<%BANNER%>

Indium Nitride Growth by Metal-Organic Vapor Phase Epitaxy (MOVPE)


PAGE 1

INDIUM NITRIDE GROWTH BY META L-ORGANIC VAPOR PHASE EPITAXY By TAEWOONG KIM A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2006

PAGE 2

Copyright 2006 by Taewoong Kim

PAGE 3

iii ACKNOWLEDGMENTS The author wishes first to thank his advisor, Dr. Timothy J. Anderson, for providing five years of valuable advice and guidance. Dr. Anderson always encouraged the author to approach his research from the high est scientific level. He is deeply thankful to his co-advisor, Dr. Olga Kryliouk, for he r valuable guidance, sincere advice, and consistent support for the past five years. Secondly, the author wishes to thank the remaining committee members of Dr. Steve Pearton and Dr. Fan Ren for their advice and guidance The author is grateful to Scott Gapinski the staff at Microfabritech, and Eric Lambers, the staff at the Major Analytical Instrumentation Center, especially for Auger characterization. Acknowledgement needs to be given to Sangwon Kang who worked with the author for the past year and provided valuable assistance. Thanks go to Youngsun Won for his useful discussion of quantum calculation and SEM characterization, and to Dr. Jianyun Shen for her assistance about how to use the ThermoCalc. The author wishes to th ank Hyunjong Park for usef ul discussion and Youngseok Kim for his kindness and friendship. Most importantly, the author is grateful to Moonhee Choi, his beloved wife, for her endless support, trust, love sacrifice and encouragement. Without her help, he would have not finished the Ph.D. course.

PAGE 4

iv The author is grateful to his mother, fath er, mother-in-law, father-in-law, sisters, and brother for providing love, supp ort and guidance throughout his life.

PAGE 5

v TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iiiLIST OF TABLES...........................................................................................................viiiLIST OF FIGURES.............................................................................................................xABSTRACT....................................................................................................................xvi i CHAPTER 1 INTRODUCTION........................................................................................................12 LITERATURE REVIEW.............................................................................................32.1 Indium Nitride (InN) and Indium Gallium Nitride (InxGa1-xN) Properties............42.1.1 Structural Properties.....................................................................................42.1.2 Physical Properties.......................................................................................72.1.3 Electrical Properties of InN..........................................................................92.1.3.1 Background Defects...........................................................................92.1.3.2 Hall mobility and Electron C oncentration in Undoped InN............102.1.4 Optical Properties of InN............................................................................142.1.5. Indium Nitride (InN) andIindium Gallium Nitride (InxGa1-xN) Applications...............................................................................................162.2 Thermodynamic Analysis and Phase Separation in the InxGa1-xN System..........182.2.1 Thermodynamic Models in Solid Solution.................................................192.2.1.1 Regular Solution Model...................................................................192.2.1.2 Bonding in Semiconductor Solid Solutions Model..........................192.2.1.3 Delta Lattice Parameter (DLP) Model for Enthalpy of Mixing.......212.2.1.4 Strain Energy Model........................................................................222.2.1.5 First-Principal Models......................................................................232.2.2 Thermodynamic Analysis of InN...............................................................242.2.3 Phase Separation in InxGa1-xN....................................................................262.3 Indium Nitride (InN) and Indium Gallium Nitride (InxGa1-xN) Growth Challenges.............................................................................................................322.3.1 Growth Temperature and V/III Ratio.........................................................322.3.2 Nitrogen Source..........................................................................................342.3.3 Carrier Gas..................................................................................................36

