grown at Cl/Ga ratio of 2.5 and 2.75, N/Ga = 250, and in H2 with nitridation.......................................................................................1857-3 SEM plan-view of GaN with Cl/Ga ratio 2, 2.5, 4, and 6 on c-Al2O3, GaN/cAl2O3, and Si(111)..................................................................................................1867-4 SEM plan-view of GaN at Cl/Ga = 19, N/Ga = 800, T = 850 C on various substrates................................................................................................................1878-1 Schematic of crack-free GaN growth on Si substrate using self-aligned InN or GaN nanorods.........................................................................................................1918-2 Lattice parameters of ZrW2O8 ( ) and HfW2O8 ( ) as a function of temperature.............................................................................................................1938-3 Crystal structures of NTE materials (a) Unit cell of ZrW2O8, with 90% thermal ellipsoids drawn. (b) Polyhedral re presentation of the structure. ZrO6 octahedra shown in white, WO4 tetrahedra shaded.................................................................1938-4 An array of linked triangles as f ound in tridymite. Rotation of one triangle causes the local environmen t to be pulled inwards................................................194

PAGE 22

xxii Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy GROWTH OF GALLIUM NITRIDE AN D INDIUM NITRIDE FILMS AND NANOSTRUCTURED MATERIALS BY HYDRIDE-METALORGANIC VAPOR PHASE EPITAXY By Hyun Jong Park December 2006 Chair: Timothy Anderson Cochair: Olga Kryliouk Major Department: Chemical Engineering Chemical equilibria analyses in the Ga/I n-H-C-Cl-N-inert system were performed to predict gas and condensed phase species that might exist in H-MOVPE of GaN and InN. It was found that carbon co-deposition an d metal droplets (Ga or In) formation can be eliminated by providing a threshold level of H2 and HCl, respectively. The transitions between deposition and etching, and between 2phase (Ga or In)N and pure (Ga or In) metal and single-phase (Ga or In)N were pred icted as a function of N/III and Cl/III molar ratios and temperature. Stress was measured by Raman spectroscopy in two GaN films (grown by MOCVD and H-MOVPE, respectively) on sapphire and a difference of 2 cm-1 in the E2 mode was observed. Subsequent measurement by XRD-SM, rocking-curves, AES, and SIMS showed inconsistent results compared to the Raman result suggesting that the Raman E2 shift may not be only related to biaxial or hydrostatic stress.

PAGE 23

xxiii Controlled growth of InN nanorods (NR) was achieved by varying the Cl/In and N/In molar ratios and growth temperatur e. The NRs were grown on a, c, r-Al2O3, GaN/cAl2O3, Si (100), and Si (111) without a templa te or external catalyst. Well-faceted wurtzite and threading dislocation-free NRs were observed by SEM and TEM. The diameters and lengths of nanor ods ranged 100 to 300 nm and ~ 1 m for 1 hr growth. XRD patterns indicated the nanorods were te xtured in the [002] direction and TEM-DP confirmed the growth axis was predomin antly [002]. Nanorods grown on GaN/c-Al2O3 occasionally showed vertical self-alignment and were epitaxially grown as judged by XRD pole-figure analysis. The nanorods ha d N-polarity as characterized by CBED, while RT PL showed a predominant peak at 1.08 eV. Raman spect roscopy showed three phonon lines (451, 496, and 596 cm-1) that were assigned to A1(TO), E2 2, and E1(LO), respectively. InN NRs-based gas sensor was fabricated that could detect H2 to 10 ppm. Crack-free, 3 m GaN films were grown on GaN/ AlGaN/Si template at 850 C although cracks developed when the thickness exceeded 7 m. It was possible, however, to grow crack-free polycrystalline 40 m thick GaN on Si using InN NRs as a buffer material.

PAGE 24

1 CHAPTER 1 INTRODUCTION 1.1 History of GaN and InN Development The technology related to GaN and InN is highly developed. Bo th are currently used as host materials in optoelectronics and electronics devices. It is informative to review the history of those materials to prope rly understand the current issues that require more study. 1.1.1 GaN Development The history of GaN crystal growth origin ates in 1938 [Juz38] when GaN was first synthesized by flowing NH3 through gallium at high temperat ures. The standard crystals for GaN powder X-ra y diffraction file ( i.e., PDF# 02-1078) were obt ained then and this data is still in use. Grimmeriss and Koelmans used the same method to grow small GaN crystals and measured PL for th e first time back in 1959 [Gri59]. GaN film was first deposited by Marusk a and Tietjen on sapphire substrate by CVD [Mar69]. Little attention, however, wa s paid to GaN's use as a semiconductor because of the difficulty in doping it p-type. GaN regained its attention in 1980s with need to develop blue emitters. Yoshida reported improvement of GaN quality by using LT-AlN buffer layer in 1983 [Yos83]. In 1986, Amano reported improved GaN surface morphology and electrical and optical proper ties by using LT-AlN buffer laye r [Ama86]. The buffer layer acted as a nucleation layer and decreased interfacial free energy to facilitate twodimensional growth by changing the properties of the surface.

PAGE 25

2 To fabricate GaN based devices electronic properties such as carrier concentrations must be controlled. The doping of wide ba nd gap materials is known to be difficult in part the probability of forming native def ects that can dominate the electronic point defect chemistry. Intrinsic GaN grows n-type with a dire ct wide (3.4 eV) bandgap. The long belief in the native n-t ype nature of GaN is due to N-vacancy donor behavior. High conductivity p-type GaN was not achieved unt il 1989. It was observed that p-type conductivity drastically improved (from 108 to 35 cm) after low electron energy beam irradiation (LEEBI). Utilizing LEEBI treated Mg doped p-type GaN and intrinsic n-type GaN, Amano fabricated the first UV-LED [Ama89]. Na kamura later established that a simple annealing in inert or vacuum environment w ill improve p-type conductivity similar to the LEEBI treatment [Nak92]. The reason of great p-type conductivity improvement by LEEBI or annealing was not clear then, but it wa s later discovered th at an Mg-H neutral species prevents Mg activation [Got95]. GaN-based Field Effect Tr ansistors (FET) [Kha93] a nd Heterojunction Bipolar Transistors (HBT) [Pan94] were also fabr icated in the early 1990s, enabled by the improved crystal quality and conductivity. The threading dislocation density of these transist ors, however, was about 109 to 1010 cm-2, or about 106 times higher than typical semiconductors. Although devices could be fabricated with the high density of dislocations, their long-term reliability was questioned. In 1994, a significant reduction of the disl ocation density was achieved by adopting lateral epitaxial overgrowth that used either a SiO2 or Si3N4 mask [Kat94]. Because

PAGE 26

3 threading dislocations tend to form parallel to the growth axis, vertical blocking of the threading dislocations improved cr ystal quality drastically (from 109 1010 to 104 105 cm-2). GaN-based devices began to be fabricated more routinely beginning in 1994. It is worth mentioning that Nakamuras two-fl ow reactor design significantly improved crystal quality [Nak94]. In this process, inert or hydroge n gas is injected vertically onto the substrate to modulate local concentra tions of precursors by reducing the boundary layer thickness. Following the invention and breakthr oughs, Nakamura reported an InGaN based 1 cd LED in 1994 [Nak94], a 10 cd LED in 1995 [Nak95a, Nak95b], a laser diode in 1996 [Nak96a], and continuous wave (CW) lasing at room temperature in 1997 [Nak97]. The intensity and life time im proved in consecutive years and Nichia Corporation reported estimated 10,000 hr life time of blue laser diodes in 1998 [Nak98]. Research related to GaN has continued, as a result of the major breakthroughs made during the 80s and 90s. The lack of a suitab le substrate material continues to create problems and the growth of large (> 2) diam eter freestanding GaN has not been easily achieved due to cracking and bowing problems. Also the heat generated from LED devices is still problematic due to the low th ermal conductivity of the sapphire substrate. Undoubtedly, the integration of GaN technology with Si would be another important step for GaN-based device development. Furthermor e the use of 12 diameter Si wafer would improve device throughput and lowe r unit device fabrication cost. 1.1.2 InN Development The first report on InN crystal growth originates from 1938 and used InF6(NH4)3 for crystallographic study just as for GaN growth [Juz38]. The powder X-ray diffraction file for InN was collected and is in curre nt use as PDF#02-1450. Several reports are

PAGE 27

4 found from the 1950s and 1960s about synthesi s of InN [Juz56, Ren58, Pas63, Sam69]. The growth methods were mainly interacti on of In compounds with ammonia or thermal decomposition of complex single precursors c ontaining direct bonding between In and N. InN powder or small crystals were usually obtained as a result. In 1972, Hovel and Cuomo [Hov72] deposited InN f ilms on sapphire and silicon by reactive RF sputtering in the growth temperature range 25 to 600 C. Hall mobility, n-type carrier concentration, and resistivity were measured as 250 50 cm2/Vs, 5 to 8 x 1018 cm-3, and 3 to 5 x 10-3 cm, respectively. Trainor and Rose grew InN by reactive evaporation and a bandgap of 1.7 eV was measured [Tra74]. Osamura re ported the entire range of InGaN by an electron beam plasma technique that used an electron beam to heat and evaporate In and Ga and a plasma to create a nitrogen dc discharge [Osa72, Osa75]. The grown InN showed bandgap energy of 1.95 eV at r oom temperature and 2.11 eV at 78 K. Puychevrier and Menoret [Puy76] grew InN by reactive cathodic sputtering and the bandgap of InN was measure as 2.07 eV at room temperature and 2.21 eV at 77 K. Then, in 1984, Tansley and Foley [Tan84] reported RF sputtering growth of InN with N2. Very high Hall mobility at room te mperature and very low background carrier concentration were measured as 2700 cm2/Vs and 5.3 x 1016 cm-3, respectively. It should be noted that even with todays advanced techniques of MOVPE and MBE, these values are not easy to obtain. In addi tion, there is no other report of InN grown by RF sputtering that demonstrated such high Hall mobility and low carrier concen tration. The typical mobility and carrier concentration of InN grown by sputtering is 100 cm2/Vs and 1020 1/cm3, respectively. The problem of Hall measurement is that sometimes it can give

PAGE 28

5 inconsistent values when the film is not uniformly deposited. However, the reason is unclear at this point. Epitaxy of InN by HVPE was first reported in 1977 [Mar77]. Indium trichloride powder was used as the indium source and am monia was used as the nitrogen source. The obtained epitaxial InN film showed background carrier concentration and Hall mobility as 2 x 1020 to 8 x 1021 cm-3 and 30 to 50 cm2/Vs, respectively. The optimal growth temperature was found to be 600 C and there was no growth at temperature higher than 670 C. Indium trichlor ide powder [Tak97b, Mar77, Sun96], indium monochloride, or both form in situ by flowing HCl over In metal depending on the source temperature [Tak97a, Sat94] grew InN by HVPE using predominantly InCl3, or InCl and showed ambiguous results on the superior precursors. Kang reported successful H-MOVPE growth of InN in a chlorinating environment [Kan04]. In this approach, the conditions fo r selective etching condi tion of In metal in addition to protecting InN film were found by thermodynamic analysis and experiments. As a result growth of InN film without evid ence of In metal dropl ets was achieved at a relatively low (2500) V/III ratio. During this study, columnar structured InN crystals were observed. It demonstrated the possibi lity of InN nanorods growth and greatly influenced the present study. MOVPE and MBE have been widely used to grow InN since the late 1980s. Single crystal InN growth by MOVPE was reporte d using trimethylindium (TMIn) and NH3 as the precursors [Mat89, Wak89, Wak90]. However, high Hall mobility ( i.e., > 300 cm2/Vs) and low background carrier concentration ( i.e., < 1019 cm-3) were not achieved until recent years.

PAGE 29

6 InN was also grown by MOMBE using TM In instead of pure In metal source [Abe93]. Electron Cyclotron Re sonance (ECR) was used to ge nerate a nitrogen plasma. The obtained Hall mobility and electron carrier concentration were 100 cm2/Vs and 1020 cm-3, respectively. Aderhold [Ade01] reported improved InN film quality by MOMBE. Hall mobility and background carrier concentration were 500 cm2/Vs and 8.8 x 1018 cm-3, respectively. The reported Hall mobility and ca rrier concentration were plotted in Figure 1-1 with recent MOVPE data [Bhu03, Yam 02, Yam04, Yam06]. Hall mobility and the carrier concentration of MOVPE grown InN were 1100 cm2/Vs and 4.5 x 1018 cm-3, respectively [Yam06]. Figure 1-1. Reported values of electric proper ties of InN (a) Hall mobility and (b) Carrier concentration with calendar ye ar [Bhu03, Yam02, Yam04, Yam06]. Interest in InN has increased since 20 02 due to the ambiguous fundamental energy bandgap. The reported bandgap energy value of InN has a wide spread: ~ 0.7 [Inu01, Dav02, Wu02, Sug03, Bri04, But05], ~ 1.1 eV [I nu01, This work], and ~ 1.9 eV [Hov72, Tya77, Nat80, Tan86, Wes88, Sul88, But02, Mot02, Had03] are commonly reported values. Figure 1-2 shows some of the InN bandgap data representing the spread of the values [But05b]. (a) (b)

PAGE 30

7 The reason for the controversial bandga p is not well established, although many believe the inclusion of oxygen (In2O3 or any form) may be responsible. High crystalline quality InN is therefore essential to determin e the controversial prope rty and clarify this ambiguity. Figure 1-2. Reported bandgap energy of InN as a function of carrier concentration [But05b]. The best reported InN films grown by MOVPE demonstrated Hall mobility and background doping as 1100 cm2/Vs and 4.5 x 1018 cm-3, respectively [Yam06]. MBE grown InN films showed better values of Hall mobility and carrier concentration as 2050 cm2/Vs and 3.49 x 1017 cm-3, respectively [Lu02a].

PAGE 31

8 InN-based electronic devices were fabricated and preliminary evidence of twodimensional electron gas (2DEG) in a field e ffect transistor (FET) was reported [Sch00, Lu02b]. Problems growing high quality InN ove rlap with those for GaN in that lattice and thermal expansion matched substrates are absent and that effici ent doping, especially p-type, is difficult. The growth of InN, however, is more challenging because of the low growth temperature and the formation of In metal dr oplets. The growth temperature and V/III ratio seem to be the most important growth parameters. Yamamoto investigated the effects of growth temperature and V/III rati o [Yam01]. They found that V/III ratio could be lowered at higher temperature due to the enhancement of the active nitrogen concentration. The suggested growth temperature was 550 to 650 C. The Hall mobility and carrier concentration were reported as 200 to 400 cm2/Vs and 1018 to 1020 cm-3, respectively [Yam01b, Yam02, Yam04]. The Ha ll mobility and carrier concentration are improving with increasing temperature up to 600 C and then showed worse results at higher temperatures. To enhance the decomposition of NH3, several techniques have been used. It is well known that the reactor geometry is very important to grow high quality material primarily for MOVPE, as demonstrated by improvement of GaN quality by the two-flow design [Nak91]. Flow modula tion [Kel00] was used to le verage the advantages of Atomic Layer Deposition (ALD). TMIn and NH3 were provided in turn, or NH3 was kept flowing and TMIn was provide in a pulsed manner. Plasma-assisted MBE and MOVPE [Wan06, Che06, Wu06, Wak89, Wak90, Sa t97] and laser-assisted MOVPE

PAGE 32

9 [Yam06, Li94, Bhu02, Bhu03] were also utilized to enhance ammonia decomposition and showed some promising results. This overview of GaN and InN thin film growth and the co rresponding electrical and optical properties demonstrates the advanc es that have been made and their potential commercial applications. It is also clear that GaN has rece ived considerably more study than InN. 1.2 Literature Review 1.2.1 Equilibrium Analysis of GaN and InN 1.2.1.1 Thermochemical Data for GaN and InN Studying phase stability is essential for predicting process limits because semiconductors grown at high temperature of ten experience thermal annealing processes for metallization, remedying implantation da mage, or dopant activation. In addition, GaN, AlN and their alloy have applications for high temperature devices thanks to their high temperature stability. Equilibrium calculations can provide pr eliminary information on appropriate processing conditions. These calculations re quire thermodynamic property values for all species and phases. Unfortunately, the prope rties for Ga-In-N system are not that well defined, especially the standard enthalpy of formation. The reported heats of formations of GaN and InN are quite scattered [Sed06]. The reported heat of formation of GaN ranges from -109.62 [Hah40] to -156.8 kJ/mol e [Ran00] and that of InN from -71.0 [Lei04] to 138.072 [Bin02]. Unland and coworkers recently reported th ermochemical data for GaN powder and measured the decomposition temperature as 1110 10 K by dynamic oscillation Thermogravimetric Analysis (TGA) and is othermal stepping T GA [Unl03]. Based on

PAGE 33

10 their assessment relative to other reported va lues, the experimental data from Unland et al. [Unl03] were adopted in this st udy for the GaNbase calculations. A similar method (TGA) was applied to InN and the) 15 298 (0K Hf, ) 15 298 (0K Sm, Cp and decomposition temperature of InN were determined [Ond02]. The equilibrium decomposition temperature may be lower than the experimentally observed temperature because of the kine tic barriers; however, the equilibrium decomposition temperature cannot be higher th an experimental data unless there is measurement error. The InN decomposition te mperature (773 K) as measured with TGA by Onderka [Ond02] is higher than Leitn ers drop calorimetry value (686 K) [Lei04]. Considerable uncertainty, however, still exists in the thermochemical data for InN. The data from SUB94 and [Lei04] will be cautiously used for InN in this study since rigorously assessed data is still not available. The reported thermochemical values for GaN and InN are tabulated in Table 1-1. 1.2.1.2 Equilibrium Calculations of GaN Growth by HVPE Koukitu [Kou98] performed chemical e quilibrium calcu lations analytically by considering only a limited numb er of species. They calculated partial pressures of the gaseous species in equilibrium with Ga N during HVPE growth with respect to temperature, input GaCl partial pressure, input V/III ratio, and H2/inert ratio. By comparing the equilibrium of Ga containing gas phase con centration with input GaCl partial pressure, the driving force for the deposition was calculated with the definition: ) (30 GaCl GaCl GaCl GaP P P P (1) The factor that represents the ratio of the H2 carrier with the inert carrier was defined as F and was given as:

PAGE 34

11 IG HCl NH H HCl NH HP P P P P P P F 2 / 1 2 / 3 ) 3 2 ( 2 / 13 2 3 2 (2) It should be noted that a way of dealing with NH3 partial decomposition was introduced here with a factor such as: NH3(g) (1-)NH3(g) + /2N2(g) + 3/2H2(g) (3) where is the mole fraction of the decomposed NH3. The value of was measured as = 0.03 at 950 C by mass spect roscopy [Ban72], but is a stro ng function of temperature. Figure 1-3 (a) shows plots of the driving force for the deposition with temperature with different F values [Kou98]. It was observed that PGa (the driving force) decreased with increasing F. Figure 1-3. Examples of equi librium calculations. (a) Drivin g force for the deposition as a function of growth temperature with various parameters for F. Total pressure: 1.0 atm, input partial pressure of GaCl: 5 x 10-3 atm, input V/III ratio: 50 and : 0.03. (b) Comparison between calculated growth rates [Kou98] and experimental data [Usu97]. The growth rate with driving force for the deposition was defined as the following formula: (a) (b)

PAGE 35

12Table 1-1. Reported values of thermochemical data for solid GaN and InN. Material ) 15 298 (0K Hf (kJ/mole) Method ) 15 298 (0K Sm (J/mole K) Method Tdecomp (K) GaN -109.62 [Hah40] -156.8 [Ran00] -156.8 16 [Unl03] -161.56 [Sed06] Combustion calorimetry Drop calorimetry TGA DFT calc 30 4 [Unl03] 36.1 [Sed06] Debye-Einstein DFT calc 920 [SUB94] 1052 [Dav01] 1435 [Lei03] 1110 10 [Unl03] InN -144.6 [Mac70] -133.8 6.3 [Vor71] -137.2 18.8 [Vor73] -130.6 [Gor77] -132.7 6.7 [Jon87] -138.07 [Bin02] -71.0 [Lei04] -78.64 [Sed06] High pressure equilibria Knudsen effusion MS Knudsen effusion MS Knudsen effusion MS Static pressure measurement Drop calorimetry DFT calc 53.7 7.1 [Vor71] 32.3 20.9 [Vor73] 51.6 [Gor77] 54.4 8.0 [Hon87] 31.6 3 [Ond02] 42.51 [Lei04] 42.5 [Sed06] Debye-Einstein PDOS/assessed DFT calc 1211 [SUB94] 773 5 [Ond02] 638 [Lei04]

PAGE 36

13 Ga gP K r (4) where r = growth rate, and Kg is the mass transfer coefficient. By adjusting the mass transfer coefficient ( i.e., Kg = 1.18 x 105 m/h atm in this case), the growth rate results matched well with experiment al data as seen in Figure 1-3 (b). Thus, it was concluded that GaN growth by HVPE is thermodynamically controlled under these ranges of growth conditions. Following similar procedures, Kumagai anal yzed HVPE growth of InN by InCl and InCl3 precursors [Kum01]. The results showed that InN growth is di fficult using InCl but possible with InCl3 along with inert or low H2/inert carrier gas as shown in Figure 1-4 [Kum01]. It could be seen that the driving force has negative values in an H2 carrier (F = 1.0), while it has slightly positive valu es in an inert carrier (F = 0.0). Figure 1-4. Driving force fo r the deposition of InN (PIn) using InCl (HVPE) and InCl3 (THVPE) as a function of growth temperature. The calculation was performed for growth unde r inter gas (F = 0.0) and hydrogen carrier gas (F = 1.0) conditions. [Kum01].

PAGE 37

14 It should be noted that no report was found that dealt with adjustable Cl/III ratio, but only at fixed values since either monochloride (Cl/III = 1) or trichloride (Cl/III = 3) was used in HVPE and THVPE techniques. Continuous variation of Cl/III ratio is possible in H-MOVPE by simply adding an independent st ream of a Cl containing species ( i.e. HCl). Thermodynamic analysis of H-MOVPE technique including various Cl/III ratios is presented in Chapter 2. 1.2.2 Stress in GaN Films Stress measurements and modeling of Ga N films is important because GaN is usually grown on foreign substrates due to th e lack of bulk GaN crystals. The traditional approach to analyze the stress in GaN films was not satisfactory because there were some cases when the expected stress occurred in an exactly opposite manner. For example, both compressive and tensile stressed GaN can be grown on SiC [Per97, Dav97, Dew98, Rie96, Wal99]. Therefore, it could be concl uded that the stress in the GaN film is not only related to the lattice and thermal expans ion mismatch, since both give stress in GaN, but also to other important factors. The stress in GaN is mainly related to th e lattice and thermal expansion mismatches with the substrate and the in itial surface properties, whic h result in a growth mode change. The origins of tens ile stress were identified as lattice and thermal expansion mismatches between GaN and substrate, Si or Mg doping, or grai n boundaries of island coalescence [Kro03]. Likewise the origin of compressive stress can be lattice and thermal expansion mismatches between GaN and substrate. The island growth mode should be avoided as it often cr eates polycrystalline materials. A buffer layer is usually used to enhance crystal quality by uniform coverage of the surface to prevent island growth.

PAGE 38

15 The traditional approach to modeling the residual stress in GaN film by lattice and thermal expansion mismatches has a flaw because GaN films are typically grown on buffer layers or by a nitridation process. The pregrowth processes change the surface properties significantly. The buffer layers are usually amorphous and the nitridated surface of sapphire is also an amorphous AlOxNy layer. Therefore, the traditional heteroepitaxy modeling such as Frank-van der Merwe and Matthews [Mer50, Mat75] cannot be applied since the existence of an amorphous interface was not considered. In addition, the lattice mismatch ( i.e., ~ 14 % for sapphire, ~ 21 % for Si, or even ~ 3.5 % for SiC) is very large; dislocations will form at the interface immediately. A visible product of the incor poration of significant stress in a film is cracking. Etzkorn and Clarke investigated the cracki ng of thick GaN films on sapphire substrate [Etz01]. A viable mechanism for cracki ng was identified as island coalescence. A model was developed to deduce the maximum thickness of crack formation with given stress. According to the model, 16 m thick crack-free GaN c ould be grown on sapphire with tensile stress in 0.14 GPa. Strain en ergy calculations relate d to crack generations were carried out by measuring the bending mome nt of GaN and sapphire. It was argued that tensile stress in GaN film was generate d at the growth temper ature after reaching a critical thickness. Kinetically limited crack healing mechanism was suggested after crack generation. Waltereit reported that GaN films grown directly on 6H-SiC showed no stress, while GaN films grown on AlN/6H-SiC showed compressive stress, although the lattice parameters and thermal expansion coefficients of AlN and SiC are similar [Wal99]. The lattice mismatch induced 3.4 % compressive stress was fully relieved in GaN directly

PAGE 39

16 grown on SiC, whereas 0.3 % compressive stress still remained in GaN grown on AlN/SiC even beyond 1 m thick. It was c oncluded that the stre ss in GaN film was mainly determined by its growth mode rather than lattice mismatch in this case. The growth mode is mainly related to the initial surface properties of the substrate. The results suggested that the adva ntage of the buffer layer is not only in averaging the lattice and thermal mismatches, but also enhanci ng the surface property. Schematic of GaN directly grown on 6H-SiC and using AlN buf fer layer is shown in Figure 1-5. Figure 1-5. Schematic of GaN growth on SiC: (a) GaN/SiC and (b) GaN/AlN/SiC growth. Note that the strain is fully relieved in (a), whereas it is only partially relaxed in (b) [Wal99]. Three dimensional growth (Volmer-Weber) governed the GaN growth without buffer layer and a number of disloc ations were shown, while tw o dimensional growth (Frankvan der Merwe) can occur with few disl ocations with AlN buffer layer [Wal99]. 1.2.2.1 In situ Stress Measurements To understand the stress evolution in GaN films, in situ stress measurements have been carried out using a Multibeam Optical Stress Sensor (MOSS) system [Flo97, Hea99]. This method determines the wa fer curvature by measuring a laser beam deflection, which is proportional to the stress. It was used to in situ measure the stress in AlGaN film grown on Al N/SiC [Aco04]. They found th at AlGaN film was initially under compressive stress and evolved into tens ile stress as the film thickness increased. The transition (from compressive to tensil e stress) thickness of AlGaN film depended on

PAGE 40

17 the AlN buffer layer growth c onditions and AlGaN fi lm Al mole fraction. For example, when V/III ratio was varied from 750 to 10600, the AlGaN initial compressive stress was varied from 1.9 to 8.7 GPa. GaN film did not show a transition to te nsile stress up to 3.8 m thick, whereas more Al content AlGaN s howed rapid transition to tensile stress. The persistent tensile st ress in MOCVD GaN on sapphire using both LT-AlN and LTGaN buffer layers were observed by MOSS at growth temperature [Hea99]. The origin of tensile stress may be from the grain boundari es of islands. It was also found that thermal annealing or temperature cycling doe s not reduce tensile stress in the film. Because the thermal expansion mismatch effect is much larger, even tually the GaN film on sapphire showed compressi ve stress after cooling. Raghavan and Redwing [Rag04] measured the persistent tensile stress in AlN grown on Si in situ across the wide temper ature range 600 to 1100 C. A sharp drop in the tensile stress was observed from 1 to 0.4 GPa while decreasing the AlN growth temperature below 800 C. This represented the transition of epitaxial AlN film to polycrystalline. GaN film was consecutivel y grown on AlN/Si. The GaN film showed slightly compressive stress at 1100 C. Krost et al. measured the stress in the GaN films on Si(111) in situ and successfully grew crack-free 7 m thick GaN on Si(111) with multiple AlN interlayers [Kro05, Dad04, Clo04, Kro03]. The sources of tensile stress were identified as grain boundaries from island coalescence (0.2 GPa/m) and Si-doping (1.6 GPa/m ) [Kro05]. The nonuniformly deposited SiN was called an in situ mask and high quality GaN could be obtained on the in situ SiN mask/AlN/Si. The in situ curvature measurement results for GaN grown on SiN/AlN/ Si are shown in Figure 1-6 [Kro03].

PAGE 41

18 Figure 1-6. In situ curvature measurements during th e growth of a ~ 6 m thick GaN layer showing the influence of AlN inte rlayers on the curvature. The sample was crack-free after growth [Kro03]. Evolution of the curvature was observed with inclusion of LT-AlN interlayers. When LT-AlN was applied, a tempor ary compressive stress was obs erved as shown in Figure 16. Nevertheless, the film was under persiste nt tensile stress as the thickness increased. Based on the results presented in Figure 1-6, it is evident that AlN has an impact for only a small amount of subsequent growth. A lthough this AlN interlayers may improve the crack-free thickness, the growth of thick (> 100 m) crack-free GaN in this way should have inherent limit. 1.2.2.2 Ex situ Stress Measurements The surface of HVPE and H-MOVPE grown GaN is often very rough. Commonly used non-destructive methods to measure the st ress in rough films or the films that have structures on them are XRD reciprocal sp ace mapping and Raman spectroscopy. In this case, the stress measurement is taken after th e sample has cooled to room temperature. XRD Reciprocal Space Mapping. The lattice parameters of a film can be precisely measured using an XRD reciprocal space mappi ng. It was observed that GaN showed strong in-plane strain although out-of-plane strain was neglig ible [Eip05]. Considerable

PAGE 42

19 thermal compressive stress was found from GaN grown by ion beam assisted MBE on LiAlO2 (100). It was found that there was 3 tim es greater compressive stress in the inplane than out-of-plane during measurement by HT-XRD in the temperature range 25 to 600 C (HT-XRD chamber temperature) [Eip05] Stress in GaN was also measured on ZnO substrate by HR-XRD [Min04]. In accordance with the in situ stress measurement results, the change of buffer layer thickness changed the film stress significantly. Thus, buffer layer material should be carefully se lected and the thickness of the buffer layer should be controlled to redu ce stress in the final film. Raman Spectroscopy. Raman spectroscopy is a well-esta blished, non-destructive, and convenient technique to detect the stress in the film, alt hough the typical accuracy of stress measured by Raman spectroscopy comp ared to ~ 5 % for HR-XRD is about 20 % [Dav97]. Among the all allowed Rama n modes, it is known that the E2 mode is strongly related to biaxial stress in Ga N [Dav97]. Between the two E2 peaks, the higher frequency mode is usually used due to the higher intensity. Biaxial stress in GaN sample can be measured by comparing the position of E2 (high) peak with that of a stress-free sample. Table 1-2 lists the Raman E2 (high) peak positions reported in the literature [Dav97, Mel97, Gil00, Kry99a, Dav97, Yam00, Dem96, Yoo97, Ben02]. It is noticed that even freestanding or high pressure grown bulk GaN samples do not have unanimous Raman E2 (high) peak that can be used for stress-free standard. The E2 mode of freestanding GaN ranged from 565 to 570 cm-1. Therefore, deciding the value of stress-free E2 (high) peak position is impossible at this point. Generally 568 cm-1 is considered as the stress-free GaN E2 (high) peak position because it is about average value from freestanding GaN samples. Therefore the stress in GaN can be determined by comparing the E2 value with

PAGE 43

20 568 cm-1. If the measured E2 frequency is higher than 568 cm-1, the film is under compressive stress qualitatively, and vice versa. Table 1-2. Reported values of the Raman E2 (high) peak positions. Sample description Raman E2 (high) peak position (cm-1) Reference Freestanding (300 m) by HVPE Substrate: Al2O3 568 [Dav97] Freestanding (100 m) by HVPE Substrate: SiC, no buffer 565 [Mel97] Freestanding by high N2 pressure 565 [Gil00] Freestanding on LiGaO2 by H-MOVPE 570 [Kry99a] 0.5 to 3 m GaN/AlN( AlGaN)/SiC 565 [Dav97] 1.8 m on AlN/Al2O3 by MOCVD 566 [Yam00] V/III (2000 to 6000) MOCVD 569 (V/III = 6000) to 571 (V/III = 2000) [Dem96] 1.5 m GaN/(10 to 85 nm)LT-GaN/cAl2O3 by MBE 565 to 572 [Yoo97] GaN/misoriented Al2O3 by MOCVD 570 to 571 [Ben02] The quantitative value of stress can be obtained using a linear proportionality factor. This quantitative ar gument may be meaningless in some sense since the stressfree value is unknown. However, two sa mples can be quantitatively compared by assuming one sample is stress-free. The reported value of this proportionality factor ranges from 2.4 to 4.2 cm-1/GPa [Dav97, Kis96]. Thus, further study is required to conclusively determine the stress from Raman peak shift quantitatively. One may expect compressive stress to be present in GaN grown on Al2O3 because the lattice and thermal expansion mismatches leads to compressive stress. However, it is not straightforward to predict whether th e overgrown GaN will have compressive or tensile stress. The Raman spectroscopy resu lts showed that the grown GaN film on cAl2O3 is less compressively stressed with increasing V/III ratio [Dem96].

PAGE 44

21 Moreover, the Raman E2 peak significantly shifted fr om 572.0 (compressive) to 565.2 cm-1 (tensile) while changing th e buffer layer thickness from 10 to 85 nm as shown in Figure 1-7 [Yoo97]. Figure 1-7. Raman peak shifts of 2 m th ick GaN with different LT-GaN buffer layer thickness; LT-GaN buffer layer thickness (a) 10 nm, (b) 50 nm, (c) 75 nm, and (d) 85 nm. [Yoo97]. This tensile to compressive st ress change is expected becau se uniform coverage of the surface at an optimal buffer layer thickness w ill lead to compressive stress from 2-D growth rather than 3-D isla nd growth that create tensile stress at the grain boundary. Since Raman spectroscopy is sensitive to the stress in the GaN film, it was used as a major technique to measure the stress in the GaN films grown on sapphire substrate and the results will be presented in Chapter 3. 1.2.3 GaN Growth on Si One of the advantages of using Si substrate instead of sapphire is its availability in large diameters. Another adva ntage of Si is that the heat generated from the device can more easily drain because Si has much hi gher thermal conductivity (145 W/m K) than sapphire (5.43 W/m K). GaN fi lms are usually grown on Si (111) substrate because the hexagonal geometry ( i.e. three-fold symmetry) of the Si (111) surface is better matched

PAGE 45

22 with the basal plane of GaN. The growth of GaN on Si ( 100) would be more challenging due to the different crystallogr aphic structures. For exampl e, it was often observed that a mixture of cubic and hexagonal GaN grown on Si (100) substrate by MBE [As00]. GaN grown on Si (100) shows multiple in-plane alig nments mainly compos ed of two in-plane domains rotated by 30 [Leb00, Sch04b]. GaN is usually grown by MOCVD on Si si nce it gives the most promising results and is suitable for mass production. HVPE can also be used to grow GaN on Si. The report of HVPE GaN growth on Si is, however limited [Mas06, Zha05, Mes05, Yu05]. A maximum crack-free thickness of HVPE GaN film grown on Si is 1.5 m on 2 inch Si(111) [Zha05]. The results suggested that the crack generation is not much related to growth chemistry, growth rate, or reactor geometry. Instead, it is more related to lattice mismatch and thermal expansion coefficient mismatch among GaN, buffer layer, and Si. The driving force for crack is the strong tensile stress in the film caused by lattice and thermal expansion differences between Ga N and Si. The origin of tensile stress within GaN film on Si substrate is not only from lattice and thermal expansion mismatches, but also from island coalescence boundaries due to the initial growth from small crystallites [Dad03]. Along with the brittleness of Si, sometimes cracks can penetrate into Si over 1 m depth [Zam02]. A number of techniques su ch as AlN buffer layer [Da d03, Lia00, Sch05], AlN/SiC buffer [Lia00], AlGaN buffer layer [Wan05] multiple AlN interl ayers [Kro03], AlGaN graded layer [Mar01, Abl05], AlGaN/HT-AlN [Sch05], AlGaN/GaN superlattice [Lia01], SiO2 patterning [Det01, Zam01, Zam02, Na o03], and Si doping in GaN [Yu06] were

PAGE 46

23 used to reduce the crack density in GaN on Si Table 1-3 lists some of the techniques used for crack-free GaN growth on Si. Table 1-3. Techniques used for growth of crack-free GaN on Si. Crack reduction technique for GaN grown on Si Growth T (C) Crack-free thickness (Crack-free area) Ref AlN/SiC buffer layer AlGaN/GaN superlattice on AlN/SiC/Si 1050 1.3 to 1.5 m (10 cm diameter) [Lia00] [Lia01] Multiple AlN interlayer on AlN buffer layer with in situ curvature measurement 1145 7 m [Kro03] AlGaN graded layer on HT(1160 C)AlN (100 nm) buffer layer 1050 2 m [Abl05] In situ SiN mask (1.5 mono layer) on Si doped AlN buffer layer(25 nm) 1165 0.75 m [Dad03] Al0.58Ga0.42N buffer layer (180 nm) at 800 C 1000 0.4 m [Wan05] AlGaN graded layer on AlN buffer layer (10 to 500 nm) 1050 0.7 to 1.6 m [Mar01] 9 m (5 m x 10 m) SiO2 patterning 1100 10 m (100 m x 100 m) [Det01] Lateral confined epitaxy (prepatterned Si) 1020 0.7 m (100 m x 100 m) [Zam02] Si doped GaN on AlN buffer layer (100 nm) 1000 < 1 m [Yu06] Crack-free 1.3 to 1.5 m GaN was successfully grown on 10 cm diameter Si (111) using AlN and SiC buffer layers [Lia00]. Fu rther study showed that high crystal quality and smooth surface were achieved by inserti on of AlGaN/GaN supe rlattice interlayers [Lia01]. The results showed that large di ameter GaN growth on Si is promising. The typical crack-free thickness of GaN gr own on large area Si is < 2 m except one report showed 7 m crack-free GaN grown on Si [Kr o03]. However, the growth of large area (> 2 diameter) freestanding (> 100 m thick) GaN on Si was not reported to date. An AlN buffer layer between GaN and Si is most widely used because the thermal expansion difference between AlN and Si is less than GaN and Si. In addition, GaN

PAGE 47

24 films on AlN will possess compressive stress during cooling that may compensate for the tensile stress. Unlike many reports of GaN growth on sa pphire that uses either LT-AlN or LTGaN as a buffer layer, reports of LT-GaN buf fer layer usage of GaN on Si were rarely found ( i.e. [Mas06b]). The benefit of LT-GaN buffer layer on sapphire may be the change of the surface property resulting in tw o-dimensional growth. On the other hand, GaN on Si is under significant tensile stress. Therefore, changing the surface property is necessary, and, in addition, artificial compre ssive stress generation is required to grow crack-free GaN film on Si. It is evident th at AlN and AlGaN are promising buffer layer materials between GaN and Si due to the comp ensation of tensile and compressive stress. It was found that the higher Al concentr ation in AlGaN buffer layer is better for subsequent GaN growth [Wan05]. Table 1-4 lists the structural properties of various substr ates for GaN epitaxy. It can be predicted that what kind of stress the GaN film will be under by comparing the lattice parameter and thermal expansion coeffi cient. Tensile and compressive stress induced by lattice mismatch and thermal expa nsion mismatch are drawn in Figure 3-1. The other widely used technique is the gr owth of an AlGaN graded layer. This method creates gradual changes in the stress in GaN films. In addition, AlGaN graded layers reduce threading disloc ations significantly [Mar01]. Other than stress engineering by buffer and interlayers, SiO2 patterning was also used to reduce cracking problems [Det01, Nao03, Zam01, Zam02] Adopted from Epitaxial Lateral Overgrowth ( ELO), grooved Si substrate was used as the substrate and a thin GaN layer was grown, followed by SiO2 patterning [Det01]. The growth of GaN on

PAGE 48

25 prepatterned Si, termed Lateral Confined Epitaxy (LCE) [Zam01] was also studied. However, this technique did not create more promising results since GaN films tend to bend upwards (concave), alt hough a crack-free region of 100 m x 100 m was reported [Zam02]. Table 1-4. Structural Pr operties of Various Substrates for GaN Epitaxy [Kan04]. Substrate Eg (eV) Lattice parameter a () Lattice parameter c () Lattice mismatch with GaN (%) Thermal expansion coefficient (10-6 K-1) a c GaN wurtzite InN wurtzite AlN wurtzite 6H-SiC wurtzite Al2O3 wurtzite ZnO wurtzite LiAlO2 tetragonal LiGaO2 orthorhombic 3.36 0.7 6.2 2.9 6.8 3.35 6.1 4.1 3.189 3.537 3.112 3.08 4.758 3.21 5.1687 5.402 5.18 5.704 4.98 15.1 12.991 5.21 6.2679 5.007 0 9.8 2.5 3.5 16.1 1.94 1.5 0.9 5.6 4.0 4.2 4.2 7.5 2.9 7.1 6 3.17 3.0 5.3 4.7 8.5 4.8 7.5 7 3C-SiC diamond GaAs zinc blende Si (111) cubic BP zinc blende 2.3 1.42 1.12 2.0 4.36 5.65 5.83 4.54 3.9 19.87 17.0 0.63 2.7 6.0 6.2 Another way of reducing cracks would i nvolve creating weak cohesion between GaN and Si, so that the film can separate du ring the cooling process. For example, ion implantation (N+) was carried out to intentionally create defective layer of AlN/Si substrate [Jam05]. This partia lly isolated the III-nitride layer and Si substrates and helped reduce the strain by up to 84 %. A novel approach of cr eating weak cohesion between GaN and Si using InN as buffer mate rial will be presented in Chapter 4. 1.2.4 Growth of InN Films and Nanostructured Materials 1.2.4.1 Growth of InN Films InN is the least studied material among the group III nitrides that are relevant to UV-visible-IR optoele ctronics. InN has high drift velocity at room temperature, which is

PAGE 49

26 a good characteristic for high speed FET applicat ions [Ole98]. In addition, tandem solar cells with Si at the bottom cell and a higher bandgap material ( i.e., InxGa1-xN 1.85 eV) on the top cell can achieve a maximu m efficiency of 32.1 % [Mat87]. Efforts have been made to grow InN crystals by MOCVD [Wak90, Sat97, Guo99, Yam94, Yam97, Yam99, Pan99, Yan02, Dra06, Mal04, Cha04, Sin04, Yam05, Yam04, Jai04, Los04, Yam05, Mal05, Hua05, Ke l00, Yam01, Sug03], MBE [Abe93, Dav99, Lu01], HVPE [Iga92], laser assi sted CVD [Bu93], as well as other less used processes. Most InN films were grown by MOCVD because MOCVD is a well-established technique especially when mass production is considered. The growth of high quality InN, however, was impeded by several factors. Low (< 700C) growth temperature is required due to the very high nitrogen vapor pressure and strong tendency to form In metal droplets. It is difficult to grow high quality material due to the reduced adatom mobility on the surface at low growth temperature. In addition, due to the high kinetic barrier for breaking N-H bonds of NH3 [Seg04, Yum81, Liu78], excessive NH3 flow is required to compensate for the amount of active nitrogen. Excessive NH3 will form H2, and H2 is known to make InN growth less thermodynamically favorable [Kou99, Hua05, Dra 06, Joh04j]. Thus, the growth rate of InN is very low compared to other group III nitrides. Another obstacle is the lack of suitable substrate for In N growth. Sapphire and Si have been widely used substrates for InN because sapphire is a good substrate for group III nitrides in that the surface of sapphire can be changed to AlOxNy by nitridation. Si has numerous advantages including the possibility of fabrica ting tandem solar cells. The lattice mismatch between InN (0001) and Si (111) is about 8 %, while the lattice

PAGE 50

27 mismatch between InN (0001) and c-Al2O3 is about 26 %. It should be noted that when a lattice mismatch is greater than 0.8 %, disl ocations will form at the interface. Thus stating that an 8 % lattice mismatch is be tter than a 26 % mismatch with InN does not have much significance when cr itical thickness is concerned. It was found in early work that although Si has a smaller lattice mismatch than sapphire, the films grown on sapphire had superi or quality after nitr idation [Yam94]. Nitridation process formed Al N on sapphire substr ate [Ven99] and AlN has a 13 % lattice mismatch (c.f. Al2O3 has 26 % lattice mismatch with InN). SixNy formation is known to prevent epitaxial growth of GaN on Si [Plo99]. Thus nitridati on of Si substrate should be avoided. InN films were grown on Si(111) substrat e without nitridation [Yan02]. To avoid SiN layer formation, TMIn was first introduced before NH3. MOCVD with double zone heating was used to facilitate thermal cracking of NH3. Growth temperature was varied from 350 to 600 C and a maximum of 6 m/hr growth rate was achieved although the films over 1 m thick cracked. For InN growth on sapphire, the optimal n itridation temperature and times were ambiguous. For example, one report show ed best results with 1000 C for 30 min [Pan99] and the other report proclaimed 1075 C for 5 min gave best results [Dra06]. To elucidate the effect of nitridation, th e most useful method may be to observe the surface evolution during the growth in situ From in situ spectroscopic ellipsometry study of InN growth by MOVPE [Dra06], it was found that the nitridation at 1050 C for 45 sec, or 1000 C for 300 sec would give the be st quality film. It may be concluded that lower nitridation temperature re quires longer nitridation time.

PAGE 51

28 The optimal growth temperature of InN is another ambiguity. In fact, the temperature is not an exact value in pr actice because it depend s on the places of thermocouples, thermocouple types, or the temperature measurement methods. Therefore, ~ 25 to 50 C of temperature error is possible. The highest reported growth temperature for InN was 750 C [Tak97]. The lowest growth temperature is not clearly reported because of the low quality of the ma terial, but it may be approximately 350 C [Yan02] for MOCVD. T ypical InN growth temperat ure is around 450 to 650 C. It seems that the crystal growers golden rule, the higher growth temperature, the better crystal quality, holds for InN case as well. The highest Hall mobility was reported from the sample grown at 600 C [Kel00]. The growth temperature at 650 C gave the best surface morphology [Yam01]. In N films grown at 620 C showed the best optical properties, whereas the samples grow n at 600 C exhibited the best electrical properties. Some reported that high growth temperature is destructive to InN because it begins to decompose. For example, surf ace morphology of the InN was best at 550 C during the 520 to 590 C experiments [Yan02]. According to the other experiments, 560 C grown samples showed the best quality among the 540 to 580 C [Jai04]. Thus, the optimal growth temperature determination is not clear yet. It should be noted that optimal growth temperature is well above the decomposition temperature of InN as in the GaN case. This result showed that InN d ecomposition and growth temperatures have a wide range of overlaps, a bout 230 C (520 C decompositi on [Dra06] to 750 C growth [Tak97]). To find optimal growth conditions for InN film by H-MOVPE, exploratory study was carried out and the results are presented in Chapter 6. It should be noted that during

PAGE 52

29 the growth study in chlorinating environment [Kan04], InN columnar structured materials were observed and the growth conditions provi ded a good starting point for the growth of nanostructured materials. 1.2.4.2 Growth of InN Nanostructured Materials The growth of InN nanostructured material is interesting sinc e the hope is that devices made of nanostructures will have superior efficiency due to quantum confinement effects. The growths of InN na norods and nanowires have been reported by several researchers [Ji05, Kim03, Che05, Qia05, Luo05, Cha05, Lia02, Yin04, Sch04, Lan04, Tan04, Sar05, Joh04, Yin04, La n04, Vad05, Zha05, Shi05, Tak97]. InN nanowires have usually been grown by thermal catalytic CVD that usually uses Au as a catalyst. For example, Liang et al. [Lia02] synthesized InN nanowires on gold patterned p-type Si (100). Pure indium foil and NH3 were used as an indium and nitrogen sources, respectively. The growth was perf ormed at 500 C for 8 hrs and 40 to 80 nm diameter nanowires were created. The band gap was 1.85 eV measured by PL. Zhang et al. [Zha02] fabricated anodic alumin a membranes (AAM) and deposited Au catalyst to grow InN nanowires by elec trodeposition. The indium and nitrogen sources were pure indium powder and amm onia. Despite the complex substrate preparation and the rather long growth time (12 hr), the InN nanowires were polycrystalline with rough surface morphology. Lan et al. [Lan04] produced InN nanorods using pure indium powder and NH3 as the indium and nitrogen sources in a hot wall quart z reactor. Au was sputtered on Si(100) substrates as the catalyst and good quality single crystal InN nanorods were produced. The band gaps of 0.766 eV and 1.9 eV were observed depending on the diameter of nanorods. It was still arguable why there exist two distinct values.

PAGE 53

30 The typical sources of In were In me tal [Vad05, Zha05, Joh04, Tan04, Lan04, Lia02], In2O3 [Tan04, Sch04, Lan04, Yin04, Luo05], InCl3 [Tak97, Kim03], or TMIn [This work]. Sometimes a novel precursor, such as indium acetylacetonate, was used [Sar05]. Ammonia was mainly used as the nitrogen source since most other nitrogen sources are extremely flammable. A less flammable nitrogen source, monomethylhydrazine (MMHy), wa s occasionally used [Tak97]. An external catalyst was not used to grow InN nanostructures in some cases [Che05, Luo05, Joh04, Vad05]; however it is possible that In liqui d acted as a selfcatalyst. For example, Johnson et al. [Joh04] reported InN nanowires on the quartz surface and indium metal surface. Indium metal and NH3 were also used as the indium and nitrogen sources, respectivel y. The growth was carried out at 700 C for 2 hrs and 50 to 100 nm diameter nanowires were ge nerated. Although th e morphology of the nanowires shows rather rough su rface, the nanowires were single crystal and the band gap was 0.80 eV as measured by PL. The growth axis was usually along the [110] direction [L ia02, Yin04, Lan04, Tan04] and sometimes the [001] direction [Kim03, Joh04], sugges ting that there was more than one growth mechanism that governs nanostructure growth. Nanostructures can be grown in a relativ ely wide range of temperature. For example, nanowires were grown at T = 600 to 750 C [Luo05, Sch04, Tak97], 565 to 590 C [Yin04], 525 C [T an04], 500 C [Lan04]. Since InN nanowires may be of higher qua lity than InN films, there was strong motivation to determine the bandgap. Ho wever, the bandgap measured from InN nanowires still showed wide ranges, although a majo rity of the results ce ntered at 0.8 eV.

PAGE 54

31 Both 1.9 eV and 0.8 eV bandgaps were detect ed as evidenced by the brown colored (30 to 50 nm diameter) and black colored (50 to 100 nm diameter) nanowires [Lan04]. In an other case, both room temperature PL and op tical absorption spectru m showed 0.8 eV of bandgaps from InN nanowires [Joh04a]. The bandgap was also measured from InN nanowires and two distinctive peaks of 0.8 eV and 1.9 eV were detected from the InN and In2O3 mixture [Vad03]. A higher bandgap (1.9 eV) was still detected from InN nanowires grown on Anodic Alumina Membrane (AAM), using direct reaction of In metal with NH3 [Zha05]. Nanowires grown by nitridation of In2O3 powders without catalyst showed a 1.7 eV bandgap [Luo05]. Another medium bandgap of 1.1 eV (FWHM 105 meV) was measured by PL from aligned wurtzite polycrystalline InN nanofingers [Ji05]. A bandgap of 1.1 eV was also obtained by calculation from InN nanotubes based on Density Functional Theory (DFT) considering stability and electronic structures of single walled (SW) InN nanot ubes [Qia05]. On the other hand, other calculations showed 0.8 to 0.9 eV [Bec02] from wurtzite InN. Further investigation is requi red to conclude th e correct bandgap of InN as well as the reasons for the wide range of bandgap values. In this study, InN nanostructured material growth optimizations were carried out by H-MOVPE and the results are presented in Chapter 5. As an exploratory work, GaN nanostructu red material was tested to grow by HMOVPE and some results are shown in Chap ter 7 followed by recommended future work in Chapter 8.

PAGE 55

32 CHAPTER 2 CHEMICAL EQUILIBRIUM ANALYSIS OF H-MOVPE SYSTEM 2.1 Introduction Although the Hydride Metalo rganic Vapor Phase Epita xy (H-MOVPE) technique has proven to be a promising and versat ile technique to synthesize III-V compound semiconductors, the chemistry of H-MOVPE is not currently well understood. Experimental definition of the gas phase and heterogeneous reac tions are noticeably absent, mainly due to the difficulty of deali ng with corrosive gases, such as HCl. In addition, hot furnace walls and wall depositi on during growth also obstruct the use of optical in situ measurements. An alternative wa y to study the chemical reactions involving the H-MOVPE t echnique is through simulations, and the most straightforward is chemical equilibrium analysis. Thermodynamic analysis can give insight into which gas phase and condensed phase species will be present without detailed knowledge of molecular structures or reaction kinetics. Because reaction kinetics is excluded, the result in some cases may not represent the actual chemistry. It can, however, give an approximate idea of which species will be dominant under various conditions. Reed [Ree02] calculated the mole fractions of gas and condensed phase species at equilibrium in the H-MOVPE system for Ga N growth by varying temperature, Cl/Ga ratio, and NH3 flow rates. It was found that th e calculated results matched well with experiments when Ga(l) is excluded from the sy stem. It was also found that Cl acted as a gallium sink and H (from NH3) acted as a carbon scavenger.

PAGE 56

33 In this chapter, more species have been added into Reeds database. The newly incorporated species includes not only Ga, but also In containing species. The database also contains metalorganics and adducts. The aspects of two ma in reactions can be compared: Group III chloride formation versus metalorganic decompositions. One of the goals for this study is to see whether H-MOVPE is clos er to MOVPE or HVPE after metalorganics + HCl reactions. If the resu lts are known, the consecutive chemistry will be clearer because both MOVPE an d HVPE chemistry are well studied. 2.2 Thermochemical Data Collections Great care should be taken when selecti ng and adding new thermochemical data. The inclusion of one erroneous data point or exclusion of one important species may render the calculation results inapplicable. Most data for well-identified traditional species in the Ga-In-C-H-Cl-N system are fr om the Thermo-Calc SUB94 database. Most data for well-known species from SUB94 have passed the self consistency test as well as 2nd and 3rd law verifications with experimental data. Species that are commonly included in a complex equilibrium analysis of subsystems in the Ga-In-H-C-Cl-N system are listed in Table 2-1. In this Table, a number of gas phase species are categorized by Ga, In, and Cl containing species, with other species such as hydrocarbons, and the species that are only composed of C, N, or H. Condensed phases are also organized in the same way as the gas phase. It should be noted that although some species may not exis t in the H-MOVPE system, they were not excluded from the database. Trace species were traditionally excluded to save computation time. Now, with advancements in computing power, there is no need to exclude non-significant species, as long as the correct thermochemical data are used. In

PAGE 57

34 addition to the data from SUB94, critically assessed data for GaN and InN were used from [Unl03] and [Lei04], respectively. Table 2-1. Commonly considered species in Ga-In-H-C-Cl-N system from SUB94 database. Ga containing species Ga, Ga2, GaH, GaCl, GaCl2, Ga2Cl2, Ga2Cl4, Ga2Cl6, GaCl3 In containing species In, In2, InH, InCl, InCl2, In2Cl2, InCl3, In2Cl4, In2Cl6 Cl containing species (excluding III chlorides) CCl, CHCl, CH2Cl, CH3Cl, CNCl, CCl2, CHCl2, CH2Cl2, CCl3, CHCl3, CCl4, C2Cl, C2HCl, C2H3Cl, C2H5Cl, C2Cl2, C2H2Cl2, C2H4Cl2, C2Cl3, C2HCl3, C2H3Cl3, C2Cl4, C2H2Cl4, C2Cl5, C2HCl5, C2Cl6, C6H5Cl, Cl, HCl, Cl2 Other species CH, HCN, CH2, CH3, CH4, CN, CN2, C2, C2H, C2NH, C2H2, C2H3, C2H4, C2H5, C2H6, C2N, C2N2, C3, C3H, C3NH, C3H4, C3H6, C3H8, C3N, C4, C4H10, C4H2, C4H4, C4H6, C4H8, C4N, C4N2, C5, C5NH, C5H8, C5N, C6H6, C6N, C6N2, C7NH, C7N, C8N, C8N2, C9NH, C10N, C10N2, C11NH, C11N, C12H26, H, NH, N3H, H2, NH2, N2H2, N, N2, N3, NH3, N2H4 Gas Phase Ga containing species GaCl3(l), Ga(l), GaCl3(s), GaN(s)*, Ga(s) In containing species InCl(l), InCl3(l), In2Cl3(l), In3Cl4(l), In4Cl7(l), In(l), InCl(s), InCl(s2), InCl2(s), InCl3(s), In2Cl3(s),In3Cl4(s), In4Cl7(s), InN(s)**, In(s) Cl containing species CCl4(l), C6H5Cl(l), NH4Cl(l), NH4Cl(s1), NH4Cl(s2) Condensed Phases Other species C(l), C(graphite), Diamond GaN(s) thermochemical data was replaced with assessed data in [Unl03] ** InN(s) thermochemical data was replaced with assessed data in [Lei04] The data for metalorganics (MOs) and a dducts were obtained from Przhevalskii et al. [Prz98] and listed in Table 2-2. Since some of thermochemical data for MOs and adducts are calculated, it is necessary to validate the data. Values of H298, S298, and Cp for all the species used in the calculations plus Al containing species are listed in Appendix A.

PAGE 58

35 Table 2-2. Additionally included gas phase species from [Prz98]. Ga containing species Ga(CH3)3, GaCH3, Ga(CH3)3NH3, GaCH3NH, GaNH3, [Ga(CH3)2NH2]3, GaCl3NH3, GaH2, GaH3, (GaCH3NH3)3, (GaN)3* Gas Phase In containing species In(CH3)3, InCH3, InH2, InH3 The gas phase ring-shaped (GaN)3 species was eliminated due to inconsistency in the results. See section 2.3.1. 2.3 Chemical Equilibrium Calculations There are two main ways of computationall y determining the equilibrium state: stoichiometric and non-stoichiometric algor ithms [Smi68]. In the stoichiometric approach, the total Gibbs energy of the system is minimized by solving the set of linear equations produced by introducing the stoichio metric constraints [Mey84]. In the nonstoichiometric approach, the system of non linear equations produced for the equilibrium expressions for each reaction in an i ndependent set of reaction is solved. In this work, Thermo-Calc is used to solve for the equilibrium states. Users need to define atoms and provide initial atomic ratios, temperature, and pressure for equilibrium calculations. The software will generate all possible molecules by combination of defined atoms, as long as they exist in the da tabase. After that, the mole fractions of all the species are calculated and iterated to find the minimum total Gibbs energy of the system. Therefore, it is impor tant to make sure that all th e thermochemical data in the database are valid. One way of checking the validity is by computing well-known equilibria. 2.3.1 Thermochemical Data Verification Starting with established databases, the sp ecies list was expande d by review of the literature for the Ga-In-C-H-Cl-N system. To verify the newly included data, well-

PAGE 59

36 known problems were solved to check the cons istency. For that purpose, the In-N and Ga-N phase diagrams were generated. Figure 2-1. Phase diagram of In-N system s at P = 0.1 MPa and experimental InN decomposition data. The decomposition temperatures of In N were measured as 638 K by drop calorimeter [Lei04], 773 K by thermogravim etric analysis (TGA) [Ond02], and 868 K by thermogravimetry-differential scanning calor imetry (TG-DSC) [Gao03]. The calculated decomposition temperature of InN is 1211 K us ing SUB94 database. In addition, the maximum experimental InN growth temperature is around 973 K (700 C) according to Chapter 5 and 6 of InN growth study. Ther efore, assessment of InN Gibbs energy is required for equilibrium calculations At this point, the data from [Lei04] is used since it was critically assessed data with experiment s. However, it should be noted that the decomposition temperature and growth temper ature have a large overlap due to the kinetic barriers. When equilibrium calculati on is needed including InN at the growth temperature, the Gibbs energy of InN should be lowered to consider kinetic barrier for

PAGE 60

37 decomposition. In this case, the data from [SUB94] was used as presented in Section 2.3.2.4. All the newly included species gave resu lts consistent with previously known results for GaN, except for the ring-shaped trimer (GaN)3(g) species. Figure 2-2 shows the calculated G values for Ga(l) + 1/2 N2(g) = GaN(s) and Ga(l) + 1/2 N2(g) = 1/3(GaN)3(g) reactions. The G value for (GaN)3(g) formation reaction is significantly lower than GaN(s) formation reaction showing that (GaN)3(g) is more stable species than GaN(s). However, the existence of (GaN)3(g) was never confirmed by experiment and the thermochemical data was calculated using estimated parameters [Prz98]. Therefore, it would be reasonable to concl ude that a kinetic barrier ex ists for formation of (GaN)3 or the data for (GaN)3 is incorrect. Ga-N phase diagram was generated excluding (GaN)3(g), as shown in Figure 2-3. Figure 2-2. Calculated Grxn for GaN(s) and (GaN)3(g) formation reactions.

PAGE 61

38 Figure 2-3. Phase diagram of Ga-N systems at P = 0.1 MPa. The data for GaN is from [Unl03] Otherwise, all the new sp ecies including MOs and a dducts from [Prz98] were added to the database and used for th e subsequent equilibrium calculations. 2.3.2 Complex Chemical Equilibrium Calculations 2.3.2.1 Ga-C-H-Cl-Inert System It will be informative to see the gas ph ase species after the reaction between TMG and HCl in N2 carrier gas since no experimental resu lts are available. Hence, the Ga-CH-Cl-Inert (N2) system was analyzed first. The schematic of the H-MOVPE inlet is shown in Figure 2-4. Figure 2-4. Schematic of the inlet of H-MOVPE technique for GaN growth. 1110 K [ Unl03 ]

PAGE 62

39 TMG, HCl, and N2 (inert) were mixed to form ga s phase species before reacting with NH3. Although NH3 is present in the outer tube, it does not interact with TMG and HCl until a certain residence time, since they ar e separated by the quartz wall. Therefore, only the TMG + HCl + N2 system will be only considered here. The base conditions of molar flow rates in the source zone for GaN growth by HMOVPE are listed in Table 2-3. It should be noted that the atomic ratios are not independent, since TMG is composed of 1 Ga, 3 C, and 9 H atoms; HCl is composed of 1 H and 1 Cl atom. Therefore, X(C) = 3 x X(Ga) and X(H) = 9 x X(Ga) + X(Cl) conditions must be satisfied. Table 2-3. Base inlet conditions for sources for GaN growth and atomic mole fractions for calculation. Precursors Flow rate Flow rate/0.7 Atom # of Atoms X(Atom) Mole fraction TMG 0.7 sccm 1 Ga 1 X(Ga) 0.00066 HCl 0.7 sccm 1 C 3 X(C) 0.00198 H 10 X(H) 0.00660 Cl 1 X(Cl) 0.00066 Inert 1500 X(Inert) 0.99010 Inert (N2) 1050 sccm 1500 Total 1514 Total 1 N2 as a product of NH3 decomposition was treated as an active N source, which may react with a Ga species to form GaN in equilibrium calculations. For example, GaN(s) can be formed by Ga(l) and N2(g) in equilibrium calculation. The source of reactive N is only NH3 and its intermediate products. The carrier gas, N2, was thus treated as an iner t species, thus all N in the inlet NH3 was available for reaction; including the product N2. Helium (He) was used instead of carrier N2. The use of different inert gas such as Ne or Ar would give the same results since they are not partic ipating in the reaction. The role of inert gas in equilibrium calculati on is related to gas mixture entropy.

PAGE 63

40 Figure 2-5. Equilibrium partial pressures of gas phase molecules and mole fractions of condensed phases with respect to the temperature in Ga-C-H-Cl-Inert system.

PAGE 64

41 In addition, total pressure and system temperature values P = 1 atm and T = 573 K were used as the base conditions for calcu lations of the source zone performance. Figure 2-5 shows the calculated equilibrium partial pressures of gas phase species in the Ga-C-H-Cl-Inert system, with respect to the temperature. In this figure Cl/Ga was set to 1 and the calculation axis was T = 430 to 1000 K. The most predominant Ga containing ga s species was GaCl above 530 K and GaCl3 below this temperature. The amount of GaCl exceeded the amount of CH4 at 580 K. The partial pressure of the second dominant species, GaCl3, decreased at temperatures higher than 500 K. It was also observed that the amount of HCl ga s increased up to 600 K and did not change at temp eratures higher than 600 K. The partial pressure of monomethylgallium (MMG; GaCH3), the most dominant organometallic species, increased with temperature. However, the amount of MMG was about 10-6 times less than GaCl even at 1000 K. Therefore, the e ffect of using TMG inst ead of liquid Ga as the Ga source is negligible, ot her than adding C to the system. In this sense it is concluded that the H-MOVPE technique is closer to traditional HVPE rather than the MOCVD technique. Two condensed phase were present at equilibrium; C (graphite) and Ga(l). Experimental observation of black depositi on in H-MOVPE inlet confirmed that the black deposition is composed of C (graphite) hollow tubes filled w ith Ga [Par05]. To prevent black deposition, a small amount of H2 should be added to N2 carrier gas, which is consistent with Parks result [Par05] Figure 2-6 shows the equilibrium partial pressures and condensed phase mole numbers with respect to X(H) /{X(Inert)+X (H)}.

PAGE 65

42 Figure 2-6. Equilibrium partial pressures of gas phase molecules and mole fractions of condensed phases with respect to X(H) /{X(Inert) + X(H)} in Ga-C-H-Cl-Inert system at P = 1 atm, T = 573 K, Cl/Ga = 1.

PAGE 66

43 For this calculation, T was set at 573 K and X(H) was varied from 0.0066 to 0.9901 (not 0 to 1) because H and iner t gas are always present with NH3 and HCl (10 % HCl and 90 % N2 (Inert)). It was noted that the amount of C (graphite) decreased rapidly with increasing H2, primarily through formation of CH4. These thermodynamic calculation results agreed well with the expe rimental results from Park [Par 05]. It is also noted that small amount of Ga(l) exists. With increasing amount of H2, the partial pressures of the gas phase species did not change significantly. It is also noted that the major species part ial pressures showed a gradual increase with increasing H fraction. It should be noted th at the amount of GaCl3 changed significantly with both temperature and Cl/Ga ratio, but not with H2 since the partial pre ssure of GaCl or GaCl3 exceeded that of HCl, the most dominant Ga or Cl containing species. The equilibrium partial pressures of the gas phase molecules and condensed phase mole fractions with respect to Cl/Ga ratio are shown in Figur e 2-7. The amount of GaCl and GaCl3 crossed at around Cl/Ga = 2. Furtherm ore, the amount of HCl prevailed over GaCl3 at Cl/Ga = 4. The experimental results from Reed [Ree02] show ed that the growth rate remained nearly constant from Cl/Ga = 0 to 3, then declined rapidly, falling to zero at near HCl/Ga = 4. This also is reasonabl y matched with thermodynamic predictions. From the experimental observation of the gr owth rate, GaCl appeared to be the main precursor to grow GaN rather than GaCl3. In the condensed ph ase, the amount of C (graphite) did not change with HCl. In fact, the only C and Cl containing gas phase species is CH3Cl, which has very low partial pressure at Cl/Ga = 4.

PAGE 67

44 Figure 2-7. Equilibrium partial pressures of gas phase molecules and mole fractions of condensed phases with respect to X(Cl)/ X(Ga) ratio 1 to 4 in Ga-C-H-Cl-Inert system at P = 105 Pa, T = 573 K.

PAGE 68

45 Therefore, to prevent C contamination, H2 carrier gas or, at least, a mixture of H2 gas is recommended. In addition, the amount of Ga (liquid) rapidly droppe d with Cl/Ga > 1. According to these calculati on results, small amounts of H2 (~ 4 %) and excess HCl (Cl/Ga > 1) are required to grow Ga droplet-free and C-free GaN films. 2.3.2.2 Ga-C-H-Cl-N System The growth of GaN by H-MOVPE in H2 carrier gas can be considered as a Ga-CH-Cl-N system, as shown in Figure 2-8. To calculate the growth conditions of GaN, the Gibbs energy of GaN should be modified, as GaN is thermodynamically unstable at typical growth temperature. For example, the calculated GaN d ecomposition temperature was 1055 K using the Gibbs energy of GaN recommended by Unland et al. [Unl03]. It was found that the decomposition temperature of GaN linearly increases with decreasing the Gibbs energy of GaN by Thermo-Calc calculations. Assuming that the decomposition temperature of GaN is 1500 K, 49 kJ/mole was subtracted from the Gibbs energy of GaN [Unl03]. Figure 2-8. Schematic of H-MOVPE for GaN growth: Ga-C-H-Cl-N system. N is present and H2 is used as a carrier gas in the system compared to Figure 2-4. The experimental inlet conditions for Ga N growth were from [Ree02], including TMG = 2.2 sccm, HCl = 4.4 sccm, NH3 = 500 sccm, and H2 = 1600 sccm. Therefore, the

PAGE 69

46 Cl/Ga and N/Ga ratios were 2 and 230, resp ectively. Given an initial atomic mole fraction of Ga, (X(Ga) = 4.2 x 10-4 = base), the other elements atomic mole fractions are calculated by X(C) = 3 x X(Ga), X(Cl) = 2 x X(Ga), X(N) = 230 x X(Ga), and X(H) = 1 X(Ga) X(C) X(Cl) X(N). The growth-e tch transition temperatures were thereafter calculated by changing Cl/Ga ratio from 0 to 10. Figure 2-9. Calculated growth-e tch transition temperature as a function of Cl/Ga ratio. N/Ga ratio was set at 230. Figure 2-9 shows the calculated growth-e tch transition temperature. It was observed that Ga(l) and GaN coexisted when Cl/Ga < 0.1 and that the upper growth temperature decreased w ith increasing Cl/Ga ratio. It is also noted that carbon appeared under certain conditions, but is not listed in this figure. The maximum growth-etch transition temperature was 1371 K at Cl/Ga = 0 in case of N/Ga = 230. Further calculations were carried out to focus on the Ga(l) GaN coexistence conditions by varying N/Ga ratio (200 300, 400, 500, and 600) and Cl/Ga ratio (continuously from 0 to 0.5), as seen in Figure 2-10. The decomposition temperature increased from 1371 to 1395 K with increasing N/Ga ratio from 200 to 600, as the N partial pressure increased. (GaN) (gas)

PAGE 70

47 Figure 2-10. Calculated transi tion of growth-etch and Ga(l) formation as a function of temperature and Cl/Ga ratio. N/Ga ratio was varied from 200 to 600. This is expected since adding excess N relative to Ga will form the more stable GaN. It was found that the Ga(l) + GaN coexistence regime becomes narrower with increasing N/Ga ratio and completely disappears at N/Ga = ~ 600. This is the N/Ga ratio thermodynamic limit to avoid Ga droplet in MOCVD (Cl/Ga = 0). Since the N/Ga ratio for GaN growth is typically much higher th an 600 in MOCVD, Ga(l) was not usually observed. A more or less linearly decreasi ng trend of growth-etch transition temperature was observed when Cl/Ga was higher than 0.2 re gardless of N/Ga ratio. As Cl is added to the system, it primarily appears as GaCl at the elevated temperature, thus making less available for formation of GaN. 2.3.2.3 In-C-H-Cl-Inert System Similar calculations were performed to analyze the inlet of H-MOVPE for InN growth in inert ambient. Figure 2-11 shows the schematic of the inlet for InN growth. TMI first reacts with HCl in N2 (inert) carrier gas before reacting with NH3. Therefore the TMI + HCl + N2 (inert), or the In-C-H-Cl-Iner t (He) system was analyzed to (gas only) (GaN + gas)

PAGE 71

48 determine the equilibrium gas and condensed phas e species. The base flow rates of TMI, HCl, and N2 were 0.7 sccm, 0.7 sccm, and 1050 sccm (6.3 sccm with 10% HCl + 200 sccm diluted with HCl + 269 sccm with TM I + 200 sccm diluted with TMI + 375 sccm carrier = 1050 sccm), respectively. Therefore, Cl/In = 1, Inert/In = 1500, P = 1 atm, and T = 573 K were used as the base conditions for calculation. It should be noted that C/In ratio is always 3, because TMI is composed of 1 In, 3 C, and 9 H. The number of H atoms is coupled with 9 TMI and 1 HCl. In other words, the relationship X(C) = 3 x X(In) and X(H) = 9 x X(In) + X(Cl) should be always satisfied. The initial atomic mole fractions for In-C-H-Cl-Inert syst em are tabulated in Table 2-4. Figure 2-11. Schematic of the inlet of H-MOVPE technique for InN growth. Table 2-4. Typical growth c onditions for InN and atomic mo le fractions for calculation. Precursors Flow rate Flow rate/0.7 Atom # of Atoms X(Atom) Mole fraction TMI 0.7 sccm 1 In 1 X(In) 0.00066 HCl 0.7 sccm 1 C 3 X(C) 0.00198 H 10 X(H) 0.00660 Cl 1 X(Cl) 0.00066 Inert 1500 X(Inert) 0.99010 Inert (N2) 1050 sccm 1500 Total 1514 Total 1 Figure 2-12 shows the equilibrium partial pressures and condensed phase mole numbers with respect to temp erature at Cl/In = 1. The calculation axis was T = 430 to 1000 K.

PAGE 72

49 The most predominant In containing sp ecies was InCl. However, the second dominant In containi ng species was not InCl3, as compared to the second most abundant gas species in the Ga based system, GaCl3. Instead of InCl3, In2Cl2 (the dimer of InCl) had higher partial pressure (105 times higher than InCl3) at T = 480 to 700 K. The partial pressure of InCl3 decreased at temperatures higher than 480 K. Although TMI and MMI thermochemical data were included during the calculation, they were not observed in the results. The lowest limit of partial pressure calculation was set as 10-17 atm, thus it can be inferred that TMI and MMI have less than 10-17 atm partial pressure. As a result, it can be concluded that H-MOVPE is closer to HVPE than MOCVD in both InN and GaN growth cases. For condensed phases, carbon (graphite) a nd In (liquid) were dominant and the amount of In(l) decreased with temperature. Deposition of a black ma terial in the inlet was occasionally observed during the growth of InN especially when the inlet temperature was too high. Even though a detailed analysis of this black material at the inlet during InN growth was not carried out, it likely consists of C (graphite) and In droplets. To avoid C and In contaminati ons, similar methods can be used such as addition of H2 and excess HCl for C and In elimination, respectively. The effect of H2 carrier gas was analyzed by incr easing the initial atomic H mole fractions. Figure 2-13 shows the equilibri um partial pressures and the amount of condensed phase species with respect to the relative amount of H compared with inert gas, or X(H)/{X(H) + X(inert)}. Compari ng Figures 2-13 and 2-5 it is noted that the amount of Ga(l) is about an or der of magnitude greater than that of In(l) at similar conditions.

PAGE 73

50 Figure 2-12. Equilibrium partial pressures of gas phase molecules and mole fractions of condensed phases with respect to the temperature in In-C-H-Cl-Inert system.

PAGE 74

51 Figure 2-13. Equilibrium partial pressures of gas phase molecules and mole fractions of condensed phases with respect to X(H) /{X(Inert) + X(H)} in In-C-H-Cl-Inert system.

PAGE 75

52 Figure 2-14. Equilibrium partial pressures of gas phase molecules and mole fractions of condensed phases with respect to X(Cl)/ X(In) ratio 1 to 4 in In-C-H-Cl-Inert system.

PAGE 76

53 The gas phase species did not change greatly, except for H2 and other inert gases. An increase of InH was observed; however, th e amount of InH was extremely small (10-15 atm), thus the role of InH was negligible. Figure 2-13 shows the equilibrium partial pressures and the amount of condensed phase species with respect to the relative amount of H comp ared with inert gas, or X(H)/{X(H) + X(inert)}. Comp aring Figures 2-13 and 2-6 it is noted that the amount of Ga(l) is about an order of magnitude greater th an that of In(l) at similar conditions. The gas phase species did not change greatly, except for H2 and other inert gases. An increase of InH was observed; however, the am ount of InH was extremely small (10-15 atm), thus the role of InH was negligible. One of the advantages of H-MOVPE techni que is the variation of Cl/In ratio. Therefore, it is interesting to s ee the effect of Cl/In ratio. The Cl/In ratio was varied from 1 to 4 and the results are shown in Figure 214. The predominant In containing species were InCl and In2Cl4 (dimer of InCl2) through Cl/In = 1 to 4 and the amount of HCl, InCl, and In2Cl4 reversed at Cl/In > 2. Therefore, an etching effect is expected due to the HCl when Cl/In > 2. The amount of In(l) drops rapidly near Cl/In = 1, and no In(l) exists at Cl/In > 1 and the amount of C (graphite) did not change with Cl/In ratio. To eliminate C contamination, H2 should be added in the system. 2.3.2.4 In-C-H-Cl-N-Inert System Thermodynamic calculations were carried out to understand the gas phase and condensed phase species for InN growth by H-MOVPE. Figure 2-15 shows the schematic of the H-MOVPE system using TMI, HCl, NH3, and N2 (inert) as precursors. This can be thought as an In-C -H-Cl-N-Inert system. It w ill be informative to see the

PAGE 77

54 InN growth-etch conditions with change in temperature, Cl/In, and N/In molar ratios since they are the most im portant growth parameters. Figure 2-15. Schematic of H-MOVPE inle t with In-C-H-Cl-N -Inert system for thermodynamic calculations. N is present in the system compared to Figure 2-11. In the equilibrium growth-etch regime cal culation, the thermochemical data for solid InN was taken from SUB94 database inst ead of using the data from [Lei04], since the decomposition temperature of InN (686 K) from [Lei04] is too low for InN. Therefore, InN is thermodynamically more unstable at growth temperature with expressions proposed by [Lei04] as compared to that from the SUB94 database. This is equivalent to lowering the Gibbs energy by about 68.5 kJ/mole. The initial process parameters for InN growth used in this calculation were TMIn = 1 sccm, HCl = 4 sccm, NH3 = 250 sccm, and Inert = 2286 sccm. Actual process parameters were about 0.7 times less than the given values, for example TMIn = 0.7 sccm. However, they were in creased by 1/0.7 times for ease of calculation. Therefore, the initial atomic mole fract ions were X(In) = 3.024 x 10-4, X(C) = 3 x X(In), X(Cl) = 4 x X(In), X(N) = 250 x X(In), X(H) = 9 x X(In ) + X(Cl) + 3 x X(N), and X(Inert) = 1 X(In) X(C) X(Cl) X(N) X(H). To carry out the calculation C l/In and N/In ratios were varied and the growth-etch transition temperatures were determined.

PAGE 78

55 As shown in Figure 2-16, the growth-etc h transition temperature decreased with increasing Cl/In ratio. When Cl/In ratio > 1, the added HCl to the system shifts the equilibrium compositions of InClx in the vapor phase, and thus it lowered the transition Figure 2-16. Calculated growth and etch regime with resp ect to temperature and Cl/In ratio. N/In ratio was varied from 100 to 7000. temperature. As the Cl/In ratio decreases towards unity, the Cl is now not in large excess but approaches the stoichiometry of the domin ant vapor phase In species, InCl. Thus a small decrease in the Cl/In ratio frees a relatively higher propor tion of the In for participation in growth. For Cl/In ratio less th an unity, there is not su fficient Cl to retain the In in the vapor phase and In is now ava ilable to form InN. The temperature of the horizontal line for a given N/ In ratio is the thermal decomposition temperature, which increases with increasing N/In as expected. In(l) was observed as a product of thermal decomposition reaction below the horizontal lines. The companion plots of tran sition temperature vs. N/In atomic ratio are shown in Figure 2-17. The results suggest that the tr ansition temperature incr eased with increasing 600 700 800 900 1000 1100 1200 012345678910 Cl/In ratioTemperature (K)N/In = 100 N/In = 250 N/In = 500 N/In = 1000 N/In = 7000 N/In = 100 N/In = 250 N/In = 500 N/In = 1000 N/In = 7000 Growth No growth (Etch) In(l) + InN + Gas (GaN + gas) Etch (gas only)

PAGE 79

56 N/In ratio to an asymptotic limit. It is not ed that the calculations were only performed for seven HCl/TMIn ratios, each grea ter than unity. Increasing the N at constant HCl and TMIn input increases the ability of N to compete with Cl for the In. Increasing the N through the addition of NH3 also increased the H in the system, which also competes with In for the Cl to increase the driving force for InN formation. Figure 2-17. Calculated growth and etch regime with resp ect to temperature and Cl/In ratio. Cl/In ratio was varied from 1.1 to 10. Figure 2-18 (a) shows the expa nded views of the near horiz ontal lines at high N/In ratios: 4.0 x 104 to 7.0 x 104. It was observed that both Cl/I n and N/In ratios affect In droplet formation significantly. When N/In ratio was 4.0 x 104, Cl/In ratio should be higher than 0.4 to avoid In dr oplet formation. When N/In ratio was increased to 5.0 x 104, the minimum Cl/In ratio to avoid In droplet formation was decreased to 0.3, and so on. Figure 2-18 (b) shows the extended calcula tion results that include N/In ratio from 100 to 7.8 x 104 and Cl/In ratio from 0 to 1. Hi gher N/In ratio lo wered the minimum Cl/In ratio to avoid In droplet formation, which is consistent with the previous results. It was noted that at N/In ratio ~ 7.8 x 104, no In droplet existed, even though no HCl was present (Cl/In ratio 0).

PAGE 80

57 Figure 2-18. Calculated results of the growth, no growth (etch), and In droplet regimes (a) Cl/In = 0 to 0.5, N/In = 4.0 x 104 to 7.0 x 104, (b) Cl/In = 0 to 1, N/In = 1.0 x 103 to 7.8 x 104, and (c) In droplet etching conditions at T = 1153 K, P = 105 Pa. (c) (a) (b) Cl/In ratio

PAGE 81

58 Therefore, In droplet formation can be avoide d when the N/In ratio is higher than 7.8 x 104. Experimental results of InN growth by MOCVD showed that In droplets formed when the NH3/TMIn ratio was less than 1.6 x 104 [Mat97]. Although In droplets were not observed when N/In rati o was higher than 1.6 x 104, the rough surface may be related to In droplet fo rmation. The surface of InN was mirror-like when N/In ratio was higher than 8.0 x 104, which may represent the complete removal of In droplets during growth [Mat97]. The decomposition temperature of InN did not change while the N/In ratio changed from 3.0 x 104 to 7.8 x 104; however, the minimum Cl/In ratio to avoid In droplets decreased at higher N/In ratio. These results are shown in Figure 2-18(c) It is clear that In droplet formation can be avoided by eith er increasing N/In rati o or increasing Cl/In ratio. Both of these changes provide reactant for In(l) (N to form InN or Cl to form InCl(g)). A key finding is that the determin ation of the minimum N/In ratio, 7.8 x 104, to avoid In droplet formation by the MO CVD technique (Cl/ In ratio = 0). In the following, the decomposition temperature change was examined when the N/In ratio was higher than 7.8 x 104. Figure 2-19 (a) shows the result when the Cl/In ratio was 0. Notice that the N/In ratio ax is is a logarithmic scale. The transition temperature decreased with increasing N/In ratio, as long as N/In > 7.8 x 104. The reason for this phenomenon is related to the gas phase species, InH. The partial pressures of InH (PInH), in fact, decreased with increas ing N/In ratio. For example, PInH were 7.975 x 10-7, 9.456 x 10-8, and 1.125 x 10-8 atm for N/In ratio 105, 106, and 107, respectively. This is expected since the initial atom ic mole fraction of In, X(In), was reduced to increase the N/In ratio. The initial X(In) values were 2.486 x 10-6, 2.499 x 10-7, and 2.5 x 10-8 for

PAGE 82

59 N/In ratio 105, 106, and 107, respectively. Therefore, the ratios of PInH to X(In) were taken to compare reasonable with the relative effect of InH. PInH/X(In) ratios were 0.321, 0.378, and 0.450 atm for N/In ratio 105, 106, and 107, respectively. Thus, it can be concluded that the relative amount of InH increased with incr easing N/In ratio. Figure 2-19. Calculated growth-e tch transition temper atures (a) In droplet etch conditions when no HCl was present (b) growth-etc h transition temperature at high N/In ratios: 105, 106, and 107. It is also observed from Figure 2-19 (a) that the grow th transition temperature lowered with increasing N/In ra tio. Thus the maximum grow th temperature is lower for very high N/In ratio. It is known that the growth rate of InN d ecreases with excess NH3 (a) InN 10 2 10 3 10 4 10 5 10 6 N/In ratio (b) Etch (gas only) Growth (InN + gas)

PAGE 83

60 flow. The results of this calculation may e xplain the reason of growth rate decrease thermodynamically. Figure 2-19 (b) shows the calcu lated growth-etch transiti on temperatures at three high values of N/In ratio (105,106, and107). No In droplet was observed in this excessively high N/In ratio due to the InH form ation. The transition from growth to etch occurs at a lower temperature; the higher the va lue of N/In as long as Cl/In < 2. For high Cl/In, the excess Cl ties up the In or InCl to minimize the effect of increased N/In. 2.3.2.5 NH3 Partial Decomposition Ammonia is the most widely used nitrogen precursor for Group III Nitride growth. Homogeneous decomposition of NH3 to N2 and H2 is thermodynamically favored and it completes at around 673 K as shown in Figure 2-20. Figure 2-20. Equilibrium partial pressures for decomposition of NH3 at 1 atm total pressures calculated by Thermo-Calc. It is known, however, that homogeneous decomposition of NH3 does not occur easily due to the very high kinetic barrier to break the initial N-H bonds [Seg04, Yum81,

PAGE 84

61 Liu78]. One approach to account this fact in equilibrium cal culations is to measure the Gibbs energy of NH3 to shift the yield curve to high temperatures. The homogeneous partial decomposition of NH3 can be written as following. NH3 = (1 ) NH3 + /2 N2 + 3/2 H2 where is the degree of completion of NH3 decomposition reaction. It was experimentally observed that NH3 decomposes to an ex tent ~ 3 % at 950 C as measured by mass spectroscopy [Ban72]. To achieve this conversion at the temperature at 950 C, 170 kJ/mole ha s to be subtracted from the NH3 Gibbs energy. Figure 2-21. Magnitude of the Gibbs energy correction required to achieve a value for partial decomposition () at 950 C. This energy may be considered as the kinetic barrier of NH3 decomposition to make NH3 decompose 3 % at 950 C. A desired value at 950 C can be achieved by modifying the Gibbs energy of NH3 by subtracting various amounts as shown in Figure 2-21. A different value of at 950 C can be assumed that will give a different value by which to modify the Gibbs energy of NH3. The calculated values for the decomposition curves as a function of T are shown in Figure 2-22. The given in this figure represents the value of the conversion along this curve at 950 C. 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 -200-150-100-500 NH3 Gibbs energy correction (kJ/mole)

PAGE 85

62 Figure 2-22. NH3 mole fraction with respect to te mperature at different values of 2.4 Conclusions Additional thermochemical data for th e Ga-In-H-C-Cl-N system including MOs and adducts were added to the current da tabase [Ree02] and validated. Complex equilibrium calculations were performed and concluded that H-MOVPE is more related to HVPE than MOCVD. It was concluded that H2 should be included to elimin ate C deposition in the inlet by thermodynamic analysis and experimental observations. In addition, excess HCl (Cl/III > 1) should be present in the system to eliminate Ga or In metal droplets in the inlet. It was determined that monochlorides ( i.e., GaCl and InCl) are predominant species at Cl/III < 2. Excess Cl/III calculations showed that GaCl3 is predominant when Cl/Ga > 2, whereas HCl prevailed over InCl and In2Cl2 at Cl/In > 2 at 573 K. However, monochlorides were always dominant at growth condition for both GaN and InN.

PAGE 86

63 High N/III, low Cl/III, and low temperature generally provide more stable conditions for III nitrides, except for the excessive N/III case. The decomposition temperature of GaN and InN increased with high N/III ratios. Ga(l) + GaN and In(l) + InN coexistence regimes were observed under th e low Cl/III ratio conditions. It was determined that for Ga(l) removal in MOCV D (Cl/Ga = 0), the N/Ga ratio should exceed 600, whereas for In(l) removal in MOCVD (Cl/In = 0), the N/ In ratio should be higher than 7.8 x 104. The results may explain the excessive ammonia flow requirement for InN growth to avoid In droplet formation. Finally the partial decomposition of NH3 was considered and the Gibbs energy of NH3 was reduced to control decomposition e fficiency. The kinetic barrier of NH3 decomposition (170 kJ/mole) was calc ulated by considering 3 % of NH3 decomposition at 950 C.

PAGE 87

64 CHAPTER 3 STRESS DETERMINATION STUDIES OF GALLIUM NITRIDE FILM ON SAPPHIRE 3.1 Introduction Due to the lack of bulk si ngle crystals, GaN films ar e usually grown on foreign substrates, such as sapphire or SiC. Stress and dislocations are inevitable because of lattice and thermal expansion mismatches be tween GaN and foreign substrates. Excess stress may cause cracks and bows of the GaN films and dislocations may act as carrier traps resulting in low-mobili ty materials. Hence, a st ress and dislocation study is important to improve device performance. The most widely used substrate for GaN gr owth is sapphire, due to its thermal and chemical stability, AlOxNy compliant layer formation at th e surface after n itridation, and relatively low cost. SiC is a substrate that provides better lattice and thermal expansion match and thermal and chemical stability, in ad dition, but the cost of SiC is still very high, restricting commercial use. Therefor e, focusing on studying the stress of GaN film on sapphire is helpful as it is more cost-eff ective than SiC and will lead to improvement in GaN/sapphire-based device reliability. The stress state of a film can be measured by several techniques. Lattice parameter changes ( i.e., strain), curvature, or vibrationa l frequency changes are some of the properties that can be readily measured to ascertain the stress of the film. Those parameters are usually obtained by XRD reci procal space mapping, optical or stylus profilometry, laser reflectance, Rama n spectroscopy, and photoluminescence.

PAGE 88

65 XRD is a non-destructive and direct met hod to measure the la ttice parameters of the film. XRD -2 scan (also known as powder XRD, XRD 2scan, or low resolution XRD) can determine d-spacing para llel to the surface of the films; however, the lattice parameters obtained in this way cannot be used for stress measurement because the 2 value can shift depending on the sample height and the tilt on the sample mount during the XRD measurement. In addition, wh en a single crystal is used, there are only peaks from the lattice spacing parallel to th e surface. For example, if a single GaN crystal was measured by LR-XRD, there will be only GaN (002) a nd GaN (004) peaks, which can only be used to calculate out-of-pla ne lattice parameters, while in-plane lattice parameters cannot be obtained. Therefor e, XRD reciprocal sp ace mappings should be carried out to measure lattice para meters precisely to calculate th e strain of the film. It is time-consuming to measure the lattice parame ters of the samples by XRD reciprocal space mapping due to the alignments steps. In addition, the lattice parameters obtained by XRD-SM may not represent the surface prop erties because of the X-ray penetration depth. The stress measured by XRD should be considered as the av erage value from the X-ray interaction volume. X -ray penetration depth depends not only on the material, but also X-ray incidence angle (). As a result, precise dete rmination of X-ray penetration depth is not simple. Curvature of the film can be measured by optical profilometer, stylus profilometer (also known as alpha step), laser reflectance, or other techniques, such as XRD rocking curves with displacing the sample position. The disadvantage of op tical profilometer and laser reflectance is their sensitivity towa rds surface roughness. Since both techniques

PAGE 89

66 utilize the reflected light from the surface, if the surface is ro ugh or has structures on it, it is difficult to measure th e curvature of the film. Curvature can also be measur ed by stylus profilometer. It requires direct contacts between the tip and the surface. While the surface roughness is not a significant obstacle to using this technique, direct contact may damage the surface. In addition, if the film has fabricated structures on it, it is not possible to use st ylus profilometer for curvature measurement. Nevertheless, stylus profilome ter may be the quickest and easiest way to measure the curvature and the roughness of the film, though it is not recommended for samples that will be damaged by direct contact. XRD rocking curves with displacing the sample position is a non-destructive and surface roughness-independent technique th at measures the curvature of the crystallographic planes, rather than the surface. Although it is time-consuming to measure the curvature of the film due to the alignment steps, this may be the only way to non-destructively measure the curvature of rough samples. The general disadvantage of curvature measurements is that they cannot measure small stresses in flat films. In other words, the stress mu st be large enough to create curvatures within the film. Th erefore, small stresses in film s without curvature should be measured by more sensitive techniques, such as Raman spectroscopy. Raman spectroscopy is well-established as a non-destructive and relatively rapid method to measure the average value of film stress. Raman spectroscopy operating in confocal mode can show the peak shifts along the thickness. The results can be used to deduce the stress variation along the depth of the sample.

PAGE 90

67 Another way of studying stress of the film is through modeling. According to the critical thickness theory, the lattice paramete r of the overgrown film will turn into its unstrained value once dislocations are formed The lattice parameter variation along the thickness, however, was experimentally observe d, even well above the critical thickness. The result cannot be predicted if the traditiona l concept of critical thickness or Frank-van der Merwes semi-infinite model is used. Shen-Danns layer-bylayer growth model allows for continuous changes in lattice parameters along the thickness, as well as dislocation density reduction as the film becomes thicker. This chapter will explore how the stress of the film was measured by vibration frequency changes (Raman E2 phonon peak shifts), lattice parameter change (XRD space map), and curvature (profilometer, XRD rock ing curves). Extensive characterizations such as XRD rocking curves, XRD pole-figur e, AES, and SIMS were carried out to determine the crystal structure and chemical compositions. For lattice strain modeling, Shen-Danns layer-by-layer grow th model was adopted and co mpared with experimental data. Although the model used only lattice parameters and el astic constants of the film and substrate, the results matche d well with experimental data. 3.2 Effects of Lattice and Th ermal Expansion Mismatches Before examining the stress measurements and modeling of GaN films, it would be instructive to explore the main sources of film stress, which are lattice and thermal expansion mismatches. The effects of thes e sources of film stress were studied as follows. 3.2.1 Lattice Mismatch at Growth Temperature To calculate the stress due to the lattice mi smatch during the grow th, it is important to consider the lattice parameters at the growth temperature, ra ther than at room

PAGE 91

68 temperature. The lattice parameters of GaN, sapphire, Si, and SiC at high temperature would be larger than the parameters at room temperature due to th e thermal expansion. For convenience, calculations were performe d assuming the growth temperature is 1300 K and room temperature is 300 K. Although th e growth experiments of this study were mostly performed at 1123 K, the calculation results would give some idea of lattice parameters change at high temperature. The definition of linear thermal expansion coefficient is given as equation (1): 0L L T (1) where is the linear thermal expansion coefficient, T is the temperature difference, L is the linear variation of th e lattice parameter, and L0 is the lattice parameter at room temperature. The definition of misfit, f, is given as equation (2): s f sa a a f (2) where as and af stand for the lattice parameters of substrate and film. The lattice parameters of GaN and wi dely-used substrates such as c-Al2O3, Si(111), and 6H-SiC at 300 K and 1300 K ar e tabulated in Table 3-1. Table 3-1. Lattice parameters of GaN and widely used substrat es at 300 and 1300 K. Structure a300K () TEC a (x 10-6/K) a1300K () Misfit (%) at 300 K Misfit (%) at 1300 K GaN Wurtzite 3.189 5.6 3.207 0 0 c-Al2O3 Rhombohedral 4.758 *(2.747) 7.5 4.794 *(2.768) 16.09 15.86 Si (111) Diamond 5.431 *(3.841) 2.7 5.445 *(3.851) + 16.97 + 16.72 6H-SiC Wurtzite 3.086 4.2 3.093 3.34 3.69 30 rotation was considered.

PAGE 92

69 The results show that the lattice misfits calcul ated at room temperature and at the growth temperature are not pa rticularly different. To see the critical thickness at the grow th temperature (1300 K, in this case), the GaN/6H-SiC system was considered. The critical thickness can be approximately expressed by equation (3). 1 ln ) 1 ( 8b h f b hc c (3) where b is Burgers vector, is Poissons ratio, and f is the misfit. Plugging in the typical values such as b (Burgers vector) af = 3.2 = 0.2, and f = 0.03 (in the case of SiC), gives hc = 2.2 which is less than one mono-layer. In this case, dislocations will form at the interface. The other substrates such as Si and sapphire will have even smaller critical thickness. Therefore, critical thic kness does not have much meaning in the case of heteroepitaxy of GaN on any currently widely used substrates. In fact, a critical thickness may only exist when the lattice misfit is smaller than 0.8 % and the thickness is less than 10 monolayer scales (< 5 nm). 3.2.2 Thermal Expansion Mismatch Most materials compress during the cooling process, except for some rare negative thermal expansion coefficient materials (i.e., see section 8.4). Hence, both GaN and substrates compress during cooli ng at different ratios due to differences in their thermal expansion coefficients. Calc ulated thermal strains during cooling ranging from 1300 to 300 K are listed in Table 3-2 for GaN as well as c-Al2O3, Si(111) and 6H-SiC. 30 rotated lattice parameters for c-Al2O3 and Si(111) were used for practical comparison with GaN basal plane and denoted in parentheses.

PAGE 93

70 Table 3-2. Lattice constants at 1300 and 300 K and thermal strain (T = 1000 K) of GaN and widely used substrates. Structure a1300K () a300K () Compressive strain during cooling Difference with GaN GaN Wurtzite 3.207 3.189 0.56 % 0 c-Al2O3 Rhombohedral 4.794 *(2.768) 4.758 *(2.747) 0.76 % 0.20 % Si (111) Diamond 5.445 *(3.851) 5.431 *(3.841) 0.29 % + 0.27 % 6H-SiC Wurtzite 3.093 3.086 0.42 % + 0.14 % 30 rotation was considered. 3.2.3 Combination of Lattice and Thermal Expansion Mismatches Figure 3-1 summarizes the stress cause d by thermal expansion and lattice mismatches assuming T = 1000 K. It should be noted that all the lattice parameters that were from different crystal structures were modified (i.e., 30 rotation) for better comparison with GaN since GaN often showed 30 rotation with respect to sapphire. Figure 3-1. Substrates for GaN plotted with thermal strain versus lattice parameters at room temperature. Net unstrained line

PAGE 94

71 It is predicted that c-Al2O3 compressed more than GaN during cooling. As a result, GaN films grown on c-Al2O3 substrates usually show compressive stress due to both thermal and lattice stress. In contrast, Ga N films grown on Si subs trates show tensile stress because they compress less than GaN film during cooling, in addition to tensile stress caused by lattice mismatch at growth temperature. This figure suggests that both 6H-SiC and AlN are good substrat es for GaN, due not only to close lattice parameters, but also the closest thermal expansion coeffi cients. In addition, the compensation of thermal tensile stress and lattice compressive st ress will be beneficial to grow stress-free GaN material. A net unstrained line in th e figure represents complete compensation of stress by equivalent contributions of lattice and thermal strain. It should be noted that thermal expansi on and lattice mismatches are not the only parameters that dominate tensile and compressive stress in the film. Use of buffer layers or nitridation will change the surface propert ies considerably. If the surface properties change, the growth mode may change. Te nsile stress at the grain boundaries during islands coalescence mode would be significant, especially if the proper buffer layer is not used. For example, GaN grown on Al2O3 shows cracks due to the excessive tensile stress, though compressive stre ss was present due to the lattice and thermal expansion mismatch [Etz01]. 3.3 Stress Measurements of GaN Films on Sapphire The stress was measured in two sample s grown by Mastro during comparison study of HVPE and MOCVD films. (p 64 in [Mas01]). One sample was a 3 m thick GaN film grown on sapphire by H-MOVPE and the othe r was a 1 m thick GaN film grown on sapphire by MOCVD. It should be noted that the term MOCVD here refers to hot wall H-MOVPE without HCl fl ow at atmospheric pressure, a nd should not be considered as

PAGE 95

72 typical MOCVD, which often uses cold walls at lower pressure. In other words, both samples were grown by H-MOVPE technique with and without HCl inclusion on sapphire substrate without buffer layer growth. Table 3-3 lists the growth conditions of both samples in this study. Table 3-3. The growth conditions of th e two samples for stress measurements. Samples Growth T (C) Cl/Ga NH3 flow rate (sccm) Buffer layer H-MOVPE GaN 950 2 500 Not used MOCVD GaN 950 0 500 Not used [Mas01] 3.3.1 GaN Structural and Compositional Studies To validate the structural quality of th e film and ensure the film was grown epitaxially; the typical approach uses HR-X RD rocking curves a nd XRD pole figures. Figure 3-2. XRD and -2 rocking curve of MOCVD and H-MOVPE grown GaN on sapphire, XRD FWHM; -rocking curve (MOCVD: 1261 arcsec, H-MOVPE: 1790 arcsec); -2 rocking curve (MOCVD: 92 arcsec, H-MOVPE: 122 arcsec). GaN-MOCVD

PAGE 96

73 Figure 3-2 shows XRD and -2 rocking curves of both the MOCVD and HMOVPE grown samples. The FWHM of -rocking curves of MOCVD and H-MOVPE grown samples were 1261 and 1790 arcsec, re spectively. The homogeneity of lattice spacing in the crystal is measured by XRD -2 rocking curve and the homogeneity of the normal axis of the lat tice planes are measured by -rocking curve. Typically it is more likely to have uniform lattice spacing th an uniform direction of lattice planes. Therefore, the FWHM of -2 rocking curves are usuall y smaller than that of -rocking curve. The FWHM values of 122 and 92 arcsec were obtaine d from MOCVD and HMOVPE grown samples, respectively. There is no clear criterion of crystallinity by only the rocking curve method, though higher crystal quality results in smaller FWHM XRD pole figure is often used to check if the film is grown epitaxial ly. Figure 3-3 shows XRD pole figures of the (116) sapphire and MOCVD growth GaN (112). A 30 rotation of GaN film w ith respect to sapphire substrate was clearly seen because the lattic e mismatch can be reduced significantly from 49 % to 14 % in this way. The results conf irmed that the GaN films were indeed grown epitaxially. Figure 3-3. Pole figures for the (116) sapphire su bstrate (2 theta = 57.490) and (112) GaN by MOCVD (2 theta = 69.185). The GaN in-plane axis is rotated by 30 with respect to the sapphire axis. 30

PAGE 97

74 Similar results were found with the Ga N grown by H-MOVPE (F igure 3-4). A 30 rotation with respect to sapphire and clear epitaxial growth we re confirmed. It is noted that the shapes of the peaks from sapphire and GaN are different. The peaks from sapphire substrate are much sharper in shape compared to the peaks from GaN, because the crystal quality of sapphi re is superior to GaN. Figure 3-4. Pole figures for the (116) sapphire su bstrate (2 theta = 57.490) and (112) GaN by H-MOVPE (2 theta = 69.185). The GaN in-plane axis is rotated by 30 with respect to the sapphire axis. Auger Electron Spectroscopy (AES) was used for chemical composition analysis. Figure 3-5 shows the AES surface scan and depth profile of GaN on sapphire grown by MOCVD. Ga (52.8 %), N (30.1 %), C (10.7 %), and O (6.3 %) were detected on the surface. Surface C and O quickly disappeared within 1 min of sputtering. Since the thickness of the GaN was reasonably thick (~ 1 m), the typical sputtering rate (~ 100 /min) would require 100 min of sputtering time to fully pr ofile. To reduce sputtering time, the raster size was decreased by a fact or of 5, so the spu ttering rate increased fivefold. In addition, the detection sequen ce was reduced to decrease sampling time. The results show that the GaN films were ma inly composed of only Ga and N from the surface to the interface. Ad sorbed C and O on the surface were apparently from air exposure. 30

PAGE 98

75 Figure 3-5. AES surface scan and depth pr ofile of GaN on sapphire grown by MOCVD Kinetic Energy (eV) dN(E) Min: -3638Max: 2760 50250450650850105012501450165018502050 Ga1 52.8 % O1 6.3 % N1 30.1 % C1 10.7 %Time (mins.) Min: 0Max: 42933 0 02468101214161820 O1 C1 Ga1 N1 O1 C1 Ga1 N1 Ga1 N1

PAGE 99

76 Figure 3-6. AES surface scan and depth profile of GaN on sapphire grown by HMOVPE. Kinetic Energy (eV) dN(E) Min: -3411Max: 2711 50250450650850105012501450165018502050 Ga1 53.4 % O1 9.0 % N1 26.4 % C1 11.2 %Time (mins.) Min: 0Max: 45691 005101520253035404550 O1 C1 Ga1 N1 O1 C1 Ga1 N1 C1 Ga1 N1

PAGE 100

77 Similar AES surface scanning followed by depth profile was done on H-MOVPE GaN on sapphire as shown in Fi gure 3-6. The surface chemical compositions were Ga (53.4 %), N (26.4 %), C (11.2 %), and O (9.0 %). The thickness of GaN was 3 m. Therefore, the fivefold increased sputtering ra te was again used by reducing raster size. Uniform Ga and N profiles were obtained from the surface to the inte rface once more. It is noteworthy that the sputtering rate was indeed ~ 500 / min. Only adsorbed surface C and O existed just as in the previous case. 3.3.2 Stress Measurements by Raman Spectroscopy The Raman shifts of the stress sensitive GaN E2 mode have been reported by several researchers, and the reported values cover the la rge range, from 565 to 572 cm-1 [Mel97, Tab96], showing the differences of the residual stress in the epitaxial film. It would be helpful to have the freestanding GaN E2 mode as a standard since there would be no substrate effect. Th e reported shifts of the E2 mode for freestanding GaN samples, however, show a rela tively wide range from 565 to 570 cm-1[Mel97, Tab96, Dav97, Age97]. Freestanding GaN films grown by HVPE technique may c ontain high impurity levels (mainly oxygen), resulting in hydrostatic stress. The E2 mode of stress-free GaN is known as 568 cm-1 at room temperature that was measured from a thick (50 70 m) GaN film grown by chloride-hydride vapor phase epitaxy (CHVPE) [Dav98]. More discussion about Raman E2 peak position is presented in Table 1-2 in Chapter 1. The residual stress in both epitaxial films wa s measured by Raman spectroscopy. The depth-dependent Raman measuremen ts of MOCVD and H-MOVPE deposited GaN films were carried out by a collaborator (Se ok-Ki Yeo in Dr. Chinho Parks group, Yeung Nam University, Korea) using a Reni shaw System-2000 (located in Miryang

PAGE 101

78 National University, Korea) in confocal mode The laser output power was set at 20 mW with an excitation wavelength of 514.5 nm. The focused beam size was approximately 2 m, accumulation time was 10 sec, and a CCD de tector was employed. By controlling the slit width and CCD pixel size, signals fr om different depths could be resolved. Figure 3-7 shows the measured Raman shifts of GaN E2 mode of (a) MOCVD GaN surface, (b) H-MOVPE GaN surface, a nd (c) the depths profiles. Figure 3-7. Raman E2 peak shifts at the surface (a ) MOCVD, (b) H-MOVPE, and (c) depth profile. In the case of the MOCVD grown film, the E2 peak varied from 568 cm-1 (GaN surface) to 569 cm-1 (GaN/sapphire inte rface). In the case of the H-MOVPE grown sample, however, the E2 peak varied from 570 cm-1 (GaN surface) to 575 cm-1 (GaN/sapphire interface). This indicates that the H-MOVPE film was in a state that was more compressive than that of the MOCVD sample. It is commonly assumed that HVPE 567 568 569 570 571 572 573 574 575 576-0.300.30.60.91.21.51.82.12.42.733.33.6Distance from the Surface [m]Raman Shift (E2 mode) [cm-1] MOCVD H-MOVPE GaN SurfaceGaN/sapphire interface (a) (b) (c)

PAGE 102

79 grown epitaxial film may ha ve higher oxygen content because the growth takes place at atmospheric pressure, whereas MOCVD grown film usually contains less oxygen due to reduced pressure. However, MOCVD in th is case is referring HMOVPE without HCl. No difference in oxygen content was expected since the same reacto r was used at the same pressure. The AES results agreed with this assumption. It is reported that a biaxial (c and a axes) stress of 1 GPa shifts the E2 Raman mode by 2.7 .3 cm-1 [Dav97] and a hydrostatic stress of 1 GPa shifts it by 4.17 cm-1 [Per92], and these values were used to calculate th e stress of the epitaxi al film. The MOCVD grown GaN was considered as stress-free due to the Raman E2 position appearing at 568 cm-1. The H-MOVPE grown sample, however, may contain compressi ve stress judged by high frequency shift (568 570 cm-1) of E2 peak. If the 2 cm-1 of shift was purely due to the biaxial stress, the st ress in the GaN film would be 0.74 GPa. Likewise if the 2 cm-1 of shift was caused by only hydrostatic stre ss, the stress in the GaN would be 0.48 GPa. Therefore, other stress measurements should be carried out to compare the values and find the origin of stress. The curvature and the lattice parameters of the epitaxial film were measured to cross-check the stress values obtained by the other techniques. Stoneys equation [Sto09] was used to calculate the curvature of the substrates, and Shens formula [She02] was used to estimate lattice parameters afte r obtaining the stress value from Raman spectroscopy. 3.3.2.1 Curvature Calculations From the Raman spectroscopy results, the cu rvature of the film was calculated and compared to the other measurements. It was shown that the stress of the 3 m HMOVPE GaN film on sapphire s ubstrate is substantial and thus substrate bowing was

PAGE 103

80 expected. The relation between the film stress and substrate curvature () is given by Stoneys equation, = 6f hf /(Ms hs 2) [Sto09]. In this expression, f is the biaxial stress in the film, hf hs are the thicknesses of the film (3 m: H-MOVPE, 1 m: MOCVD) and substrate (300 m: sapphire), respectively. Ms is the substrate biaxial modulus, which is 602 GPa for sapphire [Hel92]. The radius of curvature was calculat ed as 4.1 m by using 0.74 GPa (biaxial stress). Sim ilarly, the radius of curvatur e was calculated as 6.3 m by assuming 0.48 GPa (hydrostatic stress). 3.3.2.2 Lattice Parameter Calculations From the stress measured by Raman spectro scopy, the lattice parameters were calculated with proper assumptions. Shen et al. [She02] organized equations for hexagonal GaN epitaxial thin films assuming isot ropic strain in the epitaxial film. These equations predict the changes in the stress of the film with respect to the growth plane when both the growth plane and the underlyi ng substrate are hexagonal. Substituting and rearranging the origin al equation gives: 5 0 12 11 0 0) (sE s s a a a (7) The range in the reported values for th e elastic compliances in equation (7) are s11 = 0.00297 to 0.00502 and s12 = -0.00066 to -0.00121 GPa-1 [Dav98. Tal96, Pol96, Sav78]. The range-centered values s11 = 0.003995 and s12 = -0.000935 were used in the calculation. The values of a0 are the unstrained GaN latti ce parameters (3.189 for MOCVD and 3.067 for H-MOVPE film s assuming hydrostatic strain) and a is the lattice parameter of the strained GaN epitaxial films that is a function of stress (Es). The lattice parameter value of strained GaN assumi ng biaxial stress will be much smaller than 3.067

PAGE 104

81 Now that we have calculated curvature and lattice parameters by stress data from Raman spectroscopy, these values will be compared with other techniques. 3.3.3 Lattice Parameter Measurements by XRD Reciprocal Space Mapping The most straightforward method to measur e the strain of the films is measuring the lattice parameters of the film. XRD reciprocal space mappi ng is an effective tool to measure the lattice parameters. In XRD reciprocal space mapping, a series of -2 scans are collected with changing sample tilt offsets (). Any strain of ep itaxial layer would be shown as an asymmetrical shape of contour lines. The crystallographic plane spacing, d, of a hexagonal structure is given by equation (4). 2 2 2 2 2 23 4 1 c l a k hk h d (4) where d is the lattice spacing, h, k, and l are the Miller indices (h k l), a and c are the inplane and out-of-plane lattice pa rameters, respectively. To measure the lattice parameters of the film accurately, at least two non-para llel crystallographic plan es should be chosen and space maps taken. The Philips XPert MRD instrument was used to measure the reciprocal space mapping by using the triple ax is monochrometer for exclusive use of Cu K 1 spectrum. Braggs law is give n as equation (5). sin 2 d (5) where is the wavelength of the X-ray excitation source (Cu K 1 at 1.54056 ), d is the lattice spacing that is gi ven as equation (4), and is the Bragg angle that is to be measured by XRD space map. Figure 3-8 shows XRD reciprocal space maps of

PAGE 105

82 symmetry peak (002) and asymmetry peak (114) from the GaN film grown by HMOVPE. The measured 2 values of (002) and (114) peaks were 34.56460 and 99.97260, respectively. Plugging in the values to the above equations gave lattice parameters a = 3.187 and c = 5.186 which are similar to literature values. Figure 3-8. XRD reciprocal space mapping of H-MOVPE grown GaN (002) and (114) peaks. Similar measurements were performed on the MOCVD grown GaN sample as shown in Figure 3-9. The values of 2 = 34.55500 and 99.941870 were obtained for ( 002 ) (114)

PAGE 106

83 (002) and (114) peaks, respectively to give lattice parameters a = 3.188 and c = 5.187 It should be noted that the lattice para meters obtained by reciprocal space mapping are average values of the film from the X-ray interaction volume i.e., the reciprocal space mapping cannot measure the depth profile. An XRD reciprocal spa ce mapping, however, can give qualitative information on variations of the lattice parameters with depth. Figure 3-9. XRD reciprocal space mapping of MOCVD gr own GaN(002) and (114) peaks. Thus the c-axis, the upper line spacing is sm aller and lower line spacing is wider. The results suggest that the c-axis lattice parame ter had various values. The vertical cross (002) (114)

PAGE 107

84 section of the reciprocal sp ace mapping can be thought as a -2 rocking curve. Because it is shifted to a higher value, it can be concluded that the distribution of the lattice parameters will be on the smaller side (i.e., under compressive stress). Table 3-4 presents selected reported la ttice parameters of GaN [Mel97]. The values obtained by XRD-SM showed a = 3.187 and c = 5.186 for H-MOVPE GaN and a = 3.188 and c = 5.187 for MOCVD GaN. These values are similar to bulk GaN crystal values. Therefore, it was c oncluded that the GaN film grown by H-MOVPE is almost strain free but under slight compressive strain (not enough to measure the lattice strain quantitatively), whereas the fi lm grown by MOCVD is strain free. Table 3-4. GaN a and c lattice parameters measured at 300 K (). Samples description a () c () Thin (<1.5 m) GaN layers grown by HVPE on SiC [Mel97] 3.1937-3.1969 5.1786 -5.1811 Bulk GaN crystal grown by HVPE [Mel97] 3.1893 (Ga face) 3.1879 (N face) 5.18560 (Ga face) 5.18534 (N face) Annealed bulk GaN crystal grown by HVPE [Mel97] 3.1890 5.18500 Thick GaN layers grown by HVPE on sapphire [Det92] 3.1892+ 0.0009 5.1850+ 0.0005 Bulk GaN crystals grown at high pressure [Les96] 3.1890+ 0.0003 (Ga face) 3.1881-3.1890 (N face) 5.1864+ 0.0002 (Ga face) 5.1856-5.1864 (N face) [Mel97] 3.3.4 Curvature Measurements by XRD Rocking Curve The most widely used method for measuring residual stress of the film is curvature measurement. In situ curvature measurements are usually carried out using laser reflectance. In addition, there are a number of ex situ curvature measurements, which include optical and stylus profilometer (alpha step), laser reflectance, and XRD rocking curve. Curvature measurement by reading an XRD rocking curve is useful regardless of

PAGE 108

85 the surface roughness. Since the surface of HVPE and H-MOVPE grown films are often rough, measuring the XRD rocking curve is the preferred choice. The curvature measurement procedure is illustrated in Figure 3-10. The XRD rocking curve was initially obtained with th e normal alignment procedure. After the alignment of the rocking curve, the sample pos ition was changed precis ely. If there was curvature in the sample, the rocking curv e was then realigned, and the value of was obtained. The different values can be used to calcula te the curvature of the sample. The radius of curvature can be calculated by equation (6). x R (6) where R is the radius of curvature, x is the displacement of the sample, and (radians) is the difference of omega values upon displacement of the sample. Figure 3-10. The procedure of determining curvature by measurements of XRD rocking curves.

PAGE 109

86 The XRD rocking curve measurement method was used to determine the curvature of the 3 m thick GaN grown by H-MOVPE. By displacing the sample position precisely by 5 mm in the x-direction, the value changed from 17.26163 to 17.28132, as shown in Figure 3-11. Using the given equati on, the calculated radius of curvature of 14.5 m was obtained, which is pr actically a flat surface. The results agreed well with the measurements by XRD reciprocal space mapping in that there is not significant strain in the H-MOVPE grown GaN film. Figure 3-11. Curvature measurements by XRD rocking curve by displacing the HMOVPE grown GaN/c-Al2O3 sample in x-direction by 5 mm. Table 3-5. Stress, lattice parameter, and cu rvature of the H-MOVPE GaN with depth. Distance from the surface [m] Stress [GPa] by Raman factor Radius of Curvature by Raman Radius of Curvature by XRD RC Lattice parameter by Raman Lattice parameter by SM 0 (surface) and 0.6 (bulk): two points showed identical E2 shift 0.74 (biaxial) 0.48 (hydrostatic) 4.1 m (biaxial) 6.3 m (hydrostatic) 14.5 m 3.067 3.187 The radius of curvature calculated from Raman spectroscopy showed 4.1 to 6.3 m. It was also measured by XRD rocking curve (14.5 m) and stylus profilometer (15 m).

PAGE 110

87 Considering these results, it appears that the cu rvature of the film is ~ 15 m and thus is practically flat. However, the lattice para meters deduced from the stress measurement by Raman spectroscopy and the ones directly from XRD reciprocal sp ace mapping exhibited inconsistent results. Therefore, the stre ss of the film measured by Raman spectroscopy may not be from lattice strain (i.e., a change in lattice paramete r). To elucidate the reason for the existence of compressive stre ss measured by Raman spectroscopy without changing lattice parameters, the chemical compositions were examined. No significant difference between the MOCVD and H-MOVPE samples was observed by both XRD and AES anal ysis as in Section 3.3.1. Therefore, secondary ion mass spectroscopy (SIMS) was used to measure the chemical compositions since AES is not as sensitive as SIMS if impurities compose less than 1% of the matrix. SIMS depth profiles were acquired with a Perkin-Elmer PHI 6600 SIMS system using a 5 keV Cs primary ion beam and negative secondary acqui sition. The current intensity was set at 158 or 90 nA and the raster size was 350 x 350 m2. Several profiles for each sample were obtained from various sampling spots. Quantification of impurities by SIMS requires using standards with known concentrati ons of the species in the same matrix as the samples of interest, because SIMS is a very matrix dependent technique. In other words, when using SIMS, the sensitivity of the same element varies with matrix composition. When standards are not availabl e, it is possible to compare the intensity levels between samples by measuring the relative intensity of the impurity and a matrix ion. Therefore, to compare the oxygen level be tween samples, the ratios of O intensity to the GaN intensity were calculated for each sample.

PAGE 111

88 Figure 3-12 shows SIMS oxygen depth pr ofiles of MOCVD and H-MOVPE grown GaN films. It is clear that the O concentr ation of MOCVD film is higher than in the HMOVPE film. Regardless of the higher oxygen content, the MOCVD GaN samples showed no stress by Raman spectroscopy and XR D. On the other hand, the H-MOVPE GaN samples with lower oxygen content showed significant compressi ve stress measure by Raman spectroscopy, but no stress by XRD. It is known that Raman E2 peak shift is mainly related to the biaxial stress or hydrostatic stress in the film th at may be related to the oxygen concentration. It a ppears that, however, the Raman E2 peak shift may not be related to biaxial stress or oxygen concentr ation in this case, because the lattice parameters measured by XRD did not change. Figure 3-12. SIMS oxygen depth profiles of MOCVD and H-MOVPE grown GaN films. The stress-free Raman E2 peak positions are still in question, whether it is 570 (HMOVPE) or 568 cm-1 (MOCVD). It was expected that the MOCVD GaN would have less oxygen as it showed an E2 phonon peak at 568 cm-1, and was thus assumed to be stress-free GaN. However, H-MOVPE GaN showed a much lower oxygen concentration

PAGE 112

89 with a Raman E2 phonon peak at 570 cm-1, which is supposed to indicate significant compressive stress. Another approach is to consider 570 cm-1 as the stress-free E2 mode and that 568 cm-1 indicates tensile stress that is relate d to the growth mode It should be noted that the term MOCVD here is not the same as typical MOCVD, which produces superior quality materials, but H-MOVPE without HCl, utilizing a hot wall in atmospheric pressure. The MOCVD grown GaN may be growing by island coalescence growth mode although there is no evidence of island growth and structural quality confirmed it is epitaxially grown film. It is known that significant tensile stress generates at the grain boundaries of is lands. If H-MOVPE GaN was stress-free and MOCVD GaN film is under tensile stress, this may make sense. However, the reason for this discrepancy is not clear yet. It should be noted that these measurements were performed on almost 6 years old samples. Therefore some oxidation/degradat ion may have taken place and this may caused the inconsistent results. 3.4 Stress Modeling of GaN on Sapphire After measuring stress in GaN film on sapphire it is useful to cal culate the stress in GaN film using traditional and modified st ress modeling. By comparing the calculated results with experiments, the most important factors to induce st ress in film can be identified. Fran-van der Merwes semi-inf inite model and Shen-D anns Layer-by-layer growth model were applied in this study and compared with experiments. 3.4.1 Frank-van der Merwes Semi-infinite Model Misfit dislocation density was calculated using the Frank-van der Merwes semiinfinite overgrowth model, as represented in Figure 3-13. In this model, the strain will be completely relaxed by dislocation formation after p/2 thickness, and the film has no effect

PAGE 113

90 on the substrate beyond this thickness. Here p, the Vernier period of mi sfit dislocation, is defined as equation (8). f s f sa a a a p (8) where as and af are the lattice parameters of film and substrate, respectively. Figure 3-13. Frank-van der Merwes semi-infinite overgrowth model. In this model, only pure edge dislocati ons were considered and only form if sufficient strain energy is available to pr ovide the formation energy. The value of p/2 is ~ 9.3 nm for GaN on 6H-SiC. The p/2 value of the other substr ates such as sapphire and Si will be even smaller than this. Table 3-6. Calculated misfit dislocation dens ity of GaN on widely used substrates with Vernier period. p () Misfit dislocation density (cm-2) GaN 0 c-Al2O3 20.221 2.4 x 1013 Si (111) 19.177 2.7 x 1013 6H-SiC 87.011 1.3 x 1012 Table 3-6 shows calculated Vernier periods with misfit dislocation density of GaN on c-Al2O3, Si (111), and 6H-SiC. The Vernie r period can be thought as 1-D misfit

PAGE 114

91 dislocation density. Therefor e 2-D misfit dislocation dens ity was calculated from the square of the Vernier pe riod by assuming uniform dist ribution of misfit. It should be noted that this simple mode l agreed fairly well with experiments for the case of a Ge deposit on Si(001) [Mar03]. Ho wever, the variation of in-plane lattice parameters of GaN grown on AlN/Al2O3 was experimentally observed by XRD up to 1 m thickness [Kim99]. Therefore, Frank-va n der Merwes semi-inf inite model could not be applied to the GaN/Al2O3 system at device-re levant thicknesses. 3.4.2 Layer-by-Layer Growth Model A better model that describes gradual cha nges in lattice parameters and gradual creation of misfit dislocation was developed by Shen and Dann [She06] and illustrated in Figure 3-14. Figure 3-14. Shen-Danns layer-by-layer model. When overgrowth material is deposited on a foreign substrate, it will experience homogeneous stress without form ing dislocation as long as th e thickness is smaller than the critical thickness. This stress will proport ionally increase with increasing numbers of layers until finally the material will form dislocations to relax the stress when the Dislocation forming Homogeneous stress build-up Critical thickness T T T Homogeneous stress build-up Edge dislocation substrate strained film (not fully relaxed) strained film substrate strained film substrate strained film (not fully relaxed) strained film (not fully relaxed)

PAGE 115

92 thickness reaches a critical thickness. Dislocations will form periodically, but the inplane lattice parameters do not necessarily become their unstrained values, unlike Frankvan der Merwes semi-infinite model. An underlying assumption is that the bonds at the interface are much weaker than the bonds in the overgrown films and the substrates. After it reaches a critical thickness, the la ttice constant will be closer to the bulk (unstrained) material. The consequent growth after the critical th ickness will be a similar process. Here it was treated that the disloc ation formed layer as a new substrate. The new substrate will match the lattice of the overgrowt h film better than the original substrate. Therefore, the critical thickness will be thicke r. Nevertheless, it will experience the same stress relaxation mechanism by forming dislo cations again, until it reaches the second critical thickness through the accumulation of stress. Stress (homogeneous strain energy) pr oportionally increases with increasing numbers of layers, while the dislocation (p eriodic strain energy) number accumulates. The sum of these two energies will have a minimum at certain poi nt as the number of layers is increased. Finding this minimum en ergy with respect to the sum of strain and dislocation energy will give gradual lattice parameter changes. 3.4.2.1 Strain Energy (Homogeneous Stress) Calculations The procedure to obtain the strain en ergy of overgrowth Ga N on sapphire and LGO has been calculated by Shen et al. [She02]. Using the latti ce parameters provided in Table 3-7 and elastic constants in Table 3-8 [Tak96], the stra in energy calculations of a GaN film grown on AlN/Al2O3 were performed. Since both film and substrate are hexagonal and the basal plane of each material is used in epitaxial growth, the strain in x and y directions are symme trical and can be simply written as equation (9).

PAGE 116

93 0 0a a ay x (9) where a and a0 are the lattice parameters of the growth layer and th e substrate in the basal plane, respectively. Following Shens work [She02], the strain energy per unit volume is expressed as equation (10). 12 11 2 2 0 0 12 111 s s a a a s s Ux (10) Table 3-7. Lattice constants () of GaN, AlN, and Al2O3. Material a c GaN (wurtzite) 3.189 5.185 AlN (wurtzite) 3.112 4.980 Al2O3 (wurtzite) 4.758 12.991 Table 3-8. Elastic stiffness coefficients cij (GPa) and compliances sij (1/GPa) of GaN, AlN, and Al2O3. c11 c12 c13 c33 c44 s11 s12 s44 GaN 374 106 70 379 101 0.00297-0.00077 0.0099 AlN 424 103 71.3 455 138 0.00255-0.00057 0.0072 Al2O3 495 160 115 497 146 0.00232-0.00066 0.0068 [Tak96] To compare the calculated results with experimental results, AlN buffer layer on Al2O3 substrate was considered as the substrat e, because AlN buffer layers are widely used between GaN and Al2O3 substrate. Kim et al. [Kim99] employed X-ray diffraction to investigate the strain relaxation of GaN/ AlN on sapphire. A thin 32 layer of AlN was grown on a sapphire s ubstrate followed by a subsequent GaN layer. a0 = 3.084 inplane lattice constant of the 32 film of AlN on sapphire was measured as their experimental data. This la ttice constant of AlN on Al2O3 was considered as the starting value for a0 in the strain energy calculation.