PAGE 6

vi 2.4 Indium Nitride (InN) Growth Techniques............................................................372.4.1 Chemical Vapor Deposition (CVD)...........................................................372.4.1.1 Metal-Organic Vapor Phase Epitaxy (MOVPE)..............................382.4.1.2 Hydride Vapor Phase Epitaxy (HVPE)............................................412.4.1.3 Plasma Enhanced Chemi cal Vapor Deposition (PECVD)...............422.4.2 Molecular Beam Epitaxy (MBE) a nd Metalorganic Molecular Beam Epitaxy (MOMBE)....................................................................................442.4.3 Atomic Layer Deposition (ALD)...............................................................452.5 Substrate Materials...............................................................................................462.5.1 Sapphire Substrate (Al2O3) (0001).............................................................472.5.2 Silicon (Si) Substrate..................................................................................492.5.3 Gallium Nitride (GaN) and Alumin ium Nitride (AlN) Substrate..............502.5.4 Other Substrates..........................................................................................512.5.5 Buffer Layer...............................................................................................522.6 Summary for Growth of In N on Different Substrate............................................532.6.1 Growth on Sapphire (Al2O3) Substrate.......................................................532.6.2 Growth on Silicon (Si) Substrate................................................................552.6.3 Growth on Gallium Arsenide (GaAs) Substrate.........................................562.6.4 Growth on Gallium Phosphorus (GaP) Substrate.......................................572.6.5 Growth on Gallium Nitride (GaN ) and Alumimum Nitride (AlN) Substrate.....................................................................................................582.7 Overview...............................................................................................................593 THERMODYNAMIC ANALYSIS OF InN AND InxGa1-xN MOVPE GROWTH..603.1 Thermodynamic Analysis of InN and InxGa1-xN..................................................603.1.1 Reaction Mechanism and Kinetics of InN MOVPE...................................603.1.2 Pressur-Temperature (P-T) Phase Diagram of InxGa1-xN and Phase Separation in InxGa1-xN..............................................................................653.2 Quantum Calculation of Phase Separation in InxGa1-xN......................................703.2.1 Boundary Passivation Method with Hydrogen...........................................704 CALCULATION OF THE CRITICAL THIC KNESS OF InN ON GaN, AlN, Si, AND Al2O3.................................................................................................................764.1 Calculation of Critical Thickne ss of InN by Matthews’ Method.........................774.2 Calculation of Critical Thickness of InN by van der Merwe’s Method...............804.3 Calculation of Critical Thickness of InN by the Methods of Shen, Jesser, and Wilsdorf................................................................................................................865 Indium Nitride (InN) GROWTH BY METAL-ORGANIC VAPOR PHASE EPITAXY (MOVPE)..................................................................................................935.1. Indium Nitride (InN) Growth Optimization........................................................935.1.1. Substrate Selection....................................................................................945.1.1.1. Sapphire (c-Al2O3 (0001))...............................................................965.1.1.2. Gallium Nitride (GaN/c-Al2O3 (0001))...........................................97

PAGE 7

vii 5.1.1.3. Silicon (Si (111)).............................................................................975.1.2. Substrate Preparation Procedure................................................................985.1.3. Metal-Organic Vapor Phase Epitaxy (MOVPE) Reactor..........................985.1.4. Growth Chemistry and Conditions for InN Growth..................................995.1.5. Indium Nitride (InN) Growth and Optimization.....................................1025.1.5.1. Influence of Growth Temperature.................................................1025.1.5.2. Influence of Substrate Nitridation.................................................1105.1.5.3. Influence of N/In Ratio.................................................................1155.1.5.4. Influence of Buffer Layer and Morphological Study....................1185.1.5.5. Influence of Pressure.....................................................................1255.1.5.6. Optical and Electrical Properties...................................................1275.1.5.7. Summary.......................................................................................1295.1.6. Indium Nitride (InN) Droplet Formation................................................1305.1.7. Annealing Effect......................................................................................1365.2. Computational Fluid Dynamic Analysis of the Flow of NH3 and Proposed Inlet Tube Modification to Improve Flow Pattern of NH3................................1385.3. Inlet Tube Modification and Growth Results....................................................1456 CONCLUSIONS......................................................................................................154LIST OF REFERENCES.................................................................................................158BIOGRAPHICAL SKETCH...........................................................................................179