PAGE 117

94 3.4.2.2 Dislocation Energy (Periodic Strain Energy) Calculation Van der Merwe [Mer63] has derived the di slocation energy per unit area and Jesser [Jes67] reorganized the result as equation (11). Since the dislocation energy per unit area is assumed to be symmetrical, one di mensional dislocation was defined as xEd with unit of J/m. This value is multiplied by the speci fic area (area per mole) to have energy per mole unit. G G G P b G where b G Ex x x x x x x x x d x 2 1 1 1 2 2 / 1 2 2 / 1 2 2 11 ) 1 ( 2 ]}, 2 ) 1 ( 2 ln[ ) 1 ( 1 { 4 (11) G1 indicates the shear modulus of AlN and G2 denotes that of Al2O3. G without a subscript represents the shear modulus of GaN, is Poissons ratio and bx (Burgers vector) is given as equation (12): 1 2 2 12 a a a a bx (12) Finally, P is the Vernier periodicity of dislocations and denoted by equation (13): 1 2 2 1a a a a P (13) The required elastic properties for the calculations are listed in Table 3-9. Table 3-9. Shear moduli, Poisson's ratios, and lattice parameters of GaN, AlN, and Al2O3. G a () GaN 1.01 x 1011 0.2578 3.189 AlN 1.38 x 1011 0.2224 3.112 Al2O3 1.46 x 1011 0.2848 4.758 [Tak96]

PAGE 118

95 3.4.2.3. Total Strain Energy Calculations The total strain energy is defined as the sum of homogeneous strain energy, periodic strain (dislocation) energy, and interface energy as equation (14). Etotal = Estrain + Edisloc + Einterface (14) The strain energy per unit area and the dislo cation energy were calc ulated by equations (10) and (11), respectively. The interfaci al energy was neglected because the area between the layers was assumed to be neg ligible compared with total strain and dislocation energies. As th e number of layers increases, the strain energy proportionally increases as in equation (15): Estrain = n 6,230 kJ/mol 2 (15) where n = the number of layers and 1/(s11+s12)GaN = 454.1 GPa, or 6,230 kJ/mol was used in the calculation. It should be noted that is only a functi on of the lattice parameter of GaN for a given substrate. Hen ce, strain energy is onl y a function of lattice parameter change. The dislocation ener gy of GaN on AlN/Al2O3 was calculated by equation (11) and the sum of strain energy and th e dislocation energy was take n as total energy (stress in the GaN/AlN/Al2O3 system). As the number of layers increases, the total energy decreases. At certain point, the total energy will be minimized by the adequate contribution of strain and dislocation energy as shown Figure 3-15. From this calculation, the lattice constant that results in a minimum of the total energy could be found and is often turned the critical thickness. The calculated lattice constant was used as a new substrates latti ce constant and the same steps were repeated

PAGE 119

96 until it reached the unstrained GaN lattice cons tant, 3.189 The lattice constants and the minimum total energy with the number of la yers are listed in Table 3-10. The term critical thickness in Table 3-10 represents the thickness of GaN film from the substrate/film interface. Figure 3-15. Minimum energy calculati on of stress + dislocation energy. Table 3-10. Calculated latti ce constants and total energy. Number of layers Lattice consta nt () Critical Thickness () Etotal (J/mol) 1 3 3.135 10.45474 18740 4 10 3.157 15.62296 12425 11 23 3.167 36.38896 8328 24 44 3.174 67.52497 6299 45 77 3.178 109.0171 4624 78 125 3.18 171.2573 3608 126 186 3.182 249.0612 3603 187 269 3.184 316.4641 2509 270 399 3.186 430.5291 1961 400 703 3.188 674.2136 1436 The in-plane lattice constant versus the thickness of GaN layer are compared with experimental data [Kim99] in Figure 3-16. In their study, sets of GaN films were grown on Al2O3 substrates using AlN buffer layer by MBE. The lattice parameters of thin AlN

PAGE 120

97 and GaN films were determined by ex situ X-ray reciprocal space mapping. Poissons ratio for GaN was applied to calculate the out-of-plane lattice constants affected by lateral deformation. As in-plane lattice constants increase, th e out-of-plane lattice parameters decrease. Figure 3-16. In-plane lattice c onstants of GaN on AlN/sapphire. The calculated total strain energy at the interface was added to the Gibbs energy of GaN and the decomposition temperature of GaN was lowered by 44 K from 1110 K to 1066 K by Thermo-Calc. The unstrained Ga N thermodynamic data were taken from [Unl03]. By adding the stra in energy to the Gibbs ener gy of GaN, the decomposition temperature of strained thin GaN film on various substrates can be predicted. 3.5 Conclusions The effect of lattice mismatch was calcula ted at growth temperature and showed that critical thickness is only important when smaller lattice mi smatch system is considered. Critical thickne ss calculation does not have much meaning when GaN film

PAGE 121

98 on any widely used substr ates such as SiC, Al2O3, and Si were concerned since misfit dislocations forms at the interface. Effects of thermal expansion mismatch and lattice mismatch at growth temperature were considered and it was found that vari ous substrates woul d induce compressive and/or tensile stresses on GaN, due to the combination of lattice and thermal expansion mismatches. Several stress measurements were ca rried out on two GaN samples grown on sapphire substrate by H-MOVP E and MOCVD. Although Ra man spectroscopy is known to be a good technique to m easure the stress of the materi als, considerable stress difference between the two samples showed that Raman spectroscopy in this case may not be an adequate technique to measure the stress of the film due to the inconsistent results with other techniques. Lattice parameters of Ga N were precisely measured by HR-XRD reciprocal space mapping and no si gnificant difference between the two films were confirmed. Further curvature m easurement by XRD and stylus profilometer confirmed the samples were almost completely flat. The crystal structure and chemical compositions of the samples were charac terized by XRD rocking curves, XRD pole figures, and AES, but no signi ficant difference was observe d. Only SIMS showed varying oxygen concentrations between the tw o samples. The peak position of Raman E2 mode may be related to not only the strain of the film, but also oxygen concentration although the results showed obvious contradiction. It is not clear at this point whether Raman peak shifts were mainly related to th e oxygen concentration or other factors such as growth mode difference, which may cause tensile stress at the gr ain boundaries. Most of all, analyzing almost 6 year old samp le was not a good approach since the samples

PAGE 122

99 may be partially experienced degradation and oxidation. Similar pr ocedures with fresh samples will provide fair comparison of various stress measurement techniques. Frank-van der Merwes semi-infinite mode l was used and it was found that the model is not applicable to the GaN/Al2O3 system at device-relevant thicknesses. Strain and dislocation energy mode ling was carried out using Shen-Danns layer-by-layer growth model, based on total energy mini mization, which agreed well with the experimental data. The decomposition temper ature of strained GaN was decreased by 44 K due to the inclusion of the total strain energy.

PAGE 123

100 CHAPTER 4 GROWTH OF GALLIUM NITRIDE ON SILICON BY H-MOVPE 4.1 Introduction The growth of GaN films on a Si single crystal wafer has been pursued for over 30 years (i.e., [Mor73]). Although the quality of GaN on Si was not good in the early attempts, there was motivation to grow GaN on Si since Si is th e highest quality and lowest cost substrate material available. Large size, high quality, both p and n-type doping, and low cost are some of the advantages of Si substrates. The availability of Si substrates in large size (maximum 12 inch diameter) is especially attractive when mass production of high-volume, low co st GaN/Si-based devices such as LEDs are considered. The two main goals of growing GaN on Si are: (i) growth of thin (4 to 5 m) crack-free GaN film on Si. and (ii) gr owth of freestanding ( > 100 m ) GaN using Si wafers as a foundation. The former is needed for GaN/Si based devices such as LEDs. The large diameter of Si will reduce device fabricati on costs compared to sapphire substrates. Moreover, the heat generated from the LEDs would drain more easily through Si because it has a much higher thermal conductivity (145 W/mK) than Al2O3 (5.43 W/mK). The growth of GaN on sapphire has been limited to 2 diameter wafers. Although large diameters of sapphire are avai lable, the wafer bow on large diameter wafers produced by the mismatch prevents device processing. Furthermore, sapphire is not electrically conducting so top-side contacts are needed. Larger diameter growth of thick GaN on Si or sapphire has been impeded due to the sign ificant cracking problems. If freestanding

PAGE 124

101 GaN could be grown on 12 inch Si wafer w ithout cracking and bowing, it will be an important breakthrough for GaN-based device development. A GaN film on Si is under tensile stress during growth and under additional tensile stress during the cooling due to the lattice and thermal expansion differences. Tear-like cracks can form to accommodate excess tensile stress. Therefore, the growth of crackfree GaN film on Si substrate is a challeng e. The reported maximum thickness of crackfree GaN grown on Si with multiple AlN interlayers was 7 m, as measured by in situ curvature measurement. [Kro03]. In this chapter, the H-MOVPE technique will be introduced first. Chemical reactions and a reactor schematic will be shown to give insight into the H-MOVPE technique. For GaN film growth on Si, ba re Si substrates and GaN/AlGaN/AlN/Si templates were used. The templates were grown by MOCVD tec hniques and provided by Nitronex. The goal of the pres ent study is to grow both th in and thick GaN films on Si substrates by H-MOVPE. By utilizing a relatively low growth temperature (850 C) rather than MOVPE growth temperature (typically 1050 C), thermal stress can be reduced by ~ 20 %. For crack-free, thick GaN growth, InN was used as a buffer material since InN is known to be a softer material th an GaN and Si. To st udy structural effects, InN columnar films, nanorods (d = 250 nm), thick nanorods (d = 500 nm), and microrods were tested as buffer materials to give a compliant layer between GaN and Si. 4.2 H-MOVPE Growth Technique 4.2.1 Chemical Reactions of H-MOVPE Technique Hydride Metalorganic Vapor Phase Ep itaxy (H-MOVPE) has proven to be a promising technique to grow GaN films on LiGaO2, LiAlO2, and Si utilizing the high

PAGE 125

102 growth rate and easy precursor switching afforded by this process [Ree02, Mas01, Kry99]. Trimethylgallium (TMG), ammonia (NH3), and 10 % HCl (balance 90 % N2) were used as the precursors. N2, H2, or 4% H2 (balance 96 % N2) were typically used as carrier gases. A thermodynamic analysis was carried out to better understand the expected gas phase species and condensed phases in the Ga-C-H-Cl-N-Inert a nd In-C-H-Cl-N-Inert systems. The results are presented in Chapter 2. The overall reactions for GaN gr owth by H-MOVPE technique are: Ga(CH3)3(g) + HCl(g) + H2(g) = GaCl(g) + 3CH4(g) (R1) Ga(CH3)3(g) + HCl(g) = GaCl(g) + CH4(g) + C2H6(g) (R2) Ga(CH3)3(g) + 3HCl(g) = GaCl3(g) + 3CH4(g) (R3) GaCl(g) + NH3(g) = GaN(s) + HCl(g) + H2(g) (R4) GaCl3(g) + NH3(g) = GaN(s) + 3HCl(g) (R5) TMG reacts with HCl in a H2 ambient and produces pr esumably GaCl and CH4 (R1). When H2 is not present, GaCl can still form with CH4 and C2H6 as by-products (R2). TMG can react with excess HCl (Cl/Ga 3) and form GaCl3 (R3) especially at low temperature (< 500 K) as shown in thermodynami c analysis (Figure 25). Both GaCl and GaCl3 can react with NH3 and form GaN with HCl and/or H2 as by-products (R4, R5). It is also known, however, that it if difficult to thermally decompose NH3 at low temperature because of slow kinetics. 4.2.2 H-MOVPE Reactor Schematics The H-MOVPE reactor schematic is s hown in Figure 4-1. There are six individually controllable furnace zones, as s hown in Figure 4-1 (a). The furnace elements

PAGE 126

103 have a capability of heating to 1300 C, but the practical growth zone temperature limit was 1000 C because quartz may soften over 1000 C. The typical growth temperature for GaN was 850 C for high temperature growth and 560 C for low temperature growth, as these were the optimized growth temperatures suggested by Reed [Ree02]. Figure 4-1. Schematics of H-MOVPE (a) bird eye view of the enti re reactor, (b) the source and growth zones with temperatur e profile in the source zone [Ree02]. To prevent TMG decomposition before it reacts with HCl, the maximum inlet temperature is maintained below 400 C, as shown in Figure 4-1 (b). The source zone consists of tri-concentric quartz tubes. The input gase s do not meet each other until a certain residence time passes. Ideally, the reaction products of the source zone, i.e., GaCl Six individually controllable furnaces (a) (b)

PAGE 127

104 or GaCl3, will react with the NH3 on the surface of the substrat e. In the less ideal case, GaN is formed in the gas phase and GaN particles will drop on the substrate surface, which then act as 3-dimensional nuclei. Ther efore, it is very important to control the source zone conditions (i.e., inlet gas ratios, concentric inlet tilting, the distance from the inlet to the substrate) to give optimal and reproducible epitax ial growth conditions. Figure 4-2 shows the Process Flow Di agram (PFD) of the H-MOVPE system located in the Microfabritech, research area at the University of Flor ida. Its capabilities include 3 bubblers, 3 toxic/ corrosive gas cylinders, H2, N2, and an additional carrier gas such as He or Ar. House N2 is always provided by liquid N2 boil-off to prevent lines from clogging and to keep the reactor free of moisture. There are a number of adva ntages to H-MOVPE. Quic k switching of the input sources is essential for multiquantum well (MQW) structure or atomic layer deposition (ALD). In traditional HVPE, although HCl flow is turned off to stop GaCl supply, a thin GaClx layer on the liquid Ga can st ill provides GaCl to the growth zone. As a result, rapid switching of the group III source is re stricted. The high growth rate (50 m/hr maximum) is also an advantage when thick film growth is desired. The large surface area of the hot quartz wall also helps NH3 cracking because NH3 is known to crack more easily on the surface. But N radical recombination will produce inert N2, and thus must be minimized before reaching the substrate. The freedom to control six individual temperature zones is advantageous to adjust the gas phase reactions. H-MOVPE has some disadvantag es as well. The costs of metalorganics are much higher than pure metal sources and the grow th rate (< 50 m/hr) is not as high as traditional HVPE ( > 100 m/hr).

PAGE 128

105 Figure 4-2. Process Flow Diagram (PFD) of H-MOVPE system [Ree02].

PAGE 129

106 The surface of an H-MOVPE-grown sample is usually rough compare to a MOCVD surface because of the high grow th rate and the tendency to grow N-terminated GaN, although this surface can be smoothed by ch emical-mechanical polishing (CMP) or subsequent growth of Ga -terminated GaN by MOCVD. 4.3 Thick GaN Growth on Al2O3 Before starting the growth of GaN on Si substrates, thick GaN films were grown on Al2O3 and GaN/Al2O3 substrates to produce reference materials to compare with GaN films grown on Si substrate. The GaN/Al2O3 template was grown by MOCVD and provided by Uniroyal optoelectronics. The thickne ss of the MOCVD-grown GaN f ilm was 5 m as measured by cross-sectional SEM. The thick GaN growth was performed at atmospheric pressure in a H2 ambient. Before the growth, HCl was applied in the reactor for 10 sec to clean the substrate surface in situ. The TMG flow rate was maintained at 3.2 sccm, HCl (10 % HCl, 90 % N2) flow rate was 4.8 sccm (Cl/Ga = 1.5), NH3 flow rate was 2000 sccm (N/GA =570), and H2 carrier flow rate was 1600 sccm. During the cooling process, NH3 was provided to prevent GaN decomposition until the substrate temperature was below 300 C. The crystal quality of the Uniroyal star ting GaN/sapphire materials was assessed by an XRD -2 scan and HR-XRD -rocking curve. Figure 4-3 shows the XRD results of the templates. It is noteworthy that the GaN (002) peak shows a secondary peak at around 2 = 31.1 due to the Cu K radiation. This secondary peak appears only when the film quality is excellent and should not be confused with GaN (100) that appears at 2 = 32.4 The intense GaN (002) peak from the near-perfect la ttice spacing is the

PAGE 130

107 reason for this artifact. XRD also showed weak Al2O3 substrate peaks because the GaN film thickness was almost the same as the X-ray penetration dept h (~ 5 m for GaN [Pas01]). Figure 4-3. XRD -2 scan and -rocking curve of as received GaN/Al2O3 template from Uniroyal Optoelectronics. denotes the secondary peak due to Cu K radiation. 101520253035404550556065707580859095100 2-thetaIntensity (a.u.)GaN (002) GaN (004) A l2O3(00 12) A l2O3(006) *GaN (002)

PAGE 131

108 Figure 4-4. SEM images of GaN f ilm (a) Cross-sectional SEM of 125 m thick HMOVPE GaN film on GaN/Al2O3 template, (b) SEM plan-view of the same sample. (c) Plan-view of LT-HVPE smoothing layer. A maximum thickness of 125 m was obtaine d during the 10 hr gr owth. The film showed no cracks, although the surface was very rough, as shown in Figure 4-4 (a) and (b). The rough surface is due to the high gr owth rate and N-terminated surface of GaN observed in HT-HVPE growth. Ther efore LT-HVPE GaN was grown at 560 C to smooth the extremely rough surface as seen in Figure 4-4 (c). Although the structural quality of LT-GaN may be lower than GaN grown at higher temperature, the surface became smoother as observed by SEM. Th e surface of LT-GaN, however, is still not smooth enough to be measured by AFM to obtain RMS roughness. Subsequent growth of high quality MOCVD layer on LT-HVPE la yer should be a good approach to smooth the surface and produce freestanding GaN substrate that can be used for epitaxial growth. Another approach to improve the surface r oughness without sacrificing the crystal quality (a) 10 m (b) (c) 10 m

PAGE 132

109 is chemical mechanical polishing (CMP). Thes e steps, however, were not carried out in this work. The bowing of the film was measured by pr ofilometer. The calculated radius of curvature was about 10 m, which is practically flat. (see Chapter 3 for discuss on method) Figure 4-5. XRD of GaN film (a) XRD -2 scan of thick (45 m) GaN on GaN/Al2O3, (b) XRD -rocking curve of GaN (002) peak; FWHM = 780 arcsec. Figure 4-5 (a) shows XRD -2 scan of a 45 m thick GaN on GaN/Al2O3 template. Only the intense GaN (002) and GaN (004) peaks were observed consistent with a single crystal orientation. Figure 4-5 (b) shows typical XRD -rocking curve of GaN (002) peak. The 780 arcsec FWHM i ndicates that the grow n GaN film has high structural quality. AES analysis was performed to determine the chemical composition of the GaN film (Figure 4-6). The surface of the as grown GaN film showed oxygen and carbon peaks due to the surface adsorbed molecules dur ing air exposure. The concentrations of C and O dropped significantly during the sputte ring. Sputtering was carried out with 3 keV Ar+ ion beam. Assuming a typical 10 nm /min sputtering rate, the thickness of C contained layer was ~ 10 nm. After sputtering, Ga, N, and O were the main components of the GaN film. (a) (b) 202530354045505560657075808590 2-thetaIntensity (a.u.)GaN(002) GaN(004)

PAGE 133

110 Figure 4-6. Auger Elect ron Spectroscopy of 45 m thick GaN film (a) surface scan (b) sputtering depth profile (c) su rface scan after sputtering. The concentration of O (9.9 %) in the f ilm was uniform as judged by AES sputtering depth profile after 4 min sputtering. Since a standard material that contains known value of oxygen in GaN is absent, the quantitative anal ysis by AES is not reliable. The extreme roughness (~ 0.8 m) of the film may be responsib le for the high oxygen concentration because sputtering only removes part of the f ilm when the film is very rough [Wan03]. In addition, the sputtered surface oxygen can re -adsorb due to the slow sputtering rate. The surface oxygen can be completely removed using higher sputtering rate as shown in Figure 3-6. Otherwis e, GaN film had a uniform compos ition as judged by the AES depth profile. Table 4-1 lists the electr ical properties of 45 m thick GaN film grown on Uniroyal GaN/Al2O3 template by Hall measurement. It showed intrinsic n-type material with a ( a ) ( c ) Time (mins.) Min: 0Max: 13047 0 02 4 6 8 10 12 14 O1 C1 Ga1 O1 C1 Ga1 N1 ( b )

PAGE 134

111 carrier concentra tion of 2.6 x 1020 cm-3 and Hall mobility (45.9 cm2/Vs). The electrical property measurements show that the H-M OVPE GaN films have very high background carrier concentration due to th e O contamination or N-vacancy. Table 4-1. Electrical properties of GaN film grown on GaN/Al2O3 template. Sheet Resistance [/square] Sheet Hall Coefficient [cm2/C] Type Thickness [m] Carrier concentration [1/cm3] Hall Mobility [cm2/Vs] 0.11727 5.3823 n 45 2.6 x 1020 45.9 Four-point probe was used to measure th e sheet resistance of the as received GaN/Al2O3 template and the GaN film s grown for 3 hrs on GaN/Al2O3. The results are shown in Table 4-2. Table 4-2. Sheet resistance and resist ivity measured by four-point probe. Sample name Sheet Resistance(/sq) Thickness (m) Resistivity ( cm) 3 hr growth on GaN/Uniroyal 3.3 30.3 0.01 GaN/Uniroyal as received 20.0 5 0.01 Sheet resistance is a property that depe nds on the thickness of the material, while resistivity is a material property that do es not depend on the thickness. Both films showed 0.01 cm resistivity. This fairly low re sistivity shows that the GaN film was fairly conducting. 4.4 GaN Growth on Si 4.4.1 Growth of Thin and Thick GaN on Si 4.4.1.1 Nitronex GaN/Si template The growth of high-quality thin (3 to 4 m) GaN on Si is important for GaN/Sibased device applications. Utilizing this relatively low (850 C) growth temperature, GaN could be grown with a de vice relevant thickness (3 to 4 m), and crack-free GaN on the GaN/AlGaN/AlN/Si template provided by Nitronex. In part, the lower growth

PAGE 135

112 temperature reduced the thermal expansi on effects by ~ 20 % over that of a high temperature (1050 C) growth. Figure 4-7. X-SEM image of GaN on Si using AlGaN graded layer (provided by Nitronex). Although ~ 7 m thick crack-free GaN on Si has been reported [Kro03] incorporating multiple AlN interlayers, the maximum thickness of a commercially available crack-free GaN layer on Si is about 1 m [We03]. In this case [We03], graded AlGaN interlayers were used between Ga N and Si. Trimethylaluminum (TMA) and ammonia (NH3) were used to grow the AlN buffer layer, and trimethylgallium (TMG) was added into the system to grow compos itionally graded AlGaN layer. Finally, the GaN layer was grown on the AlGaN graded layer at 1050 C. The thickness of the GaN film was 1 m and that of Al GaN graded layer was about 0.5 m as shown in Figure 4-7. XRD -2 scan (Figure 4-8 (a)) shows severa l AlGaN graded layer peaks with a strong substrate peak. The sharp and multiple AlGaN peaks indicate that the Ga/Al ratio in the graded AlGaN layer was not continuous ly changing but had di screte values. The GaN (002) peak of the received film determined by FWHM of the XRD -rocking curve, shows high crystal quality (800 arcsec), as shown in Figure 4-8 (b), but not as high

PAGE 136

113 quality as the GaN film grown on Al2O3. This material was used as a substrate for thick GaN growth. Figure 4-8. XRD -2 scan and HR-XRD -rocking curve of as received GaN/Si from Nitronex, FWHM = 0.222 (800 arcsec). (a) (b) 25303540455055606570758 0 2-thetaIntensity (a.u. ) Si (111) GaN (002) GaN (004) AlGaN graded layer (002) A lGaN graded laye r (004) FWHM = 800 arcsec

PAGE 137

114 4.4.1.2 Growth of Thin GaN Using Nitronex Template Under the selected growth conditions (Cl/Ga = 1.5, N/Ga = 570, T = 850 C), a 10 to 15 m/hr growth rate is observed. Film thicknesses were measured by cross sectional SEM from cleaved samples. To achieve devi ce relevant 3 to 4 m thick GaN, 15 min growth time was performed. Figure 4-9 show s the plan and cross-sectional views of crack-free, 2 m thick Ga N grown on GaN/Si (111). Figure 4-9. SEM plan and cr oss sectional views of 2 m H-MOVPE + 1 m MOVPE crack-free GaN grown on Al GaN/Si graded layer. Figure 4-10. XRD -2 scan of 2 m H-MOVPE + 1 m MOVPE crack-free GaN grown on AlGaN/Si graded layer. No cracks were observed in the 1 cm x 1 cm sample presumably due to the relatively low (850 C) growth temperature. The rough surface is a typical feature of HMOVPE. An average roughness of 830 nm wa s measured by profilometry (Dektak II). 1015202530354045505560 2-thetaIntensity (a.u.)Si (111) GaN (002) GaN (101) GaN (102)

PAGE 138

115 Figure 4-10 shows XRD -2 scan of 3 m thick GaN film (2 m HVPE + 1 m GaN template) on AlGaN/Si substrate. Intense GaN (002), small (101), (102) peaks, and Si (111) substrate peaks show that the GaN film was highl y textured along the [002] axis. 4.4.1.3 Growth of Thick GaN Using the Nitronex Template Thick (> 20 m) GaN growth was carried ou t using the same procedure but with a longer growth time of 30 min to 4 hrs. Most samples grown in H-MOVPE and thicker than 7 m on bare Si substrates and GaN/ Si templates showed cracks regardless of cooling rate or annealing process, as exemplified seen in Figure 4-11. Figure 4-11. SEM plan-views of thick (55 m) GaN films grown on GaN/Si templates (4 hr) with (a) fast cooling and (b) slow cooling. Note two micrographs are at different magnification. Cracks can be generated duri ng growth or cooling due to the excess tensile stress caused by large lattice and thermal expansion differences. It was observed that the cracks penetrated through the Si subs trate and separation occurred inside the Si, as shown in Figure 4-12. Similar phenomena often occur during the growth of GaN on GaN/AlGaN/AlN/Si template. Examining the cracked pieces s howed that the strong cohesion between GaN and Si (or AlN and Si in GaN/AlGaN/AlN/Si template case), as well as the brittleness of (a) (b)

PAGE 139

116 Si, were responsible for cracking to take place in pre interior of the Si wafer. The nanoindentation hardness of the GaN, AlN, and Si are 20 [Now99], 18 [Yon02] and 14 GPa [Suz96], respectively. This brittleness of Si added with the large tens ile stress (chapter 3) created by the lattice mismatch and thermal expansion differences makes the growth of crack-free GaN on Si even more challenging. Figure 4-12. Schematic of crack generation of GaN on Si and GaN on Nitronex template. Cracking penetration to Si we re observed in both cases. To determine the crystal quality, XRD -2 scan was performed using a Philips APD 3720. The resulting intense GaN(002) and GaN(004) peaks are shown in Figure 413. GaN Si Si GaN Si Si GaN AlN AlGaN g raded la y e r GaN AlN AlGaN g raded la y e r Si Si Si

PAGE 140

117 Figure 4-13. XRD -2 scan of thick (55 m) GaN film on GaN/Si template for 4 hr. Figure 4-14. XRD -rocking curve of GaN (002) peak of 55 m thick GaN. FWHM = 0.248 (892 arcsec). The result implies that the d-spacings of (002) and (004) planes are uniform. The -rocking curve result is shown in Figure 4-14. The FWHM of the -rocking curve of the GaN (002) peak is 0.248 (892 arcsec). These two results, XRD -2 scan and rocking curve, confirm that the grown GaN film is a high quality crystal (similar to the received GaN/Si template), although the film was cracked. 102030405060708090 2-thetaIntensity (a.u.)GaN (002) GaN (004) 16.6 16.8 17.0 17.2 17.4 17.6 17.8 18.0 Omega 0 2500 10000 22500 counts/s 17.29141 ; 31577.98 16.55890 ; 706.67 18.03890 ; 873.83 0.2479FWHM = 892 arcsec

PAGE 141

118 4.4.2 Growth of Thick GaN on Si Using InN Interlayers Thick GaN films grown on Si demonstrate significant cracking problems due to the excess tensile stress produced by large lattice and TEC mism atch between the GaN film and Si substrates. To overcome cracking probl ems, different techniqu es have been tried, including multiple AlN interlayers [Sch06, Ma s06, Yu06, Jam05, Zha05, Rei03, Lia01, Ven05], AlGaN graded layers [Mar01, We03, Abl05, Wan05], patterned Si [Zam01], and in situ SiN masking (non-uniform deposition) [Dad03]. All the methods showed some decrease of bowing and cracking, but no me thod successfully produ ced crack-free thick (> 20 m) films because there still remains exce ssive tensile stress, as well as strong cohesion between GaN (or AlN buffer layer) and Si. This investigator is not aware of any report of using InN as a buffer layer. To avoid cracks, InN was used as a buffe r material to provide a weak bond between Si and the interfacial layer, since InN is known to be a soft er material than GaN and Si (Nano indentation hardness: 11 [Edg97], 20 [Now 99], and 14 GPa [Suz96] for InN, GaN, and Si, respectively). Also the bond strength of In-N (7.7 eV) is similar to that of Si-Si (7 eV) and less than the Ga-N (8.9 eV) or Al-N (11.5 eV) and Si-N (10.5 eV). The bond strength of Si-Si is the weakest. Therefore, the cracking penetration to Si substrate is expected and was observed as shown in Figure 4-12. Taking advantage of the low decompositi on temperature and softness of InN, indirect/weak contacts between GaN film and Si substrat e could be established via the InN interlayer. InN buffer materials a nd GaN films were all grown by H-MOVPE technique. Easy switching of precursors ma de it possible to grow a variety of InN structures and GaN films wit hout transferring to another r eactor. InN was grown on Si,

PAGE 142

119 followed by a thin low temperature (LT)-GaN to prevent decomposition of the InN, and finally high temperature (HT )-GaN. The growth schematic to obtain crack-free thick GaN on Si is illustrated in Figure 4-15. Figure 4-15. The schematic to obtain thick GaN on Si(111) using InN buffer materials. After InN growth at 560 to 600 C, the temperature was reset to 560 C and a low temperature (LT)-GaN was depos ited at 560 C for 10 to 30 min to cover the InN before starting growth of the thick HT-GaN film. After LT-GaN deposition, the temperature was increased to 850 C and the HT-GaN wa s grown. During the heating and cooling processes N2 was always provided. After grow th, the reactor was cooled in N2 ambient. The cooling rate was approximately 15 C/min. 4.4.2.1 InN Growth on Si(111) InN columnar film, small nanorods (d = 250 nm), large nanorods (d = 500 nm), and microrods were grown on Si substrates for comparison. Table 4-3. Growth conditions for InN buffer interlayer on Si. Growth T (C) Cl/In N/In Time Feature 560 1 2500 1 hr (a) Columnar Film 600 4 250 20 min (b) Nanorods (d = 250 nm) 600 4 250 1 hr (c) Nanorods (d = 500 nm) 650 5 250 1 hr (d) Microrods The growth conditions for the various samples are tabulated in Table 4-3. Figure 4-16 shows the plan-view of th e SEM images of each of the buffer interlayers evaluated. All InN material showed well-faceted he xagonal structures and single crystallinity as confirmed by TEM-Diffraction Pattern (DP) spot patterns as InN buffer interlayer

PAGE 143

120 presented in Chapter 5. At low growth temperature (560 C), low Cl/In (1), and high N/In ratios (2500), InN columnar polycrystal line films were grown. By changing the growth time and the conditions as indicated in Table 4-3, the other microstructures were grown. The morphology evolut ion depending upon the growth parameters are presented in details in Chapter 5. Figure 4-16. SEM plan-view of InN columnar film, small nanorods (d = 250 nm), large nanorods (d = 500 nm), and microrods grown on Si(111). 4.4.2.2 LT-GaN Growth on InN Buffer Materials/Si(111) The 4 different templates which are shown in Figure 4-16 were used as templates for subsequent GaN growth. LT (560 C)-GaN was grown on the 4 different templates for comparison of the structural effects of InN buffer material. Figure 4-17 shows the (b) Nanorods(d=250 nm) (c) Nanorods(d=500 nm) (d) Microrods (a) Columnar film

PAGE 144

121 SEM images of the LT-GaN grown on various InN crystals for 10, 20, and 30 min. The growth conditions for LT-G aN were set as T = 560 C, Cl/In = 1.5, and N/In = 570. Figure 4-17. SEM plan-view of 10, 20, a nd 30 min LT-GaN grown on (a) InN columnar film, (b) smaller nanorods (d = 250 nm), (c) larger nanorods (d = 500 nm), and (d) microrods. Significant changes in the morphology were observed as the growth time increased from 10 to 30 min. Only InN columnar film s and larger nanorods (d = 500 nm) provided uniform coverage of LT-GaN (Figure 4-17 (a) and (c)), while the smaller nanorods (d = 250 nm) and microrods demonstrated non-unifo rm coverage of LTGaN (Figure 4-17 (b) (a) (b) (c) (d) No LT-GaN No LT-GaN No LT-GaN No LT-GaN

PAGE 145

122 and (d)). From the observation of non-uniform and small InN samples it was found that GaN deposition occurred mainly on the InN surface rather than on the Si substrate. Figure 4-18 (a) shows a wider area SEM plan-view of 30 min LT-GaN grown on InN nanorods. LT-GaN started to coalescence after 30 min, and an interesting embossed pattern could be seen. The cross-secti onal view (Figure 4-18 (b)) shows that the thickness of the LT-GaN was ~ 4 m and that voids were formed at the interface. XRD -2 scan (Figure 4-18 (c)) show ed GaN (002), InN (002), and InN (101) peaks, as well as the Si (111) substrate peak. This data s hows that InN nanorods still exist after 30 min LT-GaN growth. This is expected since 560 C is well below the decomposition temperature of InN. No In metal was detected by XRD. Figure 4-18. LT-GaN deposited on InN nanorods (d = 500 nm) on Si(111) for 30 min (~ 4 m) (a) plan-view of LT-GaN on InN nanor ods, (b) cross-sectional view (c) XRD -2 scan (c) 5 m Si ( 111 ) 3 m voids (a) (b)

PAGE 146

123 4.4.2.3 Thick HT-GaN Growth on LT-GaN/InN/Si(111) Thick GaN films were next grown at 850 C for 2 hr after 30 min LT-GaN growth on InN crystals. The growth condition for HT-GaN was the same as the base growth conditions used for growth of GaN on sa pphire (Cl/Ga = 1.5, N/Ga = 570, and T = 850 C). N2 was used as the carrier gas due to the instability of InN in H2. Figure 4-19. HT-GaN/LT-GaN gr owth on various InN structures (a) InN columnar film, (b) small nanorods (d = 250 nm), (c) la rge nanorods (d = 500 nm), and (d) microrods. The right figures show e xpanded views of the surface of HT-GaN films. (d) (c) (a) (b)

PAGE 147

124 Figure 4-19 shows the morphology of thick (~ 20 m) GaN films grown on LTGaN for each of the 4 different InN interlay er microstructure. Figure 4-19 (a) shows thick GaN grown on the InN columnar film/Si. Significant peeling of the GaN film is evident. From observation of GaN deposit ion on the surface exposed by the cracking process, it was confirmed that the cracking and peeling occurred during the growth and not during the cooling. Fi gure 4-19 (b) shows the thic k GaN grown on smaller InN nanorods (d = 250 nm) over Si (111). Cracks were observe d due to the non-uniform deposition of InN nanorods and LT-GaN. Figur e 4-19 (c) shows the best result of crackfree thick GaN, which was grown on the large InN nanorod (d = 500 nm) interlayer. Although the surface is still rough, it is clea r that dense and uniform InN nanorods are a good candidate structure for thick, crack-fr ee GaN growth on Si(111). Figure 4-19 (d) shows cracking of thick GaN fi lm on InN microrods on Si. Again, the observation of cracks correlated with the films exhi biting more pronounced surface roughness. Figure 4-20. Thick (28 m) and freestandi ng (self-separated) GaN grown on InN nanorods/Si (a) SEM plan-view of the GaN film, (b) XRD -2 scan. The GaN film occasionally self-separated from the Si substrates without effort. SEM and XRD results of freestanding and cr ack-free GaN film (28 m thick) grown on InN nanorods (d = 500 nm)/Si (111) subs trates are shown in Figure 4-20. (b) -2 341 arcsec (a)

PAGE 148

125 No cracks were observed by SEM over the wide range of the film, as seen in Figure 4-20 (a). Low resolution XRD (Figure 4-20 (b)) shows GaN (002), (004), and (103) peaks. No InN was detected presumably becau se at the growth temperature (850 C), the InN dissolved into the InxGa1-xN. The film was polycrystalli ne, but highly textured along the [002] axis with an FWHM of 341 arcsec. Figure 4-21 shows a cross-sectiona l view of crack-free thick (40 m) GaN grown on Si substrate using dense a nd large (d = 500 nm) InN nanor ods. In this case, selfseparation did not occur. InN nanorods at th e interface were not vi sible after HT-GaN. InN likely dissolved into GaN to form InxGa1-x N alloy at the interface. Another possibility is that the InN ther mal decomposed to give a high N2 pressure. This could provide a force for new crack propagation, similar to splitting Si with large H in implants. Therefore, the occasional self-separations may be related to the two preferential paths: InN decomposition leading self-separation and InxGa1-xN alloy formation for adhesive growth. Figure 4-21. Cross-sectional view of th ick GaN grown on InN nanorods/Si(111). The surface was covered by LT-HVPE laye r grown after thick HT-HVPE. InN nanorods (1 m) (Not observed after HT-GaN growth)

PAGE 149

126 Figure 4-22 compares an HT-GaN gr own rough surface with a LT-GaN capped surface. The rough surface was covered by gr owing a LT-GaN (4 m) layer after the thick film growth to smooth the surface. It is noted that th e capping layer surface resembles the LT-GaN grown on InN nanorods. The surface of LT-GaN, however, is not smooth for direct device fabricat ion. In fact, the electrical pr operties and crystal quality of the LT-GaN are inferior. Since one goal was to grow thick GaN substrate using Si, high quality MOCVD GaN shoul d be grown on this thick GaN substrate. Before MOCVD growth, CMP should be carried out to smooth the surface of HT-GaN. Figure 4-22. SEM plan-view of thick GaN gr own on InN nanorods (d=500 nm)/Si(111). (a) as grown HT-GaN surface, (b) covered LT-GaN surface. 4.5 Conclusions Crack-free and thin (3 m) GaN was successfully grown on GaN/AlGaN/Si templates utilizing a relatively low (850 C) growth temperature. Attempts to grow thicker films resulted in cracking. Cracks were observed in thick (> 7 m) GaN films that were grown either directly on Si(111) or grown on multiple AlGaN graded layers/Si templates. It was obs erved that the cracks penetrated to the Si substrate. Strong cohesion be tween GaN and Si, as well as the brittleness of Si, were responsible for the cracking. To avoid cracks, InN columnar film, nanor ods with different diameters (d = 250 nm and d = 500 nm), and microrods were grown as a buffer material on Si(111). Low (a) (b)

PAGE 150

127 temperature (560 C) GaN was deposited thereafter to prevent InN from decomposing at the higher growth temperature used to grow thick GaN. LT-GaN uniformly covered only InN columnar films and nanorods (d = 500 nm), while nanorods (d = 250 nm) and microrods demonstrated non-uni form coverage of LT-GaN. It was found that the structur e of the InN interlayer was critical to growing crackfree GaN film. Thick GaN grown on InN columnar film as well as smaller nanorods (d = 250 nm) and microrods exhibited cracks. Crack-free thick GaN film, however, was grown on the dense InN nanorods (d = 500 nm) interlayered on Si(111). Voids at the interface were found to be beneficial to growing crack-free GaN. The crack-free GaN film grown on InN nanorods/Si showed strong texturing in the [0002] axis (FWHM of rocking curve: 341 arcsec), although it wa s polycrystalline. Cr ack-free (crack-free area: 10 mm x 5 mm), 40 m thick GaN was successfully grown on Si substrates using mats of dense nanorods interlayer. GaN occasio nally self-separated from the Si substrate without effort, due to the low decomposition te mperature of InN. The results show that using InN dense nanorods (d = 500 nm) over Si is a promising technique to grow crackfree thick GaN on Si substrates.

PAGE 151

128 CHAPTER 5 GROWTH OF INDIUM NITRIDE NANORODS BY H-MOVPE 5.1 Introduction The group III nitrides are receiving consid erable attention as host semiconductors for optoelectronic and high-power, high-temp erature electronic de vice applications. Among the nitrides, InN has the smallest effective mass and hi ghest electron drift velocity, making it a particularly attractive material for high-speed electronic devices. The growth of one-dimensional (1D) semic onductors, such as na nowires and nanorods, holds the promise of improved crystal quality, as well as the ability to use quantum size effects to adjust the material properties. Therefore, coupling th e excellent transport properties of InN with the a dvantages of nanostructured materials may offer superior electronic device fabrication. GaN [Kuy04, Han97, Deb05] and InN 1D st ructures [Lia02, Lan04, Tan04] have been demonstrated using an Au catalyst to promote growth by a vapor-liquid-solid (VLS) mechanism. Closed-spaced vapor tran sport using pure indium metal or In2O3 and ammonia has been used to grow InN nanostruc tures without an external catalyst [Zha02, Joh04a, Vad05, Luo05]. It is possible that the li quid In precursor structures self-seed the growth to provide a VLS mechanism. It would be helpful for device fabrications to use chemical va por deposition (CVD) to produce InN nanostructures in a contro lled manner. The effect of input NH3/TMIn molar ratio on the growth of InN films has been studied [Joh04j, Dra06, Sin04]. Its influence on InN 1D structures, however, ha s not been reported since most studies

PAGE 152

129 [Zha02, Joh04a, Vad05, Luo05] used closed sp aced vapor transport, in which it is difficult to control the molar ratio. There ar e a limited number of reports of synthesis of InN nanostructures by CVD [Par01, Sar05], HVPE [Ond02], and MBE [Dim04, Oli03, Sto06], which can independently control th e inlet molar ratios. Nevertheless, no morphological study by changing pr ocess parameters such as N/In, Cl/In ratios, and growth temperature has been reported. In the present work, InN films and na norods were grown by H-MOVPE. An advantage of H-MOVPE is that the added HCl can prevent forma tion of In droplets [Kan04], reducing the need for excess NH3 and allowing growth of InN at a reasonable rate. It was also hoped that the surfa ce morphology would impr ove given that the deposition reaction is reversible. The prom ise of improved structural quality in InN nanorods is that, once nucleated, the ratio of the contact area to free surface area is much less than in the case of film, reducing substrate effects. This difference, in addition to the potential for dislocations terminating at th e sidewalls, should si gnificantly improve the structural quality of the mate rial. Furthermore, a hot-walle d reactor increases the extent of NH3 cracking. The surface morphology evolution was observe d with the effect s of the growth temperature, HCl/TMIn and NH3/TMIn molar ratios to determine optimal growth conditions for InN nanorods. The boundaries between InN growth and et ch regimes were calculated based on chemical equilibrium and the results were co mpared with experimental data. Extensive characterizations focusing on InN nanorods were carried out and the structural, chemical, optical, and electrical properties of InN nanorods were measured.

PAGE 153

130 As an application, the high surface to volume ratio of nanorods renders them potential candidates for sensi ng applications. Therefore, InN nanorods were fabricated with Pt nanoparticle depo sition to function as an H2 sensor and promising results were obtained. 5.2 Chemical Reactions for InN Growth by H-MOVPE H-MOVPE is a versatile tec hnique because it can indepe ndently control N/In, Cl/In molar ratios, and growth temperature. InN crystals were grown by the H-MOVPE technique. The indium and nitrogen sources were TMIn (TMI solution, Epichem) and NH3 (Anhydrous Grade 5, Matheson-Trigas), respectively and HCl (10 % HCl, 90 % N2, Air Products) was used to in situ form InCl precursor. The overall reactions of InN growth are as following: In(CH3)3(g) + HCl(g) InCl(g) + CH4(g) + C2H6(g) (source zone) (R1) InCl(g) + NH3(g) InN(s) + HCl(g) + H2(g) (deposition zone) (R2) TMIn reacts with HCl and forms InCl and by-products such as CH4 and C2H6 (R1). The reactions in H2 ambient were not considered sin ce InN is generally grown in inert environment. InCl3 and other chlorides formation reacti ons are negligible because InCl is dominant species in the wide ra nge of temperatures (T > 450 K) as shown in Figure 2-12. InCl reacts with NH3 and forms InN and produces HCl and H2 as by-products (R2). 5.3 InN Nanorods Growth Optimization During the exploratory study of the InN film growth as presented in Chapter 6, possible conditions for InN na norod growth were determined The base conditions for InN nanorod growth are listed in Table 5-1. The substrates used in this study were cAl2O3, GaN/c-Al2O3, and Si(111) although occasionally a-, r-Al2O3 and Si(100) were used.

PAGE 154

131 Table 5-1. Base conditions of InN nanorods growth. Growth T TMI flow rate HCl flow rate NH3 flow rate N2 flow rate 600 C 0.7 sccm 2.8 sccm (Cl/In = 4) 175 sccm (N/In = 250) 1600 sccm The growth parameters were varied to study morphological evolution of InN to determine optimal growth conditions for InN nanorod formation. 5.3.1 Experimental Procedure To grow InN, c-Al2O3, GaN/c-Al2O3, and Si(111) wafers were normally used as the substrates. Si (111) and c-Al2O3 substrates were cleaned by wet cleaning process such as dipping in the warm trichloroethylene, acet one, and methanol for 5 min each followed by DI water rinse and nitrogen gun dry, whereas GaN/c-Al2O3 substrate was only cleaned by nitrogen gun with static eliminator. The substrates were loaded together and InN was grown simultaneously to a void run-to-run variation. The TMIn flow rate was maintained at 0. 7 sccm, 10 % HCl flow rate was 0 to 4.2 sccm (Cl/In ratio from 0 to 6), NH3 flow rate was 70 to 5000 sccm (N/In ratio from 100 to 7000), and N2 carrier flow rate was constant at 1600 sccm. After growth, the reactor was cooled to room temperature at a rate of -15 C/min. During the cooling process NH3 was provided to prevent In N decomposition up to 300 C. The reactor pressure was maintained at ~ 1 atm. 5.3.2 Results 5.3.2.1 Effects of Growth Tempera ture and HCl/TMI Molar Ratio When the temperature was lower than 400 C or higher than 750 C, no InN growth occurred. The extent of NH3 cracking was extremely low at T < 400 C and InN decomposed rapidly around 750 C. The deco mposition is somewhat altered depending

PAGE 155

132 on etching effect of H2 from NH3 decomposition. As a result, it is not surprising that no growth occurred below 400 and above 750 C. Figure 5-1. The growth map of InN at differe nt HCl/TMI ratio with growth temperature; N/In = 250; Growth time = 1 hr; Substrate = c-Al2O3. The first column of Figure 5-1 shows the results in the case where the growth temperature was around 500 C. When HCl/T MIn = 0, In droplet formation was observed. When HCl/TMIn ratio was 1 to 2, continuous InN films were grown without indium droplets formation. Th e unreacted indium after InN fo rmation perhaps is retained in the vapor phase as volatile InClx species. As a result, no In droplets were observed.

PAGE 156

133 At higher HCl/TMIn ratios (> 3) at the same temperature (500 C), for example HCl/TMIn = 4 and 500 C, nucleation was di fficult because there was not sufficient active nitrogen at the low temperature and excess HCl leading towards etching. At these conditions sparse nanorods could be formed. The sizes of nanorods in both lengths and diameters and the number density were small because the growth rate was low at low temperature with excess HCl conditions. Interesting morphological e volution was observed at a growth temperature around 600 C (second column). When no HCl was provi ded, InN film was again grown with In droplets. At HCl/TMIn = 1 to 2, column ar structures developed leading to the polycrystalline InN films. When HCl/TMIn = 3 to 5, long and dense nanorods were grown. Sparse and well f aceted microrods were observe d at HCl/TMIn = 6. Although the growth rate was higher at 600 C, th e nucleation of InN was difficult at high HCl/TMIn ratios. Once nucleated, the crystals tended to grow larger rather than creating new nucleation sites. The third column shows the results at high growth temperature (T = 700 C). In droplets formed again when HC l was not provided. At HCl/TMIn = 1 to 2 dense nanorods were formed rather than 2-D f ilms. It is noteworthy that the effects of HCl/TMIn ratio and growth temperature are similar because they are related to InN etching and decomposition, respectively. As a result, dense nanorods were grown at high HCl/TMIn (~ 4) and intermediate growth T (~ 600 C), or intermediate HCl/TMIn (~ 1) and high growth T (~ 700 C). At HCl/T MIn = 4 and T = 700 C, sparse and long nanorods were grown. The high temperature (~ 700 C) made the nuc leation of nanorods difficult due to the decompos ition effect. Once nucleated, the nanorods tended to grow along the most favorable directions rather than forming other nucleation sites. Finally, at

PAGE 157

134 HCl/TMIn = 6, the nucleation of InN was diffi cult, so low density microrods growth was anticipated for the reason similar to the above. 5.3.2.2 Effects of NH3/TMI Molar Ratio and Substrate Material The effect of NH3/TMIn ratio, another important growth parameter, was investigated with diffe rent substrates: c-Al2O3, GaN/c-Al2O3, and Si. For comparison, those substrates were loaded together. In previous experiments, it was found that long and dense nanorods were grown when the NH3/TMIn ratio was around 250 in a temperature range 600 to 650 C and th e HCl/TMIn ratio was around 3 to 5. Figure 5-2. Scanning electron microgra phs of InN film and nanorods. NH3/TMIn were varied from 100 to 7000; Other growth conditions are HCl/TMIn = 4, T = 600 C; Growth time = 1 hr.

PAGE 158

135 It was found that the length of the nanorods after 1 hr of growth was typically ~ 1 m, with a diameter in the range 50 to 380 nm, depending on the value of the NH3/TMIn ratio. While changing the NH3/TMIn ratio (Figure 5-2), it was found that nanorods were only grown when NH3/TMIn ratio was 250. At NH3/TMIn ratio greater than 103, the diameters of the rods became larg er. At very high (N/In = 7 x 103) NH3/TMIn conditions, microrods with sharp edges or ev en continuous polycrystalline films were grown with granular surface morpho logy as shown in Figure 5-2. The results can be explained by th e amount of active nitrogen from NH3 decomposition. When hardly any NH3 was provided (N/In = 100) there was not enough active nitrogen to form InN, so that the growth of both InN films and nanorods was limited. When intermediate amounts of NH3 were provided (N/In = 250), the growth occurred in a selective way. In other words, the growth occurred in the most stable direction (in this case al ong c-axis). When abundant NH3 was supplied, there was enough active nitrogen in the system, so that growth could occur isotropically. This prevents the directional growth and rather large diameter microrods were grown (i.e., N/In = 1000). If excessive NH3 was provided, the et ching effect from H2 cannot be ignored and the growth rate decreased (i.e., N/In = 7000). Due to the similarity of the crystal structur e, it is easier to form InN nuclei on GaN. Therefore the films a nd nanorods on GaN/c-Al2O3 substrates tend to be denser than the ones on c-Al2O3 and Si(111) substrates. 5.3.2.3 Equilibrium Analysis To better understand the gr owth conditions of InN na norods, a complex chemical equilibrium analysis of the In -N-Cl-H-C-Inert system was performed. Details about the thermochemical data collection, calculation procedures, and detailed studies along with

PAGE 159

136 the equilibrium predictions are presented in Chapter 2. For comparison, experimental observations were plotted in Figure 5-3 along with the eq uilibrium predictions. The outcome from the multiple growth runs are shown on this figure, indicating whether film growth, nanorod growth, or no growth occurr ed. These data span a range of conditions (HCl/TMIn ratio 1 to 7; NH3/TMIn ratio 100 to 7000, and gr owth temperature 400 to 750 C). Figure 5-3. The calculated boundary of the gr owth and etch regimes and experimental observations. (a) deposition temperature vs HCl/TMIn ratio for selected N/In atom ratios; (b) depositio n temperature vs. N/In atomic ratio for selected HCl/TMIn ratios.

PAGE 160

137 The general trend is consistent with the equi librium predictions part icularly the results with variable Cl/In ratio. Although kinetic limitations lik ely exist for both growth and etching, particularly at low temperature, the transition point occu rs when etching and deposition reactions are in bala nce and thus might be expected to be better predicted by equilibrium calculations. There is one outlier at an HCl/TMIn ratio of 4, for which no growth is observed at conditions well into th e growth region. This particular run was at the lowest temperature (400 C), and the thermal cracking of NH3 is very difficult so that no N is available for growth [Ban72]. The conditions of several experiments a nd their outcome are provided in this figure. The general trend is consistent with the experi mental observations, although two outliers at low temperatures are noted. Th e general agreement between experimental observations and conditions predicted to yiel d no net growth or etching suggests that operating near this transition allows for fine control of the nucleati on process, and thus the opportunity to contro l the density and dimensions of the nanostructures. Of course, kinetic limitations will be present and influen ce the actual process. It is unclear what mechanism initiates nanorod growth; it is al so unclear by what mechanism the bulk nanorods grow. The cooperation of In liqui d is improbable sin ce In liquid was not observed by SEM study. In addition, the existe nce of In liquid is unlikely in HCl ambient since InClx formation is dominant reaction in chlorinating environment at high temperature. Once stable nuclei form on th e substrate surface, epitaxial growth is expected for conditions of moderate latti ce mismatch because epitaxy will generate a minimum energy. Therefore, InN nanorods showed very high structural quality as judged by TEM-DP (Figure 5-12).

PAGE 161

138 5.4 Properties of InN Nanorods The grown InN nanorods were characterized by various technique s to examine the morphology, crystallinity, chemical compositi ons, optical, and electrical properties. 5.4.1 Morphology Field Emission Scanning El ectron Microscopy (FE-SEM) was used to observe the surface morphology after growth. Figure 5-4 sh ows selected images of InN nanorods at given conditions. Well faceted hexagonal pi llar structures are clearly seen. The diameters ranged 100 to 300 nm and the le ngths were ~ 1 m for 1 hr growth. Figure 5-4. Scanning electron micrographs of InN nanorods grown at optimal conditions; Cl/In = 4, N/In = 250, T = 600 C, for 1 hr growth on Si(111).

PAGE 162

139 5.4.2 Crystallinity Figure 5-5. XRD -2 scan of InN nanorods grown on various substrates (a) a-Al2O3, (b) c-Al2O3, (c) r-Al2O3, (d) Si(100), (e) Si (111), (f) GaN/c-Al2O3, and (g) PDF#50-1239 (powder diffraction file).

PAGE 163

140 XRD -2 scans of nanorods grown on a, c, r-sapphire, Si(100), Si(111), and GaN/c-Al2O3 substrates was performe d using a Phillips APD-3720 diffractometer. It is known that cubic phase GaN could be grown on Si(100) and non-polar GaN could be grown on aor rsapphire [Cra02, Pas04]. The hope was that different substrate may provide nanorods that have different phases or polarity. The patterns show only diffract ed peaks attributed to hexa gonal InN (Figure 5-5.). Furthermore, the XRD results show that within the detection limits, neither crystalline indium oxide (In2O3), nor metallic indium was present. Although individual nanorods were si ngle crystalline (confirmed by TEMDiffraction Pattern), the patterns showed only peaks related to InN. If they were randomly oriented the pattern should match the powder pattern. It was found, however, that the nanorods were not randomly oriented as compared to the powder pattern shown in Table 5-2 and Figure 5-5 (g). Table 5-2. Intensity ratio comparison with XR D powder pattern with different substrates. Substrate/PDF#Intensity ratio (%) (100)/(101) Intensity ratio (%) (002)/(101) Intensity ratio (%) (101)/(101) Intensity ratio (%) (102)/(101) InN nano on a-Al2O3 49 74 100 29 InN nano on c-Al2O3 39 107 100 22 InN nano on r-Al2O3 37 69 100 30 InN nano on Si (100) 32 114 100 33 InN nano on Si (111) N/A 59 100 24 InN nano on GaN/c-Al2O3 9 (lowest) 130 (highest)100 11(lowest) PDF#50-1239 44 41 100 18

PAGE 164

141 Comparing the powder pattern (PDF#50-1239) to the nanorods, it was concluded that the intensity of the InN (002) peak was consistent ly higher than the InN powder pattern, regardless of the substrate type. In N nanorods grown on GaN substrates showed the highest intensity of InN (002) peak, wher eas the intensities of (100) and (102) were the lowest compared to the na norods grown on the other substr ates. This is a definite feature of texturing in the [002] direction because the in tensity of the (002) peak from randomly oriented crystals (i.e., powder) should be less than half compared with the (101) peak. The results sugge st that a GaN substrate may be a good substrate for growth of aligned nanorods. 5.4.3 Self-alignment Controlled growth of uniformly-alig ned InN nanorods would have certain advantages for advanced InN-based devices su ch as LEDs, laser diodes, and sensors. Aligned nanorods, however, have only been observed on GaN/c-Al2O3 substrate. The range of conditions for the growth of aligned nanorods is still under investigation. A possible growth mechanism may be InxGa1-xN formation at the interface since it is epitaxially grown but the evid ence is not conclusive at this point. Figure 5-6 shows SEM images of al igned InN nanorods grown on GaN/c-Al2O3 substrate without an external catalyst or template. Nanorods can be aligned in both dense and sparse manners as shown in Figure 5-6 (a) and (b). The edge shapes of the nanorods are either sharp (a), (b) or flat (c ). The substrates were all GaN/c-Al2O3 and no alignment was observed on the other substrates, such as Al2O3 and Si. XRD -2 scan was carried out to characteri ze the d-spacing perpendicular to the surface. Only InN (0002) and (0004) peaks we re detected with the substrate GaN (0002)

PAGE 165

142 and (0004) peaks seen in Figure 5-7. The results showed th at the aligned nanorods were epitaxially grown along the [0002] directi on, perpendicular to the GaN surface. To see the alignment degree of nanorods with respect to the growth direction, XRD -rocking curve was obtained with FWHM = 1.29, as shown in Figure 5-8. Although the FWHM is much wider than in the case of a film, it is noteworthy that the nanorods are indeed aligned in the [0001] direction and that the grow th definitely occurred along the [0001] axis. Figure 5-6. Scanning Elect ron Micrographs of ali gned InN nanorods grown on GaN/Al2O3 substrates. (a) sparse nanorods w ith sharp tips (b) dense nanorods with sharp tips (c) dense nanorods with flat tips. (a) (b) (c)

PAGE 166

143 Figure 5-7. XRD scan of a ligned InN nanorods on GaN/c-Al2O3 substrates. Figure 5-8. XRD -rocking curve of aligned InN nanorods on GaN/Al2O3. To see rotational alignment with respect to the substrates, XRD pole figures were taken. Figure 5-9 shows XRD pole figures of self-ali gned InN nanorods on GaN/c-Al2O3 substrate. XRD pole figures were obtained by using a Philips MRD X'Pert System. As seen in the figure, the InN ( 103) plane was selected to demo nstrate the pole figures. The X-ray detector angle (2) and the incidence X-ray beam angle () were set at 57.012 and 28.506 respectively. For pole figure measurement, -tilt was changed from 0 to 90 while rotating the sample 360 around the -axis. Coincidentally, the c-Al2O3 (116) FWHM = 1.29 InN (0002) GaN (0002) InN (0004) GaN (0004)

PAGE 167

144 plane has similar d-spacing (1.602 ) as the InN (103) plane (1 .614 ). Although the Al2O3 (116) plane has similar d-spacing, it wa s not parallel to the InN (103) plane. Therefore, both peaks were observed at different in the pole figures. Figure 5-9. XRD pole figure of self-aligned InN nanor ods (Substrate: GaN/c-Al2O3). It was observed that the GaN (103) peak appeared at the same as InN (103) in the first figure. Moreover, the prev ious pole figure results on GaN/Al2O3 (Figures 3-8 and 39) showed that the GaN (112) plane was rotated 30 with respect to Al2O3 (116) plane. This confirms that InN nanorods were grow n on GaN without rotati on, but that both InN and GaN have rotated 30 with respect to the Al2O3 substrate. This is the evidence that InN nanorods are aligned not only vertically but also in-plane; that is epitaxially. The schematic of the (103) plane with respect to rotation axis is illustrated in Figure 5-10. Figure 5-10. InN (103) plane and -axis for pole figure measurement. ) 00 1 ( ) 001 ( ) 103 (

PAGE 168

145 Self-alignment only occurred when GaN subs trates were used. Therefore it is suggested that the possible growth mechanism may be InGaN formation at the interface. The nucleation started from InN nano-crystal lines formed on a GaN substrate. InGaN may form at the interface by inter-diffusion of In and Ga. Those InGaN nuclei will be epitaxial, so that the growth w ill replicate the same crystal structure of the substrate. The concentration of In will gradually increase during the InN growth with c-axis and graded InGaN nanorods would serve as the base. Ther efore the growth axis would be the c-axis and vertically and rotationally self-a ligned InN nanorods could be grown. 5.4.4 Growth Axis and Structural Properties The structural properties of InN nanorods (grown in optimal conditions) were analyzed by TEM. The insert in Figure 5-11 (a ) shows the splitting of the reflections in the [101 0] and [112 0] selected area electron diffracti on patterns (SADP). The schematic illustrates that the splitting results from the intersection of the Ewald sphere with the streaks of diffracting intensity extending thr ough each reflection. The three streaks in reciprocal space, which originate from the well-defined hexagonal shape of the nanorod, are perpendicular to the NR facets. Depending on the zone axis orientation, the Ewald sphere intersects either two or three streak s, thus enabling determination of the facet plane. A bright field (BF) image of the NR is presented at the botto m of Figure 5-11 (a). Electron diffraction patterns c onfirmed that the nanorods ar e single crystalline with a [0001] growth axis and (101 0) facet planes. It is interes ting to note that this type of faceting appears to be different from face ting in wurtzite ZnO nanorods, where facets were found to predominantly reside on the (112 0) planes [Din04, Dav99]. TEM also showed that the shape of nanorod tips is affected by growth conditions: nanorods exhibited either flat top or pyramid-truncated shape (Figure 5-4), i.e., Figure 5-11 (a),

PAGE 169

146 with side facets residing on the (101 1) planes. Darkand bright -field images Figure 5-11 (b) revealed that nanorods are threading disl ocation-free with sparse planar defects. Figure 5-11. Transmission el ectron micrographs of an InN nanorod (a) splitting of reflections and (BF) bright field image at the bottom, (b) sparse planar defects by darkand bright-field image. These defects extended across individual na norods and can be attributed to the stacking faults residing on the (0001) planes. The distance of crystallographic planes can be precisely analyzed by measuring the distances between the transmitted spot and diffracted spots of TEM-DP as described by the following relation. Rdhkl = L Sparse planar defects (arrows) (b) (a)

PAGE 170

147 where R is the measured distance between tr ansmitted spot and diffracted spot [cm], dhkl is the spacing of the crystallographic planes (h k l) [], is the wavelength of the electron beam [], and L is the camera distance [cm]. The wavelength of the electron beam depends on the acceleration voltage that is typically 100 to 200 kV in TEM. For instance, the wavelength of electron beam is 0.037 at 100 kV and 0.024 at 200 kV. Figure 5-12. TEM and SAD of InN nanorods (WZ) with growth axes (a) [0002 ], (b) [1 010], and (c) [112 4]. Rather than separately usi ng electron beam wavelength () and camera distance (L), a camera constant (L) is used to calculat e the distance of the crys tallographic planes in (b) (a) (c)

PAGE 171

148 practice. Since L [ cm] is constant, dhkl [] can be directly obtained by measuring R [cm]. Figure 5-12 shows transmission electron micrographs with selected area diffraction (SAD), in other words, TEM diffraction pattern (DP). The distances between transmitted spots and diffracted spots were measured and converted to the distance between the crystallographic planes. The results clearly showed that an individual InN nanorod was a single crystalline with wurtzite structur e and a growth axis mainly along [0002 ], as shown in Figure 5-12 (a). Although a majority of the nanorods were wurtzite with a [0002 ] growth axis, some exceptions were found. A few wurtzite InN nanorods samples showed 20 off [1 010] (Figure 5-12 (b)) and [112 4] (Figure 5-12 (c)) growth axes other than [0002 ]. Figure 5-13. SEM image of wurtzite InN nanor od grown on the side wall of a nanorod. A similar occurrence was observed by SEM, as shown in Figure 5-13. The InN nanorods nucleated on the side wall of th e nanorods and elongated tilted in the [1 010] direction. Therefore, growth along the c-axis is not the on ly mechanism to explain the elongation of nanorods, alt hough it was found that [0002 ] was the main growth axis, as confirmed by both TEM and XRD.

PAGE 172

149 Interestingly, zincblende InN nanorods a few degrees off the [2 20] growth axis and body-centered cubic In2O3 with a [4 02] growth axis were found, as shown in Figure 5-14. The zincblende InN phase was usually observed in a mixture with the wurtzite phase in thin films [Spe06, Tsu05, And05, Yam98jcg]. Therefore, the existence of pure InN zincblende without wurtzite phase inclusion is a unique result. Figure 5-14. TEM and SAD of InN and In2O3 nanorods (a) InN (ZB) growth axis = a few degrees off [2 20] (b) In2O3 (BCC) growth axis = [4 02]. Few In2O3 (BCC) nanorods were rarely observed during the TEM analysis. Although the source of the oxygen is uncle ar, the high crystal quality of In2O3 suggested that the oxygen existed during the growth. It also showed that the oxygen would not incorporate in the InN crystals by replacing N, but rather by forming In2O3 crystals. This is expected since the crystal structures of InN and In2O3 are intrinsically different. (a) (b)

PAGE 173

150 Oxygen may incorporate in InN through the in terstitial sites; however the maximum amount of oxygen concentration in In N is known to be ~ 0.3 % [Yos03]. 5.4.5 Polarity Wurtzite group III nitrides usually have spontaneous polarization effects, since most of time they are grown along the polari zed axis. These polarization effects are beneficial in some cases and should be a voided in other cases. For example, 2 dimensional electron gas (2DEG) formations at the AlGaN/GaN interface enhance the mobility of high electron m obility transistors (HEMT) si gnificantly [Smo01]. On the other hand, it may change the wavelength of LEDs (red shift), or decrease the intensity of LEDs due to the charge separation induced by polarization effect [Im98]. Another benefit of polarity study is that it may illuminate the growth mechanism. It is known that the GaN films grown by MOCVD and HVPE t echniques have different polarity because the growth mechanism is different. For ex ample, the GaN films grown by MOCVD have Ga polarity with a smooth surface, and vi ce versa. The growth mechanism of InN nanorods is not understood yet. The study of InN nanorods polarity may help to clarify the growth mechanism of InN nanorods. Convergent beam electron diffraction (CBED) is a very good tool to determine crystal structure and polarity. Unlike conve ntional selected area diffraction (SAD) that uses a parallel incident beam, CBED uses a convergent beam of el ectrons and generates diffraction discs instead of spots. The discs contains valuable information and can be used to determine thickness, lattice parameters, phase change, point and space group, Burgers vector, etc. Therefore, CBED wa s carried out during HR-TEM analysis to determine the polarity of an InN nanorod.

PAGE 174

151 Figure 5-15. CBED of the InN nanorod. As in Figure 5-15, it was determined that InN nanorods grow in the in c-axis and have N-polarity, just as film s grown by H-MOVPE. This was expected since there is always excess HCl in the system. HCl preferab ly reacts with In and forms InCl. As a result, N always remains on top. 5.4.6 Chemical Composition The surface of InN nanorods were mainly composed of indium and nitrogen without chlorine, as confirmed by AES (Figure 5-16) sp ectrum, (detection limit ~ 1%). The carbon in the AES spectrum was assumed to be from the surface adsorption during air exposure. For sputtering, 30 sec of 3 keV Ar+ beam with a 30 sputtering angle was used. Significant Ar+ ion beam damage occurred on the surface of InN nanorods.

PAGE 175

152 Figure 5-16. AES spectrum of surface scan of as grown InN nanorods at T = 600 C, HCl/TMIn = 4, NH3/TMIn = 250, 1 hr growth on c-sapphire. Figure 5-17. Scanning electron micrographs w ith AES of InN nanorods before and after AES sputtering damage. In: 31.6 % N: 33.1 % C: 19.2 % O: 16.1 % In: 79.6 % N: 0.2% C: 0 % O: 20.2 %

PAGE 176

153 Consequently, In droplets form ation was observed as shown in Figure 5-17. It is noted that the concentration of In increased from 31.6 % to 79.6 % after sputtering. Nitrogen is no longer detected by AES only In and O. Th e results are consiste nt with literature [Kos04] because the Ar+ ion beam damages th e InN surface at a sputtering angle lower than 60. The ion beam incidence angle should be adjusted for InN sputtering. The suggested sputtering angle was > 80; however that will only reduce the sputtering rate and the inherent problem will still exist. Figure 5-18. EDS line-scan of an InN nanorod grown on Si substrate. The chemical composition of an InN nanor od on Si substrate was analyzed by EDS line-scan as shown in Figure 518. The nanorods are clearly co mposed of In and N. No oxygen or carbon was detected by EDS. 5.4.7 Photoluminescence The bandgap energies of InN films a nd nanorods were measured by room temperature PL. As shown in Figure 5-19, maximum PL emissions were obtained around 1.08 eV with a significant left shoulder.

PAGE 177

154 Figure 5-19. Room temperature (300 K) PL of InN nanorods grown on Si substrate and InN film on GaN substrate, maximu m intensity observed around 1.08 eV. This value matches well with the value of MBE grown sa mple, 1.1 eV [Inu01], where they estimated the bandgap energy by absorp tion coefficient squa red plots. Although their characterization technique and the growth method we re different from this work, their value for bandgap energy is similar. A 1.1 eV bandgap was also theoretically suggested based on density-functional theory (DFT) calculations [D av02]. This value could be considered as a po ssible InN bandgap, although furthe r research is required. The differences in optical properties betw een the film and the nanorods were not substantial, because the diameters of nanorods were too large to exhibit a quantum size effects. The left shoulder of PL from nanor ods showed that there are lots of continuous peaks from 0.7 to 1.0 eV that may repres ent mixture of lower bandgap InN (pure) and higher bandgap InN (containing oxygen).

PAGE 178

155 Low temperature PL was performed to observe optical property changes with respect to temperatur e (Figure 5-20). Figure 5-20. Low Temperature (6 K to 300 K) PL of InN nanorods on Si substrate (a) whole detection range, (b) magnified range between 730 to 850 meV from 23 to 100 K region. It is known that a low temperature the PL spectrum gives more detailed information about the near-band-edge region, donor-accepto r region, and the defect-band region. The near-band-edge region includes free exc itons (X), donor bound excitons (DX) and acceptor bound excitons (AX). The PL specify one Si substrate shows that all the "classic" lines are present (free exciton, bound, exciton, exciton droplet). These lines disappear as the temperature increase, as th ey should. The bandgap of these lines is around 1.1 eV, which is the bandgap of Si. At low temperature (6 K), it is seen that the PL emitting from the Si substrate. It nearly disappears by 23 K, but reappear upon in creasing the temperatur e, it was seen that this line reappear. This is very unusual and it s value (1.1 eV) suggests that this is Si, but (a) (b)

PAGE 179

156 that the temperature behavior is completely di fferent. So it is either InN, or InN somehow affects Si in a way that it begins to emit light. It is noteworthy that from 40 to 100 K an obvious peak appeared and disappeared at 767 meV. This result is consistent w ith Lan et als report [Lan04]. 5.4.8 Cathodoluminescence The bandgap energy of InN nanorods grown on sapphire was also measured by CL at 7 K, as shown in Figure 5-21. Figure 5-21. CL spectra of InN nanorods grown on sapphire. The e-beam energy and current were set at 10 keV and 50 A, respectively. To remove the surface oxides and possible In metal or In2O3 contaminations, the samples were wet etched in 1 % aqueous HCl solu tion for 10, 30, and 60 min each. Slight improvements in the intensity were observed wi th longer etch time, but the effect was not significant. A bandgap of 0.83 eV was detect ed by CL. Note that e-beam was 10 keV much higher than Ar+ in AES. Therefore, th ere may be very similar effect happening 80001000012000140001600018000 E-beam Energy=10 keV E-beam current= 50 A no etch 10 min etch 30 min etch 60 min etch T=7 KCL Intensity (arb. units)Wavelength (A) 1500 (0.83 eV) ( )

PAGE 180

157 damage and In formation or simply InN decomposition during these characterizations. The wet etching before CL was not a good appro ach since there is no In for as grown InN nanorods. The wet etching should have been tried after CL or AES measurement to remove In droplets on the surface. 5.4.9 Raman Spectroscopy Raman spectroscopy is a well-establishe d non-destructive technique to study vibrational phenomena in solids. Since inelastic light scattering in crystals is susceptible to selection rules, this technique can be conveniently used to identify crystalline structures and evaluate material quality a nd composition. Similar to AlN and GaN, InN crystal with a 2H (wur tzite) structure belongs to the space group C4 6V and has two molecules per unit cell. Group theory predic ts eight zone-center optical phonons, namely 1A1 [transversal optical (TO)], 1A1 [longitudinal optical (LO)], 2B1 (optically inactive or silent), 1E1(TO), 1E1(LO), and 2E2. The first order Raman spectrum of InN NRs deposited directly on Si substrate is shown in the bottom spectrum of Figure 5-22. Three phonon lines were clearly observed, despite the noise, at 451 cm-1, 496 cm-1, and 596 cm-1, which can be assigned to A1(TO), E2 2, and E1(LO), respectively. The line at 522 cm-1 is the first order phonon of the Si substrate. The measured phonon energy values are between 3 cm-1 and 8 cm-1 larger than the values reported in the literature [Dav99, Che05]. These shifts may result from stress and/or heat reduction, due to the samp le morphology and experimental conditions, respectively. It was observed that measurem ents performed with small laser spots and higher laser power resulted in damage to the sample. It was also observed that the E2 2 phonon is more susceptible to stress, consistent with the larger ener gy difference reported in the present work [Kim05]. The differen t relative intensities of the observed phonon

PAGE 181

158 lines, as compared with reporte d data, do not result from break ing of the selection rules, but mostly from the near random orientation of the nanorods. A typical Raman spectrum of the InN NRs deposited on GaN/Si substrat e is represented in the top spectrum of Figure 5-22. Note the similarity of both spectr a, with the exception of the additional lines at 531 cm-1 and 561 cm-1, which are assigned to the GaN phonons A1(TO) and E1(TO), respectively. The sharpness of the InN lines is consistent with good crystalline quality of the nanorods. Figure 5-22. Room-temperature Raman scatte ring of InN nanorods deposited on Si and GaN/Si substrates. 5.4.10 Electrical Properties It is essential to measure the electrical properties of InN nanorods because one of the ultimate goals will be fabr icating devices. Carrier concen tration is one of the most commonly measured electrical properties determined by Hall measurement. Hall measurements were carried out on InN films and nanorods grown on c-Al2O3 substrates after CL measurement. Since Hall measurement is known to damage the 400500600700 InN/SiE1(TO) A1(TO) SiE1(LO)E2 2 Intensity (a.u.)Raman Scattering (cm-1)A1(TO)InN/GaN/Si

PAGE 182

159 surface of the sample during metal contact preparation, it is customary to do Hall measurement as the last step. Indium contac ts were made to the surface of InN films and InN nanorods. Since the numerical density of the nanorods was very high, the nanorods were treated as a thin film. Table 5-3. Electrical properti es of InN films and nanorods. Table 5-3 shows some electrical properties of the InN films and nanorods. Both film and nanorods were unintentionally ntype doped. The carrier concentration measured by Hall is 1.35 x 1020 cm-3 for film and 2.87 x 1020 cm-3 for nanorods. Considering the structures betw een pillar structured films and a bed of nanorods, similar Hall measurement results were expected. The rather high carrier c oncentration indicates that the grown InN films and nanorods ma y have oxygen incorporation during the CL measurement. It appears that InN is very sensitive to electron beam, intense light sources, or just air exposure and forms In metal or In2O3 on the surface. This should prevent the correct measuremen ts of properties of InN. It would be interesti ng to measure the electrical propert ies of a single nanorod. This, however, was not carried out since the le ngth of nanorods is still too short (~ 1 m) for electrical measurement. 5.4.11 Field Emission Properties A sharp edge is beneficial for field em ission applications. It was occasionally observed that the edges of InN micro/nanorods are very sharp, as shown in Figure 5-23. Sample Sheet Resistivity [/sq] Sheet Hall Coefficient [cm2/C] Type Carrier concentration [1/cm3] Hall Mobility [cm2/Vs] InN film/Al2O3 9.2582 4.6047x102 n 1.35x1020 49.7 InN nanorods/Al2O3 7.0065 2.1749x102 n 2.87x1020 31.1

PAGE 183

160 Therefore, it is interesting to measur e the field emission properties of these micro/nanorods. To determine if InN could be used as field emitters, InN nanorods on Si were used as a cathode and placed under high voltage to measure the threshold field. During the measurement, the pressure was kept at 4 x 10-9 Torr, anode-cathode spacing (d) = 65 m with circular anode 1.25 mm in diameter. 10 I-Vs were taken during ramp up and down each time and very reproducible results were obtained. However, a threshold field of 32 V/m was found at 1 nA. The typical valu e of the threshold field is ~ 5 V/m for singlewall carbon nanotube (SWNT), ~ 4 V/m for multi-walled carbon nanotube (MWNT), > 8 V/m for diamond, and > 25 V/m for amorphous carbon [Bon98]. The threshold field of InN nanorods (32 V/m) is higher than that of car bon nanotubes. Therefore, InN nanorods do not look promising for field emi ssion applications because it will require higher voltages to produce field emissions. Figure 5-23. Scanning Electron Micrograph of sharp edge InN na norods grown on Si (single NR on the left) and on GaN/Al2O3 substrates.

PAGE 184

161 Figure 5-24. I-V curves and Fowler-N ordheim plots of InN nanorods on Si. Figure 5-24 shows Fowler-Nor dheim (F-N) plots measur ing the I-V curve with schematic of F-N tunneling. The current created by quantum mech anical tunneling was derived in 1928 by Fowler and Nordheim. There are variable expressions of F-N relations and one simple expression can be written as following after WKB (Wigner, Kramers, Brillouin) approximation [Bon98]. y = -50258x 11.925 R2 = 0.9986 -36 -35 -34 -33 -32 -31 -30 -29 0.000350.000370.000390.000410.000430.000450.00047 1/VLn(I/V2)

PAGE 185

162 I = A(2E2/)exp(-B3/2/E) (1) E = V/d (2) where A is a constant, B = 6.83 x 109 [V eV-3/2 m-1], is the work function, E is the macroscopic electric field, is the field enhancement factor (geometrical factor), V is the applied potential, and d is the dist ance between the anode and cathode. By assuming that free electron theory applies, A = 6.2 x 106 (/)1/2/( + ) [A/cm2]. By rearranging equations (1) and (2), Id2/(A2V2) = exp(-B3/2d/V) (3) Ln(CI/V2) = D/V (4) Ln(I/V2) = D/V Ln(C) (5) where C = d2/A2 and D = -dB3/2/ which are constants. The most important feature of F-N relations is that the current generated by field emissions is proportional to the exponential of 3/2 power of work function of the material which was derived starting from the tim e-independent Schrdinger equation. The straight-line relation deduced by e quation (5) shows the generated current is due to the field emission effect, also cal led Fowler-Nordheim tunneling. Since V is usually much larger than I in field emission case, the Ln(I/V2) term is negative. In addition, B and C are all positive constants, so that D has a negative value. Therefore, the y intercept and slope all have negative va lues. The slope (D) of the Fowler-Nordheim plot is related to anode-cathode spacing, the work function of the material, and the geometrical factor. The ge ometrical factor for SWNT ranges from 2500 to 10000 and for MWNT ranges from 1000 to 3000 [Bon98]. The work function of InN is unknown to date. Assuming the work func tion of InN is ~ 4 eV (since the work function of GaN and

PAGE 186

163 AlN are 4.3 and 3.7 eV, respectively [He06]), th e geometrical factor was calculated to be 78, which is very low compared to CNTs. The results, however, are in good agreement with the F-N relation. Although InN nanor ods did not show promising electrical properties for field emitter, the high surface to volume ratio of dense InN nanorods may be used as sensor applications since the e xposed surface area is much larger than thin film. 5.5 Pt-dispersed InN Nanorods for Sel ective Detection of Hydrogen at Room Temperature Utilizing the high surface to volume ratio of InN nanorods, very sensitive sensors could be fabricated. In this study, a mat of InN nanorods wa s used for selective detection of hydrogen at room temperature us ing Pt coatings to promote H2 dissociation. 5.5.1 Experimental Procedure InN nanorods were grown on c-Al2O3 substrates by H-MOVPE. In selected samples, the nanorods were coated with Pt nanoparticles deposited by sputtering. Figure 5-25. SEM images and schematic of hydr ogen gas sensor made of Pt nanoparticle dispersed InN nanorods.

PAGE 187

164 A shadow mask was used to pattern sputte red Ti/Al/Au electrodes contacting both ends of multiple nanorods on the c-Al2O3 substrates. The separati on of the electrodes was ~ 500 m. Au wires were bonded to the contact pad for currentvoltage (I-V) measurements performed at 25 C in a range of ambient (N2, O2, N2O, ND3 or 10 to 250 ppm H2 in N2). Note that no underlying thin film of InN was observed at the growth conditions of these tested na norods. The schematic of this device is shown in Figure 525. 5.5.2 Results The I-V characteristics from the uncoated multiple nanorods were linear with typical currents of < 0.6 mA at an applied bias of 0.5 V as shown in Figure 5-26. Figure 5-26. I-V characteristics from uncoated and Pt-coated InN nanorods. After Pt coatings, the resistance of the nanorods was slightly higher. This may be due to the introduction of sputter damage that decr eased the conductivity of the nanorods. The

PAGE 188

165 current for the uncoated nanorods was not affected by the measurement ambient, i.e., they showed no response to hydrogen. Figure 5-27 shows the time dependence of current (top) or relative resistance change (bottom) of the Pt-dispersed multiple InN nanorods as the gas ambient is switched from N2 to various concentrations of H2 in air (10 to 250 ppm) and then back to air as time proceeds. Figure 5-27. I-Time plot of 10 to 250 ppm H2 test by Pt-InN na norods (left) and |R|/R(%)-Time plot of 10 to 250 ppm H2 test by Pt-InN nanorods (right). There are several aspects of the data. Note that the addition of the Pt coating on the nanorods now produces a strong response to the presence of hydrogen. The addition of the Pt appears to be effective in catalytic dissociation of the H2 to atomic hydrogen. There was no response of either t ype of nanorod to the presence of O2 in the ambient at room temperature. The nanorod resistance still changes at least 15 min after the introduction or removal of the hydrogen. Th e change in resistan ce during exposure to hydrogen was slower in the begi nning and the rate resistan ce change reached a maximum at ~ 15 min of exposure time. This could be due to some of the Pt becoming covered with native oxides, which is removed by exposure to hydrogen. Since the available surface Pt for catalytic chemical absorption of hydrogen increased after the removal of

PAGE 189

166 the oxide, the rate of resistance change incr eased. However, the Pt surface gradually became saturated with hydrogen and the rate of resistance change decreased. The reversible chemisorption of reactive gases at the surface of nitr ides and oxides can produce a large and reversible variation in th e conductance of the material [Mit03]. The relative response of the Pt-coated nanorods was a strong function of H2 concentration in N2. The Pt-coated InN nanorods detected hydrogen down to 10 ppm, with relative responses of ~ 10 % at 1 00 ppm and 12 % at 250 ppm H2 in N2 after 15 min exposure. The gas sensing mechanisms suggested in the past for semiconductors include the desorption of adsorbed surface hydrogen and grain boundaries in polycrystalline materials [Mit98], exchange of charges betw een adsorbed gas species and the surface leading to changes in depletion depth [Har 95] and changes in surface or grain boundary conduction by gas adsorption/desorption [Cha94]. The detection mechanism is still not firmly established in these devices and needs further st udy. In addition, hydrogen introduces a shallow donor state in InN and this change in near-surface conductivity may also play a role [Dav03]. Figure 5-28. N2, N2O, ND3, and O2 test for InN nanorods.

PAGE 190

167 Figure 5-28 shows the time dependence of resistance changes in the Pt-coated InN nanorods as the gas ambient is switched from vacuum to N2, oxygen, nitrous oxide or ammonia (using the deuterated version), and then back to air. This data confirms the absence of sensitivity to O2. The rate of resistance cha nge for the nanorods exposed to the 250 ppm H2 in N2 was measured at different temper atures. An activation energy of ~ 12 kJ/mole was calculated from the slope of the Arrhenius plot. This value is larger than that of a typical diffusion processes. Theref ore, the dominant mechanism for this sensing process is more likely to be the chem isorption of hydrogen on the Pt surface. Moreover, the data was recorded at a power level of ~ 0.3 mW, which is low even in comparison with CNTs [Say05, Lu04]. This is attractive for long-term hydrogen sensing applications. 5.6 Conclusions InN nanorods were successfully grown by H-MOVPE by adjusting HCl/TMIn, NH3/TMIn ratios, and growth temperature. Th e optimal growth conditions were found to be HCl/TMIn = 4, NH3/TMIn = 250, and T = 600 C. This study showed that HMOVPE is a promising techni que to grow not only thic k/thin films, but also nanostructured materials. It was found that the majority of nanorods were single crystalline with wurtzite structure and that the growth axis was mainly c-axis [0002 ] with N-polarity. InN nanorods grown along different axes and zincblende InN and bcc In2O3 were rarely observed. It, however, suggests that at least 4 gr owth axes are possible for InN nanorods growth. For example, InN wurtzite nanorods growth axes can be [0002 ], [1 010], [112 4],

PAGE 191

168 and InN zincblende nanorods growth axis [220]. The growth axis for In2O3 growth was [4 02]. The main chemical components of InN nanor ods were In and N, and no O, C, and Cl was detected by EDS. AES was found not to be a suitable technique for InN because InN is vulnerable to sputtering damage. Following the growth study, vertically aligned nanorods were grown on GaN/cAl2O3. InGaN formation at the interface was suggested for the self-aligned growth mechanism. Room temperature PL showed that th e bandgap of InN films and nanorods are around 1.1 eV, although undecided peaks were obse rved at 0.767 eV at low temperature. The key finding of these studi es is that the growth tr ansitioned from a continuous film to micro-structur ed to nano-structured surfaces as th e thermodynamic driving force transitioned through the growth to etchi ng conditions. Thermodynamic equilibrium analysis demonstrated InN etch ing/growth regimes. It also showed that the nanorods' growth took place during the etch regime, wh ereas InN films were grown mainly during the growth regime. This suggests that the most stable atomic bonds persisted through the harsh growth conditions and nanorods could st ill grow along the most favorable c-axis direction. Finally, Pt-coated InN nanorods based sens or was fabricated and ppm level of H2 sensing was achieved at room temperature at very low power by util izing a large surface area to volume ratio. InN nanorods may pot entially be placed on cheap transparent substrates such as glass, making them attrac tive for low-cost sensi ng applications under very low power conditions.

PAGE 192

169 CHAPTER 6 EXPLORATORY STUDY OF INDIUM NI TRIDE FILMS GROW TH BY H-MOVPE 6.1 Experimental Procedure The growth of InN films by H-MOVPE was studied by varying the growth parameters HCl/TMIn and NH3/TMIn molar ratios and the growth temperature. A detailed description of the HMOVPE technique is presented in Chapter 4. The indium, nitrogen, and chlorine sour ces were trimethylindium (TMI solution, Epichem), NH3 (Anhydrous Grade 5, Matheson-Trigas), and HCl gas (10 % HCl in 90 % N2, Air Products), respectively. GaN/c-Al2O3 substrates were typically used in this study. The thickness of the GaN film (grown by MOVPE in Uniroyal optoelectronics) on c-Al2O3 was 5 m measured by cross-sectional SEM. Substrat es were cleaned by high pressure nitrogen gun with a static eliminator to remove poten tial particles on the surface before loading. After loading the substrates, the temperatur e of the reactor was increased at a rate of 15 C/min until the desired growth temperature was reached. HCl gas was applied in situ for 10 sec to clean the surface just prior to growth. N2 was used as the carrier gas since H2 is known to inhibit InN growth [Dra04]. Table 6-1. Base conditions of InN film growth. Growth T TMI flow rate HCl flow rate NH3 flow rate N2 flow rate 560 C 0.7 sccm 0.7 sccm (Cl/In = 1) 1750 sccm (N/In = 2500) 1600 sccm Table 6-1 shows the base c onditions of this growth st udy, which was adopted from InN film growth in a chlorina ting environment [Kan04]. It is very important to choose

PAGE 193

170 the right base conditions to reduce the number of experiments, especially when there are several growth parameters, as in this case. The HCl/TMIn ratio was changed from 0 to 6 by changing HCl flow rate from 0 to 4.2 sccm while keeping TMIn and NH3 flows and growth temperature constant at 0.7 sc cm, 1750 sccm, and 560 C, respectively. 6.2 Results 6.2.1 Effect of HCl/TMIn Molar Ratio Figure 6-1 shows SEM images of InN crys tals grown for 1 hr at HCl/TMIn = 0, 0.3, 1, and 4, while keeping th e other parameters constant (NH3/TMIn = 2500, T = 560 C). Indium metal droplets were observed when HCl was not provided (HCl/TMIn = 0). Figure 6-1. Scanning electron mi crographs of InN films (a) HCl/TMIn = 0, (b) HCl/TMIn = 0.3, (c) HCl/TMIn = 1, and (d) HCl/TMI n = 4. Other growth conditions: T = 560 C; P = 760 Torr; N/In rati o=2500; substr ate: GaN/Al2O3; growth time = 1 hr.

PAGE 194

171 When the HCl/TMIn is greater than 0.3, crystal grains ra nging from 300 to 500 nm were viewed by SEM. The faceting was more cl early seen in cases with higher HCl/TMIn ratio. By looking at the cr ystal shapes and XRD results, the InN films grown were confirmed to be polycrystalline. At HC l/TMIn = 4, nanostructured materials were observed with a well-faceted surface. Figure 6-2. Growth rate and XRD of InN (a) InN growth rate and (b) XRD scan with respect to HCl/TMIn ratio; T = 560 C, N/In = 2500, substrate = GaN/Al2O3; growth time = 1 hr. Figure 6-2 shows the variati on in the growth rate with respect to HCl/TMIn molar ratio along with XRD results. All the thic knesses were measured by cross sectional SEM. The thickness of rough film, however, is difficult to define. Therefore the growth rate determination should cont ain some error. A significant decrease in the growth rate was observed with increasing HCl/TMIn ratio. As HCl/TMIn ratio increased, the extra Cl shifted In from solid InN to the vapor phase as InClx species. Highly textured InN in the [002] direction was detected by XRD when HCl/TMIn were varied from 0.3 to 3. Interface InGaN formation possi bly caused the strong texturing in the [002] direction. 0 0.2 0.4 0.6 0.8 1 1.2 1.4 01234567 HCl/TMIn ratioGrowth Rate (m/hr) 30.030.531.031.532.0 2-thetaIntensity (a.u.) Cl/III =0.3 Cl/III =1 Cl/III =1.4 Cl/III =3 Cl/III =5 InN(002) GaN(002)(a) (b)* Nanostructured Materials

PAGE 195

172 Since GaN/Al2O3 substrates were used, the 2 range was truncated at 33.0 because of the intense GaN (002) substrate peak at 34.6. It is noted, however, that the full diffraction pattern showed the film was polycrystalline and textured. The GaN (002) peak around 31.1 was from the Cu K emission (X-ray source) denoted by an asterisk (*), which only appeared when the GaN film was of excellent quality. 6.2.2 Effect of Growth Temperature Figure 6-3 shows the plan-view of SEM imag es of InN grown at 500, 525, 560, and 600 C for 1 hr. Non-uniform films were grow n at low temperature (500 C). The films look more uniform at moderate temperat ure, between 525 and 560 C. The film exhibited 300 to 500 nm grain sizes. The su rface of the InN looked more faceted at a higher temperature (T = 600 C). Figure 6-3. SEM plan-views of InN films gr own at different temp eratures (a) 500, (b) 525, (c) 560, and (d) 600 C; Other growth conditions Cl/In ratio = 1; P = 760 Torr; N/In ratio = 2500 ; Substrate = GaN/c-Al2O3; growth time = 1 hr.

PAGE 196

173 It is difficult to judge the crystal qual ity by the shapes of the grains, but more faceted crystals tend to have better crystal qu ality from the crystal growers experience. A number of hexagonal shapes were observed. This may i ndicate that the InN grown at 600 C was of high crystal quality although fu rther clarification as TEM-DP is required to confirm high crystal quality. The growth temperature was changed from 300 to 700 C and the resultant growth rate changes and XRD results are shown in Figur e 6-4. The growth rate increased as the growth temperature increased in a kinetically limited regime, while it stayed constant in a diffusion limited regime. Unsurprisingly, th e growth rate decreased as the growth temperature was near the decomposition temperature of InN. Figure 6-4. Growth rate and XRD scans with respect to th e growth temperature (a) Loge (Normalized growth rate) vs 1000/T (K ), (b) XRD scans in 2 theta ranges from 30 to 32; growth temperatur e 300 to 700 C; growth time = 1 hr. The slope of the kinetically controlled regimes (Arrhenius plot) shown in Figure 64 (a) was used to calculate the activ ation energy by the following equation: 3030.53131.53 2 2-thetaIntensity (a.u.)300 C 400 C 500 C 525 C 550 C 575 C 600 C 700 C GaN ( 002 ) InN(002) *

PAGE 197

174 RT EaAe G/ or A T R E Galn 1 ln (1) where G is the normalized growth rate (un it less), A is a constant (also called preexponential factor), Ea is the activation energy, R is the gas constant, and T is absolute temperature. The activation energy obtained from the slope of Arrhenius plot was 14.3 kJ/mole (0.15 eV), which was considerably lo wer than typical metalorganic precursors. The typical activation energy fo r CVD growth in the kinetic regime ranges from 0.5 to 1 eV [Raa93]. The result suggested that in situ formed InCl (as a product of TMIn and HCl reaction) may be a good precursor as it has lower activation energy than typical MO precursors. It could be argued, however, that the obtained activation energy may not be accurate since only three experimental poin ts were used in this calculation. The growth rate began to decrease wh en the temperature was higher than the decomposition temperature of InN, as seen in Figure 6-4. No InN peak was observed by XRD at < 400 C as well as > 700 C. A highl y textured InN (002) peak was detected from 500 to 600 C. Therefore, it can be concluded that the possible InN growth temperature range is 400 C < T < 700 C, which is consistent with Kims result [Kim06]. 6.2.3 Effect of NH3/TMIn Molar Ratio SEM plan-view micrographs of InN grow n at different N/In ratio (namely NH3/TMIn, or N/In molar ratio) are shown in Figure 6-5. It should be noted that there was no InN XRD peak at N/In =100 because the InN non-uniformly deposited under this condition. Hexagonal shapes, however, were clea rly seen in the low N/In case by SEM. The grain sizes became smaller as the N/In ratio increased and film-like morphology was observed at N/In = 10000.

PAGE 198

175 Figure 6-5. Scanning electron micrographs of InN films grown in different NH3/TMIn ratios; other growth parameters Cl/In ratio = 1; P = 760 Torr; Growth T = 560 C; growth time = 1 hr. It was known that the growth rate of GaN did not change considerably with NH3/TMGa molar ratio [Ree02]. On the contrary, the growth rate of InN significantly changed with NH3/TMIn ratio, as shown in Figure 6-6. The NH3/TMIn molar ratio was changed from 100 to 50000. A maximum growth rate of 1.6 m/hr was achieved at the lowest NH3/TMIn ratio (0.3) and it significantly decreased with increasing NH3/TMIn ratio as shown in Figure 6-6 (a). It should be noted that not only the thickness of the film but also the morphology (grain size) of the film is changing. The comparison of growth rates with different morphol ogy films may not be appropriate. Nonetheless, comparing with the maximum growth rate in conventional MOCVD system, it is about an order of magnitude higher. The reason of this high growth rate may be due to the higher reactivit y of InCl than TMIn, a decrease in the H2 from low NH3/TMIn ratio, and/or the hot reactor wall, which facilitated NH3 decomposition. An InN (002) peak was always observed from 250 to 10000 by XRD, although the highly textured (002) peak was only detected in the N/In range 500 to 5000,

PAGE 199

176 as shown in Figure 6-6 (b). The intensity of the InN (002) peak was too low to be seen because of the decrease in the thickness at high NH3/TMIn ratios. Therefore, the intensity of XRD data was multiplied by 10 or 30 times to see the peaks more clearly, as seen in Figure 6-6 (c). Figure 6-6. Growth rate and XRD scans with respect to NH3/TMIn ratio (a) Growth rate vs. N/In (NH3/TMIn) ratio, (b) XRD scans of 2 theta range from 30 to 32 when N/In was 100 to 50000, (c) enlarged view of XRD for N/In range 7000 to 50000; Cl/In ratio = 1; P = 760 Torr; growth T = 560 C; growth time = 1 hr

PAGE 200

177 6.2.4 Effect of Buffer Layer 6.2.4.1 Surface Morphology of InN Films without Buffer Layer The surface morphology of the polycrystal line InN films was characterized by AFM, with an example shown in Figure 6-7. (1 m x 1 m scanned area). The growth conditions were NH3/TMI = 10000, HCl/TMI = 1, T = 560 C, and the growth time = 1 hr without a buffer layer. The substrate was GaN/c-Al2O3. The observed grain size was in the range of 100 to 250 nm with a rou gh surface. The RMS roughness by AFM was 34 nm, suggesting that more optimization is required to obtain a smoother surface. z = 10 nm RMS roughness = 34 nm Figure 6-7. AFM images of the InN films. NH3/TMI = 10000, HCl/TMI = 0.3, T=560 C; Substrate = GaN/c-Al2O3; growth time = 1 hr 6.2.4.2 Growth of InN Films with Low Temperature Buffer Layer Optimal buffer layer growth conditions were adopted from Kims buffer layer study [Kim06]. The buffer layer was grown at T = 450 C for 30 min followed by HTInN growth at T = 560 C, NH3/TMI = 5000, and HCl/TMI = 1 for 1 hr. The substrate used in this study was GaN/Al2O3. About 0.89 m thick HT-InN film was grown. The surface morphology and thickness are shown in Figure 6-8. To determine the crystal quality of the InN film with a buffer layer, XRD -2 scan and XRD -rocking curves were obtained. Fi gure 6-9 shows the InN film highly textured along the [002] axis.

PAGE 201

178 Figure 6-8. SEM plan-view and cros s-sectional of InN films on GaN/Al2O3 substrate; N/In = 5000, Cl/In = 1, T = 560 C, and growth time = 1 hr with 450 C buffer layer growth for 30 min. -5000 0 5000 10000 15000 20000 25000 30000 253545556575 2-thetaIntensity (a.u.)InN(002) InN(004) Figure 6-9. XRD -2 scan of InN grown on GaN/Al2O3 substrate; N/In = 5000, Cl/In = 1, T = 560 C, and growth time = 1 hr with 450 C buffer layer growth for 30 min. Although the thickness of InN film was 0.89 m, the substrate peaks related to GaN (002) and Al2O3 (006) were not observed. The abse nce of substrate peaks suggest that the films were grown along a tilted axis with respect to the c-axis of GaN/Al2O3. XRD -rocking curve (Figure 6-10) shows that the gr own InN film is high crystal quality. The crystal quality of InN, however, was not as good as that of GaN. For example, the FWHM of the -rocking curve of H-MOVPE grown InN film was 2.55 (9180 arcsec),

PAGE 202

179 while the typical FWHM of the XRD -rocking curve of GaN grown by H-MOVPE was about 780 arcsec, as presented in Figure 4-5. 10 12 14 16 18 20 me g a 0k 3k 10k 23k 40k 63k 90k counts/s FWHM = 2.55 Figure 6-10. XRD -rocking curve of InN grown on GaN/Al2O3 substrate; N/In = 5000, Cl/In = 1, T = 560 C, and growth time = 1 hr with 450 C buffer layer growth for 30 min. 6.2.5 Growth of InN Films on Al2O3 and Si InN films were grown on Al2O3 and Si substrates using the same LT-buffer layer described in the previous secti on. Sapphire is the most freque ntly used substrate for III nitride growth due to its thermal and chemi cal stability and the formation of AlN or AlOxNy at the surface with nitridatio n process is also beneficial. The advantages of a Si wafer as a substrate material for InN are numerous. Not to mention the advantages of integration w ith already well-developed Si technology, another interesting possibility is the fabr ication of tandem solar cell using InN/Si structure. Growth on Si s ubstrate is known to be more difficult than growth on Al2O3, due to the dissimilarity of the surface at omic structures and the thermal expansion mismatch. The lattice mismatch of InN on Si (8 %) is less than that of InN on Al2O3 (-26 %), but the thermal expansion coefficient mismat ch of InN on Si (52 %) is more than that of InN on Al2O3 (-34 %).

PAGE 203

180 Figure 6-11 shows the XRD -2 scan of InN film grown on c-Al2O3 substrate. The peak from InN is only from the (002) plan e showing that the film is highly textured in the [002] direction. Figure 6-11. XRD -2 scan of InN film grown on c-Al2O3 substrate; N/In = 5000, Cl/In = 1, T = 560 C, and growth time = 1 hr with 450 C buffer layer growth for 30 min. Figure 6-12 shows the -2 scan of InN film grown on Si (111) substrate for 2 hr at the same growth conditions with a 30 min LT-b uffer layer. Here it was observed that a very intense InN (002) peak w ith a small Si (111) peak, sugg esting that the film is thick (about X-ray penetration depth of ~ 5 m). 252627282930313233343536373839404142434445 2-thetaIntensity (a.u. ) Si (111) InN(002) Figure 6-12. XRD -2 scan of InN film grown on Si(111) substrate. N/In = 5000, Cl/In = 1, T = 560 C, and growth time = 2 hr with 450 C buffer layer growth for 30 min.

PAGE 204

181 Figure 6-13. XRD -2 scan comparison of InN film grow n on different substrates: N/In = 5000, Cl/In = 1, T = 560 C, and grow th time = 1 hr with 450 C buffer layer growth for 30 min. To compare the substrate effects, the XRD results were organized in logarithmic scales as shown in Figure 6-13. InN films grown on Si(111), c-Al2O3, and GaN/Al2O3 substrates were compared. The data suggests that the InN film is highly textured in the [002] axis, regardless of the substrate us ed. All samples, however, showed InN(100), InN(101), and InN(102) peaks, which suggested that the InN films were polycrystalline. 6.3 Conclusions The morphology evolution of InN was studi ed by changing the growth parameters HCl/TMIn, NH3/TMIn molar ratios, and growth temper ature. The growth rate decreased with increasing HCl/TMIn and NH3/TMIn molar ratios due to the etching effect of HCl and H2. The growth rate increased in a kinetica lly controlled regime, st ayed constant in a diffusion controlled regime, and decreased in a growth/etch transition regime with increasing growth temperature. The possibl e InN growth temperature range of 400 C < T < 700 C was determined as a result. The optimal growth condition for single crystalline InN film by H-MOVPE is still under study. More optimi zation, especially

PAGE 205

182 with the various buffer layers, should be perf ormed to get high-qua lity single crystal films with a smooth surface.

PAGE 206

183 CHAPTER 7 EXPLORATORY STUDY OF GALLIUM NI TRIDE NANORODS GROWTH BY HMOVPE InN nanorods were successfully grown, as presented in Chapter 5. Following the growth of InN nanorods, the growth of GaN nanorods was attempted, since GaN has a number of potential attractive applications For example, UVLED, high power and high temperature devices can be fa bricated utilizing the wide energy bandgap and thermal stability of GaN. 7.1 Experimental Procedure The base conditions for GaN nanorod growth are listed in Table 7-1. Growth was carried out with or without a nitridation process. As a n itridation proce ss, only 800 sccm of NH3 was provided with N2 for 15 min after growth zone temperature was increased to 850 C; c-Al2O3, GaN/c-Al2O3, and Si(111) were used as the substrates. Both N2 and H2 carrier gases were tested. The Cl/Ga molar ra tio was changed in the range 2 to 19, while N/Ga ratio and growth temper ature were set at 250 and 850 C, respectively. Table 7-1. Base conditions fo r GaN micro/nanorods growth. Growth T TMGa flow rate HCl flow rate NH3 flow rate N2 flow rate 850 C 3.2 sccm 6.4 sccm (Cl/Ga = 2) 800 sccm (N/Ga = 250) 1600 sccm 7.2 Results It is observed from viewing SEM micr ograph that a GaN film was grown on the GaN/c-Al2O3 substrate, while GaN micror ods were grown on the c-Al2O3 and Si (111) substrates at the base conditi ons (T = 850 C, Cl/Ga = 2, N/ Ga = 250) without nitridation in N2 carrier gas, as shown in the first row of micrographs in Fi gure 7-1. Scattered

PAGE 207

184 growth of GaN is expected on Si with nitr idation since GaN is difficult to nucleate on Si3N4. An increase of the Cl/Ga ratio from 2 to 3 made the nucleation of GaN less favorable, as seen in the second row of mi crographs. Thus, there is no GaN deposition on c-Al2O3 or Si(111) at Cl/Ga = 3. An intere sting worm-shaped streamline is observed on the GaN/c-Al2O3 substrates. This result suggest s that nucleation of GaN is more favorable on GaN substrates rather than Al2O3 and Si substrates when nitridation was not used to prepare the after surfaces Figure 7-1. SEM plan-view mi crographs of GaN grown at Cl/Ga ratio 2 and 3 on cAl2O3, GaN/c-Al2O3, and Si at N/Ga = 250 without nitridation in N2. To judge the effects of nitrid ation and carrier gas selecti on, a 15 min nitridation at 850 C was carried out using NH3 and H2 was used as the carrier gas instead of N2. Two HCl flow rates of 8.0 sccm (Cl/Ga = 2.5) a nd 8.8 sccm (Cl/Ga = 2.75) were used to see the Cl/Ga ratio effect while keeping TMGa and NH3 flows constant at 3.2 and 800 sccm, respectively. The SEM micrograph results are shown in Figure 7-2. It is observed again that a GaN film completely covered the GaN/c-Al2O3 substrate, while GaN was nonCl/Ga = 2 on c-Al2O3 Cl/Ga = 2 on GaN/c-Al2O3Cl/Ga = 2 on Si(111) Cl/Ga = 3 on GaN/c-Al2O3Cl/Ga = 3 on Si(111) Cl/Ga = 3 on c-Al2O3

PAGE 208

185 uniformly deposited on the c-Al2O3 and Si substrates for both values of Cl/Ga. Wellfaceted GaN microrods were observed on c-Al2O3 and Si(111) substrates, while more film-like morphology was viewed on GaN substr ate at Cl/Ga = 2.5. Short and often facetted GaN crystals are seen on the c-Al2O3 and Si(111) substrates at Cl/Ga ratio = 2.75 with more exposed surface area, whereas more film-like uniform coverage was detected on the GaN/c-Al2O3 substrate. Figure 7-2. SEM micrographs of GaN grown at Cl/Ga ratio of 2.5 and 2.75, N/Ga = 250, and in H2 with nitridation. It appeared that a nitridation step may be needed to assist nucleating GaN nanorods on Si(111). It is known that Ga N film is not to grow on Si3N4. However, if Si3N4 deposited non-uniformly on Si, it will act as in situ mask that can facilitate nucleation of GaN nanorods in small opening windows. Th e overall growth rate can be lowered by also decreasing the Ga flux, which might facilitate nucleation of smaller nuclei. Therefore, a 15 min nitrida tion was carried out at 850 C and TMGa flow was decreased 3.2 to 1.6 sccm, while maintaining the N/Ga ratio at 250 by decreasing NH3 flow rate in Cl/Ga = 2.5 on GaN/c-Al2O3Cl/Ga = 2.5 on Si(111) Cl/Ga = 2.5 on c-Al2O3 Cl/Ga = 2.75 on GaN/c-Al2O3 Cl/Ga = 2.75 on Si(111) Cl/Ga = 2.75 on c-Al2O3

PAGE 209

186 the range 800 to 400 sccm. Th e Cl/Ga ratio was then cha nged in the range 2 to 6 by increasing the HCl flow rate in the range 3.2 to 9.6 sccm wh ile keeping TMGa flow at 1.6 sccm. Figure 7-3. SEM plan-view of GaN with Cl/Ga ratio 2, 2.5, 4, and 6 on c-Al2O3, GaN/cAl2O3, and Si(111). Figure 7-3 shows SEM micrographs of grow n with variation of the Cl/Ga ratio: Cl/Ga = 2, 2.5, 4, and 6 in N2 carrier gas. Because of the nitridation process, complete Cl/Ga = 2.5 on GaN/c-Al2O3 Cl/Ga = 2.5 on Si(111) Cl/Ga = 2.5 on c-Al2O3 Cl/Ga = 4 on GaN/c-Al2O3Cl/Ga = 4 on Si(111) Cl/Ga = 4 on c-Al2O3 Cl/Ga = 6 on GaN/c-Al2O3Cl/Ga = 6 on Si(111) Cl/Ga = 6 on c-Al2O3 Cl/Ga = 2 on GaN/c-Al2O3Cl/Ga = 2 on Si(111) Cl/Ga = 2 on c-Al2O3 crac k crac k crac k

PAGE 210

187 coverage of GaN/c-Al2O3 was observed, regardless of the substrate. No significant differences were found in this experiment al set, although it appeared that more hexagonal-plate structures were found at hi gh Cl/Ga = 6 conditions and more film-like features were observed on GaN/c-Al2O3 at Cl/Ga = 2.5. Although the surface was very rough, GaN film on Si(111) substrates occasiona lly showed cracks just as in the case of growing thick GaN film. A higher Cl/Ga ratio (> 6) wa s apparently required to grow GaN nanostructures. It should be noted that the maximum flow rate of HCl was 10.0 sccm because of the MFC limit. Therefore, the TMGa flow rate was d ecreased from 1.6 to 0.5 sccm while keeping the HCl flow near the maximum allowabl e value at 9.6 sccm (Cl/Ga = 19) and NH3 flow at 400 sccm (N/Ga = 800). Figure 7-4. SEM plan-view of GaN at Cl/Ga = 19, N/Ga = 800, T = 850 C on various substrates. Interesting columnar structur es were observed in this ca se, as shown in Figure 7-4. The denoted Cl/Ga ratio of 19, however, may have a large uncertainty since the TMGa Cl/Ga = 19 on c-Al2O3 Cl/Ga = 19 on GaN/c-Al2O3 Cl/Ga = 19 on Si(111)

PAGE 211

188 flow rate of 0.5 sccm may be wrong. When TMGa flow rate (FTMGa) was determined, following formula was used: ) ( ) ( ) ( ) ( ) (2sccm F Torr P Torr P Torr P sccm FN out bubbler vap TMGa vap TMGa TMGa (1) where ) (Torr Pvap TMGais the vapor pres sure of TMGa, ) (Torr Pout bubbler is the pressure of the bubbler, and 2NFis the flow rate of the N2 carrier gas through the bubbler. The vapor pressure of TMGa is a strong function of bubbler temperature and is given by the following equation [provided by Epichem]: ) ( 1703 07 8 ) (10K T Torr P Log (2) Since the TMGa bubbler was used at room temperature (20 C), vap TMGaP= 182 Torr was calculated from the above equation. The bubbler pressure is higher than the total reactor pressure (760 800 Torr) to if it deliver s the stream. It was customary to set the pressure at ) (Torr Pout bubbler= 882 Torr by adjusting the needle valve at the bubbl er exit line. Thus, the flow rate of TMGa became ) ( 17 0 ) (2sccm F sccm FN TMGa To make FTMGa = 0.5 sccm, 2NF should be 3 sccm. It is known that when the flow rate is less than 10 % of MFC maximum, the flow rate may not be accu rate. The maximum MFC flow rate of N2 carrier gas through the bubbler was 50 sccm and thus at FN2 < 5 sccm may be correlate error in the flow rate. In addition, when a very low flow rate, such as 3 sccm, was used, it was now difficult to control bubbler pressure since the pressure is not sensitive to the aperture of the needle valve. Hence, maintaining the bubbler out line pressure at ) ( Torr Pout bubbler= 882 Torr was a challenge, and it was of ten questionable how accurately the

PAGE 212

189 pressure was being maintained. Therefore, the actual TMGa flow rate and Cl/Ga ratio may not be correct. To avoid this problem one should lower the bubbler temperature and increase the N2 flow rate accordingly or obtain new MFCs. 7.3 Conclusions An exploratory growth study of GaN nanor ods was carried out by changing growth conditions such as Cl/Ga molar ra tio, substrate nitr idation, and N2/H2 carrier gases on GaN/c-Al2O3, c-Al2O3 and Si(111) substrates. More uniform coverage was observed on GaN/c-Al2O3 than c-Al2O3 or Si(111) substrates when no nitridation was carried out or when H2 carrier gas was used. When nitridation was performed and N2 carrier gas was used, complete coverage of GaN was achieve d, regardless of th e substrate. Although GaN nanorods were not successfully grown, the morphology changes suggest that GaN nanorods could be grown at a high Cl/Ga ratio (> 6).

PAGE 213

190 CHAPTER 8 FUTURE WORK AND RECOMMENDATIONS 8.1 Nucleation and Growth Mechanism Studies of InN Nanorods Although controlled syntheses of In N nanorods were achieved by H-MOVPE without external catalyst or template, the nucleation and growth mechanism of nanorods are unclear at this point. Interestingly, ve rtical alignments of nanorods were sometimes observed on GaN/c-Al2O3; however, the results were not consistent. Initial stage of growth can be studied by taking the sample after several short runs. This may show the nucleation mechanism of na norods. The attempts were not successful since no growth was observed in 10 min run, and a number of nanorods were observed after 15 min run although the deposition was non-uniform. The interface between nanorods and substrat es needs to be characterized by TEM. The challenge is how to prepare the sample without damaging the interface. This will elucidate the growth mechanism more clearly. 8.2 Self-aligned InN Nanorods as Buffer Material for GaN on Si Crack-free, thick GaN films were succe ssfully grown on Si (111) using InN nanorods as a buffer material. The GaN film s, however, were polycrystalline, perhaps because the growth axis was non-uniform b ecause of the random arrangement of the underlying InN nanorods. Self-aligned InN nanorods were also grown on GaN substrates. If self-aligned InN nanorods c ould be grown on GaN/Si templates, the overgrown thick GaN may have an aligned growth axis that is essential for single crystal growth. The schematic of this proce dure is proposed as in Figure 8-1.

PAGE 214

191 Figure 8-1. Schematic of crack -free GaN growth on Si substr ate using self-aligned InN or GaN nanorods. 8.3 Buffer Layer Optimization for GaN growth on Si Substrate The lack of suitable substrates has been a longstanding problem for Group III nitride growth. It is obvious that buffer layer optimization including the selection of buffer layer material, growth conditions, and th ickness, are very important to grow high quality film. It was observed that adequate use of such a buffer layer could compensate for lattice and thermal expansion mismatches. In addition, the buffer layer can change the surface properties by reducing the surface fr ee energy. It turns out, however that there is no universal rules for specific buffer layer selection and growth, as by occasionally observing similar results for very different growth conditions. For example HT-AlN or LT-AlN buffer layers, and nitridatio n temperature, or nitr idation time all give promising results, depending upon the recipes for growth of the buffer layer. Therefore, additional buffer layer optimization shoul d be performed for H-MOVPE system especially on Si substrate. Thick GaN was recently grown on Si using InN nanorods as an interface material. The film quality, however, was not satisfa ctory and no single crystal GaN has been grown on Si to date using InN nanorods inte rlayer approach. The proper growth and application of buffer layers will enha nce material quality significantly. aligned

PAGE 215

192 8.4 Use of Negative Thermal Expansion Materials The growth of thick/freestanding GaN is impeded by lack of suitable substrates. The bowing and cracking problems are troublesome because of the excessive stress caused mainly by thermal expansion differen ces between GaN the substrates. If the thermal expansion of the materials could be controlled, the bowing and cracking problem can be reduced. Here the use of negative th ermal expansion materials as a buffer material was suggested. Although most materials in nature expand wh en they are heated, there are several negative thermal expansion (NTE) ma terials as listed in Table 8-1. Table 8-1. Lists of NTE materials. NTE materials Reference ZrW2O8 [Per98] [Eva97] [Yam01] [Mar96] [Eva96] [Pry96] [Sik04] [Dad97] HfW2O8 [Yam01] [Wit98] ZrV2O7 [Car01] [Mar96][Pry96] ZrMo2O8 [Car00] Sc2(WO4)3 [Sec02] Lu2(WO4)3 [Liu02] Al2(WO4)3 [Gar01] Gd2(MoO4)3 [Bri72] [Dmi03] Sm2(MoO4)3 [Dmi03] Eu2(MoO4)3 [Dmi03] Sc2(MoO4)3 [Jor99] ZrW2O8 and HfW2O8 are the most well-studied compounds in this table. Figure 82 shows the lattice parameter changes of these NTE materials with increasing temperature [Yam01]. Cubic ZrW2O8 (a = 9.157 at room temperature) and HfW2O8 (a = 9.130 at room temperature) exhibits isotropic NTE from 0.3 to 1050 K with = -8.8 x 10-6 K-1 [Eva97]. The lattice parameters of HfW2O8 and ZrW2O8 decreased linearly with temperature. It should be noted that the cubic crystal structures persisted during

PAGE 216

193 heating, that the crystals sh rink isotropically, and that no pha se transform occurred up to the decomposition temperature. The decomposition temperature of ZrW2O8 is ~ 1050 K Figure 8-2. Lattice parameters of ZrW2O8 () and HfW2O8 () as a function of temperature [Yam01]. The unit cell of ZrW2O8 is shown in Figure 8-3. The unit cell is composed of ZrO6 octahedra and WO4 tetrahedra. Figure 8-3. Crystal stru ctures of NTE materials (a) Unit cell of ZrW2O8, with 90% thermal ellipsoids drawn. (b) Polyhedral representation of the structure. ZrO6 octahedra shown in white, WO4 tetrahedra shaded [Eva96]. (a) (b)

PAGE 217

194 The NTE phenomenon is related to transv erse thermal vibrations of bridging oxygen atoms, resulting in coupled rotations of the essentially rigid polyhedral blocks of the structure [Eva96]. The schematic of th is phenomenon is shown in Figure 8-4 [Pry96]. Figure 8-4. An array of linke d triangles as found in tridym ite. Rotation of one triangle causes the local environment to be pulled inwards [Pry96]. The appropriate mixture of NTE material s with Group III nitrides can create a crystal with a controllable thermal expansion co efficient. The use of NTE materials as a buffer layer for a III-V compound semiconducto r has already been suggested [Lo01]. However, no experimental work was found usi ng an NTE buffer layer. Since cracking usually occurs during the cooling of GaN grown on Si, proper use of NTE materials could reduce the excessive tensile stress. The lattice mismatch between GaN, NTE materials, and Si are not compared at this point since the structural and physical properties of most NTE materials are still no t clearly defined. The synthesis of NTE material is usually done by dehydration of hydrated NTE material. For example, ZrW2O8 was synthesized by dehydration of ZrW2O7(OH)2 2H2O [Dad97]. The growth of NTE materials by conventional epitaxy ( i.e. MOVPE, MBE, or HVPE) was not reported to date. Thus, the deposition of NTE mate rial on Si will be a challenging task.

PAGE 218

195 8.5 Double-sided Growth of GaN on Si and Sapphire The cracking of GaN occurs because of th e excessive curvature generation during the cooling process as strong tensile stress creates concave bowing in GaN. Likewise, there will be strong convex bowing of thick GaN film grown on sapphire substrate due to the compressive stress generated during the co oling process. Doubl e-sided growth could work as the stress would cancel out if gr owth on both sides were uniform. Potential problems with this process could be the non-uni formity with respect to growth axis, the role of edge growth, and backside depos ition. Suspending the double-sided polished wafer and growing simultaneously on both sides may work; however, a creative waferholding design would be necessary.

PAGE 219

APPENDIX THERMOCHEMICAL DATA FOR AL -IN-GA-C-H-CL-N SYSTEM

PAGE 220

197Table A-1. Gas Phase Cp (J/mol K) = A + B*T + C*T2 + D*T-2 Species H298 (J/mole) S298 (J/mole K) T(K) Range A B C D Ref # 298.14 4300 20.859 -9.133E-05 2.365E-08 4.855E+04 4300 8200 17.789 -1.364E-04 1.147E-07 2.956E+07 Al 330000 164.6 8200 10000 -22.242 6.992E-03 -2.429E-07 4.073E+08 [SUB94] 298.14 1200 35.167 3.691E-03 -1.222E-06 -2.628E+05 1200 3300 37.843 -1.696E-03 1.187E-06 1.979E+05 AlC 682283 225.9 3300 6000 -12.912 2.097E-02 -1.662E-06 7.627E+07 [SUB94] 298.14 1100 42.790 2.347E-02 -8.394E-06 -1.100E+05 AlC2 675623 252.9 1100 6000 61.350 3.652E-04 -3.535E-08 -4.050E+06 [SUB94] 298.14 2400 37.039 7.829E-04 -2.009E-08 -2.341E+05 2400 5700 43.137 -3.051E-03 6.447E-07 -4.397E+06 AlCl -51007 228.0 5700 6000 -25.513 1.512E-02 -7.052E-07 2.847E+08 [SUB94] 298.14 1000 37.509 2.568E-02 -9.584E-06 -2.044E+05 1000 4300 53.579 2.334E-03 -1.386E-07 -2.376E+06 AlHCl 10522 257.2 4300 6000 60.309 1.046E-03 -1.099E-07 -3.423E+07 [SUB94] 298.14 900 36.497 5.907E-02 -2.280E-05 -4.062E+05 900 2800 74.982 4.877E-03 -8.180E-07 -6.497E+06 AlH2Cl -106345 251.1 2800 6000 82.866 7.087E-05 -5.189E-09 -1.277E+07 [SUB94] 298.14 1500 56.336 2.460E-03 -8.675E-07 -4.847E+05 AlCl2 -240874 290.4 1500 5800 56.504 4.837E-04 4.193E-08 1.210E+06 [SUB94]

PAGE 221

198Table A-1. Continued 5800 6000 1.860 1.368E-02 -8.623E-07 2.875E+08 298.14 1100 61.863 2.578E-02 -9.340E-06 -7.312E+05 AlHCl2 -351279 288.8 1100 6000 82.017 4.078E-04 -3.944E-08 -4.961E+06 [SUB94] 298.14 800 75.053 1.589E-02 -8.866E-06 -6.337E+05 AlCl3 -584505 314.4 800 6000 83.082 2.294E-05 -2.118E-09 -1.280E+06 [SUB94] 298.14 1000 22.302 1.962E-02 -7.025E-06 1.637E+05 1000 2900 40.276 -2.855E-03 1.125E-06 -3.486E+06 2900 5400 -27.458 2.753E-02 -2.689E-06 9.496E+07 AlH 249250 187.9 5400 6000 151.983 -2.006E-02 8.637E-07 -6.651E+08 [SUB94] 298.14 1000 23.058 4.018E-02 -1.383E-05 1.750E+05 1000 4900 51.217 3.467E-03 -2.858E-07 -4.819E+06 AlH2 276774 213.3 4900 6000 63.237 3.110E-04 -5.827E-08 -5.331E+07 [SUB94] 298.14 800 16.480 8.353E-02 -3.208E-05 1.353E+05 800 1800 59.098 1.900E-02 -4.354E-06 -5.461E+06 AlH3 128894 206.6 1800 6000 81.991 3.523E-04 -2.997E-08 -1.626E+07 [SUB94] 298.14 1200 34.111 3.121E-03 1.457E-06 -2.456E+05 1200 2800 36.908 5.852E-03 -1.070E-06 -3.751E+06 AlN 438829 228.4 2800 6000 32.862 4.698E-03 -4.939E-07 1.790E+07 [SUB94] 298.14 900 41.640 -4.993E-03 2.943E-06 -1.708E+05 Al2 509200 243.0 900 2800 38.855 4.500E-04 5.694E-08 1.057E+04 [SUB94] Al2C2 544985 284.6 298.14 1100 63.487 2.883E-02 -1.042E-05 -3.121E+05 [SUB94]

PAGE 222

199Table A-1. Continued 1100 6000 86.053 4.518E-04 -4.371E-08 -5.045E+06 298.14 600 161.452 4.981E-02 -3.353E-05 -1.372E+06 600 3300 182.438 3.002E-04 -4.955E-08 -2.572E+06 Al2Cl6 -1295534 475.6 3300 6000 182.950 -1.474E-05 1.507E-09 -2.904E+06 [SUB94] 298.14 3400 20.975 -3.996E-04 2.008E-07 -3.361E+03 3400 10000 23.051 1.209E-04 -4.046E-09 -1.712E+07 C 716680 158.1 10000 20000 19.445 3.079E-04 3.692E-10 1.124E+08 [SUB94] 298.14 1200 154.819 1.104E-01 -3.438E-05 -1.382E+06 1200 1600 219.829 3.021E-02 -6.437E-06 -1.430E+07 1600 2200 466.590 -1.378E-01 2.577E-05 -1.648E+08 C10N 1460000 442.6 2200 4000 256.726 1.735E-03 -1.715E-07 -2.659E+07 [SUB94] 298.14 1200 167.995 1.202E-01 -3.667E-05 -1.586E+06 1200 1600 237.757 3.529E-02 -7.527E-06 -1.579E+07 1600 2200 520.417 -1.572E-01 2.942E-05 -1.881E+08 C10N2 1200000 447.4 2200 4000 280.848 2.011E-03 -1.989E-07 -3.032E+07 [SUB94] 298.14 1200 176.833 1.356E-01 -4.157E-05 -2.097E+06 1200 1600 253.664 4.071E-02 -8.505E-06 -1.741E+07 1600 2100 690.115 -2.597E-01 4.964E-05 -2.799E+08 C11NH 1270000 454.7 2100 4000 303.357 2.876E-03 -2.892E-07 -3.466E+07 [SUB94] 298.14 1300 170.519 1.148E-01 -3.384E-05 -1.490E+06 C11N 1500000 467.9 1300 1600 239.209 3.322E-02 -6.659E-06 -1.598E+07 [SUB94]

PAGE 223

200Table A-1. Continued 1600 2200 531.517 -1.642E-01 3.088E-05 -1.965E+08 2200 4000 279.296 3.787E-03 -4.181E-07 -3.142E+07 298.14 500 -268.693 1.952E+00 -1.244E-03 6.916E+06 500 800 105.945 8.242E-01 -2.882E-04 -5.539E+06 C12H26 -290872 622.6 800 1000 107.947 7.745E-01 -2.427E-04 0 [SUB94] 298.14 1100 31.548 9.313E-03 -3.749E-06 -1.550E+05 1100 5200 35.444 1.565E-03 -1.190E-07 1.432E+05 CCl 439560 224.6 5200 6000 43.872 4.413E-04 -1.305E-07 -6.230E+07 [SUB94] 298.14 800 30.466 2.977E-02 -1.093E-05 -1.466E+05 800 2500 44.262 6.562E-03 1.354E-07 -1.626E+06 2500 4900 53.024 8.112E-03 -1.014E-06 -3.571E+07 CHCl 308280 234.9 4900 6000 105.080 -8.257E-03 4.266E-07 -1.903E+08 [SUB94] 298.14 900 31.585 4.981E-02 -1.589E-05 -1.660E+05 900 2300 56.273 1.680E-02 -3.059E-06 -4.518E+06 CH2Cl 116872 243.5 2300 6000 81.106 5.609E-04 -4.383E-08 -2.269E+07 [SUB94] 298.14 700 10.683 1.086E-01 -4.137E-05 1.205E+05 700 1500 42.093 5.199E-02 -1.237E-05 -2.831E+06 1500 4000 96.339 4.656E-03 -5.241E-07 -2.509E+07 CH3Cl -81870 234.4 4000 6000 107.137 2.036E-04 -1.270E-08 -4.380E+07 [SUB94] 298.14 1200 43.592 2.088E-02 -6.386E-06 -3.531E+05 CNCl 137947 236.3 1200 3600 60.119 2.170E-03 -1.339E-07 -4.790E+06 [SUB94]

PAGE 224

201Table A-1. Continued 3600 6000 61.563 1.360E-03 -1.740E-08 -5.294E+06 298.14 800 47.759 1.869E-02 -9.555E-06 -5.351E+05 800 3500 60.402 -2.951E-03 1.035E-06 -1.887E+06 CCl2 226230 265.0 3500 6000 29.466 1.225E-02 -9.923E-07 2.944E+07 [SUB94] 298.14 700 45.180 4.939E-02 -2.188E-05 -4.718E+05 700 2300 66.827 1.084E-02 -2.080E-06 -2.612E+06 CHCl2 73895 277.8 2300 6000 82.131 2.785E-04 -2.175E-08 -1.271E+07 [SUB94] 298.14 700 28.619 9.983E-02 -4.252E-05 -3.264E+05 700 1800 66.799 3.066E-02 -6.758E-06 -3.899E+06 CH2Cl2 -95000 270.4 1800 6000 104.520 1.076E-03 -9.072E-08 -2.356E+07 [SUB94] 298.14 800 62.886 3.551E-02 -1.789E-05 -8.211E+05 CCl3 80000 300.3 800 6000 82.821 1.304E-04 -1.361E-08 -2.788E+06 [SUB94] 298.14 700 52.487 8.638E-02 -3.671E-05 -7.247E+05 700 2300 89.033 2.025E-02 -2.395E-06 -4.189E+06 CHCl3 -102700 296.4 2300 6000 106.825 7.968E-03 1.706E-09 -1.595E+07 [SUB94] 298.14 800 87.051 3.741E-02 -1.904E-05 -1.214E+06 CCl4 -95600 309.5 800 6000 107.797 1.175E-04 -1.227E-08 -3.195E+06 [SUB94] 298.14 900 28.318 -4.643E-04 4.419E-06 5.437E+04 900 2800 21.902 1.186E-02 -1.793E-06 3.455E+05 2800 6400 32.139 5.016E-03 -5.756E-07 -4.571E+06 CH 597370 183.0 6400 12000 79.927 -6.695E-03 2.178E-07 -2.232E+08 [SUB94]