PAGE 8

viii LIST OF TABLES Table page 2-1. Lattice constants of InN...............................................................................................5 2-2. Properties of GaN and InN..........................................................................................6 2-3. Elastic constants of wurtzi te InN at room temperature...............................................7 2-4. Physical properties of InN...........................................................................................7 2-5. Carrier concentration and Hall mob ility for the different growth methods...............14 2-6. Comparison of interaction parameters calculated using various models with experimental data.............................................................................................22 2-7. Interaction parameters for va rious III-V ternary alloy systems.................................30 2-8. Properties of nitrogen precursors for MOVPE..........................................................36 2-9. Structural properties of substrates.............................................................................47 3-1. Reported reaction rate cons tants for TMIn decomposition.......................................62 3-2. Species, phases, and thermodynamic pr operties included in the analysis of MOVPE of InN................................................................................................64 3-3. Phases and species included in the analysis of MOVPE of InxGa1-xN......................67 3-4. Bond lengths for the cal culation using HF-SCF........................................................74 3-5. Calculated total energy for th ree types of different bond length...............................74 3-6. Calculated energies of InxGa1-xN with the phase separation and without phase separation.........................................................................................................75 4-1. Physical properties required for the calc ulation of the critical thickness of InN on GaN, AlN, Al2O3, and Si substrates.................................................................79 4-2. Calculated critical th ickness of InN on GaN, AlN, Al2O3, and Si substrates using Mattews’ method..............................................................................................79

PAGE 9

ix 4-3. Physical properties required for the calculation of critical thickness of InN on GaN, AlN, Al2O3, and Si substrates.................................................................82 4-4. Calculated shear moduli for InN, GaN, AlN, Al2O3, and Si materials......................82 4-5. Calculated critical thickness of InN using van der Merwe’s method........................85 4-6. Lattice constant () of InN, GaN, AlN.....................................................................88 4-7. Elastic constants cij and compliances sij of InN.........................................................89 4-8. Critical thickness ( hc) calculated of InN using three different models......................92 5-1. Structural properties of InN, GaN, Al2O3, Si, and AlN substrates............................95 5-2. Range of growth conditions examined for growth of InN.......................................101 5-3. Optimum growth temperature of In N on LT-GaN and LT-InN buffer layers on various substrates...........................................................................................110 5-4. Comparison by ESCA of Si anneals.......................................................................115 5-5. Optimum growth temperature of LT-InN buffer layer depending on Al2O3 (0001). ........................................................................................................................122 5-6. Root Mean Square (RMS) roughness fo r as-grown buffer layers and InN films....124 5-7. Growth conditions for InN......................................................................................129 5-8. Optimum growth condition of InN for Al2O3 (0001), GaN/Al2O3 (0001), Si (111)...............................................................................................................130 5-9. Density and velocity of NH3 at reactor wall and substrate......................................139 5-10. Reynolds number ( Re ) calculated in the inlet tube and in the reactor depending on temperature..............................................................................................139 5-11. Typical values of FWHM depending on different reactor systems.......................153 5-12. Reference data available for FWHM for MOVPE reactor....................................153

PAGE 10

x LIST OF FIGURES Figure page 1-1. Bandgap energies Eg of the semiconductor materials.................................................2 2-1. Lattice parameter for polycrystalline and single crystalline InN reported by different groups................................................................................................5 2-2. Carrier concentration and hall mobility reported for undoped InN film grown in a variety of technique is plotte d against the calendar year................................12 2-3. Room-temperature Hall mobility as a function of InN thickness in InN films grown by MBE, MOVPE, and MEE..............................................................13 2-4. Photoluminescence spectra for MBE grown InN......................................................15 2-5. Band gap energy for InN films as a function of carrier concentration......................16 2-6. Tetrahedral cells in a ternary III-V alloy semiconductor..........................................20 2-7. Calculated phase diagram for the MBE deposition of InN using atomic N and NH3 gases. There are three deposition modes: etching, droplet and growth..26 2-8. Free energy versus solid compositi on for a hypothetical semiconductor alloy having a large positive enthalpy of mixing. Point A and B are the bimodal points, and points C and D repr esent the spinodal points..............................27 2-9. Schematic liquid-solid ps eudobinary phase diagram................................................27 2-10. Binodal (solid) and sp inodal (dashed) curves for the InxGa1-xN system, calculated assuming a constant av erage value for the solid phase interaction parameter....................................................................................28 2-11. Schematic illustration of the key CVD steps during deposition..............................38 2-12. Schematic of horizontal cold-wall MOVPE system................................................39 2-13. Schematics of horizontal hot-wa ll hydride vapor phase epitaxy chamber..............42 2-14. Schematics of PECVD.............................................................................................43