PAGE 225

202Table A-1. Continued 12000 20000 28.186 -5.528E-04 1.154E-08 8.923E+08 298.14 1600 32.577 2.324E-02 -5.310E-06 -2.832E+05 1600 5600 58.835 1.581E-03 -1.180E-07 -1.280E+07 CHN_HCN 132000 201.8 5600 10000 74.573 -2.429E-03 1.743E-07 -8.950E+07 [SUB94] 298.14 1400 37.443 1.657E-02 -3.271E-06 -1.815E+05 1400 4500 55.585 2.519E-03 -2.635E-07 -8.741E+06 CHN_HNC 194328 205.2 4500 6000 61.803 1.154E-04 -7.045E-09 -2.073E+07 [SUB94] 298.14 900 27.632 1.579E-02 2.326E-07 1.251E+05 900 2200 29.137 1.978E-02 -3.693E-06 -1.422E+06 CH2 390421 194.9 2200 6000 57.017 9.725E-04 -8.901E-08 -2.055E+07 [SUB94] 298.14 1600 25.844 4.249E-02 -9.520E-06 6.653E+04 CH3 146300 194.0 1600 4400 73.421 3.551E-03 -3.672E-07 -2.222E+07 [SUB94] 298.14 1000 2.235 9.693E-02 -2.603E-05 6.109E+05 1000 2000 47.229 4.222E-02 -7.068E-06 -8.634E+06 CH4 -74600 186.4 2000 6000 101.131 5.369E-03 1.417E-07 -4.481E+07 [SUB94] 298.14 1000 23.312 1.363E-02 -3.673E-06 1.883E+05 1000 2800 32.866 1.388E-03 5.860E-07 -1.382E+06 2800 5500 35.625 4.582E-03 -4.432E-07 -2.986E+07 5500 9800 49.663 -1.094E-03 1.036E-07 -1.052E+07 9800 19000 26.035 3.298E-03 -1.090E-07 8.617E+07 CN 440287 202.6 19000 20000 70.671 -4.074E-04 -2.351E-08 -1.757E+09 [SUB94]

PAGE 226

203Table A-1. Continued 298.14 1500 37.081 2.479E-02 -5.990E-06 -1.455E+05 CN2_CNN 633484 232.1 1500 6000 69.647 -1.800E-03 1.302E-07 -1.465E+07 [SUB94] 298.14 700 22.682 7.464E-02 -3.710E-05 6.882E+04 700 1900 65.882 -1.608E-04 -3.596E-07 -4.263E+06 CN2_NCN 500535 226.2 1900 6000 64.096 -5.036E-04 4.123E-08 -6.875E+05 [SUB94] 298.14 500 97.481 -1.641E-01 1.184E-04 -1.381E+06 500 4300 30.083 5.243E-03 -4.986E-07 1.737E+06 4300 12000 27.970 3.882E-03 -1.934E-07 4.476E+07 C2 830457 197.1 12000 20000 86.667 -3.128E-03 4.369E-08 -1.209E+09 [SUB94] 298.14 1200 42.931 2.171E-02 -7.231E-06 -3.314E+05 C2Cl 534090 241.9 1200 6000 61.288 3.761E-04 -3.562E-08 -4.829E+06 [SUB94] 298.14 1000 55.080 2.749E-02 -7.582E-06 -7.354E+05 1000 2400 69.587 1.046E-02 -1.805E-06 -3.992E+06 C2HCl 213802 242.0 2400 6000 85.868 3.755E-04 -2.795E-08 -1.733E+07 [SUB94] 298.14 600 19.098 1.453E-01 -6.540E-05 -2.608E+05 600 1400 60.174 5.961E-02 -1.488E-05 -3.088E+06 1400 3900 118.581 5.935E-03 -6.913E-07 -2.479E+07 C2H3Cl 23000 264.0 3900 6000 131.940 2.372E-04 -1.503E-08 -4.641E+07 [SUB94] 298.14 600 6.880 2.137E-01 -9.322E-05 2.930E+04 600 1300 61.711 9.916E-02 -2.568E-05 -3.729E+06 C2H5Cl -112257 275.9 1300 3200 148.486 1.431E-02 -1.940E-06 -3.177E+07 [SUB94]

PAGE 227

204Table A-1. Continued 3200 4000 173.732 1.522E-03 -1.353E-07 -6.056E+07 298.14 1000 64.347 2.679E-02 -9.646E-06 -5.242E+05 1000 3000 83.063 2.331E-03 -3.613E-07 -4.065E+06 C2Cl2 209618 272.0 3000 6000 87.130 4.102E-05 -2.820E-09 -7.872E+06 [SUB94] 298.14 700 49.260 1.051E-01 -4.482E-05 -8.464E+05 700 1900 90.822 3.056E-02 -6.535E-06 -4.824E+06 C2H2Cl2_1_1C2H2Cl2 2300 288.2 1900 6000 129.565 1.025E-03 -8.504E-08 -2.618E+07 [SUB94] 298.14 700 42.911 1.173E-01 -5.142E-05 -7.510E+05 700 1900 90.630 3.076E-02 -6.588E-06 -5.213E+06 C2H2Cl2_CIS 4100 289.6 1900 6000 129.578 1.021E-03 -8.471E-08 -2.662E+07 [SUB94] 298.14 700 42.225 1.184E-01 -5.191E-05 -5.665E+05 700 1900 90.578 3.081E-02 -6.601E-06 -5.097E+06 C2H2Cl2_TRANS 6100 290.0 1900 6000 129.578 1.021E-03 -8.470E-08 -2.652E+07 [SUB94] 298.14 600 29.966 1.940E-01 -8.817E-05 -3.244E+05 600 1400 85.860 7.776E-02 -1.977E-05 -4.196E+06 C2H4Cl2 -130122 305.2 1400 4000 161.491 7.209E-03 -8.548E-07 -3.150E+07 [SUB94] 298.14 800 70.711 5.920E-02 -2.776E-05 -8.790E+05 800 4800 106.394 7.494E-04 -8.679E-08 -5.126E+06 C2Cl3 190279 328.2 4800 6000 108.823 -1.813E-04 1.252E-08 -1.084E+07 [SUB94] 298.14 700 64.821 9.766E-02 -4.527E-05 -8.768E+05 C2HCl3 -19100 325.0 700 2300 111.494 1.454E-02 -2.828E-06 -5.432E+06 [SUB94]

PAGE 228

205Table A-1. Continued 2300 6000 131.800 3.386E-04 -2.646E-08 -1.841E+07 298.14 700 61.517 1.576E-01 -7.049E-05 -8.827E+05 700 2000 129.253 3.521E-02 -7.349E-06 -7.260E+06 C2H3Cl3 -142298 320.1 2000 4000 172.401 2.301E-03 -2.410E-07 -3.028E+07 [SUB94] 298.14 600 75.246 1.116E-01 -6.589E-05 -6.321E+05 600 1300 117.746 1.674E-02 -5.321E-06 -3.283E+06 C2Cl4 -12427 343.4 1300 6000 132.786 3.686E-05 2.361E-09 -7.206E+06 [SUB94] 298.14 700 76.342 1.505E-01 -6.799E-05 -1.093E+06 700 1900 143.958 3.085E-02 -7.407E-06 -7.738E+06 C2H2Cl4 -149369 356.0 1900 4000 181.342 -7.737E-04 6.853E-08 -2.320E+07 [SUB94] 298.14 700 124.213 6.121E-02 -3.502E-05 -1.829E+06 C2Cl5 39000 396.5 700 6000 154.083 -1.046E-04 1.074E-08 -3.848E+06 [SUB94] 298.14 700 103.110 1.242E-01 -5.979E-05 -1.493E+06 C2HCl5 -145603 381.5 700 1500 150.772 3.083E-02 -8.381E-06 -5.178E+06 [SUB94] 298.14 800 139.890 7.572E-02 -3.826E-05 -2.025E+06 C2Cl6 -141500 398.6 800 6000 182.258 2.664E-04 -2.778E-08 -6.169E+06 [SUB94] 298.14 1000 39.083 1.139E-02 2.563E-06 -2.715E+05 1000 2100 28.122 3.126E-02 -6.686E-06 7.655E+04 2100 6400 77.129 -3.728E-03 2.647E-07 -2.723E+07 C2H 569000 209.7 6400 10000 63.159 -1.206E-04 5.032E-09 3.479E+07 [SUB94] C2NH 610438 240.6 298.14 1000 49.550 3.850E-02 -1.276E-05 -5.046E+05 [SUB94]

PAGE 229

206Table A-1. Continued 1000 3200 75.742 5.872E-03 -8.461E-07 -5.981E+06 3200 6000 86.604 1.665E-04 -1.148E-08 -1.775E+07 298.14 1700 43.333 3.194E-02 -6.585E-06 -7.339E+05 1700 4700 83.176 1.701E-03 5.040E-08 -2.274E+07 4700 9200 55.556 6.419E-03 -8.416E-08 1.634E+08 C2H2 227400 200.9 9200 10000 -13.730 1.823E-02 -6.453E-07 8.536E+08 [SUB94] 298.14 700 7.990 1.121E-01 -4.332E-05 1.722E+05 700 1500 40.616 5.243E-02 -1.233E-05 -2.801E+06 1500 3900 94.926 5.278E-03 -6.030E-07 -2.527E+07 C2H3 260000 232.8 3900 6000 106.963 2.447E-04 -1.551E-08 -4.563E+07 [SUB94] 298.14 600 -1.510 1.593E-01 -6.691E-05 2.528E+05 600 1300 36.420 8.033E-02 -2.049E-05 -2.366E+06 1300 3000 104.259 1.404E-02 -1.977E-06 -2.423E+07 C2H4 52400 219.3 3000 6000 130.538 6.132E-04 -4.348E-08 -5.492E+07 [SUB94] 298.14 700 -6.535 1.754E-01 -6.480E-05 5.870E+05 700 1400 40.091 9.217E-02 -2.263E-05 -3.852E+06 1400 3300 126.700 1.229E-02 -1.599E-06 -3.522E+07 C2H5 107000 250.5 3300 6000 151.468 5.557E-04 -3.808E-08 -6.853E+07 [SUB94] 298.14 700 -2.384 1.924E-01 -6.902E-05 3.223E+05 700 1400 45.037 1.083E-01 -2.658E-05 -4.260E+06 C2H6 -84000 229.2 1400 3200 145.619 1.528E-02 -2.028E-06 -4.044E+07 [SUB94]

PAGE 230

207Table A-1. Continued 3200 6000 175.852 6.958E-04 -4.818E-08 -7.984E+07 298.14 900 36.285 3.625E-02 -1.505E-05 -1.059E+05 900 5300 59.166 1.393E-03 -7.918E-09 -3.094E+06 C2N_CCN 804806 239.0 5300 6000 54.098 3.192E-03 -1.816E-07 8.302E+06 [SUB94] 298.14 800 35.193 4.199E-02 -1.922E-05 -8.578E+04 800 4000 60.762 8.025E-04 -1.076E-07 -3.195E+06 C2N_CNC 684924 233.8 4000 6000 62.406 -1.348E-05 1.016E-09 -5.053E+06 [SUB94] 298.14 1300 52.476 3.471E-02 -9.851E-06 -4.324E+05 C2N2 309100 242.2 1300 6000 84.877 1.590E-03 -2.865E-08 -1.049E+07 [SUB94] 298.14 1000 32.216 2.910E-02 -1.067E-05 2.006E+05 1000 2800 59.940 -5.083E-03 1.525E-06 -5.539E+06 2800 5100 -7.286 2.097E-02 -1.204E-06 1.174E+08 5100 8000 60.731 9.544E-03 -8.308E-07 -3.892E+08 C3 839958 237.6 8000 10000 211.874 -1.840E-02 6.183E-07 -1.688E+09 [SUB94] 298.14 900 48.296 4.083E-02 -1.441E-05 -5.736E+05 900 1600 62.911 1.749E-02 -3.808E-06 -2.353E+06 1600 2100 200.713 -7.828E-02 1.501E-05 -8.447E+07 C3H 630000 241.9 2100 4000 83.754 1.205E-03 -1.207E-07 -1.055E+07 [SUB94] 298.14 1300 62.591 4.404E-02 -1.206E-05 -8.946E+05 1300 1600 85.867 1.739E-02 -3.441E-06 -6.296E+06 C3NH 380000 251.7 1600 2000 362.317 -1.756E-01 3.441E-05 -1.691E+08 [SUB94]

PAGE 231

208Table A-1. Continued 2000 4000 107.372 1.695E-03 -1.735E-07 -1.461E+07 298.14 600 10.206 1.907E-01 -9.046E-05 0 C3H4_1 192129 244.0 600 1350 49.160 9.737E-02 -2.733E-05 -2.051E+06 [SUB94] 298.14 600 17.229 1.679E-01 -7.440E-05 0 C3H4_2 185435 248.2 600 1350 43.111 1.040E-01 -2.988E-05 -1.272E+06 [SUB94] 298.14 700 5.297 2.209E-01 -8.349E-05 5.983E+04 700 1400 65.858 1.114E-01 -2.731E-05 -5.566E+06 1400 3400 171.676 1.427E-02 -1.822E-06 -4.429E+07 C3H6 20418 266.7 3400 4000 199.650 1.069E-03 -8.232E-08 -8.139E+07 [SUB94] 298.14 500 -238.356 1.027E+00 -8.058E-04 5.323E+06 C3H6_1 53304 237.6 500 1350 47.235 1.457E-01 -4.106E-05 -3.758E+06 [SUB94] 298.14 600 -11.349 2.677E-01 -1.202E-04 5.443E+05 C3H6_2 20418 267.0 600 1350 41.168 1.476E-01 -4.207E-05 -2.545E+06 [SUB94] 298.14 600 -21.526 3.444E-01 -1.597E-04 5.840E+05 600 1000 49.099 1.782E-01 -4.895E-05 -3.304E+06 1000 1473.09 72.225 1.300E-01 -2.721E-05 0 1473.09 2200 -217.225 3.025E-01 -5.204E-05 1.937E+08 2200 2800 -2.181 1.193E-01 -1.388E-05 2.094E+08 2800 3400 -2055.691 9.570E-01 -1.097E-04 3.808E+09 C3H8 -103847 270.0 3400 4000 188.615 2.092E-02 0 0 [SUB94]

PAGE 232

209Table A-1. Continued 298.14 1600 53.750 2.905E-02 -5.781E-06 -4.842E+05 1600 1900 618.579 -3.745E-01 7.518E-05 -3.212E+08 C3N 609000 252.1 1900 4000 102.028 -2.584E-03 7.465E-08 -2.887E+07 [SUB94] 298.14 2500 53.395 2.800E-02 -4.956E-06 -3.620E+05 2500 7000 106.174 -3.328E-03 2.418E-07 -4.381E+07 C4 1033918 252.9 7000 10000 92.452 5.378E-04 -4.244E-08 -1.517E+07 [SUB94] 298.14 500 -116.904 7.514E-01 -4.922E-04 3.033E+06 500 1000 52.283 2.540E-01 -7.632E-05 -3.081E+06 1000 1600 293.778 -1.134E-02 7.926E-06 -6.351E+07 1600 2200 268.317 6.969E-03 4.634E-06 -5.173E+07 2200 3000 423.048 -6.055E-02 1.164E-05 -2.457E+08 C4H10_1 -126148 310.2 3000 3500 188.341 6.067E-02 -5.726E-06 0 [SUB94] 298.14 500 -184.144 9.895E-01 -7.170E-04 4.419E+06 C4H10_2 -134516 294.7 500 1350 62.001 2.403E-01 -7.081E-05 -3.848E+06 [SUB94] 298.14 500 21.763 2.252E-01 -1.717E-04 0 C4H2 472792 250.1 500 1350 75.170 5.027E-02 -1.270E-05 -1.418E+06 [SUB94] 298.14 600 23.517 2.097E-01 -1.023E-04 -3.325E+05 C4H4 304595 279.5 600 1350 68.631 1.010E-01 -2.821E-05 -2.699E+06 [SUB94] 298.14 700 9.680 2.650E-01 -1.125E-04 1.292E+05 C4H6_1 162214 293.1 700 1350 32.197 1.993E-01 -6.410E-05 0 [SUB94] C4H6_2 110165 278.8 298.14 500 -98.099 6.577E-01 -4.803E-04 2.156E+06 [SUB94]

PAGE 233

210Table A-1. Continued 500 1350 76.018 1.357E-01 -3.848E-05 -3.731E+06 C4H8 -152591 265.5 298.14 1500 21.681 2.600E-01 7.727E-05 1.839E+06 [SUB94] 298.14 500 -99.938 6.617E-01 -4.458E-04 2.485E+06 C4H8_1 -126 305.7 500 1350 53.292 2.069E-01 -6.129E-05 -3.001E+06 [SUB94] 298.14 1100 66.110 4.971E-02 -1.630E-05 -6.063E+05 1100 1600 93.493 1.395E-02 -3.074E-06 -5.497E+06 1600 2300 175.665 -4.194E-02 7.688E-06 -5.567E+07 C4N 790000 285.0 2300 4000 110.391 5.997E-04 -5.737E-08 -1.124E+07 [SUB94] 298.14 1000 77.196 6.461E-02 -2.131E-05 -7.699E+05 1000 2100 117.116 1.349E-02 -2.631E-06 -8.239E+06 C4N2 533460 290.1 2100 6000 136.050 3.163E-04 -2.447E-08 -2.046E+07 [SUB94] 298.14 900 62.972 6.340E-02 -2.458E-05 -3.726E+05 900 3500 106.597 2.957E-03 -4.243E-07 -7.496E+06 C5 1050942 271.7 3500 10000 112.171 1.322E-05 -6.633E-10 -1.313E+07 [SUB94] 298.14 1300 92.087 6.513E-02 -1.855E-05 -1.187E+06 1300 1600 127.558 2.345E-02 -4.767E-06 -8.931E+06 1600 2000 493.195 -2.323E-01 4.547E-05 -2.237E+08 C5HN 600000 303.0 2000 4000 156.227 2.051E-03 -2.096E-07 -1.938E+07 [SUB94] 298.14 500 -470.127 1.873E+00 -1.509E-03 1.074E+07 500 800 7.450 3.324E-01 -1.211E-04 -2.769E+06 C5H8_1 32928 289.8 800 1350 4.770 3.171E-01 -1.046E-04 0 [SUB94]

PAGE 234

211Table A-1. Continued 298.14 1100 67.079 9.244E-02 -3.357E-05 -2.755E+05 1100 1600 135.934 3.636E-03 -1.306E-06 -1.264E+07 C5N 830000 304.6 1600 4000 141.655 -2.127E-03 2.719E-07 -1.228E+07 [SUB94] C6H5Cl 52773 297.4 298.14 1500 179.992 1.596E-02 0 -7.792E+06 [SUB94] 298.14 500 -182.327 9.542E-01 -6.880E-04 3.619E+06 C6H6 82927 269.3 500 1350 58.568 2.359E-01 -8.006E-05 -4.823E+06 [SUB94] 298.14 1200 96.197 6.881E-02 -2.166E-05 -8.885E+05 1200 1600 136.761 1.825E-02 -3.895E-06 -8.791E+06 1600 2300 255.584 -6.182E-02 1.133E-05 -8.213E+07 C6N 1010000 337.5 2300 4000 159.212 9.606E-04 -9.347E-08 -1.643E+07 [SUB94] 298.14 1200 108.875 7.920E-02 -2.420E-05 -1.081E+06 1200 1600 155.202 2.287E-02 -4.869E-06 -1.054E+07 1600 2200 341.228 -1.038E-01 1.942E-05 -1.240E+08 C6N2 750000 342.6 2200 4000 183.123 1.329E-03 -1.320E-07 -1.991E+07 [SUB94] 298.14 1200 119.937 8.952E-02 -2.674E-05 -1.475E+06 1200 1600 167.892 3.078E-02 -6.424E-06 -1.118E+07 1600 2100 483.179 -1.863E-01 3.565E-05 -2.007E+08 C7NH 820000 353.6 2100 4000 205.569 2.196E-03 -2.204E-07 -2.491E+07 [SUB94] 298.14 1200 113.429 7.583E-02 -2.354E-05 -1.118E+06 1200 1600 158.283 2.067E-02 -4.393E-06 -1.011E+07 C7N 1060000 363.2 1600 2300 293.904 -7.063E-02 1.295E-05 -9.390E+07 [SUB94]

PAGE 235

212Table A-1. Continued 2300 4000 183.774 1.095E-03 -1.073E-07 -1.875E+07 298.14 1200 125.526 8.959E-02 -2.800E-05 -1.136E+06 1200 1600 178.856 2.370E-02 -5.024E-06 -1.172E+07 1600 2200 377.032 -1.110E-01 2.078E-05 -1.328E+08 C8N 1240000 390.1 2200 4000 207.862 1.394E-03 -1.380E-07 -2.134E+07 [SUB94] 298.14 1200 138.689 9.935E-02 -3.031E-05 -1.338E+06 1200 1600 196.805 2.875E-02 -6.109E-06 -1.322E+07 1600 2200 430.872 -1.305E-01 2.443E-05 -1.560E+08 C8N2 980000 395.1 2200 4000 232.070 1.633E-03 -1.609E-07 -2.521E+07 [SUB94] 298.14 1200 148.372 1.126E-01 -3.416E-05 -1.787E+06 1200 1600 210.770 3.575E-02 -7.463E-06 -1.429E+07 1600 2100 586.995 -2.233E-01 4.269E-05 -2.405E+08 C9NH 1050000 403.9 2100 4000 254.482 2.527E-03 -2.537E-07 -2.981E+07 [SUB94] 298.14 800 23.686 1.000E-05 -1.412E-06 -1.540E+05 800 3400 22.633 -9.689E-04 1.415E-07 3.847E+05 Cl 121302 165.2 3400 10000 20.835 -8.310E-06 4.021E-10 2.260E+06 [SUB94] 298.14 4100 37.589 1.823E-04 1.360E-07 -1.750E+05 GaCl -69622 240.3 4100 6000 25.418 2.980E-03 -3.342E-09 5.102E+07 [SUB94] 298.14 800 29.017 -1.963E-03 4.705E-06 2.588E+04 800 2400 25.299 7.871E-03 -1.320E-06 -1.616E+05 HCl -92310 186.9 2400 6000 31.673 2.508E-03 -1.856E-07 -3.868E+05 [SUB94]

PAGE 236

213Table A-1. Continued 298.14 3900 37.682 1.040E-04 1.936E-07 -1.448E+05 InCl -72148 248.3 3900 6000 14.971 6.635E-03 -3.069E-07 7.373E+07 [SUB94] 298.14 1500 36.246 1.913E-03 -4.635E-07 -2.527E+05 1500 3300 40.693 -3.208E-03 1.101E-06 -8.928E+05 3300 5600 12.227 1.468E-02 -1.649E-06 -7.633E+06 Cl2 0 223.1 5600 10000 109.552 -1.331E-02 5.566E-07 -3.130E+08 [SUB94] 298.14 2300 58.245 -4.760E-04 3.134E-07 -3.834E+05 GaCl2 -220979 302.9 2300 6000 50.476 3.846E-03 -3.312E-07 6.173E+06 [SUB94] Ga2Cl2 -220973 353.5 298.15 6000 83.100 2.245E-05 -2.684E-09 -3.018E+05 [SUB94] 298.14 3600 58.201 -3.802E-04 2.872E-07 -2.910E+05 InCl2 -201484 314.2 3600 6000 56.463 2.071E-03 -1.861E-07 -1.269E+07 [SUB94] In2Cl2 -232177 372.4 298.15 6000 83.122 1.125E-05 -1.345E-09 -2.165E+05 [SUB94] 298.14 2300 82.177 9.538E-04 -2.397E-07 -6.855E+05 GaCl3 -432625 324.5 2300 6000 83.139 1.589E-06 -1.281E-10 -8.867E+05 [SUB94] InCl3 -369693 339.2 298.15 6000 82.938 1.042E-04 -1.248E-08 -5.613E+05 [SUB94] GaCl4 -602327 419.7 298.15 6000 132.773 1.302E-04 -1.560E-08 -9.029E+05 [SUB94] In2Cl4 -579126 456.0 298.15 6000 132.848 9.229E-05 -1.106E-08 -6.844E+05 [SUB94] 298.14 3800 182.038 5.920E-04 -9.666E-08 -1.464E+06 Ga2Cl6 -962464 492.7 3800 6000 182.909 3.208E-06 -2.970E-10 -1.770E+06 [SUB94] In2Cl6 -882340 527.7 298.15 6000 182.554 1.835E-04 -2.198E-08 -1.198E+06 [SUB94] Ga 272000 169.0 298.14 600 38.711 -2.108E-02 5.921E-06 -6.770E+05 [SUB94]

PAGE 237

214Table A-1. Continued 600 1400 30.750 -1.075E-02 3.279E-06 3.019E+05 1400 6000 21.983 -6.385E-04 8.778E-08 1.984E+06 6000 10000 9.704 9.028E-04 6.806E-08 1.368E+08 298.14 1000 22.503 2.007E-02 -7.377E-06 1.424E+05 1000 3400 36.596 5.827E-04 3.138E-07 -2.151E+06 3400 5800 35.170 6.243E-03 -8.725E-07 -4.964E+07 GaH 214323 199.8 5800 6000 94.036 -9.534E-03 3.057E-07 -2.849E+08 [SUB94] 298.14 1100 33.729 1.874E-02 -4.577E-06 3.904E+04 Ga2 433600 267.1 1100 2500 40.336 8.371E-03 -1.608E-07 -6.242E+05 [SUB94] H 217999 114.7 298.15 6000 20.786 0 0 0 [SUB94] 298.14 900 22.573 2.164E-02 -8.679E-06 1.177E+05 900 3300 35.333 1.954E-03 -8.363E-08 -1.506E+06 InH 215016 207.7 3300 6000 53.482 -9.479E-04 -2.567E-07 -7.437E+07 [SUB94] 298.14 900 29.254 -2.267E-03 4.316E-06 2.116E+04 900 2600 26.349 6.226E-03 -7.612E-07 -4.852E+05 2600 6000 16.433 8.307E-03 -6.415E-07 2.451E+07 6000 16500 78.442 -4.553E-03 1.061E-07 -3.990E+08 NH 358433 181.2 16500 20000 41.987 -1.226E-03 2.052E-08 9.200E+08 [SUB94] 298.14 800 31.620 5.719E-02 -2.100E-05 -2.304E+05 N3H 294000 239.3 800 2100 57.402 1.744E-02 -3.349E-06 -3.613E+06 [SUB94]

PAGE 238

215Table A-1. Continued 2100 6000 81.038 8.198E-04 -1.461E-08 -1.877E+07 298.14 1000 31.357 -5.518E-03 4.478E-06 -1.132E+05 1000 2100 17.849 1.168E-02 -1.888E-06 2.560E+06 H2 0 130.7 2100 6000 32.051 2.146E-03 -6.857E-08 -7.122E+06 [SUB94] 298.14 900 27.174 1.655E-02 -3.012E-07 1.583E+05 900 2200 30.778 1.703E-02 -2.630E-06 -1.582E+06 NH2 190000 195.0 2200 6000 45.494 6.385E-03 -4.552E-07 -1.037E+07 [SUB94] N2H2_1_1N2H2 283962 228.2 298.14 700 11.294 7.922E-02 -3.040E-05 3.581E+05 [SUB94] 700 1700 37.418 3.308E-02 -7.209E-06 -2.187E+06 1700 6000 76.059 2.407E-03 -1.934E-07 -2.176E+07 298.14 800 10.046 7.712E-02 -2.664E-05 4.207E+05 800 1700 39.212 3.247E-02 -7.158E-06 -3.368E+06 1700 5300 78.695 4.766E-04 2.085E-07 -2.181E+07 N2H2_CIS 247885 218.4 5300 6000 52.960 8.505E-03 -4.837E-07 5.199E+07 [SUB94] 298.14 800 8.430 7.959E-02 -2.795E-05 4.766E+05 800 1700 39.063 3.215E-02 -6.990E-06 -3.423E+06 1700 4500 71.814 3.506E-03 -1.227E-08 -1.561E+07 N2H2_TRANS 211859 218.3 4500 6000 57.776 9.540E-03 -6.898E-07 -3.415E+06 [SUB94] 298.14 1100 21.218 4.574E-02 -1.085E-05 1.534E+05 1100 2600 45.828 1.963E-02 -2.846E-06 -6.582E+06 NH3 -45940 192.8 2600 6000 76.177 2.454E-03 -8.429E-08 -3.612E+07 [SUB94]

PAGE 239

216Table A-1. Continued 298.14 600 9.501 1.451E-01 -6.574E-05 1.336E+05 600 1600 51.478 5.589E-02 -1.231E-05 -2.639E+06 1600 4200 112.895 5.963E-03 -6.336E-07 -3.186E+07 N2H4 95180 238.5 4200 6000 127.366 3.171E-04 -1.953E-08 -5.990E+07 [SUB94] 298.14 600 15.352 1.054E-02 2.390E-06 1.890E+05 600 1100 8.405 3.137E-02 -1.327E-05 2.213E+05 1100 2900 41.363 -1.043E-02 1.516E-06 -5.674E+06 2900 8400 30.106 -3.387E-03 3.487E-07 -2.187E+05 In 240700 173.8 8400 10000 -55.474 1.108E-02 -3.416E-07 8.985E+08 [SUB94] 298.14 1800 35.821 1.310E-02 -1.224E-07 4.027E+04 In2 418800 285.1 1800 3000 41.188 1.183E-02 -4.893E-07 -6.095E+06 [SUB94] 298.14 2950 20.894 -1.691E-04 6.011E-08 -5.578E+03 N 472687 153.3 2950 6000 16.376 4.567E-04 1.674E-07 1.512E+07 [SUB94] 298.14 800 27.223 2.520E-03 3.236E-06 7.665E+04 800 2200 28.424 6.379E-03 -1.240E-06 -8.339E+05 N2 0 191.6 2200 6000 37.550 1.232E-05 2.535E-08 -6.855E+06 [SUB94] 298.14 800 24.402 5.021E-02 -2.048E-05 -1.233E+05 800 2200 52.350 7.104E-03 -1.439E-06 -3.739E+06 N3 436000 223.1 2200 6000 61.965 1.111E-04 -8.863E-09 -9.310E+06 [SUB94]

PAGE 240

217Table A-2. Gas Phase (Metalorga nics, Adducts, Oligomers, etc) Cp (J/mol K) = A + B*T + C*T2 + D*T3 + E*T-2 Component T (K) range H298 (J/mol) S298 (J/mol K) A B C D E Ref AlCH3 298.15-3000 70000 239.2 35.039 7.017E-02 -2.651E-05 3.625E-09 -3.492E+05 [Prz98] Al(CH3)3 298.15-3000 -87864 345.0 67.035 2.236E-01 -8.632E-05 1.206E-08 -1.996E+06 [Prz98] Al2(CH3)6 298.15-3000 -268000 499.3 153.819 4.582E-01 -1.788E-04 2.521E-08 -4.421E+06 [Prz98] GaCH3 298.15-3000 79000 250.9 33.365 7.271E-02 -2.783E-05 3.853E-09 -5.996E+05 [Prz98] Ga(CH3)3 298.15-3000 -45100 354.6 56.520 2.506E-01 -1.101E-04 1.900E-08 -1.247E+06 [Prz98] InCH3 298.15-3000 215000 256.3 37.027 6.639E-02 -2.449E-05 3.282E-09 -4.696E+05 [Prz98] In(CH3)3 298.15-3000 207000 367.5 75.349 2.121E-01 -8.065E-05 1.112E-08 -1.933E+06 [Prz98] AlNH3 298.15-3000 250000 257.6 41.860 5.102E-02 -1.546E-05 1.701E-09 4.237E+04 [Prz98] (AlN)3 298.15-3000 -525000 300.3 108.730 3.436E-02 -1.743E-05 2.986E-09 -3.663E+06 [Prz98] MMANH 298.15-3000 -190000 286.2 61.619 8.466E-02 -3.058E-05 3.995E-09 -5.076E+05 [Prz98] (MMANH)3 298.15-3000 -750000 506.7 202.220 2.815E-01 -1.080E-04 1.508E-08 -7.674E+06 [Prz98] TMANH3 298.15-3000 -246000 399.5 112.231 2.546E-01 -9.317E-05 1.267E-08 -3.771E+06 [Prz98] (DMANH2)3 298.15-3000 -760000 587.4 324.353 4.812E-01 -1.742E-04 2.341E-08 -1.257E+07 [Prz98] GaNH3 298.15-3000 200000 265.6 40.799 5.281E-02 -1.635E-05 1.842E-09 5.339E+04 [Prz98] (GaN)3 298.15-3000 -360000 322.0 122.283 1.554E-02 -8.002E-05 1.386E-08 -3.307E+06 [Prz98] MMGNH 298.15-3000 -130000 293.9 60.265 8.687E-02 -3.168E-05 4.174E-09 -3.669E+05 [Prz98] (MMGNH)3 298.15-3000 -520000 526.9 213.634 2.695E-01 -1.027E-04 1.421E-08 -7.327E+06 [Prz98] TMGNH3 298.15-3000 -170000 400.4 87.982 2.934E-01 -1.125E-04 1.557E-08 -1.451E+06 [Prz98] (DMGNH2)3 298.15-3000 -525000 573.7 264.909 5.836E-01 -2.304E-04 3.301E-08 -9.659E+06 [Prz98] GaCl3NH3 298.15-3000 -619000 376.1 99.906 7.306E-02 -2.672E-05 3.543E-09 -6.737E+05 [Prz98] GaH2 298.15-3000 261000 225.1 27.738 3.792E-02 -1.757E-05 2.814E-09 -1.971E+05 [Prz98] GaH3 298.15-3000 140000 217.3 25.648 7.203E-02 -3.346E-05 5.366E-09 1.415E+05 [Prz98] InH2 298.15-3000 301000 234.3 30.939 3.490E-02 -1.652E-05 2.694E-09 -3.536E+05 [Prz98] InH3 298.15-3000 226000 228.6 36.024 5.894E-02 -2.743E-05 4.417E-09 -7.673E+05 [Prz98]

PAGE 241

218Table A-3. Liquid Phase Cp (J/mol K) = A + B*T + C*T2 + D*T-2 Species H298 (J/mole) S298 (J/mole K) T(K) Range A B C D Ref# 298.14 465.69 64.929 8.787E-02 0 0 AlCl3(l) -670277 185.2 465.69 2000 125.520 0 0 0 [SUB94] 298.14 700 24.367 3.769E-03 5.266E-06 -1.482E+05 700 933.39 38.584 -3.706E-02 3.459E-05 -1.482E+05 Al(l) 10711 39.8 933.39 2900 31.748 0 0 0 [SUB94] CCl4(l) -128005 216.4 298.14 500 123.989 2.611E-02 0 0 [SUB94] C6H5Cl (l) 11698 194.1 298.14 410 59.932 2.938E-01 0 0 [SUB94] 298.14 457.69 132.065 1.840E-02 -7.288E-05 -3.967E+06 457.69 793.2 26.736 1.303E-01 -1.226E-05 4.184E+05 NH4Cl(l) -300018 116.8 793.2 1000 121.336 0 0 0 [SUB94] 298.14 393 35.146 4.184E-02 0 0 393 498 58.576 0 0 0 InCl(l) -170080 131.0 498 900 62.760 0 0 0 [SUB94] 298.14 351 118.407 0 0 0 GaCl3(l) -513168 167.9 351 600 128.030 0 0 0 [SUB94] 298.14 856 90.923 3.045E-02 1.528E-08 -2.951E+02 InCl3(l) -503000 173.5 856 1200 117.000 -1.345E-15 2.684E-19 -4.159E-07 [SUB94] 298.14 598 93.722 9.707E-02 0 0 In2Cl3(l) -506264 306.4 598 1000 146.440 0 0 0 [SUB94] 298.14 538 125.102 1.703E-01 0 0 In3Cl4(l) -673624 455.6 538 1000 205.016 0 0 0 [SUB94] 298.14 531 188.280 2.738E-01 0 0 In4Cl7(l) -1188256 655.5 531 1000 322.168 0 0 0 [SUB94] 200 302.89 108.229 -4.543E-01 7.115E-04 -8.799E+05 Ga(l) 5590 59.2 302.89 4000 26.069 -3.012E-04 2.410E-07 2.367E+05 [SUB94] 298.14 429.69 21.839 1.145E-02 1.272E-05 4.581E+04 In(l) 3283 65.3 429.69 3800 27.456 -1.092E-03 5.020E-07 4.234E+05 [SUB94] 298.14 465.69 64.929 8.787E-02 0 0 AlCl3(l) -670277 185.2 465.69 2000 125.520 0 0 0 [SUB94] 298.14 700 24.367 3.769E-03 5.266E-06 -1.482E+05 Al(l) 10711 39.8 700 933.39 38.584 -3.706E-02 3.459E-05 -1.482E+05 [SUB94]

PAGE 242

219Table A-3. Continued 933.39 2900 31.748 0 0 0 CCl4(l) -128005 216.4 298.14 500 123.989 2.611E-02 0 0 [SUB94] C6ClH5(l) 11698 194.1 298.14 410 59.932 2.938E-01 0 0 [SUB94] 298.14 457.69 132.065 1.840E-02 -7.288E-05 -3.967E+06 457.69 793.2 26.736 1.303E-01 -1.226E-05 4.184E+05 NH4Cl(l) -300018 116.8 793.2 1000 121.336 0 0 0 [SUB94] 298.14 393 35.146 4.184E-02 0 0 393 498 58.576 0 0 0 InCl(l) -170080 131.0 498 900 62.760 0 0 0 [SUB94] 298.14 351 118.407 0 0 0 GaCl3(l) -513168 167.9 351 600 128.030 0 0 0 [SUB94] 298.14 856 90.923 3.045E-02 1.528E-08 -2.951E+02 InCl3(l) -503000 173.5 856 1200 117.000 -1.345E-15 2.684E-19 -4.159E-07 [SUB94] 298.14 598 93.722 9.707E-02 0 0 In2Cl3(l) -506264 306.4 598 1000 146.440 0 0 0 [SUB94] 298.14 538 125.102 1.703E-01 0 0 In3Cl4(l) -673624 455.6 538 1000 205.016 0 0 0 [SUB94] 298.14 531 188.280 2.738E-01 0 0 In4Cl7(l) -1188256 655.5 531 1000 322.168 0 0 0 [SUB94] 200 302.89 108.229 -4.543E-01 7.115E-04 -8.799E+05 Ga(l) 5590 59.2 302.89 4000 26.069 -3.012E-04 2.410E-07 2.367E+05 [SUB94] 298.14 429.69 21.839 1.145E-02 1.272E-05 4.581E+04 In(l) 3283 65.3 429.69 3800 27.456 -1.092E-03 5.020E-07 4.234E+05 [SUB94]

PAGE 243

220Table A-4. Solid Phase H298 S298 Cp (J/mol K) = A + B*T + C*T2 + D*T-2 Species (J/mole) (J/mole K) T(K) Range A B C D Ref # 298.14 465.69 64.929 8.787E-02 0 0 AlCl3(s) -705632 109.3 465.69 2000 125.520 0 0 0 [SUB94] AlH3(s) -11400 30.0 298.14 500 49.383 2.408E-02 6.330E-07 -1.459E+06 [SUB94] 298.14 500 -56.593 3.221E-01 -2.772E-04 1.364E+06 500 1300 43.974 1.181E-02 -5.001E-06 -2.010E+06 1300 1900 21.657 2.379E-02 -5.375E-06 1.045E+07 AlN(s) -317984 20.2 1900 3000 55.534 -2.407E-03 4.190E-07 -7.645E+06 [SUB94] Al4C3(s) -206900 89.0 298.14 2500 148.741 3.346E-02 3.921E-10 -3.728E+06 [SUB94] 298.14 700 24.367 3.769E-03 5.266E-06 -1.482E+05 700 933.39 38.584 -3.706E-02 3.459E-05 -1.482E+05 Al(s) 0 28.3 933.39 2900 31.748 0 0 0 [SUB94] 298.14 457.69 132.065 1.840E-02 -7.288E-05 -3.967E+06 457.69 793.2 26.736 1.303E-01 -1.226E-05 4.184E+05 NH4Cl(s) -314553 94.9 793.2 1000 121.336 0 0 0 [SUB94] 298.14 393 35.146 4.184E-02 0 0 393 498 58.576 0 0 0 InCl(s) -186188 95.0 498 900 62.760 0 0 0 [SUB94] InCl2(s) -362753 122.2 298.14 509 58.576 5.021E-02 0 0 [SUB94] 298.14 351 118.407 0 0 0 GaCl3(s) -524674 135.1 351 600 128.030 0 0 0 [SUB94] 298.14 856 90.923 3.045E-02 1.528E-08 -2.951E+02 InCl3(s) -530000 142.0 856 1200 117.000 -1.345E-15 2.684E-19 -4.159E-07 [SUB94] 298.14 598 93.722 9.707E-02 0 0 In2Cl3(s) -548104 236.4 598 1000 146.440 0 0 0 [SUB94] 298.14 538 125.102 1.703E-01 0 0 In3Cl4(s) -736384 338.9 538 1000 205.016 0 0 0 [SUB94] 298.14 531 188.280 2.738E-01 0 0 In4Cl7(s) -1271936 497.9 531 1000 322.168 0 0 0 [SUB94] Diamond 1828 2.4 298.14 700 -10.399 6.646E-02 -3.708E-05 0 [SUB94]

PAGE 244

221Table A-4. Continued 700 2500 21.052 3.224E-03 -4.524E-07 -2.515E+06 2500 5700 23.836 1.116E-03 -1.692E-08 -3.989E+06 5700 6000 22.672 1.202E-03 0 0 200 302.89 108.229 -4.543E-01 7.115E-04 -8.799E+05 Ga(s) 0 40.7 302.89 4000 26.069 -3.012E-04 2.410E-07 2.367E+05 [SUB94] 298.14 429.69 21.839 1.145E-02 1.272E-05 4.581E+04 In(s) 0 57.6 429.69 3800 27.456 -1.092E-03 5.020E-07 4.234E+05 [SUB94] Table A-5. Carbon (Graphite) Cp = A + B*T + C*T-2 + D*T-3 + E*T-4 Species H298 (J/mol) S298 (J/mol K) T(K) Range A B C D E Ref # C(s) 0 5.7 298.14 6000 24.300 9.446E-04 -5.125E+06 1.586E+09 -1.440E+11 [SUB94] Table A-6. GaN and InN Cp (J/mol K) = A + B*T + C*T2 + D*T-2 + E*T-3 Species H298 (J/mole) S298 (J/mole K) T(K) Range A B C D E Ref GaN(s) -156800 30 298 4000 32.532 1.867E-02 -2.335E-06 -3.307E+05 0 [Unl03] InN(s) -71000 42.5 298 4000 43.886 8.194E-03 0 -1.007E+06 8.353E+07 [Lei04]

PAGE 245

222 LIST OF REFERENCES [Abe93] C. R. Abernathy, S. J. Pearton, F. Ren, and P. W. Wisk, J. Vac. Sci. Technol. B 11, 179 (1993). [Abl05] A. Able, W. Wegscheider, K. Engl and J. Zweck, J. Cryst. Growth 276 (2005) 415. [Aco04] J.D. Acord, S. Raghavan, D.W. Snyder and J.M. Redwing, J. Cryst. Growth 272 (2004) 65. [Ade01] J. Aderhold, V. Yu. Davydov, F. Fedler H. Klausing, D. Mistele, T. Rotter, O. Semchinova, J. Stemmer, and J. Graul, J. Cryst. Growth 222, 701 (2001). [Age97] J. Ager, T. Suski, S. Ruvinov, J. Krue ger, G. Conti, E. Weber, M. Bremser, R. Davis, and C. Kuo, Materials Research Society Symposium Proceedings, 449, 775 (1997). [Alb98] J.D. Albrecht, R.P. Wang, P.P. Rude n, M. Farahmand, K.F. Brennan, J. Appl. Phys. 83 (1998) 4777. [Alm92a] M.J. Almond, C.A. Jenkins, D.A. Rice, J. Organomet. Chem. 439 (1992) 251. [Alm92b] M.J. Almond, M.G.B. Drew, C.E. Je nkins, D.A. Rice, J. Chem. Soc. Dalton Trans. (1992) 5. [Ama86] H. Amano, N. Sawaki, I. Akasaki, Y. Toyada, Appl. Phys. Lett. 48 (1986) 353. [Ama89] H. Amano, M. Kito, K. Hiramatsu, I. Akaski, Jpn. J. Appl. Phys. 28 (1989) L2112. [Amb96] O. Ambacher, M. Brandt, R. Dimitrov, T. Metzger, M. Stutzmann, R. Fisher, A. Miehr, A. Bergmajer, G. Dollinger, J. Vac. Sci. Technol. B 14 (1996) 3532. [And05] P.A. Anderson, C.E. Kendrick, R.J. Kinsey, A. Asadov, W. Gao, R.J. Reeves, S.M. Durbin, Phys. Status So lidi c (2005), 2(7), 2320-2323. [And74] A.F. Andreeyva and O.J. Elisee jva, Z. Neorg. Chim. 13, 185 (1974). [As00] D.J. As, T. Frey, D. Sc hikora, K. Lischka, V. Cimalla, J. Pezoldt, R. Goldhahn, S. Kaiser, and W. Gebhardt, A ppl. Phys. Lett. 76 (2000) 1686. [Ban72] V.S. Ban, J. Electr ochem. Soc. 119 (1972) 762.

PAGE 246

223 [Bar93] I. Barin, Thermochemical Data of Pure Substances, 2nd ed, VCH, Weinheim, 1993. [Bec02] F. Bechstedt, J. Furthmulle r, J. Cryst. Growth 246 (2002) 315. [Ben02] M. Benyoucef, M. Kuball, D. D. Kolesk e, A. E. Wickenden, R. L. Henry, M. Fatemi and M. E. Twigg, Mater. Sci. Eng. B, 93 (2002) 15. [Bhu02] A.G. Bhuiyan, T. Tanaka, A. Yamamo to, A. Hashimoto, Phys. Status Solidi a 194, 502 (2002). [Bhu03] A.G. Bhuiyan, T. Tanaka, K. Kasa shima, A. Hashimoto, A. Yamamoto, 5th International Conference on Nitride Semiconductors (ICNS -5), Nara, Japan, May 25, 2003. [Bi04] Z.X. Bi, R. Zhang, Z.L. Xie, X.Q. Xiu, Y.D. Ye, B. Liu, S.L. Gu, B. Shen, Y. Shi, Y.D. Zheng, Mater. Lett. 58 (2004) 3641. [Bil26] W. Biltz, W. Klemm. Z. Zacmah, Anorg. Chem. 152 (1926) 276. [Bin02] M. Binnewies, E. Milke, Thermoch emical Data of Elements and Compounds, second ed, Wiley-VCH, Wenheim, 2002. [Bok00] T.H. Bok, J.H. Ye, S.F.Y. Li, J. Vac. Sci. Technol. A 18 (2000) 2542. [Bon98] J.-M. Bonard, J.-P. Salvetat, T. St ockli, W.A. de Heer, L. Forro, and A. Chatelain, Appl. Phys. Lett.73 (1998) 918. [Bra96] O. Brandt, H. Yang, H. Kostal, K. Ploog, A ppl. Phys. Lett. 69 (1996) 2707. [Bre97] E. Bretschneider, diss. University of Florida 1997. [Bri03] O. Briot, B. Maleyre, S. Ruffenach C. Pinquier, F. Demangeot, and J. Frandon, Phys. Status Solidi(c) 0, No. 7, (2003) 2851. [Bri04] O. Briot, B. Maleyre, S. Ruffenach B. Gil, C. Pinquier, F. Demangeot, J. Frandon, J. Cryst. Growth 269 (2004) 22. [Bri72] L.H. Brixner, Mater. Res. Bull. 7 (1972) 879. [Bru89] T.R. Brumleve, S.A. Mucklejohn, N.W. O`Brien, J. Chem. Thermodyn., 21 (1989) 1193. [Bu93] Y. Bu, L. Ma, M.C. Lin, J. Vac. Sci. Technol. A11 (6) (1993) 2931. [Bur97] J. Burm, K. Chu, W. Davis, W.J. Sc haff, L.F. Eastman, T.J. Eustis, Appl. Phys. Lett. 70 (1997) 464.

PAGE 247

224 [But02] K.S.A. Butcher, M. Wintrebert-Fouquet, P.P.-T. Chen, T.L. Tansley, S. Sriheaw, Mater. Res. Symp. Proc. 693 (2002) 341. [But05a] K.S.A. Butcher, M. Wintrebert-Fouquet, P.P.-T. Chen, K.E. Prince, H. Timmers, S.K. Shrestha, T.V. Shubina, S.V. Ivanov, R. Wuhrer, M.R. Phillips, B. Monemar, Phys. Status Solidi c 2 (2005) 2263. [But05b] K.S.A. Butcher, T.L. Tansley, Superlatt. Microstructures 38 (2005) 1. [Cad01] R. Cadoret, A. Trass oudaine, E. Aujol, Phys. St atus Solidi a 183 (2001) 5. [Cad97] R. Cadoret, E. Gil-Lafon, J. Phys. I France 7 (1997) 889. [Cad99] R. Cadoret, J. Cr yst. Growth 205 (1999) 123. [Car00] S. Carlson and A.M.K. A ndersen, Phys. Rev. B 61 (2000) 11209. [Car01] S. Carlson and A.M.K. Anders en, J. Appl. Crystallogr. 34 (2001) 7. [Cha04] C.-A. Chang, C.-F. Shih, N.-C. Chen, P.-H. Chang, and K.-S. Liu, Phys. Status Solidi(c) 1, 2559 (2004). [Cha05] C.-Y. Chang, G.-C. Chi, W.-M. Wang, L.-C. Chen, K.-H. Chen, F. Ren, S. J. Pearton, Appl. Phys. Lett. 87, 093112 (2005). [Cha06] C.H. Chang, M. Akilian, M.L. Scha ttenburg, Applied Optics 45 (2006) 432. [Cha94] J.F.Chang, H.H. Kuo, I.C. Leu and M.H. Hon, Sens Actuators B84 258 (1994). [Che04] J. Chen, A. Br, G. Nouet and P. Ruterana, Superlatt. Microstructures, 36 (2004) 369. [Che05a] S.-C. Chen, S.-K. Lin, K.-T. Wu, C. -P. Huang, P.-H. Chang, N.C. Chen, C.-A. Chang, H.-C. Peng, C.-F. Shih, K.-S. Liu, Microelectronics Journal 36 (2005) 428. [Che05b] J.W. Chen, Y.F. Chen, H. Lu, a nd W.J. Schaff, Appl. Phys. Lett. 87 (2005) 041907. [Che06] T.P. Chen, C. Thomidis, J. Abell, W. Li, T.D. Moustakas, Journal of Crystal Growth 288(2) (2006) 254-260. [Che91] Q. Chen and P. D. Dapkus, J. Electrochem. Soc., 138, 2821 (1991). [Che97] Y. Chen, D.E. Laughlin, X. Ma, M. Kryder, J. Appl. Phys. 81 (1997) 4380. [Chu98] S.N.G. Chu, J. Electrochem. Soc. Vol. 145 (1998) 3621. [Clo04] R. Clos, A. Dadgar and A. Krost, Phys. Status Solidi a 201 (2004) R75.

PAGE 248

225 [Cra02] M. D. Craven, S. H. Lim, F. Wu, J. S. Speck, S. P. DenBaars, Appl. Phys. Lett. 81 (2002) 469. [Dad03] A. Dadgar, M. Poschenrie der, A. Reiher, J. Blsing, J. Christen, A. Krtschil, T. Finger, T. Hempel, A. Diez, and A. Krost, Appl. Phys. Lett., 82 (2003) 28. [Dad04] A. Dadgar, F. Schulze, T. Zettler, K. Haberland, R. Clos, G. Straburger, J. Blsing, A. Diez and A. Krost, J. Cryst. Growth, 272 (2004) 72. [Dad97] M.S. Dadachov, R.M. Lambrech t, J. Mater. Chem. 7 (1997) 1867. [Dav01] A.V. Davydov, W.J. Boettinger, U.R. Kattner, and T.J. Anderson, Phys. Status Solidi a 188, 407 (2001). [Dav02a] V.Yu Davydov, A.A. Kl ochikhin, V.V. Emtsev, S.V. Ivanov, V.V. Vekshin, F. Bechstedt, J. Furthmuller, H. Harima, A. V. Mudryi, A. Hashimoto, A. Yamamoto, A.J. Aderhold, J. Graul, E.E. Haller, Phys. Status Solidib 230 (2002) R4. [Dav02b] V.Yu. Davydov, A.A. Klochikhin, V.V. Emtsev, D.A. Kurdyukov, S.V. Ivanov, V.A. Vekshin, F. Bechstedt, J. Furthmuller, J. Aderhold, J. Graul, A.V. Mudryi, H. Harima, A. Hashimoto, A. Yamamoto and E.E. Haller, Phys. Status Solidib 234, No. 3, 787 (2002). [Dav02c] V.Yu. Davydov, A.A. Kl ochikhin, R.P. Seisyan, V.V. Emtsev, S.V. Ivanov, F. Bechstedt, J. Furthmller, H. Harima, A. V. Mudryi, J. Aderhold, O. Semchinova, J. Graul, Phys. Status Solidib 229 (2002) R1. [Dav03] E. A. Davis, S. F. J. Cox, R. L. Li chti, and C. G. Van de Walle, Appl. Phys. Lett. 82 592 (2003). [Dav97] V. Y. Davydov, N. S. Averkiev, I. N. G oncharuk, D. K. Nelson, I. P. Nikitina, A. S. Polkovnikov, A. N. Smirnov, M. A. J acobsen, and O. K. Semchinova, J. Appl. Phys. 82, 5097 (1997). [Dav98a] A.V. Davydov, T. J. Anderson, El ectrochem. Soc. Proc 98-118 (1998) 38. [Dav98b] V.Yu. Davydov, Yu. E. Kitaev, I.N. Goncharuk, A.N. Smirnov, J. Graul, O. Semchinova, D. Uffmann, M.B. Smirnov, A.P. Mirgorodsky, R.A. Evarestov, Phys. Rev. B 58, 12899 (1998). [Dav99] V.Yu. Davidov, V.V. Emtsev, I.N. Goncharuk, A.N. Sminov, V.D. Petrikov, V.V. Mamutin, V.A. Vekshin, S.V. Ivanov, M.B. Smirnov, and T. Inushima, Appl. Phys. Lett. 75 (1999) 3297. [Deb05] P. Deb, H. Kim, V. Rawat, M. Oliv er, S. Kim, M. Marshall, E. Stach, and T. Sands, Nano Lett., Vol. 5, No. 9, (2005) 1847.

PAGE 249

226 [Dem96] F. Demangeot, J. Fra ndon, M. A. Renucci, O. Briot, B. Gil, R.-L. Aulombard, MRS Internet J. Nitride Semicond. Res. 1, 23(1996). [Det01] T. Detchprohm, M. Yano, S. Sano, R. Nakamura, S. Mochiduki, T. Nakamura, H. Amano and I. Akasaki, Jpn. J. Appl. Phys. Vol. 40 (2001) L16. [Det92] T. Detchprohm, K. Hiramatsu, K. Itoh, I. Akasaki, Jpn. J. Appl. Phys. 31, L1454L1456 (1992). [Dim04] E. Dimakis, G. Konstantinidis, K. Ts agaraki, A. Adikimenakis, E. Iliopoulos, A. Georgakilas, Superlatt. Mi crostructures 36 (2004) 497. [Dim05] E. Dimakis, K. Tsagaraki, E. I liopoulos, Ph. Komninou, Th. Kehagias, A. Delimitis and A. Georgakilas, J. Cryst. Growth, 278(2005) 367. [Din04] Y. Ding, P. X. Ga o, Z. L. Wang, J. Am. Chem. Soc. 126 (2004) 2066. [Dmi03] V. Dmitriev et al., J. Phys. Chem. Solids 64 (2003) 307. [Dra04] M. Drago, T. Schmidtling, C. Werner, M. Pristovsek, U.W. P ohl and W. Richter, J. Cryst. Growth 272 (2004) 87. [Dra06] M. Drago, P. Vogt, W. Richte r, Phys. Status Solidi a 203 (2006) 116. [Edg97] J.H. Edgar, C.H. Wei, D.T. Smith, T.J. Kistenmacher, W.A. Bryden, J. Mater. Sci. 8 (1997) 307. [Edw98] N.V. Edwards, M.D. Bremser, R.F. Davis, A.D. Batchelor, S.D. Yoo, C.F. Karan, and D.E. Aspnes, Appl. Phys. Lett. 73, 2808 (1998). [Eip05] E. Eiper, A. Hofmann, J.W. Gerlach, B. Rauschenbach and J. Keckes, J. Cryst. Growth, 284 (2005) 561. [Etz01] E. V. Etzkorn and D. R. Clarke, J. Appl. Phys. 89 (2001) 1025. [Eva97] J.S.O. Evans, T.A. Mary and A.W. Sleight, Physica B, 1997, 241, 311-316. [Flo97] J.A. Floro, E. Chason, S.R. Lee, R. D. Twesten, R.Q. Hwang and L.B. Freund, J. Electron. Mater. 26 (1997) 969. [Gao03] L. Gao, Q. Zhang, J. Li, J. Mater. Chem. 13 (2003) 154. [Gar01] N. Garg et al. 2001 Solid State Phys. 39. [Gil00] D.R. Gilbert, A. Novikov, N. Patrin J.S. Budai, F. Kelly, R. Chodelka, R. Abbaschian, S.J. Pearton, R. Singh, Appl. Phys. Lett.(2000), 77, 4172. [Gor77] S.P. Gordienko, B.V. Fenoc hka, Zh. Fiz. Khim. 51 (1977) 530.

PAGE 250

227 [Got95] W. Gotz, N.M. Johnson, J. Walker, D.P. Bour, H. Amano, I. Akasaki, Appl. Phys. Lett. 67 (1995) 2606. [Gra05] J. Grandal, M.A. Sanchez-Garc ia, J. Cryst. Growth 278 (2005) 373. [Gri59] H. Grimmeiss and Z. HKo elmans, Nature (London) 14a, 264 (1959). [Guo93] Q. Guo, O. Kato, J. Appl. Phys. 73 (1993) 7969. [Guo98] Q. X. Guo, M. Nishio, and H. Ogawa, A. Wakahara and A. Yoshida, Phys. Rev. B 58 (1998) 15304. [Guo99] Q. Guo, M. Nishio, H. Ogawa, A. Yo shida, Jpn. J. Appl. Phys. 38 (1999) L490. [Had03] D.B. Haddad, J.S. Takur, V.M. Naik, G.W. Auner, R. Naik, L.E. Wenger, Mater Res. Soc. Symp. Proc. 743 (2003) 701. [Had04] D.B. Haddad, H. Dai, R. Naik, C. Mo rgan, V.M. Naik, J.S. Thakur, G.W. Auner, L.E. Wenger, H. Lu, W.J. Schaff, Mate r. Res. Soc. Symp. Proc. 798 (2004) Y12.7.1. [Hag03] P.R. Hageman, V. Kirilyuk, W.H. M. Corbeek, J.L. Weyher, B. Lucznik, M. Bockowski, S. Porowski and S. Mll er, J. Cryst. Growth 255 (2003) 241. [Hah40] H. Hahn, R. Juza, Z. Anorg. Allgem. Chem. 244 (1940) 111. [Han97] W. Han, S. Fan, Q. Li Y. Hu, Science 277 (1997) 1287. [Har95] H.L. Hartnagel, A.L. Dawar, A. K. Jain and C. Jagadish, Semiconducting Transparent Thin Films (IO P Publsihing, Bristol,1995). [He06] J.H. He, R. Yang, Y.L. Chueh, L.J. Chou, L.J. Chen, Z.L. Wang, Adv. Mater 18 (2006) 650. [Hea99] S. Hearne, E. Chason, J. Han, J.A. Floro, J. Figiel, J. Hunter, H. Amano and I.S.T. Song, Appl. Phys. Lett. 74 3 (1999) 356. [Hel92] K. H. Hellwege, A. M. Hellwege, and D. F. Nelson, Landolt-Brnstien Numerical Data and Functiona l Relationships in Science and Technology (Springer, Berlin, 1979, 1992), Vols. III:11, III:29. [Hig02] M. Higashiwaki, T. Matsui, Jpn. J. Appl. Phys. 41 (2002) L540. [Hos90] M. Hoshino, J. Appl. Phys., 68 (1990) 2538. [Hov72] H.J. Hovel and J.J. Cuomo, Appl. Phys. Lett. 20 (1972) 71. [Hua01] M.H. Huang, S. Mao, H. Feick, H. Q. Yan, Y.Y. Wu, H. Kind, E. Weber, R. Russo, P.D. Yang, Science 292 (2001) 1897.

PAGE 251

228 [Hua03] M. Huang, diss. Un iversity of Florida (2003). [Hua05a] Y. Huang, H. Wang, Q. Sun, J. Chen, D.Y. Li, Y.T. Wang and H. Yang, J. Cryst. Growth 276 (2005) 13. [Hua05b] Y. Huang, H. Wang, Q. Sun, J. Chen, J.F. Wang, Y.T. Wang and H. Yang, J. Cryst. Growth 281 (2005) 310. [Hul69] F. Hulbert, J. Br. Ceram. Soc. 6 (1969) 11. [Hur04] T.-B. Hur I.J. Lee H.L. Park Y.-H. Hwang and H.-K. Kim, Solid State Comm. 130 (2004) 397. [Hwa04] J. Hwang, diss., Univ ersity of Florida 2004. [Iga92] O. Igaracshi, Jpn. J. Appl. Phys. 31 (1992) 2665. [IIIV03] III-Vs Review, Elsevier Ltd., Vol 16, No 6, August 2003. [Im98] J.S. Im, H. Kollmer, J. Off, A. Sohm er, F. Scholz and A. Hangleiter, Phys. Rev. B 57, (1998) R9435. [Inu01] T. Inushima, V. V. Mamutin, V. A. Vekshin, S. V. Ivanov, T. Sakon, M. Motokawa and S. Ohoya, J. Cryst. Growth, 227-228 (2001) 481. [Jac63] M.G. Jacko, S.J.W. Pri ce, Can. J. Chem. 41 (1963) 1560. [Jai04] A. Jain, S. Raghavan, J.M. Re dwing, J. Cryst. Growth 269 (2004) 128. [Jam05] M. Jamil, J. R. Grandusky, V. Jinda l, F. Shahedipour-Sandvik, S. Guha and M. Arif, Appl. Phys. Lett. 87 (2005) 082103. [JCP93] Inorganic Powder Da ta, 33, Joint Committee for Powder Diffraction Standards (JCPDS), International Centre for Diff raction Data (ICDD), USA, 1162, (1993). [JCP96] Joint Committee for Powder Diffrac tion Standards (JCPDS), International Centre for Diffraction Data (ICDD). (1996). [Jes67] W.A. Jesser, D. Kuhlmann-Wilsdor f, Phys. Status Solidi 19, (1967) 95. [Ji05] X.H. Ji, S.P. Lau, H.Y. Yang, S.F. Yu, Nanotec hnology 16 (12) (2005) 3069-3073. [Joh02] J.C. Johnson, H.Q. Yan, R.D. Schaller, P.B. Petersen, P.D. Yang, R.J. Saykally, Nano Lett. 2 279 (2002). [Joh04a] M.C. Johnson, C.J. Lee, E.D. Bourre t-Courchesne, S.L. Konsek, S. Aloni, W.Q. Han, and A. Zettl, Appl. Phys. Lett. 85 (2004) 5670.

PAGE 252

229 [Joh04j] M.C. Johnson, S.L. Konsek, A. Zettl, E.D. Bourret-Courchesne, J. Cryst. Growth 272 (2004) 400. [Jon87] R.D. Jones, K. Rose, J. Phys. Chem. Solids 48 (1987) 587. [Jor99] J.D. Jorgensen, Z. Hu, S. Teslic, D.N. Argyriou, S. Short, J.S.O. Evans and A.W. Sleight, Phys. Rev. B 59 (1999) 215. [Jot01] R. Jothilingam, M.W. Koch, J.B. Po sthill, G.W. Wicks, Journal of Electronic Materials, Vol 30, No 7 (2001) 821. [Juz38] R. Juza and H. Hahn, Z. Anorg. Allg. Chem. 239, 282 (1938). [Juz56] R. Juza and A. Rabenau, Z. Anorg. Allg. Chem. 285, 212 (1956). [Kan04] S.W. Kang, diss., University of Florida 2004. [Kat94] Y. Kato, S. Kitamura, K. Hiramatsu, N. Sawaki, J. Cryst. Growth 144 (1994) 133. [Kee04] K. Keem, H. Kim, G.T. Kim, J.S. Lee, B. Min, K. Cho, M.Y. Sung and S. Kim, Appl. Phys. Lett. 84 4376 (2004). [Kel00] S. Keller, I. Ben-yaacov, S.P. De nvers, and U.K. Mishra, Proceedings of the International Workshop on Nitride Semi conductors (IWN 2000), Nagoya, Japan, September 24, 2000, IPAP conference series 1, p.233. [Kha93] M.A. Khan, J.N. Kuznia, A.R. Bhatta rai, and D.T. Olsen, Appl. Phys. Lett. 62, 1786 (1993). [Kih00] M. Kihara, T. Sasaki, T. Tsuchiya and H. Sakaguchi, Proc. Int. Workshop on Nitride Semiconductors, IPAP C onf. Series 1, pp. 117-120 (2000). [Kim03] H.-M. Kim, W.C. Lee, T.W. Ka ng, K.S. Chung, C.S. Yoon, C.K. Kim, Chemical Physics Letters 380 (2003) 181. [Kim03c] H.-M. Kim, W.C. Lee, T.W. Kang, K.S. Chung, C.S. Yoon, C.K. Kim, Chemical Physics Letters 380 (2003) 181. [Kim05] J-H. Kim, J.A. Freita s, Jr., P.B. Klein, S. Jang, F. Ren, and S.J. Pearton, Eletroch. Sol-St. Lett. 8 (2005) G345. [Kim06] T.W. Kim, diss., Un iversity of Florida 2006. [Kim94] S.H. Kim, H.S. Kim, J.S. Hwang, J. G. Choi, P.J. Chong, Chem. Mater. 6 (1994) 278. [Kim99] C. Kim, I. K. Robinson, J. M young, K-H Shim, K. Kim, J. Appl. Phys. 85 (1999) 4040.

PAGE 253

230 [Kis96] C. Kiselowski, J. Kruger, S. Ruvimov, T. Suski, J. W. Ager III, E. Jones, Z. Liliental-Weber, M. Rubin, E. R. Weber, M. D. Bremser, and R. F. Davis, Phys. Rev. B 54, (1996) 745. [Kle00] C.A. Klein, J. Appl. Phys. 88 (2000) 5487. [Kol98] J.W. Kolis, S.Wilcenski, R.A. Laudi se, Mater. Res. Soc. Symp. Proc. 495 (1998) 367. [Kos04] R. Kosiba, G. Ecke, V. Cimalla, L. Spie, S. Krischok, J.A. Schaefer, O. Ambacher, W.J. Schaff, Nuclear Instrume nts and Methods in Physics Research B 215 (2004) 486. [Kou05] A. Koukitu, J. Kikuchi Y. Kangawa and Y. Kumagai, J. Cryst. Growth 281 (2005) 47. [Kou98] A. Koukitu, S. Hama, T. Taki, H. Se ki, Jpn. J. Appl. Phys. PART 137 (3A): 762-765 (1998). [Kri04] S. Krischok, V. Yanev, O. Balykov, M. Himmerlich, J.A. Schaefer, R. Kosiba, G. Ecke, I. Cimalla, V. Cimalla, O. Ambacher H. Lu, W.J. Schaff, L.F. Eastman, Surface Science 566 (2004) 849. [Kro03] A. Krost, A. Dadgar, G. Strassbur ger, and R. Clos, Phys. Status Solidi a 200 (2003) 26. [Kro05] A. Krost, A. Dadgar, F. Schulze, J. Blsing, G. Strass burger, R. Clos, A. Diez, P. Veit, T. Hempel and J. Christen J. Cryst. Growth, 275(2005) 209. [Kry05] O. Kryliouk, H.J. Park, H.T. Wang, B.S. Kang, T.J. Anderson, F. Ren and S.J. Pearton, J. Vac. Sci. Technol. B, 23(5), 1891(2005). [Kry06] O. Kryliouk, H. J. Park T. Anderson, to be published. [Kry99a] O. Kryliouk, M. Reed, T. Dann, T. Anderson and B. Chai, Mat. Sci. Eng. B59, 116-119 (1999). [Kry99b] O. Kryliouk, M. Reed, T. Dann, T. J. Anderson and B. Chai, Mat. Sci. Eng. B, 66, 26 (1999). [Kub89] K. Kubota, Y. Kobaya shi, and K. Fujimoto, J. Appl. Phys. 66, 2984 (1989). [Kum01] Y. Kumagai, K. Takemoto, A. Kouk itu, H. Seki, J. Cryst. Growth 222 (2001) 118. [Kum02] M. Senthil Kumar, G. Sonia, V. Ra makrishnan, R. Dhanasekaran and J. Kumar, Physica B: Condensed Matter, 324 (2002) 223.

PAGE 254

231 [Kum04] Y. Kumagai, J. Kikuchi, Y. Mats uo, Y. Kangawa, K. Tanaka, A. Koukitu, J.Cryst. Growth 272 (2004) 341. [Kur01] E. Kurimoto, H. Harima, A. Hashim oto, and A. Yamamoto, phys.stat.sol.(b) 228, No.1, 1 (2001). [Kur04a] K. Kuriyama, T. Tokumasu, H. Sano and M. Okada, Solid State Communications, 131 (2004) 31. [Kur04b] E. Kurimoto, M. Hangyo, H. Harima, M. Yoshimoto, T. Yamaguchi, T. Araki, Y. Nanishi, K. Kisoda, Appl. Phys. Lett. 84 (2004) 212. [Kuy04] T. Kuykendall, P.J. Pauzauskie, Y. Zhang, J. Goldberger, D. Sirbuly, J. Denlinger, and P. Yang, Nature materials 3 (2004) 524. [Lan04] Z.H. Lan, W.M. Wang, C.L. Sun, S.C. Shi, C.W. Hsu, T.T. Chen, K.H. Chen, C.C. Chen, Y.F. Chen, L.C. Chen J. Cryst. Growth 269 (2004) 87. [Lar88] C.A. Larsen, N.I. Buchan, and G.B. Stringfellow, Appl. Phys. Lett., 52, 480 (1988). [Lar90] C.A. Larsen, N.I. Buchan, S.H. Li, and G.B. Stringfellow, J. Cryst. Growth 102 (1990) 103. [Leb00] V. Lebedev, J. Jinsch ek, U. Kaiser, B. Schroter, W. Richter, Appl. Phys. Lett. 76 (2000) 2029. [Lee04] I.J. Lee, J.-Y. Kim, H.-J. Shin, H.-K. Kim, J. Appl. Phys. 95 (2004) 5540. [Lei03] J. Leitner, A. Strejc, D. Sedmi dubsk, K. Ruicka, Thermochimica Acta 401 (2003) 169. [Lei04] J. Leitner, P. Mars k, D. Sedmidubsky, K. Ruzicka, Journal of Physics and Chemistry of Solids 65 (2004) 1127. [Lei65] A. Leib, M. T. Emerson, J. P. Oliver, Inorg. Chem. 4 (1965) 1825. [Lei96] J. Leitner, J. Stejskal, P. Vonka, Mater. Lett. 28 (1996) 197. [Les96] M. Leszczynski, H. Teisseyre, T. Susk i, I. Grzegory, M. Bockowski, J. Jun, S. Porowski, K. Pakula, J. M. Baranowski C. T. Foxon, T. S. Cheng Appl. Phys. Lett. 69, 73-75 (1996). [Li04] Q.H. Li, Q. Wan, Y. X. Liang and T.H. Wang, Appl Phys. Lett. 84 (2004) 4556. [Li06] J.-B. Li, J. Liang, G. Rao, Y. Zhang, G.Y. Liu, J. Chen, Q. Liu and W. Zhang, Journal of Alloys and Compounds, In Pre ss, Corrected Proof, Available online 18 January 2006. ( http://www.elsevier.com/locate/jallcom ).

PAGE 255

232 [Li94] X. Li, B. Zhou, K.S.A. Butcher, E. Florido, N. Syakir, and T.L. Tansley, Proceedings of the Australian Comp ound Optoelectronic Materials Devices Conference, Sydney, Australia, December 12, 1994, p. 43. [Lia00] H.M. Liaw, R. Venugopal, J. Wan, R. Doyle, P.L. Fejes and M.R. Melloch, Solid-State Electronics 44 (2000) 685. [Lia01] H.M. Liaw, R. Venugopal, J. Wan a nd M.R. Melloch, Solid-State Electronics 45 (2001) 1173. [Lia02] C.H. Liang, L.C. Chen, J.S. Hwang, K.H. Chen, Y.T. Hung and Y.F. Chen, Appl. Phys. Lett. 81, 22 (2002). [Liu02] H. Liu et al. Solid State Commun. 121 (2002) 177. [Liu78] S. Liu and D.A. Stevenson, J. Electrochem. Soc. 125 (1978) 1161. [Lo01] Y.-H. Lo, F. Ejeckman, Z. Zhu, US Pat. 20010042503A1. [Los04] M. Losurdo and G. Bruno, confer ence proceedings of the ICNS-6, Bremen (2005); see otherwise: T. Ive, Brandt, M. Ramsteiner, M. Giehler, H. Kostial, and K. Ploog, Appl. Phys. Lett. 84, 1671 (2004). [Lu01] H. Lu, W.J. Schaff, J. Hwang, H. Wu G. Koley, and L.F. Eastman, Appl. Phys. Lett., Vol. 79, No. 10, 3 September (2001) 1489. [Lu02a] H. Lu, W. J. Schaff, L. F. Eastman, J. Wu, W. Walukiewicz, K. M. Yu, J. W. Auger III, E. E. Haller, and O. Ambacher, Abstract of the 44th Electronic Material Conference, Santa Barbara, CA, June 26, 2002; J. Electron. Mater. (to be published); W. J. Schaff (private communica tion, Cornell University, Ithaca, N.Y.). [Lu02b] H. Lu, W.J. Schaff, L.F. Eastman, a nd C. Wood, Mater. Res. Soc. Symp. Proc. 693, 9 (2002). [Lu04a] H. Lu, W.J. Schaff, L.F. Ea stman, J. Appl. Phys. 96 (2004) 3577. [Lu04b] Y.Lu, J. Li, H.T. Ng, C. Binder, C. Partridge and M. Meyyapan, Chem. Phys. Lett. 391 (2004) 344. [Luo05] S. Luo, W. Zhou, Z. Zh ang, L. Liu, X. Dou, J. Wang, X. Zhao, D. Liu, Y., L. Song, Y. Xiang, J. Zhou, S. Xie, small 1 (2005) 1004. [Mac70] J.B. MacChesney, P.M. Bridenbaugh, P.B. OConnor, Mater. Res. Bul. 5 (1970) 783. [Mal04a] B. Maleyre, O. Briot, and S. Ruffenach, J. Cryst. Growth 269, 15 (2004). [Mal04b] B. Maleyre, S. Ruffenach, O. Briot, B. Gil and A. Van der Lee, Superlatt. Microstructures 36 (2004) 517.

PAGE 256

233 [Mal05] B. Maleyre, S. Ruffenach, O. Briot, B. Gil, and A. van der Lee, Phys. Status Solidi c 2, 2309 (2005). [Mar01] H. Marchand, L. Zhao, N. Zhang, B. Moran, R. Coffie, U. K. Mishra, J. S. Speck, S. P. DenBaars, and J. A. Freitas, J. Appl. Phys. 89 (2001) 7846. [Mar02] D. Martin, J.F. Carlin, V. Wagner, H.J. Buhlmann, M. Ilegems, Phys. Status Solidib 194 (2002) 520. [Mar03] I.V. Markov, Crystal Growth for Beginners, World Scientific, 2nd ed. p 422 (2003). [Mar69] H.P. Maruska and J.J. Tietje n, Appl. Phys. Lett. 15, 327 (1969). [Mar77] L.A. Marasina, I.G. Pichugin, and M.Tlaczala, Krist. Tech. 12, 541 (1977). [Mar96] T.A. Mary, J.S.O. Evans, T. Vogt and A.W. Sleight, Science, 1996, 272, 90-92. [Mas01] M. Mastro, diss. Un iversity of Florida 2001. [Mas06a] M.A. Mastro, C.R. Eddy, Jr., D.K. Gaskill, N.D. Bassim, J. Casey, A. Rosenberg, R.T. Holm, R.L. Henry and M.E. Twigg, J. Cryst. Growth 287 (2006) 610. [Mas06b] M.A. Mastro, O.M. Kryliouk and T.J. Anderson, Mater. Sci. Eng. B 127 (2006) 91. [Mat02] T. Matsuoka, H. Okamoto, M. Nakao, H. Harima, E. Kurimoto, Appl. Phys. Lett. 81 (2002) 1246. [Mat75] J.W. Matthews, Epitaxial Growth Part B (1975), Academic Press, Inc. [Mat87] E.N. Matthias, B.M. Allen, IEEE Trans. Electron Devices, ED-34 (1987) 257. [Mat89] T. Matsuoka, H. Tanaka, T. Sasaki, a nd A. Katsui, Proceedings of the Sixteenth International Symposium on GaAs and Related Compounds,Karuizawa, Japan, September 25, 1989 (Institute of Physics, Bristol, 1990), p. 141. [Mat97] T. Matsuoka, in GaN and Related Materi als, edited by S. J. Pearton, Gordon and Breach, New York, 1997, pp. 53. [Maz89] D. Mazzarese, A. Tripathi, W. C. C onner, K. A. Jones, L. Calderon, D. W. Eckart, J. of Electronic Materials 18, 3 (1989)369. [Mel97] Yu.V. Melnik, K.V. Vassilevski, I.P. Nikitina, A.I. Baba nin, V. Yu. Davydov, V.A. Dmitriev, MRS Internet J. Nitride Semicond. Res. 2, 39(1997). [Mer50] J.H. van der Merwe, Proceedings : Physical Society London, A63, 616 (1950).

PAGE 257

234 [Mer63] J.H. van der Merwe, J. Appl. Phys. 34, No1 (1963) 117. [Mes05] V.N. Bessolov, V.Y. Davydov, Y.V. Zhilyaev, E.V. Konenkova, G.N. Mosina, S.D. Raevskii, S.N. Rodin, S. Sharofidinov, M.P. Shcheglov, H.S. Park, M. Koike, Technical Physic s Letters 31 (2005) 915. [Mey83] G. Meyer, R. Blachnik, Anorg. Allg. Chem., 503 (1983) 126. [Mey84] D.J. Meyer, diss., University of Florida 1984. [Min04] T. Minegishi, T. Su zuki, C. Harada, H. Goto, M.-W. Cho and T. Yao, Current Applied Physics, 4, (2004) 685. [Mit03] K.D. Mitzner, J. Sternhagen and D. W. Galipeau, Sens Actuators B93 92 (2003). [Mit98] P. Mitra, A.P. Chatterjee a nd H.S. Maiti, Mater.Lett. 35 (1998) 33. [Mor73] Y. Morimoto, K. Uchino and S. Us hio: J. Electrochem. Soc. 120 (1973) 1783. [Mot02a] K. Motoki, T. Okahisa, S. Nakahata N. Matsumoto, H. Kimura, H. Kasai, K. Takemoto, K. Uematsu, M. Ueno, Y. Kumagai, A. Koukitu, H. Seki, J. Cryst. Growth 237 (2002) 912. [Mot02b] K. Motoki, T. Okahisa, S. Nakahata N. Matsumoto, H. Kimura, H. Kasai, K. Takemoto, K. Uematsu, M. Ueno, Y. Kumaga i, A. Koukitu and H. Seki, Mater. Sci. Eng. B 93, (2002) 123. [Mot02c] Motlan, E.M. Goldys, T.L. Tans ley, J. Cryst. Growth 241 (2002) 165. [Mot18] W.R. Mott, Trans. Am. Electrochem. Soc., 34 (1918) 287. [Mou93] T. J. Mountziaris, S. Kalyanasundara m, N. K. Inge, J. Crystal Growth 131 (1993) 283. [Mur02] A.K. Murali, A.D. Barve, S.H. Risbud, Mater. Sci. Eng. B96 (2002) 111. [Nak92] S. Nakamura, T. Muskai, M. Senoh, N. Iwasa, Jpn. J. Appl. Phys. 31 (1992) L139. [Nak94] S. Nakamura, T. Mukai, M. Senoh, Appl. Phys. Lett. 64 (1994) 1687. [Nak95a] S. Nakamura, T. Muskai, M. Senoh, Appl. Phys. Lett. 67 (1995) 1868. [Nak95b] S. Nakamura, M. Senoh, N. Iwasa, S. Nagahama, T. Yamada, T. Muskai, Jpn. J. Appl. Phys. 10B (1995) L1332. [Nak96a] S. Nakamura, M. Senoh, S.I. Nagaha ma, N. Iwasa, T. Yamada, T. Matsushita, H. Kiyoku, Y.S. Ugimoto, Jpn. J. Appl. Phys. 35 (1996) L217.

PAGE 258

235 [Nak96b] S. Nakamura, M. Senoh, S.I. Nagaha ma, N. Iwasa, T. Yamada, T. Matsushita, H. Kiyoku, Y.S. Ugimoto, Appl. Phys. Lett. 68 (1996) 2105. [Nak97] S. Nakamura, M. Senoh, S. Hagahama, H. Iwasa, T. Yamada, T. Matsushita, Y. Sugimoto, and H. Kiyoku, Appl. Phys. Lett. 70 (1997)1417. [Nak98] S. Nakamura, M. Senoh, S.I. Nagahama N. Iwasa, T. Yamada, T. Matsushita, H. Kiyoku, Y.S. Ugimoto, T. Kozaki, H. Umemoto, M. Sano, K. Chocho, Appl. Phys. Lett. 72 (1998) 211. [Nao03] H. Naoi, M. Narukawa H. Miyake and K. Hiramats u, J. Cryst. Growth 248, (2003) 573. [Nat80] B.R. Natarajan, A.H. Eltoukhy, J.E. Greene, T.L. Barr, Thin Solid Films 69 (1980) 201. [Nik99] A. Nikolaev, I. Nikitina, A. Zubrilov, M. Mynbaeva, Y. Melnik, and V. Dmitriev MRS 1999 Fall Proceedings W6.5. [Nis83] J. Nishizawa, T. Kurabayashi, J. Electrochem. Soc. 130 (1983) 413. [Nou04] G. Nouet, P. Ruterana, J. Chen, B. Lei, H. Ye, G. Yu, M. Qi and A. Li, Superlatt. Microstructures, 36 (2004) 417. [Now99] R. Nowak, M. Pessa, M. Suganuma, M. Leszczynski, I. Grzegory, S. Porowski, F. Yoshida, Appl. Phys. Lett. 75 (1999) 2070. [Oka98] Y. Okamoto, J. Cr ystal Growth 191 (1998) 405. [Ole98] S.K. OLeary, B.E. Foutz, M.S. Shur, U.V. Bhapkar, L.F. Eastman, J. Appl. Phys. 83 (1998) 826. [Oli03] R.A. Oliver,C. Norenberg, M.G. Martin M.R. Castell, L. Allers, G.A.D. Briggs, Surface Science 532 (2003) 806. [Ond02] B. Onderka, J. Unland, and R. Sc hmid-Fetzer, J. Mater. Res.17 (2002) 3065. [Osa72] K. Osamura, K. Nakajima, Y. Mura kami, H. P. Shingu, and A. Ohtsuki, Solid State Commun. 11, 617 (1972). [Osa75] K. Osamura, S. Naka, and Y. Murakami, J. Appl. Phys. 46, 3432 (1975). [Oto03] Y. Otoki, M. Kihara, T. Tanaka, K. Takano, T. Kikkawa and T. Igarashi The International Conference on Compound Semiconductor Manufacturing Technology 13.3 (2003). [Pan94] J.I. Pankove, S.S. Chang, H.C. Lee, R. Molnar, T.D. Moustakas, and B. Van Zeghbroeck, Tech. Dig. Int. Elec tron Devices Meet. 94, 389 (1994).

PAGE 259

236 [Pan99] Y.C. Pan, W.H. Lee, C.K. Shu, H.C. Lin, C.I. Chiang, H. Chang, S.D. Lin, M.C. Lee, W.K. Chen, J. Appl. Phys. 38 (1999) 645. [Par01] H. Parala, A. Devi, F. Hipler, E. Maile A. Birkner, H.W. Beck er, R.A. Fischer, J. Cryst. Growth 231 (2001) 68. [Par03] C. Park, S. Han, C. Doh, S. Yeo, D. Yoon, S.-K. Hwang, K.-H. Lee, and T. Anderson, Chemical Vapor Deposition XVI (CVD-VXI) and EUROCVD 14, M. Allendorf, F. Maury, and F. Teyssa ndier, Eds., Vol. 2003-08, 92-97 (2003). [Par05] C. Park, J. Kim, D. Yoon, S. Han, C. Doh, S. Yeo, K. Lee, T. Anderson, Journal of The Electrochemical Society, 152 (2005) C298. [Pas01] R. Paszkiewicz, B. Paszkiewicz, R. Korbutowicz, J. Kozlowski, M. Tlaczala, L. Bryja, R. Kudrawiec, J. Misiewizc, Cryst. Res. Tec hnol., 36 (2001), 971. [Pas04] T. Paskova, P.P. Paskov, E. Valcheva, V. Darakchieva, J. Birch, A. Kasic, B. Arnaudov, S. Tungasmita, B. Monemar, Phys. Status Solidi(a) 201 (2004) 2265. [Pas63] J. Pastrnak and L. Souckova Phys. Status Solidi3, K71 (1963). [Per92] P. Perlin, C. J. Carrilon, J. P. Itie, A. S. Miguel, I. Grzegory, and A. Polian, Phys. Rev. B, 45, 83 (1992). [Per97] W.G. Perry, T. Zheleva, M.D. Bremser, R.F. Davis, W. Shan, and J.J. Song, J. Electron. Mater. 26, 224 (1997). [Per98] C.A. Perottani and J.A.H. da Jornada Science 280 (1998) 886. [Plo99] K.H. Ploog, O. Brandt, H. Yang, A. Trampert, J. Vac. Sci. Technol. B 16 (1999) 2229. [Pol96] A. Polian, M. Grimsditch, and I. Grzegory, J. Appl. Phys., 79, 3343 (1996). [Por95] S. Porowski, High pressure crysta llization of III-V nitr ides, Acta Physica Polonica. 87 (2) (1995) 295. [Por99] S. Porowski, MRS Internet J. N itride Semicond. Res. 4S1, G1.3 (1999). [Pra69] G. L. Pratt, Gas Kinetics, Wiley, London (1969). [Pri95] D.M. Price, J. Th erm. Anal., 45 (1995) 1285. [Prz98] I.N. Przhevalskii, S. Yu. Karpov, Yu.N. Makarov, MRS Internet J. Nitride Semicond. Res. 3, 30 (1998). [Puy76] N. Puychevrier and M. Menore t, Thin Solid Films 36, 141 (1976).

PAGE 260

237 [Qia05] Z. Qian, S. Hou, J. Zhang, R. Li, Z. Shen, X. Zhao, Z. Xue, Physica E 30 (2005) 81. [Raa93] I.J. Raaijmakers, J. Yang, Appl. Surf. Sci. 73 (1993) 31-41. [Rag02] B. Raghothamachar, W.M. Vetter, M. Dudley, R. Dalmau, R. Schelesser, Z. Sitar, E. Michaels, J.W. Kolis, J. Cryst. Growth 246 (2002) 271. [Rag04] S. Raghavan and J.M. Redwing, J. Cryst. Growth 261 (2004) 294. [Ran00] M.R. Ranade, F. Tessier A. Navrotsky, V.J. Leppert, S.H. Risbud, F.J. DiSalvo, C.M. Balkas, J. Phys. Chem. B 104 (2000) 4060. [Ran01] M.R. Ranade, F. Tessier, A. Navrot ska, R. Marchand, J. Mater. Res 16 (2001) 2824. [Ree02] M. Reed, diss., Univ ersity of Florida 2002. [Rei03] A. Reiher, J. Blsing, A. Dadgar, A. Diez and A. Krost, J. Cryst. Growth 248 (2003) 563. [Rei87] R.C. Reid, J.M. Prausnitz, B.E. Poling, The Properties of Gases and Liquids, McGraw Hill, New York (1987). [Ren58] T. Renner, Z. Anorg. Allg. Chem. 298, 28 (1958). [Rie96] W. Rieger, T. Metz ger, H. Angerer, R. Dimitrov, O. Ambacher, and M. Stutzmann, Appl. Phys. Lett. 68, 970 (1996). [Rob36] C. Robert, Helv. P hys. Acta, 9 (1936) 418. [Rod95] D.L. Rode, D.K. Gaskill, Appl. Phys. Lett. 66 (1995) 1972. [Saf97] S.A. Safvi, J.M. Redwing, M.A. Ti schler, T.F. Kuech, J. Electrochem. Soc. 144, 5 (1997) 1789. [Sam69] G.V. Samsonov, Nitridy Kiev, 1969. [San99] S.I. Sandler, Chemical and Engin eering Thermodynamics, 3rd ed, John Wiley & Sons, Inc. (1999). [Sar05] K. Sardar, F.L. Deepak, A. Govindara j, M.M. Seikh, and C.N.R. Rao, small 1 (2005) 91. [Sat94] Y. Sato and S. Sato, J. Cryst. Growth 144, 15 (1994). [Sat95] Y. Sato, S. Sato, J. Cryst. Growth 146 (1995) 262. [Sat97] M. Sato, Jpn. J. Appl Phys., Part 2 36, L658 (1997).

PAGE 261

238 [Sav78] V.A. Savastenko and A.U. Sheleg Phys. Status Solidi a, 48, K135 (1978). [Say05] I. Sayago, E. Terrado, E. Lafuente, M.C. Horillo, W.K. Maser, A.M. Benito, R. Navarro, E.P. Urriolabeita, M.T. Martinez and J. Gutierr ez, Synth. Metals 148, 15 (2005). [Sch00] W.J. Schaff, H.Lu, J. Hwang, and H. Wu, Proceedings of the Seventeenth Biennial IEEE/Cornell Conference on Adva nced Concepts in High Performance Devices, August 7, 2000, p. 225. [Sch04a] B. Schwenzer, L. Loeffler, R. Sesh adri, S. Keller, F.F. Lange, S.P. DenBaars and U.K. Mishra, J. Mater. Chem. 14 (2004), 637. [Sch04b] F. Schulze, A. Dadgar, J. Blaosing, A. Krost, J. Crysta l Growth 272 (2004) 496. [Sch06] F. Schulze, A. Dadgar, J. Blsing, T. Hempel, A. Diez, J. Christen and A. Krost, J. Cryst. Growth 289 (2006) 485. [Sec02] R.A. Secco, J. Phys. Chem. Solids 63 (2002) 425. [Sed06] D. Sedmidubsky and J. Leitne r, J. Cryst. Growth 286 (2006) 66. [Seg04] A.S. Segal, A.V. Kondratyev, S.Yu. Karpov, D. Martin, V. Wagner, M. Ilegems, J. Cryst. Growth 270 (2004) 384. [She02] J. Shen, S. Johnston, S. Shang and T. Anderson, J. Cryst. Growth 240 (2002) 6. [She06] J. Shen, Research Institute for N on-ferrous Metals, Beijing, China, personal communications (2006). [Shi05] C.F. Shih, N.C. Chen, P.H. Chang a nd K.S. Liu, J. Cryst. Growth 281 (2005) 328. [Sin04] P. Singh, P. Ruterana, M. Morales, F. Goubilleau, M. Wojdak, J.F. Carlin, M. Ilegems and D. Chateigner, Superla tt. Microstructures 36 (2004) 537. [Smi68] W.R. Smith, R.W. Missen, Cana d. J. of Chem. Eng. 46 (1968) 269. [Smo01] I.P. Smorchkova, L. Chen, T. Mates, L. Shen, S. Heikman, B. Moran, S. Keller, S.P. DenBaars, J.S. Speck and U.K. Mishra, J. Appl. Phys. 90 (2001) 5196. [Spe06] P. Specht, J.C. Ho, X. Xu, R. Armitage E.R. Weber, E. Erni, C. Kisielowski, J. Cryst. Growth 288 (2006) 225. [Sta02] E. Starikov, P. Shiktorov, V. Gruzinsk is, L. Reggiani, L. Varani, J.C. Vaissiere, Jian H. Zhao, Physica B 314 (2002) 171. [Sto06] T. Stoica, R. Meijers, R. Calarco, T. Richter, and H. Luth, J. Cryst. Growth 290 (2006) 241.

PAGE 262

239 [Sto09] G. Stoney, Proc. R. So c. London, Ser. A 82, 172 (1909). [Str99] G.B. Stringfellow, Organometallic Vapor Phase Epitaxy, Academic Press (San Diego, CA) 2nd Ed. (1999) 192. [SUB94] Thermo-Calc version K database. Stockholm Technology Park Bjrnnsvgen 21 SE-113 47 Stockholm, Sweden. [Sug03] K. Sugita, H. Takatsuka, A. Hashimot o, and A. Yamamoto, Phys. Status Solidib 240, 421 (2003). [Sul88] B.T. Sullivan, R.R. Parsons, K.L. West ra, M.J. Brett, J. Appl. Phys 64 (1988) 144. [Sun85] B. Sundman, B. Jansson, J. O. Andersson, Calphad 9 (1985) 153. [Sun96] H. Sunakawa, A. Yamaguchi, A. Kimura and A. Usui, Jpn. J. Appl. Phys., Part 2 35, L1395 (1996). [Suz96] T. Suzuki, T. Ohmura, Phil. Mag. A 74 (1996) 1073. [Tab96] A. Tabata, R. Enderlein, J. R. Leite, S.W. da Silva, J.C. Galzerani, D. Schikora, M. Kloidt and K. Lischka, J. Appl. Phys. 79, 4137 (1996). [Tak04] N. Takahashi, A. Niwa, T. Naka mura, J. Phys. Chem. Solids 65 (2004) 1259. [Tak96] Y. Takagi, M. Ahart, T. Azuhata, T. Sota, K. Suzuki, S. Nakamura, Physica B 219-220 (1996) 547. [Tak97a] N. Takahashi, J. Ogasawara, and A. Koukitu, J. Cryst. Growth 172, 298 (1997). [Tak97b] N. Takahashi, R. Matsumoto, A. Kouk itu, and H. Seki, Jpn. J. Appl. Phys., Part 2 36, L743 (1997). [Tan04] T. Tang, S. Han, W. Jin, X. Liu, C. Li, D. Zhang, C. Zhou, B. Chen, J. Han, and M. Meyyapan, J. Mat. Res. 19 423 (2004). [Tan05] T. Tanaka K. Takano, H. Fujikura, T. Mishima, Y. Kohji, H. Kamogawa, T. Meguro and Y. Otoki, The Internatio nal Conference on Compound Semiconductor Manufacturing Technology 4.1 (2005). [Tan86] T.L. Tansley, C.P. Foley, J. Appl. Phys. 59 (1986) 3241. [Tho96] A. Thon and T.F. Kuech, Appl. Phys. Lett. 69 (1996) 55. [Tou77] Y.S. Touloukian, R.K. Kriby, R.E. Taylor, and P.D. Desai, Thermophysical Properties of Matter, 12 (1977) 298. [Tra74] J.W. Trainor and K. Rose, J. Electron. Mater. 3, 821 (1974).

PAGE 263

240 [Tsu05] A. Tsuyuguchi, K. Teraki, T. Koizumi, J. Wada, T. Araki, Y. Nanishi, H. Naoi, Institute of Physics Conference Series (2005), 184 (Compound Semiconductors 2004), 239-242. [Tya77] V.A. Tyagai, A.M. Evstigneev, A.N. Krasiko, A.F. Andreeva, V.Ya. Malakhov, Sov. Phys. Semicond. 11 (1977) 1257. [Unl03] J. Unland, B. Onderka, A. Davydov, R. Schmid-Fetzer, J. Cryst. Growth 256 (2003) 33. [Usu97] A. Usui, H. Sunakawa A. Sakai, and A. Yamaguchi, Jpn. J. Appl. Phys. 36 (1997) L899. [Vad05] S. Vaddiraju, A. Mohite, A. Chin, M. Meyyappan, G. Sumanasekera, B. W. A, and M.K. Sunkara, Nano lett. 2005 Vol. 5, No. 8 1625. [Ven05] S. Venkataraj, H. K ittur, R. Drese and M. Wuttig, Thin Solid Films 514 (2006) 1. [Ven99] P. Vennegues and B. Beaum ont, Appl. Phys. Lett. 75, 4115 (1999). [Vor71] A.M. Vorobev, G.V. Evseeva, L.V. Zenkevich, Zh. Fiz. Khim. 45 (1971) 2650. [Vor73] A.M. Vorobev, G.V. Evseeva, L.V. Zenkevich, Zh. Fiz. Khim. 47 (1973) 2885. [Wak89] A. Wakahara and A. Yoshid a, Appl. Phys. Lett. 54, 709 (1989). [Wak90] A. Wakahara, T. Tsuc hiya, and A. Yoshida, J. Cryst. Growth 99, 385 (1990). [Wal04] W. Walukiewicz, S.X. Li, J. Wu, K.M. Yu, J.W. Ager III, E.E. Haller, H. Liu, W.J. Schaff, J. Cryst. Growth 269 (2004) 119. [Wal99] P. Waltereit, O. Brandt, A. Trampert M. Ramsteiner, M. Reiche, M. Qi, and K. H. Ploog, Appl. Phys. Lett. 74 (1999) 3660. [Wan03] J.Y. Wang, S. Hofmann, A. Zalar, E.J. Mittemeijer, Thin Solid Films 444 (2003) 120. [Wan04] Q. Wan, Q.H. Li, Y.J. Chen, T.H. Wang, X.L. He, J.P.Li and C.L. Lin, Appl. Phys. Lett. 84 3654(2004). [Wan05] C.-L. Wang, J.-R. Gong, W.-T. Liao, C.-K. Lin and T.-Y. Lin, Thin Solid Films 493 (2005) 135. [Wan06] X. Wang, S.-B. Che, Y. Ishitani, A. Yoshikawa, Journal of Applied Physics (2006) 99(7), 073512/1. [Wee03] T.W. Weeks, Jr., E.L. Piner, T. Ge hrke, K.J. Linthicum, United States Patent 6,617,060 September 9, 2003.

PAGE 264

241 [Wes88] K.L. Westra, R.P.W. Lawson, M.J. Br ett, J. Vac. Sci. Technol. A 6 (1988) 1730. [Wet94] C. Wetzel, D. Volm, B.K. Meyer, K. Pressel, S. Nilsson, E.N. Mokhov, P.G. Baranov, Appl. Phys. Lett. 65 (1994)1033. [Wit05] H. Witte, A. Krtschil, E. Schrenk, K. Fluegge, A. Dadgar, and A. Krost, J. Appl. Phys. 97, 043710 (2005). [Wit98] R. L. Withers, J. S. O. Evans, J. Hanson and A. W. Sleight J. Solid State Chem., 1998, 137, 161-167. [Wu02a] J. Wu, W. Walukiewicz, K.M. Yu, J. W. Ager III, E.E. Haller, H. Lu, W.J. Schaff, Y. Saito, Y. Nanishi, A ppl. Phys. Lett., 80 (2002) 3967. [Wu02b] J. Wu, W. Walukiewicz, W. Shan, K.M. Yu, J.W. Ager III, E.E. Haller, Hai Lu, W.J. Schaff, Phys. Rev. B 66 (2002) 201403. [Wu06] C.-L. Wu, C.-H. Shen, H.-Y. Chen, S. -J. Tsai, H.-W. Lin, H.-M. Lee, S. Gwo, T.-F Chuang, H.-S Chang, T. M. Hsu, Journal of Crystal Growth 288 (2006) 247. [Xu03] K. Xu and A. Yoshikawa, Appl. Phys. Lett. 83, (2003)14. [Yam00] S. Yamaguchi, M. Kari ya, S. Nitta, H. Amano and I. Akasaki, Applied Surface Science, 159-160 (2000) 414. [Yam01a] A. Yamamoto, Y. Murakami, K. Koide, M. Adachi, and A. Hashimoto, Phys. Status SolidiB 228, 5 (2001). [Yam01b] A. Yamamoto, M. Adachi, A. Hash imoto, J. Cryst. Growth 230 (2001) 351. [Yam02] A. Yamamoto T. Tanaka, K. Koide, and A. Hashimoto, phys. stat. sol. (a) 194, 510 (2002). [Yam04] A. Yamamoto K. Sugita, H.Takatsuka A. Hashimoto, V. Yu. Davydov, J. Cryst. Growth 261, 275 (2004). [Yam05] A. Yamamoto, T. Kobayashi, T. Ya mauchi, M. Sasase, A. Hashimoto, and Y. Ito, Phys. Status Solidi(c) 2, 2281(2005). [Yam06] A. Yamamoto, H. Miwa Y. Shibata, and A. Hashimoto, phys. stat. sol. (c) 3, No. 6, 1527 (2006). [Yam94a] A. Yamamoto, M. Tsujino, M. Ohkubo and A. Hashimoto, J. Cryst. Growth 137 (1994) 415. [Yam94b] A. Yamamoto, M. Tsujino, M. Ohkubo, A. Hashimoto, Sol. Energy Mater. Sol. Cells 35 (1994) 53.

PAGE 265

242 [Yam97] A. Yamamoto, Y. Yamauchi, M. O hkubo, A. Hashimoto and T. Saitoh, SolidState Electronics 41 (1997) 149. [Yam98] A. Yamamoto, T. Shin-ya, T. Sugiur a, A. Hashimoto, J. Cryst. Growth 189/190 (1998) 461. [Yam99] S. Yamaguchi, M. Kariya, S. Nitta, T. Takeuchi, C. Wetzel, H. Amano, I. Akasaki, J. Appl. Phys. 85 (1999) 7682. [Yan02a] F.-H. Yang, J.-S. Hwang, K.-H. Chen, Y.-J. Yang, T.-H. Lee, L.-G. Hwa, and L.-C. Chen, Thin Solid Films 405 (2002) 194. [Yan02b] W. Yang, P.Wang, F. Li an d K.W. Cheah, Nanotechnology 13 (2002) 65. [Yin04] L. Yin, Y. Bando, D. Golber g, M. Li, Adv. Mater. 16 (2004) 1833. [Yon02] I. Yonenaga, MRS Internet J. Nitride Semicond. Res. 7, 6(2002). [Yoo97] S.F. Yoon, X.B. Li, and M.Y. Kong, J. Cryst. Growth 180, 27 (1997). [Yos03] M. Yoshimoto, H. Yamamoto, W. Huang, H. Harima, J. Saraie, A. Chayahara, Y. Horino, Appl. Phys. Lett. 83 (2003) 3480. [Yos83] S. Yoshida, S. Misawa, and S. Gonda, Appl. Phys. Lett. 42, 427 (1983). [You00] W.T. Young, S.R.P. Silva, J.V. Anguita, J.M. Shannon, K.P. Homewood and B.J. Sealy, Diamond and Related Materials, 9 (2000) 456. [Yu05] H.Q. Yu, L. Chen, R. Zhang, X. Xiu, Z. L. Xie, Y.D. Ye, S.L. Gu, B. Shen, Y. Shi, Y.D. Zheng, Materials Science Forum 475-479 (2005) 3783. [Yu06] J.W. Yu, H.C. Lin, Z.C. Feng, L.S. Wang, S. Tripathy and S.J. Chua, Thin Solid Films 498 (2006) 108. [Yum81] M. Yumura, T. Asaba, Symp. Int. Combust. Proc. 18 (1981) 863. [Zam01] S. Zamir, B. Meyler, and J. Salzman, Appl. Phys. Lett. 78 (2001) 288. [Zam02] S. Zamir, B. Meyler, J. Salz man, J. Cryst. Growth 243 (2002) 375. [Zao81] A. Zaouk, E. Salvetat, J. Sakaya, F. Maury, G. Constant, J. Crystal Growth 55 (1981) 135. [Zha02] J. Zhang, L. Zhang, X. Peng and X. Wang, J. Mater. Chem (2002) 12, 802. [Zha05] J.X. Zhang, Y. Qu, Y.Z. Chen, A. Uddin and Shu Yuan, J. Cryst. Growth 282 (2005) 137.

PAGE 266

243 BIOGRAPHICAL SKETCH Hyun Jong Park was born on October 25, 1973, in Seoul, Korea. He is the son of Jong Soo Park and Sun Kyoung Kim, and has an elder sister, Soo Kyung Park. He graduated from Yongsan High school in 1992 where he had an opportunity to perform various chemistry experiments as a member of the Yongsan Chemistry Club. His dream of becoming a scientist started then. Following high school, he attended Hanyang University, graduating with a Bachelor of Science degree in chemical engineering in February 1999. Between 1994 and 1997, he serv ed as an artillery vehicle driver while performing military duty in Korea. In 1998, he attended the English Language Institute at the University of Florida in Gainesville wh ere he met his lovely wife, Catherine. After graduating from Hanyang Univer sity, he attended the Japane se Language Institute of Asuka Gakuin in Yokohama, Japan, in 1999. He returned to Gainesville and began a graduate program at the University of Florida, pursuing his doctoral degree in chemical engine ering in 2000. He married Catherine Jaewon Park in 2002, and had two sons, David Sangjin and Justin Sanghyun Park, in 2003 and 2005. While doing resear ch in Dr. Andersons group, his interests were the growth of gallium nitride and indium nitride films and nanostructured materials by Hydride-Metalorganic Vapor Phase Epitaxy.


Permanent Link: http://ufdc.ufl.edu/UFE0015718/00001

Material Information

Title: Growth of Gallium Nitride and Indium Nitride Films and Nanostructured Materials by Hydride-Metalorganic Vapor Phase Epitaxy
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0015718:00001

Permanent Link: http://ufdc.ufl.edu/UFE0015718/00001

Material Information

Title: Growth of Gallium Nitride and Indium Nitride Films and Nanostructured Materials by Hydride-Metalorganic Vapor Phase Epitaxy
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0015718:00001


This item has the following downloads:


Full Text












GROWTH OF GALLIUM NITRIDE AND INDIUM NITRIDE FILMS AND
NANOSTRUCTURED MATERIALS BY HYDRIDE-METALORGANIC VAPOR
PHASE EPITAXY















By

HYUN JONG PARK


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


2006




























Copyright 2006

by

Hyun Jong Park

































To Cathy, who endured and enjoyed our life in Gainesville, Florida















ACKNOWLEDGMENTS

I would like to thank Dr. Tim Anderson for being my advisor and chairman.

Although the expectations placed on me were always challenging, it led me to a level of

academic excellence that was unthinkable without his expectations.

I want to sincerely thank Dr. Olga Kryliouk for being a coadvisor and cochair. Dr.

Kryliouk showed me new ways of thinking in our daily discussions. She was a mentor,

friend, and sometime critic, who allowed me to attain a higher level of research. I am in

debt for her the effort and creativity she provided throughout my time at UF.

I thank Dr. Fan Ren, Dr. Cammy Abernathy, and Dr. Jason Weaver for being my

committee members. Their thoughtful advice is greatly appreciated.

Without the senior students' contributions, this work would not be possible. I am

truly in debt to Dr. Mike Reed who greatly modified the Hydride-Metalorganic Vapor

Phase Epitaxy system in Microfabritech from RCA hydride for VPE of InP. The

optimism and kindness he provided helped even after leaving the group were crucial to

me in solving numerous technical problems.

I would like to thank Dr. Mike Mastro for being my mentor and teaching me how

to operate the H-MOVPE system. He gave me valuable advice during my first year in

Dr. Anderson's group.

Dr. Sang Won Kang opened a new door for InN growth using the H-MOVPE

system. His creative thinking, logical approach, experience, and kindness have

influenced me. He is the best colleague I have ever had.









I am thankful for the enormous help of Woo Kyoung Kim, from whom I learned to

organize things and how to work effectively. He also helped to perform HT-XRD and

analyze the data. He was a counselor, friend, and mentor who has permanently

influenced my life.

YongSun Won is the smartest person that I have known who understands matters in

a scientific way. His way of thinking was always beyond my knowledge. I appreciate

his help and valuable discussions.

Josh Mangum embodies the right attitude towards research and people. He is a

good communicator and friend. His willingness to help was essential for me to complete

this project.

Dr. Jang Yeon Hwang and Young Seok Kim helped me with Raman spectroscopy

and hydrodynamics simulations. Their thorough understanding of the physical world

deeply influenced me.

Dr. Seokhyun Yoon helped me with SEM and EDS when I did not know how to

use them. He also shared with me a happiness of a loving family when I was alone.

Dr. Byoung Sam Kang and Hung-Ta Wang in Dr. Fan Ren's group fabricated InN

nanorods based gas sensor. The completeness of my work would not be possible without

their help.

I would like to thank my colleagues, KeeChan Kim, Oh-Hyun Kim, and Do Jun

Kim for providing helpful discussions and encouragement during my research. KeeChan

Kim and Oh Hyun Kim especially helped me automate the H-MOVPE system. Their

way of working effectively was very influential.









Without the great staff at UF, this work would not have been possible. I would like

to acknowledge James Hinnant and Dennis Vince in Chemical Engineering, Scott

Gapinski, Di Badylak, and Chuck Rowland in Microfabritech. I sincerely thank Valentin

Craciun for HR-XRD, Eric Lambers for AES and XPS, Maggie Puga-Lambers for SIMS,

Kerry Siebein for TEM at the Major Analytical Instrument Center, UF.

I also would like to acknowledge my collaboration outside UF. The support and

discussion with Dr. Chinho Park (Yeung Nam University) was greatly appreciated. He

and his students, Seok-Ki Yeo, helped analyze the Raman spectroscopy data and gave me

valuable advice. Dr. Jianyun Shen taught me how to use Thermo-Calc and strain energy

modeling. Drs. Albert Davydov and Igor Levin helped me verify the thermochemical

data and did excellent characterization samples at NIST by FE-SEM, TEM, and CBED.

Drs. Dmitry Khokhlov at Moscow State University and Timur Burbaev at Physical

Institute, Moscow carried out the PL measurements for the InN films and nanorods. Dr.

Jaime Freitas at Naval Research Lab performed the Raman spectroscopy and PL on the

InN nanorod samples. Dr. Zuzanna Liliental-Weber at Lawrence Berkeley Lab

characterized InN nanorods by HR-TEM. Dr. Talmage Tyler at the International

Technology Center measured the field emission properties of InN nanorods. Dr. Mee Yi

Ryu at the Air Force Institute of Technology performed CL and Hall measurements of

InN films and nanorods. I would not have reached this point without their enormous

help.

I thank Drs. Seong-Geun Oh, Yeong Koo Yeo, and Hong-Woo Park in Chemical

Engineering, Hanyang University for recommending me for studying in the University of

Florida. I would not have this great opportunity without their kind and thoughtful help.









I also would like to thank Andrew Wislocki for proofreading this dissertation. He

corrected grammatical errors and made the contents clearer by helpful discussions.

I cannot thank my family enough for their boundless love. My father (Jong Soo

Park), mother (Sun Kyoung Kim), and sister (Soo Kyoung Park) have been an essential

part of my life. Although they were far away in Korea, they were always with me in my

heart throughout my life as a graduate student in Gainesville.

My sons, David and Justin Park, showed me what true love is. Although I needed

more time and effort to balance family life and research, I would not trade their love for

anything in the world.

Finally, I would like to thank my wife, Catherine Park, for her endless support and

trust. I was always happy for her presence and I will feel this way forever.
















TABLE OF CONTENTS



A CK N O W LED G M EN TS ........................ ............................................ ............ ........ iv

L IST O F T A B L E S ................. .................................................... .. ........ .................. xii

LIST OF FIGURES ......... ....................... .......... ....... ............ xiv

ABSTRACT. ........................................ xxii

CHAPTER

1 IN TR OD U CTION ............................................... .. ......................... ..

1.1 History of GaN and InN Development ....................................... ................
1.1.1 G aN D evelopm ent .......................................... ...... .. ...... .............. .
1.1.2 InN D evelopm ent ............................................. .............. ........ 3
1.2 Literature R review ................... ..... ........................................ .................. .... .9
1.2.1 Equilibrium Analysis of GaN and InN..................................................9
1.2.1.1 Thermochemical Data for GaN and InN ...........................................9
1.2.1.2 Equilibrium Calculations of GaN Growth by HVPE.....................10
1.2 .2 Stress in G aN F ilm s........................................................................ ... ... 14
1.2.2.1 In situ Stress M easurem ents............... ...........................................16
1.2.2.2 Ex situ Stress M easurements ............. ............................................18
1.2.3 G aN G row th on Si ............... ...... ....................... ............... ... 2 1
1.2.4 Growth of InN Films and Nanostructured Materials............... ...............25
1.2.4.1 G row th of InN Film s ................................ ........................ .......... 25
1.2.4.2 Growth of InN Nanostructured Materials .....................................29

2 CHEMICAL EQUILIBRIUM ANALYSIS OF H-MOVPE SYSTEM.................32

2 .1 Introdu action ................. ....... .............. ...... ........... ................. 32
2.2 Therm ochem ical D ata C collections ............................................ .....................33
2.3 Chemical Equilibrium Calculations..... .................... ...............35
2.3.1 Therm ochem ical D ata V erification ........................................ .................35
2.3.2 Complex Chemical Equilibrium Calculations ...................................38
2.3.2.1 G a-C-H -Cl-Inert System ...................................... ............... 38
2.3.2.2 Ga-C-H-Cl-N System ........................................................ ......... 45
2.3.2.3 In-C-H -Cl-Inert System ......................................... ............... 47
2.3.2.4 In-C-H -Cl-N -Inert System .................................... ............... 53









2.3.2.5 NH3 Partial Decomposition............................. .. .......... ........ 60
2 .4 C o n c lu sio n s ..................................................................................................... 6 2

3 STRESS DETERMINATION STUDIES OF GALLIUM NITRIDE FILM ON
S A P P H IR E ........................................................................................................... 6 4

3.1 Introduction ................................................ ... ........................ 64
3.2 Effects of Lattice and Thermal Expansion Mismatches ............. ...............67
3.2.1 Lattice Mismatch at Growth Temperature ...........................................67
3.2.2 Thermal Expansion Mismatch.......................... .....................69
3.2.3 Combination of Lattice and Thermal Expansion Mismatches ...................70
3.3 Stress Measurements of GaN Films on Sapphire ...........................................71
3.3.1 GaN Structural and Compositional Studies.....................................72
3.3.2 Stress Measurements by Raman Spectroscopy .......................................77
3.3.2.1 Curvature Calculations............... ........ ......... .... ........... 79
3.3.2.2 Lattice Parameter Calculations............................... 80
3.3.3 Lattice Parameter Measurements by XRD Reciprocal Space
M ap p in g ...................................... ............................................. .. 8 1
3.3.4 Curvature Measurements by XRD Rocking Curve ...................................84
3.4 Stress M odeling of GaN on Sapphire...................... .......................................... 89
3.4.1 Frank-van der M erwe's Semi-infinite M odel .............................................89
3.4.2 Layer-by-Layer Growth M odel ...................................... ............... .... 91
3.4.2.1 Strain Energy (Homogeneous Stress) Calculations .......................92
3.4.2.2 Dislocation Energy (Periodic Strain Energy) Calculation ..............94
3.4.2.3. Total Strain Energy Calculations ......................................... 95
3 .5 C o n clu sio n s.................................................. ................ 9 7

4 GROWTH OF GALLIUM NITRIDE ON SILICON BY H-MOVPE.................100

4 .1 In tro d u ctio n ............. ........ ......... ................. ................................................ 10 0
4.2 H-M OVPE Growth Technique............................................ .......................... 101
4.2.1 Chemical Reactions of H-MOVPE Technique.................. ........... 101
4.2.2 H-M OVPE Reactor Schematics .................................... ............... 102
4.3 Thick G aN G row th on A 120 3.................................................................. ...... 106
4.4 G aN G row th on Si .................................................... .. .. .. .. .............. 111
4.4.1 Growth of Thin and Thick GaN on Si ................................................111
4.4.1.1 N itronex G aN /Si tem plate........................................................... 111
4.4.1.2 Growth of Thin GaN Using Nitronex Template ............................114
4.4.1.3 Growth of Thick GaN Using the Nitronex Template.....................115
4.4.2 Growth of Thick GaN on Si Using InN Interlayers .............................118
4.4.2.1 InN G row th on Si( 11) ............................................................ 119
4.4.2.2 LT-GaN Growth on InN Buffer Materials/Si(1 11)........................120
4.4.2.3 Thick HT-GaN Growth on LT-GaN/InN/Si(1 11)..........................123
4 .5 C on clu sion s................................................. ................ 12 6

5 GROWTH OF INDIUM NITRIDE NANORODS BY H-MOVPE.................. 128









5 .1 In tro d u ctio n ................. ...... ...... ........... ...... ............................ 12 8
5.2 Chemical Reactions for InN Growth by H-MOVPE ........................................130
5.3 InN Nanorods Growth Optimization .............. ............................................130
5.3.1 Experim ental Procedure ........................................ ....................... 131
5 .3 .2 R esu lts ................... ................... .. ........................... ....................13 1
5.3.2.1 Effects of Growth Temperature and HC1/TMI Molar Ratio ..........131
5.3.2.2 Effects of NH3/TMI Molar Ratio and Substrate Material .............134
5.3.2.3 Equilibrium A nalysis................................ ........................ 135
5.4 Properties of InN N anorods .................................................................... ..... 138
5.4.1 M orphology ..................................... ...... .... .... ................. 138
5 .4 .2 C ry stalling ity ............................................................... 13 9
5.4.3 Self-alignment ................................... ....... .... .. .......... 141
5.4.4 Growth Axis and Structural Properties.........................................145
5 .4 .5 P olarity .....................................................................150
5.4.6 Chem ical Com position ................................................... ........ ....... 151
5.4.7 Photolum inescence .............................................................. ............... 153
5.4.8 Cathodolum inescence ...................................................... ...... ........ 156
5.4.9 R am an Spectroscopy ........................................ .......................... 157
5.4.10 E electrical Properties........................................... .......... ............... 158
5.4.11 Field Em mission Properties............................ ......... ... ................. 159
5.5 Pt-dispersed InN Nanorods for Selective Detection of Hydrogen at Room
T em p eratu re ................... ................................................ ............... 16 3
5.5.1 Experim ental Procedure ........................................ ....................... 163
5 .5 .2 R e su lts ................................................................... 16 4
5 .6 C on clu sion s................................................. ................ 16 7

6 EXPLORATORY STUDY OF INDIUM NITRIDE FILMS GROWTH BY H-
M O V P E .............................................................................16 9

6.1 Experim ental Procedure.............................................. ............................ 169
6.2 Results........................................................... 170
6.2.1 Effect of HCl/TMIn Molar Ratio ..................................................170
6.2.2 Effect of Growth Temperature ..........................................172
6.2.3 Effect of NH3/TMIn Molar Ratio...... ............. ...............174
6.2.4 Effect of Buffer Layer ........................................................................ 177
6.2.4.1 Surface Morphology of InN Films without Buffer Layer............ 177
6.2.4.2 Growth of InN Films with Low Temperature Buffer Layer ..........177
6.2.5 Growth of InN Film s on A1203 and Si..................................................... 179
6 .3 C on clu sion s ................................................. ................. 18 1

7 EXPLORATORY STUDY OF GALLIUM NITRIDE NANORODS GROWTH
BY H-M OVPE.................. ............................. ...... .... .... .. ............... 183

7.1 Experim ental Procedure............................................. ............................. 183
7.2 Results .............................................. .................................183
7.3 Conclusions ......................................... 189









8 FUTURE WORK AND RECOMMENDATIONS ...............................................190

8.1 Nucleation and Growth Mechanism Studies of InN Nanorods ..........................190
8.2 Self-aligned InN Nanorods as Buffer Material for GaN on Si .........................190
8.3 Buffer Layer Optimization for GaN growth on Si Substrate.............................191
8.4 Use of Negative Thermal Expansion Materials.......................................192
8.5 Double-sided Growth of GaN on Si and Sapphire ........... ....... ............... 195

APPENDIX: THERMOCHEMICAL DATA FOR AL-IN-GA-C-H-CL-N
SY S T E M ...................................... ................................................... 19 6

L IST O F R E FE R E N C E S ..................................................................... ...... ................222

BIOGRAPH ICAL SKETCH ...................................................... 243















LIST OF TABLES

Table p

1-1 Reported values of thermochemical data for solid GaN and InN ............................12

1-2 Reported values of the Raman E2(high) peak positions .............................................20

1-3 Techniques used for growth of crack-free GaN on Si............................................23

2-1 Commonly considered species in Ga-In-H-C-Cl-N system................................34

2-2 Additionally included gas phase species ...... ......... ....................................... 35

2-3 Base inlet conditions for sources for GaN growth and atomic mole fractions for
calcu nation ........................................................................... 3 9

2-4 Typical growth conditions for InN and atomic mole fractions for calculation........48

3-1 Lattice parameters of GaN and widely used substrates at 300 and 1300 K.............68

3-2 Lattice constants at 1300 and 300 K and thermal strain (AT = 1000 K) of GaN
and w idely u sed sub states ............................................................ .....................70

3-3 The growth conditions of the two samples for stress measurements .......................72

3-4 GaN a and c lattice parameters measured at 300 K (A)...........................................84

3-5 Stress, lattice parameter, and curvature of the H-MOVPE GaN with depth............86

3-6 Calculated misfit dislocation density of GaN on widely used substrates with
V ernier period ........................................................................90

3-7 Lattice constants (A) of GaN, A1N, and A1203 ................................................93

3-8 Elastic stiffness coefficients c, (GPa) and compliances sij (1/GPa) of GaN, A1N,
an d A 120 3 ............................................................................9 3

3-9 Shear moduli, Poisson's ratios, and lattice parameters of GaN, A1N, and A1203.....94

3-10 Calculated lattice constants and total energy ................................. ............... 96

4-1 Electrical properties of GaN film grown on GaN/A1203 template .........................111









4-2 Sheet resistance and resistivity measured by four-point probe .............................111

4-3 Growth conditions for InN buffer interlayer on Si..............................................119

5-1 Base conditions of InN nanorods growth...... .............. ................................131

5-2 Intensity ratio comparison with XRD powder pattern with different
substrates ............................................................... ... .... ......... 140

5-3 Electrical properties of InN films and nanorods ....................................................159

6-1 Base conditions of InN film growth.................................................................... 169

7-1 Base conditions for GaN micro/nanorods growth...............................183

8-1 L ists of N T E m materials ................................................................ ..................... 192

A -i G as P hase .......................................................................... 197

A-2 Gas Phase (Metalorganics, Adducts, Oligomers, etc) ................. .. ..................217

A-3 Liquid Phase .................................... ................................ .........218

A -4 Solid Phase .............................................................................................. ........220

A -5 Carbon (Graphite) ................................................................221

A-6 GaN and InN .................................. ..... .. ...... .. ............221














LIST OF FIGURES


Figure page

1-1 Reported values of electric properites of InN (a) Hall mobility and (b) Carrier
concentration w ith calendar year.................................................................... ...... 6

1-2 Reported bandgap energy of InN as a function of carrier concentration ................7

1-3 Examples of equilibrium calculations. (a) Driving force for the deposition as a
function of growth temperature with various parameters for F (b) Comparison
between calculated growth rates and experimental data ..........................................11

1-4 Driving force for the deposition of InN (APin) using InCl (HVPE) and InC13
(THVPE) as a function of growth temperature. ............ ....................................13

1-5 Schematic of GaN growth on SiC (a) GaN/SiC and (b) GaN/AlN/SiC growth ......16

1-6 In situ curvature measurements during the growth of a 6 gsm thick GaN layer
showing the influence of AIN interlayers on the curvature. The sample was
crack-free after grow th ...................... .... ......... ............................ ............... 18

1-7 Raman peak shifts of 2 [tm thick GaN with different LT-GaN buffer layer
thickness; LT-GaN buffer layer thickness (a) 10 nm, (b) 50 nm, (c) 75 nm, and
(d ) 8 5 n m ............................................................................ 2 1

2-1 Phase diagram of In-N systems at P = 0.1 MPa and experimental InN
decom p o sition data ............ ...................................................................... .. ...... .. 36

2-2 Calculated AGrxn for GaN(s) and (GaN)3(g) formation reactions ..........................37

2-3 Phase diagram of Ga-N systems at P = 0.1 MPa................... .............................. 38

2-4 Schematic of the inlet of H-MOVPE technique for GaN growth ............................38

2-5 Equilibrium partial pressures of gas phase molecules and mole fractions of
condensed phases with respect to the temperature in Ga-C-H-Cl-Inert system.......40

2-6 Equilibrium partial pressures of gas phase molecules and mole fractions of
condensed phases with respect to X(H)/{X(Inert) + X(H)} in Ga-C-H-Cl-Inert
sy ste m .......................................... ............................. ................ 4 2









2-7 Equilibrium partial pressures of gas phase molecules and mole fractions of
condensed phases with respect to X(C1)/X(Ga) ratio in Ga-C-H-Cl-Inert
sy stem ...................................... ......................................................4 4

2-8 Schematic of H-MOVPE for GaN growth: Ga-C-H-Cl-N system. N is present
and H2 is used as a carrier gas in the system compared to Figure 2-4 ...................45

2-9 Calculated growth-etch transition temperature as a function of Cl/Ga ratio............46

2-10 Calculated transition of growth-etch and Ga(l) formation as a function of
tem perature and C l/G a ratio ............................................. ............................ 47

2-11 Schematic of the inlet of H-MOVPE technique for InN growth.............................48

2-12 Equilibrium partial pressures of gas phase molecules and mole fractions of
condensed phases with respect to the temperature in In-C-H-Cl-Inert system........50

2-13 Equilibrium partial pressures of gas phase molecules and mole fractions of
condensed phases with respect to X(H)/{X(Inert) + X(H)} in In-C-H-Cl-Inert
system ............. ... ........ ............. ........... 51

2-14 Equilibrium partial pressures of gas phase molecules and mole fractions of
condensed phases with respect to X(C1)/X(In) ratio in In-C-H-Cl-Inert system .....52

2-15 Schematic of H-MOVPE inlet with In-C-H-Cl-N-Inert system for
thermodynamic calculations. N is present in the system compared to Figure 2-
1 1 .................................................................................... . .5 4

2-16 Calculated growth and etch regime with respect to temperature and Cl/In ratio.
N/In ratio was varied from 100 to 7000 ....................................... ............... 55

2-17 Calculated growth and etch regime with respect to temperature and Cl/In ratio.
Cl/In ratio w as varied from 1.1 to 10 ............................................ ............... 56

2-18 Calculated results of the growth, no growth (etch), and In droplet regimes (a)
Cl/In = 0 to 0.5, N/In = 4.0 x 104 to 7.0 x 104, (b) Cl/In = 0 to 1, N/In = 1.0 x 103
to 7.8 x 104, and (c) In droplet etching conditions at T = 1153 K, P = 105 Pa.........57

2-19 Calculated growth-etch transition temperatures (a) In droplet etch conditions
when no HC1 was present (b) growth-etch transition temperature at high N/In
ratios: 10 106, and 107 ....................... .. .................... .................. ...........59

2-20 Equilibrium partial pressures for decomposition of NH3 at 1 atm total pressures
calculated by Therm o-C alc ............................................. ............................. 60

2-21 Magnitude of the Gibbs energy correction required to achieve a value for partial
decom position ( ) at 950 C ......................................................... ............... 61









2-22 NH3 mole fraction with respect to temperature at different values of c..................62

3-1 Substrates for GaN plotted with thermal strain versus lattice parameters at room
tem perature ..................................... ................................ ........... 70

3-2 XRD o and o-20 rocking curve of MOCVD and H-MOVPE grown GaN on
sapphire, XRD FWHM; co-rocking curve (MOCVD: 1261 arcsec, H-MOVPE:
1790 arcsec); co-20 rocking curve (MOCVD: 92 arcsec,
H -M O V PE : 122 arcsec) ................................................ .............................. 72

3-3 Pole figures for the (116) sapphire substrate (2 theta = 57.4900) and (112) GaN
by MOCVD (2 theta = 69.1850). The GaN in-plane axis is rotated by 300 with
respect to the sapphire axis............................................. .............................. 73

3-4 Pole figures for the (116) sapphire substrate (2 theta = 57.4900) and (112) GaN
by H-MOVPE (2 theta = 69.1850). The GaN in-plane axis is rotated by 300
w ith respect to the sapphire axis ........................................ ......................... 74

3-5 AES surface scan and depth profile of GaN on sapphire grown by MOCVD......... 75

3-6 AES surface scan and depth profile of GaN on sapphire grown by H-MOVPE .....76

3-7 Raman E2 peak shifts at the surface (a) MOCVD, (b) H-MOVPE, and (c) depth
p ro file ............................................................................ 7 8

3-8 XRD reciprocal space mapping of H-MOVPE grown GaN (002) and (114)
p e a k s ...................................... .................................................... 8 2

3-9 XRD reciprocal space mapping of MOCVD grown GaN(002) and (114)
p eak s ..............................................................................................8 3

3-10 The procedure of determining curvature by measurements of XRD rocking
c u rv e s ............................................................................ 8 5

3-11 Curvature measurements by XRD rocking curve by displacing the H-MOVPE
grown GaN/c-Al203 sample in x-direction by 5 mm ............................................ 86

3-12 SIMS oxygen depth profiles of MOCVD and H-MOVPE grown GaN films .........88

3-13 Frank-van der Merwe's semi-infinite overgrowth model .......................................90

3-14 Shen-D ann's layer-by-layer m odel ............................................... ............... 91

3-15 Minimum energy calculation of stress + dislocation energy................................96

3-16 In-plane lattice constants of GaN on AIN/sapphire .............................................97









4-1 Schematics of H-MOVPE (a) bird eye view of the entire reactor, (b) the source
and growth zones with temperature profile in the source zone.............................103

4-2 Process Flow Diagram (PFD) of H-M OVPE system ............................................ 105

4-3 XRD 0-20 scan and co-rocking curve of as received GaN/A1203 template from
Uniroyal Optoelectronics. denotes the secondary peak due to Cu Kp radiation 107

4-4 SEM images of GaN film (a) Cross-sectional SEM of 125 itm thick H-MOVPE
GaN film on GaN/A1203 template, (b) SEM plan-view of the same sample. (c)
Plan-view of LT-HVPE smoothing layer................... .....................108

4-5 XRD of GaN film (a) XRD 0-20 scan of thick (45 [tm) GaN on GaN/A1203, (b)
XRD co-rocking curve of GaN (002) peak; FWHM = 780 arcsec .........................109

4-6 Auger Electron Spectroscopy of 45 |tm thick GaN film (a) surface scan (b)
sputtering depth profile (c) surface scan after sputtering ................................110

4-7 X-SEM image of GaN on Si using AlGaN graded layer ................................112

4-8 XRD 0-20 scan and HR-XRD co-rocking curve of as received GaN/Si from
Nitronex, FWHM = 0.222 0 (800 arcsec) ............... ....... .... ..... ......... 113

4-9 SEM plan and cross sectional views of 2 pm H-MOVPE + 1 pm MOVPE crack-
free GaN grown on AlGaN/Si graded layer..........................................................114

4-10 XRD 0-20 scan of 2 im H-MOVPE + 1 im MOVPE crack-free GaN grown on
A lG aN /Si graded layer .......................................................... ............... 114

4-11 SEM plan-views of thick (55 [tm) GaN films grown on GaN/Si templates (4 hr)
with (a) fast cooling and (b) slow cooling. Note two micrographs are at
different m agnification ..................................................................................... 115

4-12 Schematic of crack generation of GaN on Si and GaN on Nitronex template.
Cracking penetration to Si were observed in both cases .............................116

4-13 XRD 0-20 scan of thick (55 pm) GaN film on GaN/Si template for 4 hr............17

4-14 XRD co-rocking curve of GaN (002) peak of 55 pm thick GaN. FWHM = 0.248
0 (8 9 2 arcsec)....................................................................................... ..... 1 17

4-15 The schematic to obtain thick GaN on Si(111) using InN buffer materials...........119

4-16 SEM plan-view of InN columnar film, small nanorods (d = 250 nm), large
nanorods (d = 500 nm), and microrods grown on Si(11)................................... 120









4-17 SEM plan-view of 10, 20, and 30 min LT-GaN grown on (a) InN columnar film,
(b) smaller nanorods (d = 250 nm), (c) larger nanorods (d = 500 nm), and (d)
m icrorods ..................................... ................................ ........... 121

4-18 LT-GaN deposited on InN nanorods (d = 500 nm) on Si( 11) for 30 min (- 4
[tm) (a) plan-view of LT-GaN on InN nanorods, (b) cross-sectional view (c)
XRD 0-20 scan .................................... ............................... ........122

4-19 HT-GaN/LT-GaN growth on various InN structures (a) InN columnar film, (b)
small nanorods (d = 250 nm), (c) large nanorods (d = 500 nm), and (d)
microrods. The right figures show expanded views of the surface of HT-GaN
fi lm s ......................................................................... 12 3

4-20 Thick (28 pm) and freestanding (self-separated) GaN grown on InN nanorods/Si
(a) SEM plan-view of the GaN film, (b) XRD 0-20 scan............................... 124

4-21 Cross-sectional view of thick GaN grown on InN nanorods/Si( 11). The surface
was covered by LT-HVPE layer grown after thick HT-HVPE..............................125

4-22 SEM plan-view of thick GaN grown on InN nanorods (d=500 nm)/Si(l 11). (a)
as grown HT-GaN surface, (b) covered LT-GaN surface.................................... 126

5-1 The growth map of InN at different HC1/TMI ratio with growth temperature;
N/In = 250; Growth time = 1 hr; Substrate = c-A1203 ....................................... 132

5-2 Scanning electron micrographs of InN film and nanorods. NH3/TMIn were
varied from 100 to 7000; Other growth conditions are HCl/TMIn = 4, T = 600
C; Growth time = 1 hr....................................... ............. 134

5-3 The calculated boundary of the growth and etch regimes and experimental
observations. (a) deposition temperature vs. HCl/TMIn ratio for selected N/In
atom ratios; (b) deposition temperature vs. N/In atomic ratio for selected
H Cl/TM In ratios ........... ........ ... ... ... ...... ..... .. ............ 136

5-4 Scanning electron micrographs of InN nanorods grown at optimal conditions;
Cl/In = 4, N/In = 250, T = 600 C, for 1 hr growth on Si(11) .............................138

5-5 XRD 0-20 scan of InN nanorods grown on various substrates (a) a-Al203, (b) c-
Al203, (c) r-Al203, (d) Si(100), (e) Si (111), (f) GaN/c-A1203, and (g) PDF#50-
1239 (powder diffraction file) .......................... ......... .................... 139

5-6 Scanning Electron Micrographs of aligned InN nanorods grown on GaN/A1203
substrates. (a) sparse nanorods with sharp tips (b) dense nanorods with sharp tips
(c) dense nanorods w ith flat tips ........................................ ........................ 142

5-7 XRD scan of aligned InN nanorods on GaN/c-A1203 substrates .........................143

5-8 XRD co-rocking curve of aligned InN nanorods on GaN/A1203 ............................143


xviii









5-9 XRD pole figure of self-aligned InN nanorods (Substrate: GaN/c-Al203)............144

5-10 InN (103) plane and )-axis for pole figure measurement .....................................144

5-11 Transmission electron micrographs of an InN nanorod (a) splitting of reflections
and (BF) bright field image at the bottom, (b) sparse planar defects by dark- and
b rig ht-field im ag e ................................................................. ................ ...... 14 6

5-12 TEM and SAD of InN nanorods (WZ) with growth axes (a) [0002], (b) [1010],
an d (c) [1124 ] ........................................................................ 14 7

5-13 SEM image of wurtzite InN nanorod grown on the side wall of a nanorod ..........148

5-14 TEM and SAD of InN and In203 nanorods (a) InN (ZB) growth axis = a few
degrees off [220] (b) In203 (BCC) growth axis = [402] .....................................149

5-15 C B E D of the InN nanorod........................................................................... ... .... 151

5-16 AES spectrum of surface scan of as grown InN nanorods at T = 600 OC,
HCl/TMIn = 4, NH3/TMIn = 250, 1 hr growth on c-sapphire .............................152

5-17 Scanning electron micrographs with AES of InN nanorods before and after AES
sputtering dam age ...................... .................... .. .... ......... ..... .... 152

5-18 EDS line-scan of an InN nanorod grown on Si substrate................... .............153

5-19 Room temperature (300 K) PL of InN nanorods grown on Si substrate and InN
film on GaN substrate, maximum intensity observed around 1.08 eV ................154

5-20 Low Temperature (6 K to 300 K) PL of InN nanorods on Si substrate (a) whole
detection range, (b) magnified range between 730 to 850 meV from 23 to 100 K
reg io n ...................................... .................................................. 1 5 5

5-21 CL spectra of InN nanorods grown on sapphire .............................................. 156

5-22 Room-temperature Raman scattering of InN nanorods deposited on Si and
G aN /Si substrates ........................................... ....... ........ .. ........ .. .. 158

5-23 Scanning Electron Micrograph of sharp edge InN nanorods grown on Si (single
NR on the left) and on GaN/A1203 substrates....................................................... 160

5-24 I-V curves and Fowler-Nordheim plots of InN nanorods on Si.............................161

5-25 SEM images and schematic of hydrogen gas sensor made of Pt nanoparticle
dispersed InN nanorods ................................................... ........ ............... 163

5-26 I-V characteristics from uncoated and Pt-coated InN nanorods ..........................164









5-27 I-Time plot of 10 to 250 ppm H2 test by Pt-InN nanorods (left) and |AR|/R(%)-
Time plot of 10 to 250 ppm H2 test by Pt-InN nanorods (right) ............................165

5-28 N2, N20, ND3, and 02 test for InN nanorods ............................... ............... .166

6-1 Scanning electron micrographs of InN films (a) HCl/TMIn = 0, (b) HCl/TMIn =
0.3, (c) HCl/TMIn = 1, and (d) HCl/TMIn = 4. Other growth conditions: T = 560
C; P = 760 Torr; N/In ratio = 2500; substrate: GaN/A1203;
grow th tim e = 1 hr............. .... .................................. ....... .......... .. 70

6-2 Growth rate and XRD of InN (a) InN growth rate and (b) XRD scan with respect
to H Cl/TM In ratio ...................... .................... .. .... ......... .. .... .. 171

6-3 SEM plan-views of InN films grown at different temperatures (a) 500, (b) 525,
(c) 560, and (d) 600 C; Other growth conditions Cl/III ratio = 1; P = 760 Torr;
N/In ratio = 2500; Substrate = GaN/c-A1203; growth time = 1 hr .........................172

6-4 Growth rate and XRD scans with respect to the growth temperature (a) Loge
(Normalized growth rate) vs 1000/T (K), (b) XRD scans in 2 theta ranges from
300 to 320; growth temperature 300 to 700 OC; growth time = 1 hr.......................173

6-5 Scanning electron micrographs of InN films grown in different NH3/TMIn
ratios; other growth parameters Cl/In ratio = 1; P = 760 Torr; Growth T = 560
C; grow th tim e = 1 hr .......... ...... .................... ........ ...... .. .. ........ .... 175

6-6 Growth rate and XRD scans with respect to NH3/TMIn ratio (a) Growth rate vs.
N/In (NH3/TMIn) ratio, (b) XRD scans of 2 theta range from 30 to 320 when
N/In was 100 to 50000, (c) enlarged view of XRD for N/In range 7000 to 50000;
Cl/In ratio = 1; P=760 Torr; Growth T = 560 C; growth time = 1 hr................... 176

6-7 AFM images of the InN films. NH3/TMI = 10000, HC1/TMI = 0.3, T=560 C;
Substrate = GaN/c-A1203; growth tim e = 1 hr ....................................................... 177

6-8 SEM plan-view and cross-sectional of InN films on GaN/A1203 substrate; N/In
= 5000, Cl/In = 1, T = 560 C, and growth time = 1 hr with 450 C buffer layer
grow th for 30 m in............ .... ............................................... ... 78

6-9 XRD 0-20 scan of InN grown on GaN/A1203 substrate; N/In = 5000, Cl/In = 1,
T = 560 C, and growth time = 1 hr with 450 C buffer layer growth for 30 min.178

6-10 XRD co-rocking curve of InN grown on GaN/A1203 substrate; N/In = 5000,
Cl/In = 1, T = 560 C, and growth time = 1 hr with 450 C buffer layer growth
fo r 3 0 m in ......................................... ................... ................. 17 9

6-11 XRD 0-20 scan of InN film grown on c-A1203......................................................180

6-12 XRD 0-20 scan of InN film grown on Si(111) ..................................................... 180









6-13 XRD 0-20 scan comparison of InN film grown on different substrates ..............181

7-1 SEM plan-view micrographs of GaN grown at Cl/Ga ratio 2 and 3 on c-A1203,
GaN/c-A1203, and Si at N/Ga = 250 without nitridation in N2 ...........................184

7-2 SEM micrographs of GaN grown at Cl/Ga ratio of 2.5 and 2.75, N/Ga = 250,
and in H 2 w ith nitridation ........... ......................................................... 185

7-3 SEM plan-view of GaN with Cl/Ga ratio 2, 2.5, 4, and 6 on c-A1203, GaN/c-
A 120 3, an d Si(111)................................................................................ ..... 186

7-4 SEM plan-view of GaN at Cl/Ga = 19, N/Ga = 800, T = 850 C on various
substrates ................ ................ ............. .......... .......... 187

8-1 Schematic of crack-free GaN growth on Si substrate using self-aligned InN or
G aN n an oro d s ................................................................... ................ 19 1

8-2 Lattice parameters of ZrW20 (0) and HfW20s (0) as a function of
tem perature ............................................................... ... .... ......... 193

8-3 Crystal structures of NTE materials (a) Unit cell of ZrW20s, with 90% thermal
ellipsoids drawn. (b) Polyhedral representation of the structure. ZrO6 octahedra
shown in white, W 04 tetrahedra shaded............... ............................................. 193

8-4 An array of linked triangles as found in tridymite. Rotation of one triangle
causes the local environment to be pulled inwards.............. ....... ....................194















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

GROWTH OF GALLIUM NITRIDE AND INDIUM NITRIDE FILMS AND
NANOSTRUCTURED MATERIALS BY HYDRIDE-METALORGANIC VAPOR
PHASE EPITAXY

By

Hyun Jong Park

December 2006

Chair: Timothy Anderson
Cochair: Olga Kryliouk
Major Department: Chemical Engineering

Chemical equilibria analyses in the Ga/In-H-C-Cl-N-inert system were performed

to predict gas and condensed phase species that might exist in H-MOVPE of GaN and

InN. It was found that carbon co-deposition and metal droplets (Ga or In) formation can

be eliminated by providing a threshold level of H2 and HC1, respectively. The transitions

between deposition and etching, and between 2-phase (Ga or In)N and pure (Ga or In)

metal and single-phase (Ga or In)N were predicted as a function of N/III and Cl/III molar

ratios and temperature.

Stress was measured by Raman spectroscopy in two GaN films (grown by

MOCVD and H-MOVPE, respectively) on sapphire and a difference of 2 cm1 in the E2

mode was observed. Subsequent measurement by XRD-SM, rocking-curves, AES, and

SIMS showed inconsistent results compared to the Raman result suggesting that the

Raman E2 shift may not be only related to biaxial or hydrostatic stress.









Controlled growth of InN nanorods (NR) was achieved by varying the Cl/In and

N/In molar ratios and growth temperature. The NRs were grown on a, c, r-A1203, GaN/c-

A1203, Si (100), and Si (111) without a template or external catalyst. Well-faceted

wurtzite and threading dislocation-free NRs were observed by SEM and TEM. The

diameters and lengths of nanorods ranged 100 to 300 nm and 1 |tm for 1 hr growth.

XRD patterns indicated the nanorods were textured in the [002] direction and TEM-DP

confirmed the growth axis was predominantly [002]. Nanorods grown on GaN/c-Al203

occasionally showed vertical self-alignment and were epitaxially grown as judged by

XRD pole-figure analysis. The nanorods had N-polarity as characterized by CBED,

while RT PL showed a predominant peak at 1.08 eV. Raman spectroscopy showed three

phonon lines (451, 496, and 596 cm-1) that were assigned to Ai(TO), E22, and Ei(LO),

respectively. InN NRs-based gas sensor was fabricated that could detect H2 to 10 ppm.

Crack-free, 3 |tm GaN films were grown on GaN/AlGaN/Si template at 850 C

although cracks developed when the thickness exceeded 7 |tm. It was possible, however,

to grow crack-free polycrystalline 40 |tm thick GaN on Si using InN NRs as a buffer

material.


xxiii














CHAPTER 1
INTRODUCTION

1.1 History of GaN and InN Development

The technology related to GaN and InN is highly developed. Both are currently

used as host materials in optoelectronics and electronics devices. It is informative to

review the history of those materials to properly understand the current issues that require

more study.

1.1.1 GaN Development

The history of GaN crystal growth originates in 1938 [Juz38] when GaN was first

synthesized by flowing NH3 through gallium at high temperatures. The standard crystals

for GaN powder X-ray diffraction file (i.e., PDF# 02-1078) were obtained then and this

data is still in use. Grimmeriss and Koelmans used the same method to grow small GaN

crystals and measured PL for the first time back in 1959 [Gri59].

GaN film was first deposited by Maruska and Tietjen on sapphire substrate by

CVD [Mar69]. Little attention, however, was paid to GaN's use as a semiconductor

because of the difficulty in doping it p-type. GaN regained its attention in 1980s with

need to develop blue emitters.

Yoshida reported improvement of GaN quality by using LT-A1N buffer layer in

1983 [Yos83]. In 1986, Amano reported improved GaN surface morphology and

electrical and optical properties by using LT-A1N buffer layer [Ama86]. The buffer layer

acted as a nucleation layer and decreased interfacial free energy to facilitate two-

dimensional growth by changing the properties of the surface.









To fabricate GaN based devices, electronic properties such as carrier concentrations

must be controlled. The doping of wide band gap materials is known to be difficult in

part the probability of forming native defects that can dominate the electronic point

defect chemistry. Intrinsic GaN grows n-type with a direct wide (3.4 eV) bandgap. The

long belief in the native n-type nature of GaN is due to N-vacancy donor behavior. High

conductivity p-type GaN was not achieved until 1989. It was observed that p-type

conductivity drastically improved (from 108 to 35 Qcm) after low electron energy beam

irradiation (LEEBI).

Utilizing LEEBI treated Mg doped p-type GaN and intrinsic n-type GaN, Amano

fabricated the first UV-LED [Ama89]. Nakamura later established that a simple

annealing in inert or vacuum environment will improve p-type conductivity similar to the

LEEBI treatment [Nak92]. The reason of great p-type conductivity improvement by

LEEBI or annealing was not clear then, but it was later discovered that an Mg-H neutral

species prevents Mg activation [Got95].

GaN-based Field Effect Transistors (FET) [Kha93] and Heterojunction Bipolar

Transistors (HBT) [Pan94] were also fabricated in the early 1990s, enabled by the

improved crystal quality and conductivity.

The threading dislocation density of these transistors, however, was about 109 to

1010 cm-2, or about 106 times higher than typical semiconductors. Although devices

could be fabricated with the high density of dislocations, their long-term reliability was

questioned.

In 1994, a significant reduction of the dislocation density was achieved by adopting

lateral epitaxial overgrowth that used either a SiO2 or Si3N4 mask [Kat94]. Because









threading dislocations tend to form parallel to the growth axis, vertical blocking of the

threading dislocations improved crystal quality drastically (from 109 1010 to 104 105

cm-2).

GaN-based devices began to be fabricated more routinely beginning in 1994. It is

worth mentioning that Nakamura's two-flow reactor design significantly improved

crystal quality [Nak94]. In this process, inert or hydrogen gas is injected vertically onto

the substrate to modulate local concentrations of precursors by reducing the boundary

layer thickness. Following the invention and breakthroughs, Nakamura reported an

InGaN based 1 cd LED in 1994 [Nak94], a 10 cd LED in 1995 [Nak95a, Nak95b], a laser

diode in 1996 [Nak96a], and continuous wave (CW) lasing at room temperature in 1997

[Nak97]. The intensity and life time improved in consecutive years and Nichia

Corporation reported estimated 10,000 hr life time of blue laser diodes in 1998 [Nak98].

Research related to GaN has continued, as a result of the major breakthroughs made

during the 80s and 90s. The lack of a suitable substrate material continues to create

problems and the growth of large (> 2") diameter freestanding GaN has not been easily

achieved due to cracking and bowing problems. Also the heat generated from LED

devices is still problematic due to the low thermal conductivity of the sapphire substrate.

Undoubtedly, the integration of GaN technology with Si would be another important step

for GaN-based device development. Furthermore the use of 12" diameter Si wafer would

improve device throughput and lower unit device fabrication cost.

1.1.2 InN Development

The first report on InN crystal growth originates from 1938 and used InF6(NH4)3

for crystallographic study just as for GaN growth [Juz38]. The powder X-ray diffraction

file for InN was collected and is in current use as PDF#02-1450. Several reports are









found from the 1950s and 1960s about synthesis of InN [Juz56, Ren58, Pas63, Sam69].

The growth methods were mainly interaction of In compounds with ammonia or thermal

decomposition of complex single precursors containing direct bonding between In and N.

InN powder or small crystals were usually obtained as a result. In 1972, Hovel and

Cuomo [Hov72] deposited InN films on sapphire and silicon by reactive RF sputtering in

the growth temperature range 25 to 600 OC. Hall mobility, n-type carrier concentration,

and resistivity were measured as 250 + 50 cm2/Vs, 5 to 8 x 1018 cm-3, and 3 to 5 x 10-3

Qcm, respectively. Trainor and Rose grew InN by reactive evaporation and a bandgap of

1.7 eV was measured [Tra74]. Osamura reported the entire range of InGaN by an

electron beam plasma technique that used an electron beam to heat and evaporate In and

Ga and a plasma to create a nitrogen dc discharge [Osa72, Osa75]. The grown InN

showed bandgap energy of 1.95 eV at room temperature and 2.11 eV at 78 K.

Puychevrier and Menoret [Puy76] grew InN by reactive cathodic sputtering and the

bandgap of InN was measure as 2.07 eV at room temperature and 2.21 eV at 77 K.

Then, in 1984, Tansley and Foley [Tan84] reported RF sputtering growth of InN

with N2. Very high Hall mobility at room temperature and very low background carrier

concentration were measured as 2700 cm2/Vs and 5.3 x 1016 cm-3, respectively. It should

be noted that even with today's advanced techniques of MOVPE and MBE, these values

are not easy to obtain. In addition, there is no other report of InN grown by RF sputtering

that demonstrated such high Hall mobility and low carrier concentration. The typical

mobility and carrier concentration of InN grown by sputtering is 100 cm2/Vs and 1020

1/cm3, respectively. The problem of Hall measurement is that sometimes it can give









inconsistent values when the film is not uniformly deposited. However, the reason is

unclear at this point.

Epitaxy of InN by HVPE was first reported in 1977 [Mar77]. Indium trichloride

powder was used as the indium source and ammonia was used as the nitrogen source.

The obtained epitaxial InN film showed background carrier concentration and Hall

mobility as 2 x 1020 to 8 x 1021 cm-3 and 30 to 50 cm2/Vs, respectively. The optimal

growth temperature was found to be 600 C and there was no growth at temperature

higher than 670 C. Indium trichloride powder [Tak97b, Mar77, Sun96], indium

monochloride, or both form in situ by flowing HC1 over In metal depending on the source

temperature [Tak97a, Sat94] grew InN by HVPE using predominantly InC13, or InCl and

showed ambiguous results on the superior precursors.

Kang reported successful H-MOVPE growth of InN in a chlorinating environment

[Kan04]. In this approach, the conditions for selective etching condition of In metal in

addition to protecting InN film were found by thermodynamic analysis and experiments.

As a result growth of InN film without evidence of In metal droplets was achieved at a

relatively low (2500) V/III ratio. During this study, columnar structured InN crystals

were observed. It demonstrated the possibility of InN nanorods growth and greatly

influenced the present study.

MOVPE and MBE have been widely used to grow InN since the late 1980s. Single

crystal InN growth by MOVPE was reported using trimethylindium (TMIn) and NH3 as

the precursors [Mat89, Wak89, Wak90]. However, high Hall mobility (i.e., > 300

cm2/Vs) and low background carrier concentration (i.e., < 1019 cm-3) were not achieved

until recent years.









InN was also grown by MOMBE using TMIn instead of pure In metal source

[Abe93]. Electron Cyclotron Resonance (ECR) was used to generate a nitrogen plasma.

The obtained Hall mobility and electron carrier concentration were 100 cm2/Vs and 1020

cm-3, respectively. Aderhold [AdeOl] reported improved InN film quality by MOMBE.

Hall mobility and background carrier concentration were 500 cm2/Vs and 8.8 x 1018 cm-3,

respectively. The reported Hall mobility and carrier concentration were plotted in Figure

1-1 with recent MOVPE data [Bhu03, Yam02, Yam04, Yam06]. Hall mobility and the

carrier concentration of MOVPE grown InN were 1100 cm2/Vs and 4.5 x 1018 cm-3,

respectively [Yam06].

SMBE O MOVE : ME Q MOYPE
0: HPrE : Sptuer A On 0l t | VE Spellr A: Otmr


0 1 0 1 9


01 l4 Il I0

S1 016

1970 1980 1990 2000 2010
1970 1980 1990 2000 2010
Calendaryar Catendaryear
Figure 1-1. Reported values of electric properties of InN (a) Hall mobility and (b) Carrier
concentration with calendar year [Bhu03, Yam02, Yam04, Yam06].

Interest in InN has increased since 2002 due to the ambiguous fundamental energy

bandgap. The reported bandgap energy value of InN has a wide spread: 0.7 [InuOl,

Dav02, Wu02, Sug03, Bri04, But05], 1.1 eV [InuOl, This work], and ~ 1.9 eV [Hov72,

Tya77, Nat80, Tan86, Wes88, Sul88, But02, Mot02, Had03] are commonly reported

values. Figure 1-2 shows some of the InN bandgap data representing the spread of the

values [But05b].










The reason for the controversial bandgap is not well established, although many

believe the inclusion of oxygen (In203 or any form) may be responsible. High crystalline

quality InN is therefore essential to determine the controversial property and clarify this

ambiguity.


Wu etal. [WuO2b]
o Haddad et al [Had03]
V lffe Institute [Inu01, Dav02a]
V Bio elaL Bri04]
Tr3sley ard Foley [Tan86]
0 O1her RF sputter [Hov72. MotC2c, ButO2]
DC sputtering [Tya77, NatBO, Wes88, SulBS]
O Butcher el al. [Bul5a]
A Sugita et al. ISuggD3
2.4 theoretical values of [Wal04]

2.2 o
2.0, o 0
2.0 0 0

S1.6
LU 1.4
S1.2 >1019 cm-3 carrier concentration
w a strong Moss-Burstein effect is o
1.0 expected
0.8
!--- ---
0.6
0.4
1016 1 07 1018 1019 10' 1021

Carrier Concentralion (cm-3)
Figure 1-2. Reported bandgap energy of InN as a function of carrier concentration
[But05b].

The best reported InN films grown by MOVPE demonstrated Hall mobility and

background doping as 1100 cm2/Vs and 4.5 x 108 cm-3, respectively [Yam06]. MBE

grown InN films showed better values of Hall mobility and carrier concentration as 2050

cm2/Vs and 3.49 x 1017 cm-3, respectively [Lu02a].









InN-based electronic devices were fabricated and preliminary evidence of two-

dimensional electron gas (2DEG) in a field effect transistor (FET) was reported [Sch00,

Lu02b]. Problems growing high quality InN overlap with those for GaN in that lattice

and thermal expansion matched substrates are absent and that efficient doping, especially

p-type, is difficult.

The growth of InN, however, is more challenging because of the low growth

temperature and the formation of In metal droplets. The growth temperature and V/III

ratio seem to be the most important growth parameters. Yamamoto investigated the

effects of growth temperature and V/III ratio [Yam01]. They found that V/III ratio could

be lowered at higher temperature due to the enhancement of the active nitrogen

concentration. The suggested growth temperature was 550 to 650 C. The Hall mobility

and carrier concentration were reported as 200 to 400 cm2/Vs and 1018 to 1020 cm-3,

respectively [Yam01b, Yam02, Yam04]. The Hall mobility and carrier concentration are

improving with increasing temperature up to 600 OC and then showed worse results at

higher temperatures.

To enhance the decomposition of NH3, several techniques have been used. It is

well known that the reactor geometry is very important to grow high quality material

primarily for MOVPE, as demonstrated by improvement of GaN quality by the two-flow

design [Nak91]. Flow modulation [Kel00] was used to leverage the advantages of

Atomic Layer Deposition (ALD). TMIn and NH3 were provided in turn, or NH3 was

kept flowing and TMIn was provide in a pulsed manner. Plasma-assisted MBE and

MOVPE [Wan06, Che06, Wu06, Wak89, Wak90, Sat97] and laser-assisted MOVPE









[Yam06, Li94, Bhu02, Bhu03] were also utilized to enhance ammonia decomposition and

showed some promising results.

This overview of GaN and InN thin film growth and the corresponding electrical

and optical properties demonstrates the advances that have been made and their potential

commercial applications. It is also clear that GaN has received considerably more study

than InN.

1.2 Literature Review

1.2.1 Equilibrium Analysis of GaN and InN

1.2.1.1 Thermochemical Data for GaN and InN

Studying phase stability is essential for predicting process limits because

semiconductors grown at high temperature often experience thermal annealing processes

for metallization, remedying implantation damage, or dopant activation. In addition,

GaN, A1N and their alloy have applications for high temperature devices thanks to their

high temperature stability.

Equilibrium calculations can provide preliminary information on appropriate

processing conditions. These calculations require thermodynamic property values for all

species and phases. Unfortunately, the properties for Ga-In-N system are not that well

defined, especially the standard enthalpy of formation. The reported heats of formations

of GaN and InN are quite scattered [Sed06]. The reported heat of formation of GaN

ranges from -109.62 [Hah40] to -156.8 kJ/mole [RanOO] and that of InN from -71.0

[Lei04] to 138.072 [Bin02].

Unland and coworkers recently reported thermochemical data for GaN powder and

measured the decomposition temperature as 1110 + 10 K by dynamic oscillation

Thermogravimetric Analysis (TGA) and isothermal stepping TGA [Unl03]. Based on









their assessment relative to other reported values, the experimental data from Unland et

al. [Unl03] were adopted in this study for the GaN-base calculations.

A similar method (TGA) was applied to InN and the AH0 (298.15K),

So (298.15K), Cp, and decomposition temperature of InN were determined [Ond02].

The equilibrium decomposition temperature may be lower than the experimentally

observed temperature because of the kinetic barriers; however, the equilibrium

decomposition temperature cannot be higher than experimental data unless there is

measurement error. The InN decomposition temperature (773 K) as measured with TGA

by Onderka [Ond02] is higher than Leitner's drop calorimetry value (686 K) [Lei04].

Considerable uncertainty, however, still exists in the thermochemical data for InN. The

data from SUB94 and [Lei04] will be cautiously used for InN in this study since

rigorously assessed data is still not available. The reported thermochemical values for

GaN and InN are tabulated in Table 1-1.

1.2.1.2 Equilibrium Calculations of GaN Growth by HVPE

Koukitu [Kou98] performed chemical equilibrium calculations analytically by

considering only a limited number of species. They calculated partial pressures of the

gaseous species in equilibrium with GaN during HVPE growth with respect to

temperature, input GaCl partial pressure, input V/III ratio, and H2/inert ratio. By

comparing the equilibrium of Ga containing gas phase concentration with input GaCl

partial pressure, the driving force for the deposition was calculated with the definition:

Ga PGaCI (PGaCI + GaC) (1)

The factor that represents the ratio of the H2 carrier with the inert carrier was defined as F

and was given as:










F = 1/2(2PH +3PNH3 + PHC )
7 +3/2P 1/2P 7 (2)
PH2 +3/2PNH3 +1/2PHcI +PIG

It should be noted that a way of dealing with NH3 partial decomposition was introduced

here with a factor a such as:

NH3(g) (1-a)NH3(g) + a/2N2(g) + 3a/2H2(g) (3)

where a is the mole fraction of the decomposed NH3. The value of a was measured as a

= 0.03 at 950 C by mass spectroscopy [Ban72], but is a strong function of temperature.

Figure 1-3 (a) shows plots of the driving force for the deposition with temperature

with different F values [Kou98]. It was observed that APGa (the driving force) decreased

with increasing F.

(a) EMPi:l.Oati P'ha:d.OxlO atm V/Ml:50
S F0.0 1.0 :0.03 (b)

5100x10
100 In experimemal lWusIur
calculated |Ko-98


F0.0 5
0.1 o0
0.24
0.6 40 Team. :1000'C
t.o P0 m: 0.26 atm
20 F:110
,: 0.03

1O.0lo -- -- 2.0xo10 4.00xx10 6.0 8.0~xt0 1.OXIO 1.2x10U
500 600 700 800 900 100 0 1 1200 n = e GC ,.
Tcmpraturc CC) Input partial pressure of GaCL (atm)

Figure 1-3. Examples of equilibrium calculations. (a) Driving force for the deposition as
a function of growth temperature with various parameters for F. Total
pressure: 1.0 atm, input partial pressure of GaCl: 5 x 10-3 atm, input V/III
ratio: 50 and a: 0.03. (b) Comparison between calculated growth rates
[Kou98] and experimental data [Usu97].

The growth rate with driving force for the deposition was defined as the following

formula:











Table 1-1. Reported values of thermochemical data for solid GaN and InN.

Material AH) (298.15K) Method S (298.15K) Method Tdecomp (K)
(kJ/mole) (J/mole K)

GaN -109.62 [Hah40] Combustion calorimetry
920 [SUB94]
-156.8 [RanOO] Drop calorimetry
1052 [DavOl]
1435 [Lei03]
-156.8 16 [Unl03] TGA 30 4 [Unl03] Debye-Einstein 1110 + 10 [Unl03]
-161.56 [Sed06] DFT calc 36.1 [Sed06] DFT calc

InN -144.6 [Mac70] High pressure equilibria
-133.8 6.3 [Vor71] Knudsen effusion MS 53.7 7.1 [Vor71]
-137.2 18.8 [Vor73] Knudsen effusion MS 32.3 20.9 [Vor73]
-130.6 [Gor77] Knudsen effusion MS 51.6 [Gor77]
54.4 8.0 [Hon87]
-132.7 6.7 [Jon87] Static pressure measurement 1211 [SUB94]
-138.07 [Bin02]
31.6 3 [Ond02] Debye-Einstein 773 5 [Ond02]
-71.0 [Lei04] Drop calorimetry 42.51 [Lei04] PDOS/assessed 638 [Lei04]
-78.64 [Sed06] DFT calc 42.5 [Sed06] DFT calc










r = KgAPGa (4)


where r = growth rate, and Kg is the mass transfer coefficient.

By adjusting the mass transfer coefficient (i.e., Kg = 1.18 x 105 ipm/h atm in this case), the

growth rate results matched well with experimental data as seen in Figure 1-3 (b). Thus,

it was concluded that GaN growth by HVPE is thermodynamically controlled under these

ranges of growth conditions.

Following similar procedures, Kumagai analyzed HVPE growth of InN by InCl and

InC3 precursors [Kum01]. The results showed that InN growth is difficult using InCl but

possible with InC13 along with inert or low H2/inert carrier gas as shown in Figure 1-4

[Kum01]. It could be seen that the driving force has negative values in an H2 carrier (F =

1.0), while it has slightly positive values in an inert carrier (F = 0.0).

InN HVPE and THVPE
XPi: 1.0 atm, PinCi or Plncl,: 5.0x10"3 atm,
VIIII: 50, F: 0.0 1.0, a: 0.0
0.005

THVPE, F: 0.0

HVPE, F: 0.0
HVPE, F: 1.0


0-

-0.005




/-0.THVPE, F: 1.0
-0.010
500 600 700 800 900
Temperature (*C)
Figure 1-4. Driving force for the deposition of InN (APIn) using InCl (HVPE) and InC13
(THVPE) as a function of growth temperature. The calculation was
performed for growth under inter gas (F = 0.0) and hydrogen carrier gas (F =
1.0) conditions. [KumOl].









It should be noted that no report was found that dealt with adjustable Cl/III ratio,

but only at fixed values since either monochloride (Cl/III = 1) or trichloride (Cl/III = 3)

was used in HVPE and THVPE techniques. Continuous variation of Cl/III ratio is

possible in H-MOVPE by simply adding an independent stream of a Cl containing

species (i.e., HC1). Thermodynamic analysis of H-MOVPE technique including various

Cl/III ratios is presented in Chapter 2.

1.2.2 Stress in GaN Films

Stress measurements and modeling of GaN films is important because GaN is

usually grown on foreign substrates due to the lack of bulk GaN crystals. The traditional

approach to analyze the stress in GaN films was not satisfactory because there were some

cases when the expected stress occurred in an exactly opposite manner. For example,

both compressive and tensile stressed GaN can be grown on SiC [Per97, Dav97, Dew98,

Rie96, Wal99]. Therefore, it could be concluded that the stress in the GaN film is not

only related to the lattice and thermal expansion mismatch, since both give stress in GaN,

but also to other important factors.

The stress in GaN is mainly related to the lattice and thermal expansion mismatches

with the substrate and the initial surface properties, which result in a growth mode

change. The origins of tensile stress were identified as lattice and thermal expansion

mismatches between GaN and substrate, Si or Mg doping, or grain boundaries of island

coalescence [Kro03]. Likewise, the origin of compressive stress can be lattice and

thermal expansion mismatches between GaN and substrate. The island growth mode

should be avoided as it often creates polycrystalline materials. A buffer layer is usually

used to enhance crystal quality by uniform coverage of the surface to prevent island

growth.









The traditional approach to modeling the residual stress in GaN film by lattice and

thermal expansion mismatches has a flaw because GaN films are typically grown on

buffer layers or by a nitridation process. The pregrowth processes change the surface

properties significantly. The buffer layers are usually amorphous and the nitridated

surface of sapphire is also an amorphous AlOxNy layer. Therefore, the traditional

heteroepitaxy modeling such as Frank-van der Merwe and Matthews [Mer50, Mat75]

cannot be applied since the existence of an amorphous interface was not considered. In

addition, the lattice mismatch (i.e., ~ 14 % for sapphire, 21 % for Si, or even ~ 3.5 %

for SiC) is very large; dislocations will form at the interface immediately.

A visible product of the incorporation of significant stress in a film is cracking.

Etzkorn and Clarke investigated the cracking of thick GaN films on sapphire substrate

[EtzOl]. A viable mechanism for cracking was identified as island coalescence. A

model was developed to deduce the maximum thickness of crack formation with given

stress. According to the model, 16 |tm thick crack-free GaN could be grown on sapphire

with tensile stress in 0.14 GPa. Strain energy calculations related to crack generations

were carried out by measuring the bending moment of GaN and sapphire. It was argued

that tensile stress in GaN film was generated at the growth temperature after reaching a

critical thickness. Kinetically limited crack healing mechanism was suggested after crack

generation.

Waltereit reported that GaN films grown directly on 6H-SiC showed no stress,

while GaN films grown on A1N/6H-SiC showed compressive stress, although the lattice

parameters and thermal expansion coefficients of AIN and SiC are similar [Wal99]. The

lattice mismatch induced 3.4 % compressive stress was fully relieved in GaN directly









grown on SiC, whereas 0.3 % compressive stress still remained in GaN grown on

AIN/SiC even beyond 1 pm thick. It was concluded that the stress in GaN film was

mainly determined by its growth mode rather than lattice mismatch in this case. The

growth mode is mainly related to the initial surface properties of the substrate. The

results suggested that the advantage of the buffer layer is not only in averaging the lattice

and thermal mismatches, but also enhancing the surface property. Schematic of GaN

directly grown on 6H-SiC and using A1N buffer layer is shown in Figure 1-5.



/Ga1 GT N /G-
-I ---\ ----------I--------I-------
T-- (G AIN T

(a) 6H-SiC(0001) fb) 6H-SiC(0001)
Figure 1-5. Schematic of GaN growth on SiC: (a) GaN/SiC and (b) GaN/AlN/SiC
growth. Note that the strain is fully relieved in (a), whereas it is only partially
relaxed in (b) [Wal99].

Three dimensional growth (Volmer-Weber) governed the GaN growth without buffer

layer and a number of dislocations were shown, while two dimensional growth (Frank-

van der Merwe) can occur with few dislocations with A1N buffer layer [Wal99].

1.2.2.1 In situ Stress Measurements

To understand the stress evolution in GaN films, in situ stress measurements have

been carried out using a Multibeam Optical Stress Sensor (MOSS) system [Flo97,

Hea99]. This method determines the wafer curvature by measuring a laser beam

deflection, which is proportional to the stress. It was used to in situ measure the stress in

AlGaN film grown on AIN/SiC [Aco04]. They found that AlGaN film was initially

under compressive stress and evolved into tensile stress as the film thickness increased.

The transition (from compressive to tensile stress) thickness of AlGaN film depended on









the A1N buffer layer growth conditions and AlGaN film Al mole fraction. For example,

when V/III ratio was varied from 750 to 10600, the AlGaN initial compressive stress was

varied from 1.9 to 8.7 GPa. GaN film did not show a transition to tensile stress up to 3.8

jtm thick, whereas more Al content AlGaN showed rapid transition to tensile stress.

The persistent tensile stress in MOCVD GaN on sapphire using both LT-A1N and LT-

GaN buffer layers were observed by MOSS at growth temperature [Hea99]. The origin

of tensile stress may be from the grain boundaries of islands. It was also found that

thermal annealing or temperature cycling does not reduce tensile stress in the film.

Because the thermal expansion mismatch effect is much larger, eventually the GaN film

on sapphire showed compressive stress after cooling.

Raghavan and Redwing [Rag04] measured the persistent tensile stress in A1N

grown on Si in situ across the wide temperature range 600 to 1100 C. A sharp drop in

the tensile stress was observed from 1 to 0.4 GPa while decreasing the A1N growth

temperature below 800 C. This represented the transition of epitaxial A1N film to

polycrystalline. GaN film was consecutively grown on AIN/Si. The GaN film showed

slightly compressive stress at 1100 C.

Krost et al. measured the stress in the GaN films on Si( 11) in situ and successfully

grew crack-free 7 gm thick GaN on Si( 11) with multiple A1N interlayers [Kro05,

Dad04, Clo04, Kro03]. The sources of tensile stress were identified as grain boundaries

from island coalescence (0.2 GPa/gm) and Si-doping (1.6 GPa/gm) [Kro05]. The non-

uniformly deposited SiN was called an "in situ mask" and high quality GaN could be

obtained on the in situ SiN mask/AlN/Si. The in situ curvature measurement results for

GaN grown on SiN/AlN/Si are shown in Figure 1-6 [Kro03].






18


0.0 -__.- wA.._ A change in curvature:

-05- SN LT-AIN
E LT-AIN t

S-1.0- LT AIN

3 -1.5-1 i ^(|b

LT-AIN
-2.0-

0 50 100 150 200
Time (min)
Figure 1-6. In situ curvature measurements during the growth of a 6 pm thick GaN
layer showing the influence of AIN interlayers on the curvature. The sample
was crack-free after growth [Kro03].

Evolution of the curvature was observed with inclusion of LT-AIN interlayers. When

LT-A1N was applied, a temporary compressive stress was observed as shown in Figure 1-

6. Nevertheless, the film was under persistent tensile stress as the thickness increased.

Based on the results presented in Figure 1-6, it is evident that AIN has an impact for only

a small amount of subsequent growth. Although this AIN interlayers may improve the

crack-free thickness, the growth of thick (> 100 [tm) crack-free GaN in this way should

have inherent limit.

1.2.2.2 Ex situ Stress Measurements

The surface of HVPE and H-MOVPE grown GaN is often very rough. Commonly

used non-destructive methods to measure the stress in rough films or the films that have

structures on them are XRD reciprocal space mapping and Raman spectroscopy. In this

case, the stress measurement is taken after the sample has cooled to room temperature.

XRD Reciprocal Space Mapping. The lattice parameters of a film can be precisely

measured using an XRD reciprocal space mapping. It was observed that GaN showed

strong in-plane strain although out-of-plane strain was negligible [Eip05]. Considerable









thermal compressive stress was found from GaN grown by ion beam assisted MBE on y-

LiAlO2 (100). It was found that there was 3 times greater compressive stress in the in-

plane than out-of-plane during measurement by HT-XRD in the temperature range 25 to

600 C (HT-XRD chamber temperature) [Eip05]. Stress in GaN was also measured on

ZnO substrate by HR-XRD [Min04]. In accordance with the in situ stress measurement

results, the change of buffer layer thickness changed the film stress significantly. Thus,

buffer layer material should be carefully selected and the thickness of the buffer layer

should be controlled to reduce stress in the final film.

Raman Spectroscopy. Raman spectroscopy is a well-established, non-destructive, and

convenient technique to detect the stress in the film, although the typical accuracy of

stress measured by Raman spectroscopy compared to 5 % for HR-XRD is about 20 %

[Dav97]. Among the all allowed Raman modes, it is known that the E2 mode is strongly

related to biaxial stress in GaN [Dav97]. Between the two E2 peaks, the higher frequency

mode is usually used due to the higher intensity. Biaxial stress in GaN sample can be

measured by comparing the position of E2(high) peak with that of a stress-free sample.

Table 1-2 lists the Raman E2(high) peak positions reported in the literature [Dav97,

Mel97, Gil00, Kry99a, Dav97, Yam00, Dem96, Yoo97, Ben02]. It is noticed that even

freestanding or high pressure grown bulk GaN samples do not have unanimous Raman

E2(high) peak that can be used for stress-free standard. The E2 mode of freestanding GaN

ranged from 565 to 570 cm-1. Therefore, deciding the value of stress-free E2(high) peak

position is impossible at this point. Generally 568 cm- is considered as the stress-free

GaN E2(high) peak position because it is about average value from freestanding GaN

samples. Therefore the stress in GaN can be determined by comparing the E2 value with









568 cm-1. If the measured E2 frequency is higher than 568 cm-1, the film is under

compressive stress qualitatively, and vice versa.

Table 1-2. Reported values of the Raman E2(high) peak positions.
Sampe dn Raman E2(high) peak Reference
Sample description .-1 Reference
position (cm )
Freestanding (300 pm) by HVPE
Substrate: A1203
Freestanding (100 im) by HVPE
565 [Me197]
Substrate: SiC, no buffer
Freestanding by high N2 pressure 565 [Gil00]
Freestanding on LiGaO2 by H-MOVPE 570 [Kry99a]
0.5 to 3 um GaN/AlN(AlGaN)/SiC 565 [Dav97]
1.8 im on AIN/A1203 by MOCVD 566 [Yam00]
V/III (2000 to 6000) MOCVD 569 (V/ 6000) to [Dem96]
571 (V/III= 2000)
1.5 Lm GaN/(10 to 85 nm)LT-GaN/c- 565 to 572 [Yoo97]
565 to 572 [Yoo97]
A1203 by MBE
GaN/misoriented A1203 by MOCVD 570 to 571 [Ben02]

The quantitative value of stress can be obtained using a linear proportionality

factor. This quantitative argument may be meaningless in some sense since the stress-

free value is unknown. However, two samples can be quantitatively compared by

assuming one sample is stress-free. The reported value of this proportionality factor

ranges from 2.4 to 4.2 cm-'/GPa [Dav97, Kis96]. Thus, further study is required to

conclusively determine the stress from Raman peak shift quantitatively.

One may expect compressive stress to be present in GaN grown on A1203 because

the lattice and thermal expansion mismatches leads to compressive stress. However, it is

not straightforward to predict whether the overgrown GaN will have compressive or

tensile stress. The Raman spectroscopy results showed that the grown GaN film on c-

A1203 is less compressively stressed with increasing V/III ratio [Dem96].










Moreover, the Raman E2 peak significantly shifted from 572.0 compressivee) to 565.2

cm-1 (tensile) while changing the buffer layer thickness from 10 to 85 nm as shown in

Figure 1-7 [Yoo97].




1000 4-7









.1o uon /on so
20 400 600 goo
Raman Shift (cm"')
Figure 1-7. Raman peak shifts of 2 pm thick GaN with different LT-GaN buffer layer
thickness; LT-GaN buffer layer thickness (a) 10 nm, (b) 50 nm, (c) 75 nm,
and (d) 85 nm. [Yoo97].

This tensile to compressive stress change is expected because uniform coverage of the

surface at an optimal buffer layer thickness will lead to compressive stress from 2-D

growth rather than 3-D island growth that create tensile stress at the grain boundary.

Since Raman spectroscopy is sensitive to the stress in the GaN film, it was used as

a major technique to measure the stress in the GaN films grown on sapphire substrate and

the results will be presented in Chapter 3.

1.2.3 GaN Growth on Si

One of the advantages of using Si substrate instead of sapphire is its availability in

large diameters. Another advantage of Si is that the heat generated from the device can

more easily drain because Si has much higher thermal conductivity (145 W/m K) than

sapphire (5.43 W/m K). GaN films are usually grown on Si (111) substrate because the

hexagonal geometry (i.e., three-fold symmetry) of the Si (111) surface is better matched









with the basal plane of GaN. The growth of GaN on Si (100) would be more challenging

due to the different crystallographic structures. For example, it was often observed that a

mixture of cubic and hexagonal GaN grown on Si (100) substrate by MBE [As00]. GaN

grown on Si (100) shows multiple in-plane alignments mainly composed of two in-plane

domains rotated by 300 [Leb00, Sch04b].

GaN is usually grown by MOCVD on Si since it gives the most promising results

and is suitable for mass production. HVPE can also be used to grow GaN on Si. The

report of HVPE GaN growth on Si is, however, limited [Mas06, Zha05, Mes05, Yu05].

A maximum crack-free thickness of HVPE GaN film grown on Si is 1.5 [pm on 2 inch

Si( 11) [Zha05]. The results suggested that the crack generation is not much related to

growth chemistry, growth rate, or reactor geometry. Instead, it is more related to lattice

mismatch and thermal expansion coefficient mismatch among GaN, buffer layer, and Si.

The driving force for crack is the strong tensile stress in the film caused by lattice

and thermal expansion differences between GaN and Si. The origin of tensile stress

within GaN film on Si substrate is not only from lattice and thermal expansion

mismatches, but also from island coalescence boundaries due to the initial growth from

small crystallites [Dad03]. Along with the brittleness of Si, sometimes cracks can

penetrate into Si over 1 [tm depth [Zam02].

A number of techniques such as A1N buffer layer [Dad03, LiaOO, Sch05], AIN/SiC

buffer [LiaOO], AlGaN buffer layer [Wan05], multiple A1N interlayers [Kro03], AlGaN

graded layer [Mar01, Abl05], AlGaN/HT-AIN [Sch05], AlGaN/GaN superlattice [Lia01],

Si02 patterning [DetOl, ZamOl, Zam02, Nao03], and Si doping in GaN [Yu06] were









used to reduce the crack density in GaN on Si. Table 1-3 lists some of the techniques

used for crack-free GaN growth on Si.

Table 1-3. Techniques used for growth of crack-free GaN on Si.
Crack reduction technique for GaN grown on Si Growth Crack-free thickness Ref
T (C) (Crack-free area)
AIN/SiC buffer layer 1050 1.3 to 1.5 .im [Lia00]
AlGaN/GaN superlattice on AIN/SiC/Si (10 cm diameter) [Lia0l]
Multiple A1N interlayer on A1N buffer layer 1145 7 [im [Kro03]
with in situ curvature measurement
AlGaN graded layer on HT(1160 C)A1N 1050 2 [m [Abl05]
(100 nm) buffer layer
In situ SiN mask (1.5 mono layer) on Si 1165 0.75 [m [Dad03]
doped A1N buffer layer(25 nm)
Al0.58Ga0.42N buffer layer (180 nm) at 800 OC 1000 0.4 tm [WanO5]
AlGaN graded layer on A1N buffer layer (10 1050 0.7 to 1.6 pm [Mar01]
to 500 nm)
SiO2 patterning 1100 9 im [DetO1]
(5 lm x 10 [m)
10 pm
(100 m x 100 pm)
Lateral confined epitaxy (prepatterned Si) 1020 0.7 [m [Zam02]
(100 m x 100 um)
Si doped GaN on A1N buffer layer (100 nm) 1000 < 1 pm [Yu06]

Crack-free 1.3 to 1.5 ptm GaN was successfully grown on 10 cm diameter Si (111)

using A1N and SiC buffer layers [LiaOO]. Further study showed that high crystal quality

and smooth surface were achieved by insertion of AlGaN/GaN superlattice interlayers

[Lia01]. The results showed that large diameter GaN growth on Si is promising.

The typical crack-free thickness of GaN grown on large area Si is < 2 [im except

one report showed 7 pm crack-free GaN grown on Si [Kro03]. However, the growth of

large area (> 2" diameter) freestanding (> 100 [im thick) GaN on Si was not reported to

date.

An A1N buffer layer between GaN and Si is most widely used because the thermal

expansion difference between A1N and Si is less than GaN and Si. In addition, GaN









films on A1N will possess compressive stress during cooling that may compensate for the

tensile stress.

Unlike many reports of GaN growth on sapphire that uses either LT-A1N or LT-

GaN as a buffer layer, reports of LT-GaN buffer layer usage of GaN on Si were rarely

found (i.e., [Mas06b]). The benefit of LT-GaN buffer layer on sapphire may be the

change of the surface property resulting in two-dimensional growth. On the other hand,

GaN on Si is under significant tensile stress. Therefore, changing the surface property is

necessary, and, in addition, artificial compressive stress generation is required to grow

crack-free GaN film on Si. It is evident that A1N and AlGaN are promising buffer layer

materials between GaN and Si due to the compensation of tensile and compressive stress.

It was found that the higher Al concentration in AlGaN buffer layer is better for

subsequent GaN growth [Wan05].

Table 1-4 lists the structural properties of various substrates for GaN epitaxy. It

can be predicted that what kind of stress the GaN film will be under by comparing the

lattice parameter and thermal expansion coefficient. Tensile and compressive stress

induced by lattice mismatch and thermal expansion mismatch are drawn in Figure 3-1.

The other widely used technique is the growth of an AlGaN graded layer. This

method creates gradual changes in the stress in GaN films. In addition, AlGaN graded

layers reduce threading dislocations significantly [Mar01].

Other than stress engineering by buffer and interlayers, SiO2 patterning was also

used to reduce cracking problems [DetOl, Nao03, ZamOl, Zam02]. Adopted from

Epitaxial Lateral Overgrowth (ELO), grooved Si substrate was used as the substrate and a

thin GaN layer was grown, followed by SiO2 patterning [DetO 1]. The growth of GaN on









prepatterned Si, termed Lateral Confined Epitaxy (LCE) [ZamOl] was also studied.

However, this technique did not create more promising results since GaN films tend to

bend upwards (concave), although a crack-free region of 100 pm x 100 pm was reported

[Zam02].

Table 1-4. Structural Properties of Various Substrates for GaN Epitaxy [Kan04].
Substrate Eg Lattice Lattice Lattice Thermal
(eV) parameter parameter c mismatch expansion
a (A) (A) with GaN coefficient
(%) (10-6K-1)
a c
GaN wurtzite 3.36 3.189 5.18 0 5.6 3.17
InN- wurtzite 0.7 3.537 5.704 9.8 4.0 3.0
AIN wurtzite 6.2 3.112 4.98 2.5 4.2 5.3
6H-SiC wurtzite 2.9 3.08 15.1 3.5 4.2 4.7
A1203 wurtzite 6.8 4.758 12.991 16.1 7.5 8.5
ZnO wurtzite 3.35 3.21 5.21 1.94 2.9 4.8
LiA102- tetragonal 6.1 5.1687 6.2679 1.5 7.1 7.5
LiGaO2 orthorhombic 4.1 5.402 5.007 0.9 6 7
3C-SiC diamond 2.3 4.36 3.9 2.7
GaAs zinc blende 1.42 5.65 19.87 6.0
Si (111)- cubic 1.12 5.83 17.0 6.2
BP zinc blende 2.0 4.54 0.63 _

Another way of reducing cracks would involve creating weak cohesion between

GaN and Si, so that the film can separate during the cooling process. For example, ion

implantation (N ) was carried out to intentionally create defective layer of AIN/Si

substrate [Jam05]. This partially isolated the III-nitride layer and Si substrates and

helped reduce the strain by up to 84 %. A novel approach of creating weak cohesion

between GaN and Si using InN as buffer material will be presented in Chapter 4.

1.2.4 Growth of InN Films and Nanostructured Materials

1.2.4.1 Growth of InN Films

InN is the least studied material among the group III nitrides that are relevant to

UV-visible-IR optoelectronics. InN has high drift velocity at room temperature, which is









a good characteristic for high speed FET applications [Ole98]. In addition, tandem solar

cells with Si at the bottom cell and a higher bandgap material (i.e., InxGal-xN 1.85 eV) on

the top cell can achieve a maximum efficiency of 32.1 % [Mat87].

Efforts have been made to grow InN crystals by MOCVD [Wak90, Sat97, Guo99,

Yam94, Yam97, Yam99, Pan99, Yan02, Dra06, Mal04, Cha04, Sin04, Yam05, Yam04,

Jai04, Los04, Yam05, Mal05, Hua05, Kel00, YamOl, Sug03], MBE [Abe93, Dav99,

LuOl], HVPE [Iga92], laser assisted CVD [Bu93], as well as other less used processes.

Most InN films were grown by MOCVD because MOCVD is a well-established

technique especially when mass production is considered.

The growth of high quality InN, however, was impeded by several factors. Low (<

7000C) growth temperature is required due to the very high nitrogen vapor pressure and

strong tendency to form In metal droplets. It is difficult to grow high quality material due

to the reduced adatom mobility on the surface at low growth temperature. In addition,

due to the high kinetic barrier for breaking N-H bonds ofNH3 [Seg04, Yum81, Liu78],

excessive NH3 flow is required to compensate for the amount of active nitrogen.

Excessive NH3 will form H2, and H2 is known to make InN growth less

thermodynamically favorable [Kou99, Hua05, Dra06, Joh04j]. Thus, the growth rate of

InN is very low compared to other group III nitrides.

Another obstacle is the lack of suitable substrate for InN growth. Sapphire and Si

have been widely used substrates for InN because sapphire is a good substrate for group

III nitrides in that the surface of sapphire can be changed to AlOxNy by nitridation. Si has

numerous advantages including the possibility of fabricating tandem solar cells. The

lattice mismatch between InN (0001) and Si (111) is about 8 %, while the lattice









mismatch between InN (0001) and c-A1203 is about 26 %. It should be noted that when a

lattice mismatch is greater than 0.8 %, dislocations will form at the interface. Thus

stating that an 8 % lattice mismatch is better than a 26 % mismatch with InN does not

have much significance when critical thickness is concerned.

It was found in early work that although Si has a smaller lattice mismatch than

sapphire, the films grown on sapphire had superior quality after nitridation [Yam94].

Nitridation process formed A1N on sapphire substrate [Ven99] and A1N has a 13 % lattice

mismatch (c.f. A1203 has 26 % lattice mismatch with InN). SixNy formation is known to

prevent epitaxial growth of GaN on Si [Plo99]. Thus nitridation of Si substrate should be

avoided.

InN films were grown on Si( 11) substrate without nitridation [Yan02]. To avoid

SiN layer formation, TMIn was first introduced before NH3. MOCVD with double zone

heating was used to facilitate thermal cracking of NH3. Growth temperature was varied

from 350 to 600 oC and a maximum of 6 pm/hr growth rate was achieved although the

films over 1 [m thick cracked.

For InN growth on sapphire, the optimal nitridation temperature and times were

ambiguous. For example, one report showed best results with 1000 C for 30 min

[Pan99] and the other report proclaimed 1075 C for 5 min gave best results [Dra06].

To elucidate the effect of nitridation, the most useful method may be to observe the

surface evolution during the growth in situ. From in situ spectroscopic ellipsometry

study of InN growth by MOVPE [Dra06], it was found that the nitridation at 1050 C for

45 sec, or 1000 C for 300 sec would give the best quality film. It may be concluded that

lower nitridation temperature requires longer nitridation time.









The optimal growth temperature of InN is another ambiguity. In fact, the

temperature is not an exact value in practice because it depends on the places of

thermocouples, thermocouple types, or the temperature measurement methods.

Therefore, 25 to 50 OC of temperature error is possible. The highest reported growth

temperature for InN was 750 C [Tak97]. The lowest growth temperature is not clearly

reported because of the low quality of the material, but it may be approximately 350 C

[Yan02] for MOCVD. Typical InN growth temperature is around 450 to 650 C.

It seems that the crystal grower's golden rule, "the higher growth temperature, the

better crystal quality," holds for InN case as well. The highest Hall mobility was

reported from the sample grown at 600 oC [Kel00]. The growth temperature at 650 C

gave the best surface morphology [YamO1]. InN films grown at 620 C showed the best

optical properties, whereas the samples grown at 600 oC exhibited the best electrical

properties. Some reported that high growth temperature is destructive to InN because it

begins to decompose. For example, surface morphology of the InN was best at 550 C

during the 520 to 590 oC experiments [Yan02]. According to the other experiments, 560

C grown samples showed the best quality among the 540 to 580 C [Jai04]. Thus, the

optimal growth temperature determination is not clear yet. It should be noted that

optimal growth temperature is well above the decomposition temperature of InN as in the

GaN case. This result showed that InN decomposition and growth temperatures have a

wide range of overlaps, about 230 C (520 C decomposition [Dra06] to 750 C growth

[Tak97]).

To find optimal growth conditions for InN film by H-MOVPE, exploratory study

was carried out and the results are presented in Chapter 6. It should be noted that during









the growth study in chlorinating environment [Kan04], InN columnar structured materials

were observed and the growth conditions provided a good starting point for the growth of

nanostructured materials.

1.2.4.2 Growth of InN Nanostructured Materials

The growth of InN nanostructured material is interesting since the hope is that

devices made of nanostructures will have superior efficiency due to quantum

confinement effects. The growths of InN nanorods and nanowires have been reported by

several researchers [Ji05, Kim03, Che05, Qia05, Luo05, Cha05, Lia02, Yin04, Sch04,

Lan04, Tan04, Sar05, Joh04, Yin04, Lan04, Vad05, Zha05, Shi05, Tak97].

InN nanowires have usually been grown by thermal catalytic CVD that usually uses

Au as a catalyst. For example, Liang et al. [Lia02] synthesized InN nanowires on gold

patterned p-type Si (100). Pure indium foil and NH3 were used as an indium and nitrogen

sources, respectively. The growth was performed at 500 oC for 8 hrs and 40 to 80 nm

diameter nanowires were created. The band gap was 1.85 eV measured by PL.

Zhang et al. [Zha02] fabricated anodic alumina membranes (AAM) and deposited

Au catalyst to grow InN nanowires by electrodeposition. The indium and nitrogen

sources were pure indium powder and ammonia. Despite the complex substrate

preparation and the rather long growth time (12 hr), the InN nanowires were

polycrystalline with rough surface morphology.

Lan et al. [Lan04] produced InN nanorods using pure indium powder and NH3 as

the indium and nitrogen sources in a hot wall quartz reactor. Au was sputtered on Si(100)

substrates as the catalyst and good quality single crystal InN nanorods were produced.

The band gaps of 0.766 eV and 1.9 eV were observed depending on the diameter of

nanorods. It was still arguable why there exist two distinct values.









The typical sources of In were In metal [Vad05, Zha05, Joh04, Tan04, Lan04,

Lia02], In203 [Tan04, Sch04, Lan04, Yin04, Luo05], InC13 [Tak97, Kim03], or TMIn

[This work]. Sometimes a novel precursor, such as indium acetylacetonate, was used

[Sar05]. Ammonia was mainly used as the nitrogen source since most other nitrogen

sources are extremely flammable. A less flammable nitrogen source,

monomethylhydrazine (MMHy), was occasionally used [Tak97].

An external catalyst was not used to grow InN nanostructures in some cases

[Che05, Luo05, Joh04, Vad05]; however it is possible that In liquid acted as a self-

catalyst. For example, Johnson et al. [Joh04] reported InN nanowires on the quartz

surface and indium metal surface. Indium metal and NH3 were also used as the indium

and nitrogen sources, respectively. The growth was carried out at 700 oC for 2 hrs and 50

to 100 nm diameter nanowires were generated. Although the morphology of the

nanowires shows rather rough surface, the nanowires were single crystal and the band

gap was 0.80 eV as measured by PL.

The growth axis was usually along the [110] direction [Lia02, Yin04, Lan04,

Tan04] and sometimes the [001] direction [Kim03, Joh04], suggesting that there was

more than one growth mechanism that governs nanostructure growth.

Nanostructures can be grown in a relatively wide range of temperature. For

example, nanowires were grown at T = 600 to 750 C [Luo05, Sch04, Tak97], 565 to 590

C [Yin04], 525 C [Tan04], 500 C [Lan04].

Since InN nanowires may be of higher quality than InN films, there was strong

motivation to determine the bandgap. However, the bandgap measured from InN

nanowires still showed wide ranges, although a majority of the results centered at 0.8 eV.









Both 1.9 eV and 0.8 eV bandgaps were detected as evidenced by the brown colored (30

to 50 nm diameter) and black colored (50 to 100 nm diameter) nanowires [Lan04]. In an

other case, both room temperature PL and optical absorption spectrum showed 0.8 eV of

bandgaps from InN nanowires [Joh04a]. The bandgap was also measured from InN

nanowires and two distinctive peaks of 0.8 eV and 1.9 eV were detected from the InN

and In203 mixture [Vad03]. A higher bandgap (1.9 eV) was still detected from InN

nanowires grown on Anodic Alumina Membrane (AAM), using direct reaction of In

metal with NH3 [Zha05]. Nanowires grown by nitridation of In203 powders without

catalyst showed a 1.7 eV bandgap [Luo05]. Another medium bandgap of 1.1 eV

(FWHM 105 meV) was measured by PL from aligned wurtzite polycrystalline InN

nanofingers [Ji05]. A bandgap of 1.1 eV was also obtained by calculation from InN

nanotubes based on Density Functional Theory (DFT) considering stability and electronic

structures of single walled (SW) InN nanotubes [Qia05]. On the other hand, other

calculations showed 0.8 to 0.9 eV [Bec02] from wurtzite InN.

Further investigation is required to conclude the correct bandgap of InN as well as

the reasons for the wide range of bandgap values.

In this study, InN nanostructured material growth optimizations were carried out by

H-MOVPE and the results are presented in Chapter 5.

As an exploratory work, GaN nanostructured material was tested to grow by H-

MOVPE and some results are shown in Chapter 7 followed by recommended future work

in Chapter 8.














CHAPTER 2
CHEMICAL EQUILIBRIUM ANALYSIS OF H-MOVPE SYSTEM

2.1 Introduction

Although the Hydride Metalorganic Vapor Phase Epitaxy (H-MOVPE) technique

has proven to be a promising and versatile technique to synthesize III-V compound

semiconductors, the chemistry of H-MOVPE is not currently well understood.

Experimental definition of the gas phase and heterogeneous reactions are noticeably

absent, mainly due to the difficulty of dealing with corrosive gases, such as HC1. In

addition, hot furnace walls and wall deposition during growth also obstruct the use of

optical in situ measurements. An alternative way to study the chemical reactions

involving the H-MOVPE technique is through simulations, and the most straightforward

is chemical equilibrium analysis.

Thermodynamic analysis can give insight into which gas phase and condensed

phase species will be present without detailed knowledge of molecular structures or

reaction kinetics. Because reaction kinetics is excluded, the result in some cases may not

represent the actual chemistry. It can, however, give an approximate idea of which

species will be dominant under various conditions.

Reed [Ree02] calculated the mole fractions of gas and condensed phase species at

equilibrium in the H-MOVPE system for GaN growth by varying temperature, Cl/Ga

ratio, and NH3 flow rates. It was found that the calculated results matched well with

experiments when Ga(l) is excluded from the system. It was also found that Cl acted as a

gallium sink and H (from NH3) acted as a carbon scavenger.









In this chapter, more species have been added into Reed's database. The newly

incorporated species includes not only Ga, but also In containing species. The database

also contains metalorganics and adducts. The aspects of two main reactions can be

compared: Group III chloride formation versus metalorganic decompositions. One of the

goals for this study is to see whether H-MOVPE is closer to MOVPE or HVPE after

metalorganics + HC1 reactions. If the results are known, the consecutive chemistry will

be clearer because both MOVPE and HVPE chemistry are well studied.

2.2 Thermochemical Data Collections

Great care should be taken when selecting and adding new thermochemical data.

The inclusion of one erroneous data point or exclusion of one important species may

render the calculation results inapplicable. Most data for well-identified traditional

species in the Ga-In-C-H-Cl-N system are from the Thermo-Calc SUB94 database. Most

data for well-known species from SUB94 have passed the self consistency test as well as

2nd and 3rd law verifications with experimental data.

Species that are commonly included in a complex equilibrium analysis of

subsystems in the Ga-In-H-C-Cl-N system are listed in Table 2-1. In this Table, a

number of gas phase species are categorized by Ga, In, and Cl containing species, with

other species such as hydrocarbons, and the species that are only composed of C, N, or H.

Condensed phases are also organized in the same way as the gas phase. It should be

noted that although some species may not exist in the H-MOVPE system, they were not

excluded from the database. Trace species were traditionally excluded to save

computation time. Now, with advancements in computing power, there is no need to

exclude non-significant species, as long as the correct thermochemical data are used. In
















Ga containing
species


addition to the data from SUB94, critically assessed data for GaN and InN were used

from [Unl03] and [Lei04], respectively.

Table 2-1. Commonly considered species in Ga-In-H-C-Cl-N system from SUB94
database.


Ga, Ga2, GaH, GaC1, GaCl2, Ga2C12, Ga2Cl4, Ga2Cl6, GaCl3


Gas Phase


In containing In, In2, InH, InCl, InC12, In2Cl2, InC13, In2Cl4, In2C16
species

Cl containing CC1, CHC1, CH2C1, CH3C1, CNC1, CC12, CHC12, CH2C12,
species CC13, CHC13, CC14, C2C1, C2HC1, C2H3C1, C2H5C1, C2C12,
(excluding III C2H2C12, C2H4C12, C2C13, C2HC13, C2H3C13, C2C14, C2H2C14,
chlorides) C2C15, C2HC15, C2C16, C6H5C1, Cl, HC1, Cl2

Other species CH, HCN, CH2, CH3, CH4, CN, CN2, C2, C2H, C2NH, C2H2,
C2H3, C2H4, C2H5, C2H6, C2N, C2N2, C3, C3H, H, C3 3H4,
C3H6, C3H8, C3N, C4, C4Hlo, C4H2, C4H4, C4H6, C4H8, C4N,
C4N2, C5, C5NH, C5H8, C5N, C6H6, C6N, C6N2, C7NH, C7N,
CsN, C8N2, C9NH, CloN, CloN2, C11NH, C11N, C12H26, H,
NH, N3H, H2, NH2, N2H2, N, N2, N3, NH3, N2H4


GaCl3(1), Ga(l), GaCl3(s), GaN(s)*, Ga(s)


Ga containing
species


Condensed In containing InCl(1), InCl3(1), In2C3(1), In3C4(1), In4C7(1), In(l), InCl(s),
Phases species InCl(s2), InCl2(s), InC13(s), In2Cl3(s),In3Cl4(s), In4Cl7(s),
InN(s)**, In(s)

Cl containing CC14(1), C6H5Cl(1), NH4C1(1), NH4Cl(sl), NH4Cl(s2)
species
Other species C(1), C(graphite), Diamond

* GaN(s) thermochemical data was replaced with assessed data in [Unl03]
** InN(s) thermochemical data was replaced with assessed data in [Lei04]

The data for metalorganics (MOs) and adducts were obtained from Przhevalskii et

al. [Prz98] and listed in Table 2-2. Since some of thermochemical data for MOs and

adducts are calculated, it is necessary to validate the data. Values of AH298, S0298, and

Cp for all the species used in the calculations plus Al containing species are listed in

Appendix A.










Table 2-2. Additionally included gas phase species from [Prz98].
Gas Phase Ga Ga(CH3)3, GaCH3, Ga(CH3)3NH3, GaCH3NH, GaNH3,
containing [Ga(CH3)2NH2]3, GaCl3NH3, GaH2, GaH3, (GaCH3NH3)3,
species (GaN)3*
In containing In(CH3)3, InCH3, InH2, InH3
species


* The gas phase ring-shaped (GaN)3 species was eliminated due to inconsistency in the
results. See section 2.3.1.

2.3 Chemical Equilibrium Calculations

There are two main ways of computationally determining the equilibrium state:

stoichiometric and non-stoichiometric algorithms [Smi68]. In the stoichiometric

approach, the total Gibbs energy of the system is minimized by solving the set of linear

equations produced by introducing the stoichiometric constraints [Mey84]. In the non-

stoichiometric approach, the system of nonlinear equations produced for the equilibrium

expressions for each reaction in an independent set of reaction is solved.

In this work, Thermo-Calc is used to solve for the equilibrium states. Users need to

define atoms and provide initial atomic ratios, temperature, and pressure for equilibrium

calculations. The software will generate all possible molecules by combination of

defined atoms, as long as they exist in the database. After that, the mole fractions of all

the species are calculated and iterated to find the minimum total Gibbs energy of the

system. Therefore, it is important to make sure that all the thermochemical data in the

database are valid. One way of checking the validity is by computing well-known

equilibria.

2.3.1 Thermochemical Data Verification

Starting with established databases, the species list was expanded by review of the

literature for the Ga-In-C-H-Cl-N system. To verify the newly included data, well-










known problems were solved to check the consistency. For that purpose, the In-N and

Ga-N phase diagrams were generated.

P = 0.1MPa
2500- i -
gas
S11211 K ISUB94
S>s868 K [Gao03]
j 3773 K [Ond02]
S1500 gas iq 63 K Le04]
Lii
S1211 K

1000-

I X 638 K

50- InN + gas InN + In

InN InN + In(s)
0 ---- i i ----
S0 0.2 0.4 0.6 0.8 1.0
MOLEFRPCTION IN
Figure 2-1. Phase diagram of In-N systems at P = 0.1 MPa and experimental InN
decomposition data.

The decomposition temperatures of InN were measured as 638 K by drop

calorimeter [Lei04], 773 K by thermogravimetric analysis (TGA) [Ond02], and 868 K by

thermogravimetry-differential scanning calorimetry (TG-DSC) [Gao03]. The calculated

decomposition temperature of InN is 1211 K using SUB94 database. In addition, the

maximum experimental InN growth temperature is around 973 K (700 C) according to

Chapter 5 and 6 of InN growth study. Therefore, assessment of InN Gibbs energy is

required for equilibrium calculations. At this point, the data from [Lei04] is used since it

was critically assessed data with experiments. However, it should be noted that the

decomposition temperature and growth temperature have a large overlap due to the

kinetic barriers. When equilibrium calculation is needed including InN at the growth

temperature, the Gibbs energy of InN should be lowered to consider kinetic barrier for









decomposition. In this case, the data from [SUB94] was used as presented in Section

2.3.2.4.

All the newly included species gave results consistent with previously known

results for GaN, except for the ring-shaped trimer (GaN)3(g) species. Figure 2-2 shows

the calculated AG values for Ga(l) + 1/2 N2(g) = GaN(s) and Ga(l) + 1/2 N2(g) =

1/3(GaN)3(g) reactions. The AG value for (GaN)3(g) formation reaction is significantly

lower than GaN(s) formation reaction showing that (GaN)3(g) is more stable species than

GaN(s). However, the existence of (GaN)3(g) was never confirmed by experiment and

the thermochemical data was calculated using estimated parameters [Prz98]. Therefore,

it would be reasonable to conclude that a kinetic barrier exists for formation of (GaN)3 or

the data for (GaN)3 is incorrect. Ga-N phase diagram was generated excluding

(GaN)3(g), as shown in Figure 2-3.



80
60
40








-100


-120
500 600 700 800 90 1000 1100 1200 1300 1400 1500

Temperature (K)
Figure 2-2. Calculated AGr for GaN(s) and (GaN)3(g) formation reactions.
-120-------- --------- --



Figure 2-2. Calculated AG" for GaN(s) and (GaN)3(g) formation reactions.



















l r"iC-


TMERMl I: 11. I 11*1. 221 P=O 1 MPa
H. 1 L -L I I I I I


GaN + gas
| GaN + liq
" _I GaN"


-'i I I --""I III


S 0.1 0.2 .3 I4 0.5 .6 D .7 0.8 j.s 1.
MDLE FRACT-i C.A

Figure 2-3. Phase diagram of Ga-N systems at P = 0.1 MPa. The data for GaN is from
[Unl03]

Otherwise, all the new species including MOs and adducts from [Prz98] were

added to the database and used for the subsequent equilibrium calculations.

2.3.2 Complex Chemical Equilibrium Calculations

2.3.2.1 Ga-C-H-Cl-Inert System

It will be informative to see the gas phase species after the reaction between TMG

and HC1 in N2 carrier gas since no experimental results are available. Hence, the Ga-C-

H-Cl-Inert (N2) system was analyzed first. The schematic of the H-MOVPE inlet is

shown in Figure 2-4.


N2 (Inert) --
TMG

HCI


Ga-C-H-Cl-lnert system


NH c of te it of H t f G

Figure 2-4. Schematic of the inlet of H-MOVPE technique for GaN growth.


gas




gas + liq


1110 K[Unl03]


2:KIII"


LU
I









TMG, HC1, and N2 (inert) were mixed to form gas phase species before reacting

with NH3. Although NH3 is present in the outer tube, it does not interact with TMG and

HC1 until a certain residence time, since they are separated by the quartz wall. Therefore,

only the TMG + HC1 + N2 system will be only considered here.

The base conditions of molar flow rates in the source zone for GaN growth by H-

MOVPE are listed in Table 2-3. It should be noted that the atomic ratios are not

independent, since TMG is composed of 1 Ga, 3 C, and 9 H atoms; HC1 is composed of 1

H and 1 Cl atom. Therefore, X(C) = 3 x X(Ga) and X(H) = 9 x X(Ga) + X(C1) conditions

must be satisfied.

Table 2-3. Base inlet conditions for sources for GaN growth and atomic mole fractions
for calculation.
Precursors Flow rate Flow Atom # of Atoms X(Atom) Mole fraction
rate/0.7
TMG 0.7 sccm 1 Ga 1 X(Ga) 0.00066
HCI 0.7 sccm 1 C 3 X(C) 0.00198
Inert (N2) 1050 sccm 1500 H 10 X(H) 0.00660
C1 1 X(C1) 0.00066
Inert 1500 X(Inert) 0.99010
Total 1514 Total 1

N2 as a product of NH3 decomposition was treated as an active N source, which

may react with a Ga species to form GaN in equilibrium calculations. For example,

GaN(s) can be formed by Ga(l) and N2(g) in equilibrium calculation. The source of

reactive N is only NH3 and its intermediate products. The carrier gas, N2, was thus

treated as an inert species, thus all N in the inlet NH3 was available for reaction;

including the product N2.

Helium (He) was used instead of carrier N2. The use of different inert gas such as

Ne or Ar would give the same results since they are not participating in the reaction. The

role of inert gas in equilibrium calculation is related to gas mixture entropy.










10-2

S10-4
1 0l

S10-8
i

oa-
.
a 10-14


10G14


lOra1 1-----ZIA -,--t--1
400 500 600 700 800 900 1000
Temperature (K)


100

10-1

c 10-2
0

10-3


~ 10-5

10-a


10-7 1 i i I 1I
400 500 600 700 800 900 1000
Temperature (K)
Figure 2-5. Equilibrium partial pressures of gas phase molecules and mole fractions of
condensed phases with respect to the temperature in Ga-C-H-Cl-Inert system.









In addition, total pressure and system temperature values P = 1 atm and T = 573 K

were used as the base conditions for calculations of the source zone performance.

Figure 2-5 shows the calculated equilibrium partial pressures of gas phase species

in the Ga-C-H-Cl-Inert system, with respect to the temperature. In this figure Cl/Ga was

set to 1 and the calculation axis was T = 430 to 1000 K.

The most predominant Ga containing gas species was GaCl above 530 K and GaC13

below this temperature. The amount of GaCl exceeded the amount of CH4 at 580 K.

The partial pressure of the second dominant species, GaC13, decreased at temperatures

higher than 500 K. It was also observed that the amount of HCl gas increased up to 600

K and did not change at temperatures higher than 600 K. The partial pressure of

monomethylgallium (MMG; GaCH3), the most dominant organometallic species,

increased with temperature. However, the amount of MMG was about 10-6 times less

than GaCl even at 1000 K. Therefore, the effect of using TMG instead of liquid Ga as

the Ga source is negligible, other than adding C to the system.

In this sense it is concluded that the H-MOVPE technique is closer to traditional

HVPE rather than the MOCVD technique.

Two condensed phase were present at equilibrium; C (graphite) and Ga(l).

Experimental observation of black deposition in H-MOVPE inlet confirmed that the

black deposition is composed of C (graphite) hollow tubes filled with Ga [Par05]. To

prevent black deposition, a small amount of H2 should be added to N2 carrier gas, which

is consistent with Park's result [Par05]. Figure 2-6 shows the equilibrium partial

pressures and condensed phase mole numbers with respect to X(H)/{X(Inert)+X(H)}.



























0 0.2 0.4 0.6


0.8 1.0


X(H)l(X(Inert)+X(H)}


SGas ~ I


Ga (liquid)


Is


0 0.2 0.4 0.6 0.8 1.0


X(H)I(X(Inert)+X(H))
Figure 2-6. Equilibrium partial pressures of gas phase molecules and mole fractions of
condensed phases with respect to X(H)/{X(Inert) + X(H)} in Ga-C-H-Cl-Inert
system at P = 1 atm, T = 573 K, Cl/Ga = 1.


100

10-2


5 10-6

W 10-8
Q. 10-10


S10-
10-14

10-16


rr
.-- --,--""- --"dl./

S_~~ cH










( I


100

!-;

S10-2
1-10 3i

510-4
S10.5


10 -


10-8





-









For this calculation, T was set at 573 K and X(H) was varied from 0.0066 to 0.9901

(not 0 to 1) because H and inert gas are always present with NH3 and HC1 (10 % HC1 and

90 % N2 (Inert)). It was noted that the amount of C (graphite) decreased rapidly with

increasing H2, primarily through formation of CH4. These thermodynamic calculation

results agreed well with the experimental results from Park [Par05]. It is also noted that

small amount of Ga(l) exists.

With increasing amount of H2, the partial pressures of the gas phase species did not

change significantly. It is also noted that the major species partial pressures showed a

gradual increase with increasing H fraction.

It should be noted that the amount of GaC13 changed significantly with both temperature

and Cl/Ga ratio, but not with H2 since the partial pressure of GaCl or GaC13 exceeded that

of HC1, the most dominant Ga or Cl containing species.

The equilibrium partial pressures of the gas phase molecules and condensed phase

mole fractions with respect to Cl/Ga ratio are shown in Figure 2-7. The amount of GaCl

and GaC13 crossed at around Cl/Ga = 2. Furthermore, the amount of HC1 prevailed over

GaC13 at Cl/Ga = 4. The experimental results from Reed [Ree02] showed that the growth

rate remained nearly constant from Cl/Ga = 0 to 3, then declined rapidly, falling to zero at

near HCl/Ga = 4. This also is reasonably matched with thermodynamic predictions.

From the experimental observation of the growth rate, GaCl appeared to be the main

precursor to grow GaN rather than GaC13. In the condensed phase, the amount of C

(graphite) did not change with HC1. In fact, the only C and Cl containing gas phase

species is CH3C1, which has very low partial pressure at Cl/Ga = 4.









100

10-2

- 104
E
1if
1 0-6

S10-8

CL 10.10

10-14
Ja-



10-16



100

10-1
10-2


10-5-
106-
10-7-

10-8-
109-
10-10-


1 2 3 4
X(CI)IX(Ga)


1 2 3 4
X(CI)IX(Ga)
Figure 2-7. Equilibrium partial pressures of gas phase molecules and mole fractions of
condensed phases with respect to X(C1)/X(Ga) ratio 1 to 4 in Ga-C-H-Cl-Inert
system at P = 105 Pa, T = 573 K.


Gas 1


C (graphite)




0


9.


.
-









Therefore, to prevent C contamination, H2 carrier gas or, at least, a mixture of H2 gas is

recommended. In addition, the amount of Ga (liquid) rapidly dropped with Cl/Ga > 1.

According to these calculation results, small amounts of H2 (- 4 %) and excess HCI

(Cl/Ga > 1) are required to grow Ga droplet-free and C-free GaN films.

2.3.2.2 Ga-C-H-Cl-N System

The growth of GaN by H-MOVPE in H2 carrier gas can be considered as a Ga-C-

H-Cl-N system, as shown in Figure 2-8. To calculate the growth conditions of GaN, the

Gibbs energy of GaN should be modified, as GaN is thermodynamically unstable at

typical growth temperature. For example, the calculated GaN decomposition temperature

was 1055 K using the Gibbs energy of GaN recommended by Unland et al. [Unl03]. It

was found that the decomposition temperature of GaN linearly increases with decreasing

the Gibbs energy of GaN by Thermo-Calc calculations. Assuming that the

decomposition temperature of GaN is 1500 K, 49 kJ/mole was subtracted from the Gibbs

energy of GaN [Unl03].


H2 -

NH3
HCI
Ga-C-H-CI-N
TMG j- system
HCI
NH3

H,
H 2 ----------------------------------------...


Figure 2-8. Schematic of H-MOVPE for GaN growth: Ga-C-H-Cl-N system. N is
present and H2 is used as a carrier gas in the system compared to Figure 2-4.

The experimental inlet conditions for GaN growth were from [Ree02], including

TMG = 2.2 sccm, HC1 = 4.4 sccm, NH3 = 500 sccm, and H2 = 1600 sccm. Therefore, the










Cl/Ga and N/Ga ratios were 2 and 230, respectively. Given an initial atomic mole

fraction of Ga, (X(Ga) = 4.2 x 10-4 = base), the other elements' atomic mole fractions are

calculated by X(C) = 3 x X(Ga), X(C1) = 2 x X(Ga), X(N) = 230 x X(Ga), and X(H) = 1

- X(Ga) X(C) X(C1) X(N). The growth-etch transition temperatures were thereafter

calculated by changing Cl/Ga ratio from 0 to 10.


1400
1371 K
1350
1300 \ Gal) + GaN
S coexistence
1250 / CI/Ga < 0.1
1200
S1150 Etch (gas)
E 1100
1050
1000 ;/
1000 Growth A230
950 (GaN)
900
0 2 4 6 8 10
Cl/Ga ratio
Figure 2-9. Calculated growth-etch transition temperature as a function of Cl/Ga ratio.
N/Ga ratio was set at 230.

Figure 2-9 shows the calculated growth-etch transition temperature. It was

observed that Ga(l) and GaN coexisted when Cl/Ga < 0.1 and that the upper growth

temperature decreased with increasing Cl/Ga ratio. It is also noted that carbon appeared

under certain conditions, but is not listed in this figure. The maximum growth-etch

transition temperature was 1371 K at Cl/Ga = 0 in case of N/Ga = 230.

Further calculations were carried out to focus on the Ga(l) GaN coexistence

conditions by varying N/Ga ratio (200, 300, 400, 500, and 600) and Cl/Ga ratio

(continuously from 0 to 0.5), as seen in Figure 2-10. The decomposition temperature

increased from 1371 to 1395 K with increasing N/Ga ratio from 200 to 600, as the N

partial pressure increased.










1400
N/Ga = 500
S1390 -- NIGa = 400
1380 N dscompos = 300Etch
S1370 N/Ga =200 (gas only)
1360 Z 2 Z Z
.Q 1350
a 1340 C
1330 Ga(l)+ GaN
: Growth
1320 (GaN + gas)
1310 \
1300
0 0.1 0.2 0.3 0.4 0.5
Cl/Ga ratio
Figure 2-10. Calculated transition of growth-etch and Ga(l) formation as a function of
temperature and Cl/Ga ratio. N/Ga ratio was varied from 200 to 600.

This is expected since adding excess N relative to Ga will form the more stable GaN. It

was found that the Ga(l) + GaN coexistence regime becomes narrower with increasing

N/Ga ratio and completely disappears at N/Ga = 600. This is the N/Ga ratio

thermodynamic limit to avoid Ga droplet in MOCVD (Cl/Ga = 0). Since the N/Ga ratio

for GaN growth is typically much higher than 600 in MOCVD, Ga(l) was not usually

observed. A more or less linearly decreasing trend of growth-etch transition temperature

was observed when Cl/Ga was higher than 0.2 regardless of N/Ga ratio. As Cl is added

to the system, it primarily appears as GaCl at the elevated temperature, thus making less

available for formation of GaN.

2.3.2.3 In-C-H-Cl-Inert System

Similar calculations were performed to analyze the inlet of H-MOVPE for InN

growth in inert ambient. Figure 2-11 shows the schematic of the inlet for InN growth.

TMI first reacts with HC1 in N2 (inert) carrier gas before reacting with NH3. Therefore

the TMI + HC1 + N2 (inert), or the In-C-H-Cl-Inert (He) system was analyzed to






48

determine the equilibrium gas and condensed phase species. The base flow rates of TMI,

HC1, and N2 were 0.7 sccm, 0.7 sccm, and 1050 seem (6.3 sccm with 10% HC1 + 200

sccm diluted with HC1 + 269 sccm with TMI + 200 sccm diluted with TMI + 375 sccm

carrier = 1050 sccm), respectively. Therefore, Cl/In = 1, Inert/In = 1500, P = 1 atm, and

T = 573 K were used as the base conditions for calculation. It should be noted that C/In

ratio is always 3, because TMI is composed of 1 In, 3 C, and 9 H. The number ofH

atoms is coupled with 9 TMI and 1 HC1. In other words, the relationship X(C) = 3 x

X(In) and X(H) = 9 x X(In) + X(C1) should be always satisfied. The initial atomic mole

fractions for In-C-H-Cl-Inert system are tabulated in Table 2-4.




N2 (Inert) -------------
TMI In-C-H-Cl-Inert system

HC1 ------------------------------------
HCI

NH3
Figure 2-11. Schematic of the inlet of H-MOVPE technique for InN growth.

Table 2-4. Typical growth conditions for InN and atomic mole fractions for calculation.
Precursors Flow rate Flow Atom # of Atoms X(Atom) Mole fraction
rate/0.7
TMI 0.7 sccm 1 In 1 X(In) 0.00066
HC1 0.7 sccm 1 C 3 X(C) 0.00198
Inert (N2) 1050 sccm 1500 H 10 X(H) 0.00660
C1 1 X(C1) 0.00066
Inert 1500 X(Inert) 0.99010
Total 1514 Total 1

Figure 2-12 shows the equilibrium partial pressures and condensed phase mole

numbers with respect to temperature at Cl/In = 1. The calculation axis was T = 430 to

1000 K.









The most predominant In containing species was InCl. However, the second

dominant In containing species was not InC13, as compared to the second most abundant

gas species in the Ga based system, GaC13. Instead of InC13, In2C12 (the dimer of InCl)

had higher partial pressure (105 times higher than InC13) at T = 480 to 700 K. The partial

pressure of InC13 decreased at temperatures higher than 480 K. Although TMI and MMI

thermochemical data were included during the calculation, they were not observed in the

results. The lowest limit of partial pressure calculation was set as 10-17 atm, thus it can be

inferred that TMI and MMI have less than 10-17 atm partial pressure. As a result, it can

be concluded that H-MOVPE is closer to HVPE than MOCVD in both InN and GaN

growth cases.

For condensed phases, carbon (graphite) and In (liquid) were dominant and the

amount of In(l) decreased with temperature. Deposition of a black material in the inlet

was occasionally observed during the growth of InN especially when the inlet

temperature was too high. Even though a detailed analysis of this black material at the

inlet during InN growth was not carried out, it likely consists of C (graphite) and In

droplets. To avoid C and In contaminations, similar methods can be used such as

addition of H2 and excess HC1 for C and In elimination, respectively.

The effect of H2 carrier gas was analyzed by increasing the initial atomic H mole

fractions. Figure 2-13 shows the equilibrium partial pressures and the amount of

condensed phase species with respect to the relative amount of H compared with inert

gas, or X(H)/{X(H) + X(inert)}. Comparing Figures 2-13 and 2-5 it is noted that the

amount of Ga(l) is about an order of magnitude greater than that of In(l) at similar

conditions.









100

10-2

%- 10-
E
. 10-

10-8

CL 10.10'

i 10-12
EL
.10-14


10-16 i -----' 1 1--- r- I--
400 500 600 700 800 900 1000
Temperature (K)


100

10-1

10-2
0


10-3
o -5


10-6 In (liquid)

10-7 ,
400 500 600 700 800 900 1000
Temperature (K)
Figure 2-12. Equilibrium partial pressures of gas phase molecules and mole fractions of
condensed phases with respect to the temperature in In-C-H-Cl-Inert system.










100


10-8-


InCI1
InlllI


10-10_
10-12-

10-14- Ck V___ ----
10-14 ,
0 02 04 06 08 1


0


X(H)/{X(Inert)+X(H)}


10-1

10-2

10(-3

10-4

10-s

10"-


10-7 -
10-8,,,,

0 0.2 0.4 0.6 0.8 1.0
X(H)/{X(Inert)+X(H))
Figure 2-13. Equilibrium partial pressures of gas phase molecules and mole fractions of
condensed phases with respect to X(H)/{X(Inert) + X(H)} in In-C-H-Cl-Inert
system.


.









10o
Inert I -
1 a2 ____ H| -,







I %10- -


0.10-10 ----2--------
i" 1 C C2H4


10-' -

10-18
1 2 3 4
X(CI)/X(ln)
IV











Gasl -
10-1 Gas-1



KC
10-2 -
10-3-----------
0 10 -









| 10-5
a 10-6









10-8 1
10 -s9 B


1 2 3 4
X(CI)/X(In)
Figure 2-14. Equilibrium partial pressures of gas phase molecules and mole fractions of
condensed phases with respect to X(C1)/X(In) ratio 1 to 4 in In-C-H-Cl-Inert
system.
system.









The gas phase species did not change greatly, except for H2 and other inert gases. An

increase of InH was observed; however, the amount of InH was extremely small (1015

atm), thus the role of InH was negligible.

Figure 2-13 shows the equilibrium partial pressures and the amount of condensed

phase species with respect to the relative amount of H compared with inert gas, or

X(H)/{X(H) + X(inert)}. Comparing Figures 2-13 and 2-6 it is noted that the amount of

Ga(l) is about an order of magnitude greater than that of In(l) at similar conditions. The

gas phase species did not change greatly, except for H2 and other inert gases. An increase

of InH was observed; however, the amount of InH was extremely small (10-15 atm), thus

the role of InH was negligible.

One of the advantages of H-MOVPE technique is the variation of Cl/In ratio.

Therefore, it is interesting to see the effect of Cl/In ratio. The Cl/In ratio was varied from

1 to 4 and the results are shown in Figure 2-14. The predominant In containing species

were InCl and In2C14 dimerr of InC12) through Cl/In = 1 to 4 and the amount of HC1,

InCl, and In2C14 reversed at Cl/In > 2. Therefore, an etching effect is expected due to the

HC1 when Cl/In > 2. The amount of In(l) drops rapidly near Cl/In = 1, and no In(l) exists

at Cl/In > 1 and the amount of C (graphite) did not change with Cl/In ratio. To eliminate

C contamination, H2 should be added in the system.

2.3.2.4 In-C-H-Cl-N-Inert System

Thermodynamic calculations were carried out to understand the gas phase and

condensed phase species for InN growth by H-MOVPE. Figure 2-15 shows the

schematic of the H-MOVPE system using TMI, HC1, NH3, and N2 (inert) as precursors.

This can be thought as an In-C-H-Cl-N-Inert system. It will be informative to see the









InN growth-etch conditions with change in temperature, Cl/In, and N/In molar ratios

since they are the most important growth parameters.


N2 (Inert) ....................

NH3
HCI
() In-C-H-Cl-N-lnert
TMI system
HCI
NH3
N2 (Inert) / H2" ---------
Figure 2-15. Schematic of H-MOVPE inlet with In-C-H-Cl-N-Inert system for
thermodynamic calculations. N is present in the system compared to Figure
2-11.

In the equilibrium growth-etch regime calculation, the thermochemical data for

solid InN was taken from SUB94 database instead of using the data from [Lei04], since

the decomposition temperature of InN (686 K) from [Lei04] is too low for InN.

Therefore, InN is thermodynamically more unstable at growth temperature with

expressions proposed by [Lei04] as compared to that from the SUB94 database. This is

equivalent to lowering the Gibbs energy by about 68.5 kJ/mole.

The initial process parameters for InN growth used in this calculation were TMIn =

1 sccm, HCI = 4 sccm, NH3 = 250 sccm, and Inert = 2286 sccm. Actual process

parameters were about 0.7 times less than the given values, for example TMIn = 0.7

sccm. However, they were increased by 1/0.7 times for ease of calculation. Therefore,

the initial atomic mole fractions were X(In) = 3.024 x 10-4, X(C) = 3 x X(In), X(C1) = 4 x

X(In), X(N) = 250 x X(In), X(H) = 9 x X(In) + X(C1) + 3 x X(N), and X(Inert) = 1 -

X(In) X(C) X(C1) X(N) X(H). To carry out the calculation Cl/In and N/In ratios

were varied and the growth-etch transition temperatures were determined.









As shown in Figure 2-16, the growth-etch transition temperature decreased with

increasing Cl/In ratio. When Cl/In ratio > 1, the added HC1 to the system shifts the

equilibrium compositions of InClx in the vapor phase, and thus it lowered the transition


1200
N/In = 7000

1100 N/In = 1000
N/In = 500
\ "*N/In = 250
1000 N/n =250 N/In = 7000
N/In = 100
N/ln = 1000
Etch (gas only)
900 N/In = 500





Growth (GaN + gas)



Cl/In ratio
Figure 2-16. Calculated growth and etch regime with respect to temperature and Cl/In
ratio. N/In ratio was varied from 100 to 7000.

temperature. As the C1/In ratio decreases towards unity, the C1 is now not in large excess

but approaches the stoichiometry of the dominant vapor phase In species, InCl. Thus a

small decrease in the C1/In ratio frees a relatively higher proportion of the In for

participation in growth. For Cl/In ratio less than unity, there is not sufficient Cl to retain

the In in the vapor phase and In is now available to form InN. The temperature of the

horizontal line for a given N/In ratio is the thermal decomposition temperature, which

increases with increasing N/In as expected. In(l) was observed as a product of thermal

decomposition reaction below the horizontal lines.

The companion plots of transition temperature vs. N/In atomic ratio are shown in

Figure 2-17. The results suggest that the transition temperature increased with increasing










N/In ratio to an asymptotic limit. It is noted that the calculations were only performed for

seven HCl/TMIn ratios, each greater than unity. Increasing the N at constant HC1 and

TMIn input increases the ability of N to compete with Cl for the In. Increasing the N

through the addition of NH3 also increased the H in the system, which also competes with

In for the Cl to increase the driving force for InN formation.






1 800 0



| 800
Etch Cl"Cn = 1


750
75 Growh "ClIn 10

700
0 1000 2000 3000 4000 6Oo EGoo 7000 B0
Nfln raot
Figure 2-17. Calculated growth and etch regime with respect to temperature and Cl/In
ratio. Cl/In ratio was varied from 1.1 to 10.

Figure 2-18 (a) shows the expanded views of the near horizontal lines at high N/In

ratios: 4.0 x 104 to 7.0 x 104. It was observed that both Cl/In and N/In ratios affect In

droplet formation significantly. When N/In ratio was 4.0 x 104, Cl/In ratio should be

higher than 0.4 to avoid In droplet formation. When N/In ratio was increased to 5.0 x

104, the minimum Cl/In ratio to avoid In droplet formation was decreased to 0.3, and so

on. Figure 2-18 (b) shows the extended calculation results that include N/In ratio from

100 to 7.8 x 104 and Cl/In ratio from 0 to 1. Higher N/In ratio lowered the minimum

Cl/In ratio to avoid In droplet formation, which is consistent with the previous results. It

was noted that at N/In ratio 7.8 x 104, no In droplet existed, even though no HC1 was

present (Cl/In ratio 0).
















1150 InN I li
InN Iln[l] C
1145 Ir.In I

Ii 1n
1135

1130

1125 IIN InN lnlN

1120
0 0-1 0-2 0.3 0.4 0-5
Clln ratio
1160 -


1155

S1150

S1145

8 1140

x 1135

z 1130

S1125

1120


Sl N/In= 10000
N/In = 500

N/ln-
4N/In= 20(0
z Zz
I" I n n

Co o 0 /In 1000 Do 0t
I 0

0 0.2 0.4 0.6 0.8 1
Cl/In ratio


In droplet selective etching conditions at 1153 K
90000

80000

No n(l) InN
70000
0
60000
S60000 n ratio
30000 to 78000
50000
In(l) + InN
40000

30000
0 0.1 0.2 0.3 0.4 0.5 0.6
Cl/n ratio
Figure 2-18. Calculated results of the growth, no growth (etch), and In droplet regimes
(a) Cl/In = 0 to 0.5, N/In = 4.0 x 104 to 7.0 x 104, (b) Cl/In = 0 to 1, N/In = 1
x 103 to 7.8 x 104, and (c) In droplet etching conditions at T = 1153 K, P = 1
Pa.


1153 K
N In = 1nnnn 7Rnr


Decomposition
0n f


No In droplet









Therefore, In droplet formation can be avoided when the N/In ratio is higher than 7.8 x

104. Experimental results of InN growth by MOCVD showed that In droplets formed

when the NH3/TMIn ratio was less than 1.6 x 104 [Mat97].

Although In droplets were not observed when N/In ratio was higher than 1.6 x 104, the

rough surface may be related to In droplet formation. The surface of InN was mirror-like

when N/In ratio was higher than 8.0 x 104, which may represent the complete removal of

In droplets during growth [Mat97].

The decomposition temperature of InN did not change while the N/In ratio changed

from 3.0 x 104 to 7.8 x 104; however, the minimum Cl/In ratio to avoid In droplets

decreased at higher N/In ratio. These results are shown in Figure 2-18(c). It is clear that

In droplet formation can be avoided by either increasing N/In ratio or increasing Cl/In

ratio. Both of these changes provide reactant for In(l) (N to form InN or Cl to form

InCl(g)). A key finding is that the determination of the minimum N/In ratio, 7.8 x 104, to

avoid In droplet formation by the MOCVD technique (Cl/In ratio = 0).

In the following, the decomposition temperature change was examined when the

N/In ratio was higher than 7.8 x 104. Figure 2-19 (a) shows the result when the Cl/In

ratio was 0. Notice that the N/In ratio axis is a logarithmic scale. The transition

temperature decreased with increasing N/In ratio, as long as N/In > 7.8 x 104. The reason

for this phenomenon is related to the gas phase species, InH. The partial pressures of InH

(PInH), in fact, decreased with increasing N/In ratio. For example, PInH were 7.975 x 10-7,

9.456 x 10-8, and 1.125 x 10-8 atm for N/In ratio 105, 106, and 107, respectively. This is

expected since the initial atomic mole fraction of In, X(In), was reduced to increase the

N/In ratio. The initial X(In) values were 2.486 x 10-6, 2.499 x 10-7, and 2.5 x 10-8 for










N/In ratio 105, 106, and 107, respectively. Therefore, the ratios of PInH to X(In) were

taken to compare reasonable with the relative effect of InH. PInH/X(In) ratios were 0.321,

0.378, and 0.450 atm for N/In ratio 105, 106, and 107, respectively. Thus, it can be

concluded that the relative amount of InH increased with increasing N/In ratio.


(a) 116o
1150

1140 Etch
Etch
1130

i 1120 InN

1110 InN + In(l) coexistence No

1100 Growth Growth
1090
78000
108
102 103 104 105 106
N/In ratio

(b) 1200

1150 ._ o
1 1100
Ioso e, Etch
S1000 (gas only)

I 950
S0oo Growth
850 (InN + gas)
850
800
0 1 2 3 4 5
CI/ln ratio
Figure 2-19. Calculated growth-etch transition temperatures (a) In droplet etch conditions
when no HC1 was present (b) growth-etch transition temperature at high N/In
ratios: 105, 106, and 107.

It is also observed from Figure 2-19 (a) that the growth transition temperature

lowered with increasing N/In ratio. Thus the maximum growth temperature is lower for

very high N/In ratio. It is known that the growth rate of InN decreases with excess NH3










flow. The results of this calculation may explain the reason of growth rate decrease

thermodynamically.

Figure 2-19 (b) shows the calculated growth-etch transition temperatures at three

high values of N/In ratio (105,106, andl07). No In droplet was observed in this

excessively high N/In ratio due to the InH formation. The transition from growth to etch

occurs at a lower temperature; the higher the value of N/In as long as Cl/In < 2. For high

Cl/In, the excess Cl ties up the In or InCl to minimize the effect of increased N/In.

2.3.2.5 NH3 Partial Decomposition

Ammonia is the most widely used nitrogen precursor for Group III Nitride

growth. Homogeneous decomposition of NH3 to N2 and H2 is thermodynamically

favored and it completes at around 673 K as shown in Figure 2-20.




0. S..- H2



0.-
I-0 I-


CL 0- .
0. 1 j NH,

500 1000 1500 2000 2500 3000
Temperature ( K)
Figure 2-20. Equilibrium partial pressures for decomposition of NH3 at 1 atm total
pressures calculated by Thermo-Calc.

It is known, however, that homogeneous decomposition of NH3 does not occur

easily due to the very high kinetic barrier to break the initial N-H bonds [Seg04, Yum81,










Liu78]. One approach to account this fact in equilibrium calculations is to measure the

Gibbs energy of NH3 to shift the yield curve to high temperatures.

The homogeneous partial decomposition ofNH3 can be written as following.

NH3 =(1 a) NH3 + a/2 N2 + 3a/2 H2

where a is the degree of completion of NH3 decomposition reaction.

It was experimentally observed that NH3 decomposes to an extent 3 % at 950 C

as measured by mass spectroscopy [Ban72]. To achieve this conversion at the

temperature at 950 C, 170 kJ/mole has to be subtracted from the NH3 Gibbs energy.

1 i --------------- i ,------
0.9



t 0.5
0.4
0.3 ----
0.2 1- -' --- -
0.1
0
-200 -150 -100 -50 0
NH3 Gibbs energy correction (kJ/mole)

Figure 2-21. Magnitude of the Gibbs energy correction required to achieve a value for
partial decomposition (a) at 950 'C.

This energy may be considered as the kinetic barrier of NH3 decomposition to make NH3

decompose 3 % at 950 OC. A desired a value at 950 OC can be achieved by modifying

the Gibbs energy of NH3 by subtracting various amounts as shown in Figure 2-21. A

different value of a at 950 OC can be assumed that will give a different value by which to

modify the Gibbs energy of NH3. The calculated values for the decomposition curves as

a function of T are shown in Figure 2-22. The a given in this figure represents the value

of the conversion along this curve at 950 OC.












0.90.03
Soa = 0.03

0.8 I
=0.7
o .3




0.8123 K
0.50 500 700 00 1100 1300 1500 1700 1900
0 a 0 5









Temperature (K)

Figure 2-22. NH3 mole fraction with respect to temperature at different values of ca.
2.4 Conclusions








Additional thermochemical data for the Ga-In-H-C-Cl-N system including MOs

and adducts were added to the current database [Ree02] and validated. Complex

equilibrium calculations were performed and concluded that H-MOVPE is more related
z 0.3 =0.7










to HVPE than MOCVD.
1223 K

300 500 700 900 1100 1300 1500 1700 1900







It was concluded that H should be included to eliminate C deposition in the inlet









by thermodynamic analysis and experimental observations. In addition, excess HC1
Figure/III > 1) should be prevention withe system to eliminate Ga or In metal droplets in the
2.4 Conclusions

















inlet.
Additional thermined that monochl data forides (i.e., Ga-In-H-C-Cd I-N system includinant species
and adducts were added to the current database [ReeO2] and validated. Complex









at C/III < 2. Excess C/III calculations were performed and concluded that Ga-MOPE is predominant when Crel/Ga >
to HVPE than MOCVD.

It was concluded that H2 should be included to eliminate C deposition in the inlet

by thermodynamic analysis and experimental observations. In addition, excess HCl

(Cl/Ill > 1) should be present in the system to eliminate Ga or In metal droplets in the

inlet.

It was determined that monochlorides (i.e., GaCl and InCl) are predominant species

at Cl/Ill < 2. Excess Cl/Ill calculations showed that GaC13 is predominant when Cl/Ga >

2, whereas HC1 prevailed over InCl and In2C12 at Cl/In > 2 at 573 K. However,

monochlorides were always dominant at growth condition for both GaN and InN.









High N/III, low Cl/III, and low temperature generally provide more stable

conditions for III nitrides, except for the excessive N/III case. The decomposition

temperature of GaN and InN increased with high N/III ratios. Ga(l) + GaN and In(l) +

InN coexistence regimes were observed under the low Cl/III ratio conditions. It was

determined that for Ga(l) removal in MOCVD (Cl/Ga = 0), the N/Ga ratio should exceed

600, whereas for In(l) removal in MOCVD (Cl/In = 0), the N/In ratio should be higher

than 7.8 x 104. The results may explain the excessive ammonia flow requirement for InN

growth to avoid In droplet formation.

Finally the partial decomposition of NH3 was considered and the Gibbs energy of

NH3 was reduced to control decomposition efficiency. The kinetic barrier of NH3

decomposition (170 kJ/mole) was calculated by considering 3 % of NH3 decomposition

at 950 oC.














CHAPTER 3
STRESS DETERMINATION STUDIES OF GALLIUM NITRIDE FILM ON
SAPPHIRE

3.1 Introduction

Due to the lack of bulk single crystals, GaN films are usually grown on foreign

substrates, such as sapphire or SiC. Stress and dislocations are inevitable because of

lattice and thermal expansion mismatches between GaN and foreign substrates. Excess

stress may cause cracks and bows of the GaN films and dislocations may act as carrier

traps resulting in low-mobility materials. Hence, a stress and dislocation study is

important to improve device performance.

The most widely used substrate for GaN growth is sapphire, due to its thermal and

chemical stability, AlOxNy compliant layer formation at the surface after nitridation, and

relatively low cost. SiC is a substrate that provides better lattice and thermal expansion

match and thermal and chemical stability, in addition, but the cost of SiC is still very

high, restricting commercial use. Therefore, focusing on studying the stress of GaN film

on sapphire is helpful as it is more cost-effective than SiC and will lead to improvement

in GaN/sapphire-based device reliability.

The stress state of a film can be measured by several techniques. Lattice parameter

changes (i.e., strain), curvature, or vibrational frequency changes are some of the

properties that can be readily measured to ascertain the stress of the film. Those

parameters are usually obtained by XRD reciprocal space mapping, optical or stylus

profilometry, laser reflectance, Raman spectroscopy, and photoluminescence.









XRD is a non-destructive and direct method to measure the lattice parameters of

the film. XRD 0-20 scan (also known as powder XRD, XRD 20-co scan, or low

resolution XRD) can determine d-spacing parallel to the surface of the films; however,

the lattice parameters obtained in this way cannot be used for stress measurement because

the 20 value can shift depending on the sample height and the tilt on the sample mount

during the XRD measurement. In addition, when a single crystal is used, there are only

peaks from the lattice spacing parallel to the surface. For example, if a single GaN

crystal was measured by LR-XRD, there will be only GaN (002) and GaN (004) peaks,

which can only be used to calculate out-of-plane lattice parameters, while in-plane lattice

parameters cannot be obtained. Therefore, XRD reciprocal space mappings should be

carried out to measure lattice parameters precisely to calculate the strain of the film. It is

time-consuming to measure the lattice parameters of the samples by XRD reciprocal

space mapping due to the alignments steps. In addition, the lattice parameters obtained

by XRD-SM may not represent the surface properties because of the X-ray penetration

depth. The stress measured by XRD should be considered as the average value from the

X-ray interaction volume. X-ray penetration depth depends not only on the material, but

also X-ray incidence angle (co). As a result, precise determination of X-ray penetration

depth is not simple.

Curvature of the film can be measured by optical profilometer, stylus profilometer

(also known as alpha step), laser reflectance, or other techniques, such as XRD rocking

curves with displacing the sample position. The disadvantage of optical profilometer and

laser reflectance is their sensitivity towards surface roughness. Since both techniques









utilize the reflected light from the surface, if the surface is rough or has structures on it, it

is difficult to measure the curvature of the film.

Curvature can also be measured by stylus profilometer. It requires direct contacts

between the tip and the surface. While the surface roughness is not a significant obstacle

to using this technique, direct contact may damage the surface. In addition, if the film

has fabricated structures on it, it is not possible to use stylus profilometer for curvature

measurement. Nevertheless, stylus profilometer may be the quickest and easiest way to

measure the curvature and the roughness of the film, though it is not recommended for

samples that will be damaged by direct contact.

XRD rocking curves with displacing the sample position is a non-destructive and

surface roughness-independent technique that measures the curvature of the

crystallographic planes, rather than the surface. Although it is time-consuming to

measure the curvature of the film due to the alignment steps, this may be the only way to

non-destructively measure the curvature of rough samples.

The general disadvantage of curvature measurements is that they cannot measure

small stresses in flat films. In other words, the stress must be large enough to create

curvatures within the film. Therefore, small stresses in films without curvature should be

measured by more sensitive techniques, such as Raman spectroscopy.

Raman spectroscopy is well-established as a non-destructive and relatively rapid

method to measure the average value of film stress. Raman spectroscopy operating in

confocal mode can show the peak shifts along the thickness. The results can be used to

deduce the stress variation along the depth of the sample.









Another way of studying stress of the film is through modeling. According to the

critical thickness theory, the lattice parameter of the overgrown film will turn into its

unstrained value once dislocations are formed. The lattice parameter variation along the

thickness, however, was experimentally observed, even well above the critical thickness.

The result cannot be predicted if the traditional concept of critical thickness or Frank-van

der Merwe's semi-infinite model is used. Shen-Dann's layer-by-layer growth model

allows for continuous changes in lattice parameters along the thickness, as well as

dislocation density reduction as the film becomes thicker.

This chapter will explore how the stress of the film was measured by vibration

frequency changes (Raman E2 phonon peak shifts), lattice parameter change (XRD space

map), and curvature (profilometer, XRD rocking curves). Extensive characterizations

such as XRD rocking curves, XRD pole-figure, AES, and SIMS were carried out to

determine the crystal structure and chemical compositions. For lattice strain modeling,

Shen-Dann's layer-by-layer growth model was adopted and compared with experimental

data. Although the model used only lattice parameters and elastic constants of the film

and substrate, the results matched well with experimental data.

3.2 Effects of Lattice and Thermal Expansion Mismatches

Before examining the stress measurements and modeling of GaN films, it would

be instructive to explore the main sources of film stress, which are lattice and thermal

expansion mismatches. The effects of these sources of film stress were studied as

follows.

3.2.1 Lattice Mismatch at Growth Temperature

To calculate the stress due to the lattice mismatch during the growth, it is important

to consider the lattice parameters at the growth temperature, rather than at room









temperature. The lattice parameters of GaN, sapphire, Si, and SiC at high temperature

would be larger than the parameters at room temperature due to the thermal expansion.

For convenience, calculations were performed assuming the growth temperature is 1300

K and room temperature is 300 K. Although the growth experiments of this study were

mostly performed at 1123 K, the calculation results would give some idea of lattice

parameters change at high temperature.

The definition of linear thermal expansion coefficient is given as equation (1):

AL
aAT = (1)
L0

where a is the linear thermal expansion coefficient, AT is the temperature difference, AL

is the linear variation of the lattice parameter, and Lo is the lattice parameter at room

temperature. The definition of misfit, is given as equation (2):

as af
f (2)
as

where as and af stand for the lattice parameters of substrate and film.

The lattice parameters of GaN and widely-used substrates such as c-A1203, Si(1 11),

and 6H-SiC at 300 K and 1300 K are tabulated in Table 3-1.

Table 3-1. Lattice parameters of GaN and widely used substrates at 300 and 1300 K.
Structure a300K (A) TEC (a a1300K Misfit (%) Misfit (%)
(x 10-6/K) (A) at 300 K at 1300 K
GaN Wurtzite 3.189 5.6 3.207 0 0
c-A1203 Rhombohedral 4.758 7.5 4.794 16.09 15.86
*(2.747) *(2.768)
Si (111) Diamond 5.431 2.7 5.445 + 16.97 + 16.72
*(3.841) *(3.851)
6H-SiC Wurtzite 3.086 4.2 3.093 -3.34 -3.69
* 300 rotation was considered.









The results show that the lattice misfits calculated at room temperature and at the growth

temperature are not particularly different.

To see the critical thickness at the growth temperature (1300 K, in this case), the

GaN/6H-SiC system was considered. The critical thickness can be approximately

expressed by equation (3).


h = --n- -fs l+1l (3)
hc 87t(lb+v)f [b)+ 1] (3)

where b is Burger's vector, v is Poisson's ratio, andfis the misfit. Plugging in the typical

values such as b (Burger's vector) af = 3.2 A, v = 0.2, andf= 0.03 (in the case of SiC),

gives he = 2.2 A, which is less than one mono-layer. In this case, dislocations will form

at the interface. The other substrates such as Si and sapphire will have even smaller

critical thickness. Therefore, critical thickness does not have much meaning in the case

of heteroepitaxy of GaN on any currently widely used substrates. In fact, a critical

thickness may only exist when the lattice misfit is smaller than 0.8 % and the thickness is

less than 10 monolayer scales (< 5 nm).

3.2.2 Thermal Expansion Mismatch

Most materials compress during the cooling process, except for some rare negative

thermal expansion coefficient materials (i.e., see section 8.4). Hence, both GaN and

substrates compress during cooling at different ratios due to differences in their thermal

expansion coefficients. Calculated thermal strains during cooling ranging from 1300 to

300 K are listed in Table 3-2 for GaN as well as c-A1203, Si( 11) and 6H-SiC. 30 0

rotated lattice parameters for c-A1203 and Si( 11) were used for practical comparison

with GaN basal plane and denoted in parentheses.









Table 3-2. Lattice constants at 1300 and 300 K and thermal strain (AT =
and widely used substrates.


1000 K) of GaN


Structure a1300K a300K Compressive strain Difference
(A) (A) during cooling with GaN
GaN Wurtzite 3.207 3.189 0.56 % 0
c-A1203 Rhombohedral 4.794 4.758 0.76 % -0.20%
*(2.768) *(2.747)
Si (111) Diamond 5.445 5.431 0.29 % +0.27 %
*(3.851) *(3.841)
6H-SiC Wurtzite 3.093 3.086 0.42 % + 0.14 %
* 300 rotation was considered.

3.2.3 Combination of Lattice and Thermal Expansion Mismatches

Figure 3-1 summarizes the stress caused by thermal expansion and lattice

mismatches assuming AT = 1000 K. It should be noted that all the lattice parameters that

were from different crystal structures were modified (i.e., 30 rotation) for better

comparison with GaN since GaN often showed 30 0 rotation with respect to sapphire.

082 7


072


062


052


0.42


032


0.22


2.5 2.6 2.7 2.8 2.9 3.0 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 4.0
Lattice parameter (A)

Figure 3-1. Substrates for GaN plotted with thermal strain versus lattice parameters at
room temperature.


A120s Lattice tensile
Lattice compressive I L
.- Net unstrained line
Thermal compressive Thermal compressive
G4N InN
.. .... .... .. ......... .


I ; Thermal tensile
. Thermal tensile6H-Sic
ITh l Lattice tensile


SSKI 11)
IZnL Si(111)c
Lattice compressive <* j









It is predicted that c-A1203 compressed more than GaN during cooling. As a result,

GaN films grown on c-Al203 substrates usually show compressive stress due to both

thermal and lattice stress. In contrast, GaN films grown on Si substrates show tensile

stress because they compress less than GaN film during cooling, in addition to tensile

stress caused by lattice mismatch at growth temperature. This figure suggests that both

6H-SiC and A1N are good substrates for GaN, due not only to close lattice parameters,

but also the closest thermal expansion coefficients. In addition, the compensation of

thermal tensile stress and lattice compressive stress will be beneficial to grow stress-free

GaN material. A net unstrained line in the figure represents complete compensation of

stress by equivalent contributions of lattice and thermal strain.

It should be noted that thermal expansion and lattice mismatches are not the only

parameters that dominate tensile and compressive stress in the film. Use of buffer layers

or nitridation will change the surface properties considerably. If the surface properties

change, the growth mode may change. Tensile stress at the grain boundaries during

islands coalescence mode would be significant, especially if the proper buffer layer is not

used. For example, GaN grown on Al203 shows cracks due to the excessive tensile

stress, though compressive stress was present due to the lattice and thermal expansion

mismatch [EtzOl].

3.3 Stress Measurements of GaN Films on Sapphire

The stress was measured in two samples grown by Mastro during comparison study

ofHVPE and MOCVD films. (p 64 in [MasO1]). One sample was a 3 gm thick GaN film

grown on sapphire by H-MOVPE and the other was a 1 gm thick GaN film grown on

sapphire by MOCVD. It should be noted that the term "MOCVD" here refers to hot wall

H-MOVPE without HC1 flow at atmospheric pressure, and should not be considered as









typical MOCVD, which often uses cold walls at lower pressure. In other words, both

samples were grown by H-MOVPE technique with and without HC1 inclusion on

sapphire substrate without buffer layer growth. Table 3-3 lists the growth conditions of

both samples in this study.

Table 3-3. The growth conditions of the two samples for stress measurements.
Samples Growth T (oC) Cl/Ga NH3 flow rate (sccm) Buffer layer
H-MOVPE GaN 950 2 500 Not used
MOCVD GaN 950 0 500 Not used
[MasOl]
3.3.1 GaN Structural and Compositional Studies

To validate the structural quality of the film and ensure the film was grown

epitaxially; the typical approach uses HR-XRD rocking curves and XRD pole figures.

counts/s

GaN-HVPE


1000- omeg-rocm g GaN-MOCVD









107


omega-2tdeta-rocking

16.0 16.5 17.0 17.5 18.0 18.5

Figure 3-2. XRD co and 0-20 rocking curve of MOCVD and H-MOVPE grown GaN on
sapphire, XRD FWHM; co-rocking curve (MOCVD: 1261 arcsec, H-MOVPE:
1790 arcsec); 0-20 rocking curve (MOCVD: 92 arcsec, H-MOVPE: 122
arcsec).









Figure 3-2 shows XRD co and 0-20 rocking curves of both the MOCVD and H-

MOVPE grown samples. The FWHM of o)-rocking curves of MOCVD and H-MOVPE

grown samples were 1261 and 1790 arcsec, respectively. The homogeneity of lattice

spacing in the crystal is measured by XRD 0-20 rocking curve and the homogeneity of

the normal axis of the lattice planes are measured by co-rocking curve. Typically it is

more likely to have uniform lattice spacing than uniform direction of lattice planes.

Therefore, the FWHM of 0-20 rocking curves are usually smaller than that of o -rocking

curve. The FWHM values of 122 and 92 arcsec were obtained from MOCVD and H-

MOVPE grown samples, respectively.

There is no clear criterion of crystallinity by only the rocking curve method, though

higher crystal quality results in smaller FWHM. XRD pole figure is often used to check

if the film is grown epitaxially. Figure 3-3 shows XRD pole figures of the (116) sapphire

and MOCVD growth GaN (112). A 300 rotation of GaN film with respect to sapphire

substrate was clearly seen because the lattice mismatch can be reduced significantly from

49 % to 14 % in this way. The results confirmed that the GaN films were indeed grown

epitaxially.


-300


n '.,--- K ,. .- ,



,...-.- -. _. _.- .. ." .* *

Figure 3-3. Pole figures for the (116) sapphire substrate (2 theta = 57.4900) and (112)
GaN by MOCVD (2 theta = 69.1850). The GaN in-plane axis is rotated by
300 with respect to the sapphire axis.









Similar results were found with the GaN grown by H-MOVPE (Figure 3-4). A 300

rotation with respect to sapphire and clear epitaxial growth were confirmed. It is noted

that the shapes of the peaks from sapphire and GaN are different. The peaks from

sapphire substrate are much sharper in shape compared to the peaks from GaN, because

the crystal quality of sapphire is superior to GaN.



S- ,300

,I I --II





Figure 3-4. Pole figures for the (116) sapphire substrate (2 theta = 57.4900) and (112)
GaN by H-MOVPE (2 theta = 69.1850). The GaN in-plane axis is rotated by
300 with respect to the sapphire axis.

Auger Electron Spectroscopy (AES) was used for chemical composition analysis.

Figure 3-5 shows the AES surface scan and depth profile of GaN on sapphire grown by

MOCVD. Ga (52.8 %), N (30.1 %), C (10.7 %), and O (6.3 %) were detected on the

surface. Surface C and O quickly disappeared within 1 min of sputtering. Since the

thickness of the GaN was reasonably thick (- 1 .im), the typical sputtering rate (- 100

A/min) would require 100 min of sputtering time to fully profile. To reduce sputtering

time, the raster size was decreased by a factor of 5, so the sputtering rate increased

fivefold. In addition, the detection sequence was reduced to decrease sampling time.

The results show that the GaN films were mainly composed of only Ga and N from the

surface to the interface. Adsorbed C and O on the surface were apparently from air

exposure.








Min: -3638 Max: 2760


dN(E)1 301
10. %
6.3 %


N1
30.1 %

50 250 450 650


Min: 0 Max: 42933

Gal


Gal
850 10 .01 0 1450 1650 1850 2050
Kinetic Energy (eV)


Gal


Gal


0 2 4 6 8 10
Time (mins.)


12 14 16 18 20


Figure 3-5. AES surface scan and depth profile of GaN on sapphire grown by MOCVD









Min: -3411 Max: 2711


9.0 %


26.4 %


50 250 450 650


850 105tO'"250 1450 1650 1850 2050
Kinetic Energy (eV)


Min: 0 Max: 45691


GN



N1


Gal







N1







C1


0 5 10 15 20 25 30 35 40 45 50
Time (mins.)
Figure 3-6. AES surface scan and depth profile of GaN on sapphire grown by H-
MOVPE.


dN(E)


Gal









Similar AES surface scanning followed by depth profile was done on H-MOVPE

GaN on sapphire as shown in Figure 3-6. The surface chemical compositions were Ga

(53.4 %), N (26.4 %), C (11.2 %), and O (9.0 %). The thickness of GaN was 3 im.

Therefore, the fivefold increased sputtering rate was again used by reducing raster size.

Uniform Ga and N profiles were obtained from the surface to the interface once more. It

is noteworthy that the sputtering rate was indeed 500 A/min. Only adsorbed surface C

and O existed just as in the previous case.

3.3.2 Stress Measurements by Raman Spectroscopy

The Raman shifts of the stress sensitive GaN E2 mode have been reported by

several researchers, and the reported values cover the large range, from 565 to 572 cm-1

[Mel97, Tab96], showing the differences of the residual stress in the epitaxial film. It

would be helpful to have the freestanding GaN E2 mode as a standard since there would

be no substrate effect. The reported shifts of the E2 mode for freestanding GaN samples,

however, show a relatively wide range from 565 to 570 cm1 [Mel97, Tab96, Dav97,

Age97]. Freestanding GaN films grown by HVPE technique may contain high impurity

levels (mainly oxygen), resulting in hydrostatic stress. The E2 mode of stress-free GaN is

known as 568 cm- at room temperature that was measured from a thick (50 70 [tm)

GaN film grown by chloride-hydride vapor phase epitaxy (CHVPE) [Dav98]. More

discussion about Raman E2 peak position is presented in Table 1-2 in Chapter 1. The

residual stress in both epitaxial films was measured by Raman spectroscopy.

The depth-dependent Raman measurements of MOCVD and H-MOVPE deposited

GaN films were carried out by a collaborator (Seok-Ki Yeo in Dr. Chinho Park's group,

Yeung Nam University, Korea) using a Renishaw System-2000 (located in Miryang