PAGE 11

xi 2-15. Perspective views in (2 2 1) unit cell: (a) along [0001] direction in a rhombohedral unit cell; (b) along the (0001) direction in hexagonal unit cell................................................................................................................48 2-16. Common facets of sapphire crystals: (a) view down c -axis; (b) surface planes......48 2-17. Perspective views of Si along various directions: (a) [001]; (b) [011]; (c) [111]...49 2-18. Perspective views of wurtzite Ga N along various directions: (a) [0001]; (b) [112 0]; (c) [101 0].......................................................................................50 2-19. Perspective views of zincblende GaN along various directions: (a) [100] (111 unit); (b) [110] (2 2 2 units); (c) [111] (2 2 2 units)...........................51 3-1. Calculated P-T phase diagram for InN at X(In) = 5.31212 10-6, X(N) = 0.24998, X(H) = 0.75000, X(C) = 1.59364 10-5 and V/III = X(N)/X(In) = 50,000..65 3-2. Relation between indi um mole fraction (x) of InxGa1-xN and the flow rate ratio of the sum of group III source of TMI and TEG................................................68 3-3. Calculated P-T phase diagram for In0.3Ga0.7N at X(In)=1.87328 10-5, X(Ga)=3.05276 10-5, X(N)=0.111, X(H)=0.8887, X(C)=2.39364 10-4 and the data points ( ) are from the measuremen ts observed by Matsuoka..68 3-4. Thermodynamically calculated miscibility gap of InxGa1-xN grown by MOVPE and the data points ( ) are from the measurements observed by Piner et al .....................................................................................................................70 3-5. Flow chart of the HF-SCF procedure........................................................................72 3-6. Structures used to com pute the total energy for the InxGa1-xN vs. indium mole fraction...........................................................................................................73 4-1. Schematic representation of the forma tion of misfit dislocations: (a) unstrained lattice; (b) thickness of the film is less than hc; (c) thickness of the film is greater than hc misfit dislocations are generated............................................77 4-2. Model of epitaxial interface between tw o semi-infinite crystals resolved in a sequence of misfit dislocations spaced at an average distance p ...................81 4-3. Homogeneous strain energy, Ehs for the 1st InN epilayer and dislocation energy, Ed vs. the misfit, f on GaN substrate..............................................................83 4-4. Homogeneous strain energy, Ehs for the 1st InN epilayer and dislocation energy, Ed vs.the misfit, f on AlN substrate................................................................84

PAGE 12

xii 4-5. Homogeneous strain energy, Ehs for the 1st InN epilayer and dislocation energy, Ed vs. the misfit, f on Al2O3 substrate............................................................84 4-6. Homogeneous strain energy, Ehs for the 1st InN epilayer and dislocation energy, Ed vs. the misfit, f on Si substrate...................................................................85 4-7. Total energy Et, strain energy Es, interfacial energy Ei vs. lattice constant of InN for 1st epilayer InN on GaN substrate.............................................................89 4-8. Total energy Et, strain energy Es, interfacial energy Ei vs. lattice constant of InN for 1st epilayer InN on AlN substrate.............................................................90 4-9. Total energy Et, strain energy Es, interfacial energy Ei vs. lattice constant of InN for 1st epilayer InN on Al2O3 substrate...........................................................90 4-10. Total energy Et, strain energy Es, interfacial energy Ei vs. lattice constant of InN for 1st epilayer InN on Si substrate........................................................91 5-1. Schematic for the depos ition of the (0001)//(0001), [010 1]//[ 10 2 1] GaN/Al2O3 system.............................................................................................................96 5-2. Planes of Si (111) substrate.......................................................................................96 5-3. Image and schematic of horizontal, cold-wall MOVPE reactor system....................99 5-4. Indium Nitride (InN) growth seque nce for each of the three substrates.................101 5-5. X-ray Diffraction (XRD) -2 scans for InN/LT-GaN on (a) Al2O3 (0001) at N/In = 3000, T = 450, 550, 650, and 750 oC................................................102 5-6. X-ray Diffraction (XRD) -2 scan for InN/LT-GaN on (a) Al2O3 (0001) at N/In = 3000, T = 450, 550, 650, and 750 oC. Pure In was removed by etching with HCl.......................................................................................................104 5-7. X-ray Diffraction (XRD) -2 scan for InN/LT-GaN on Al2O3 (0001) at N/In = 50,000, T = 530, 550, and 570 oC................................................................105 5-8. X-ray Diffraction (XRD) -2 scan for InN/LT-InN on Al2O3 (0001) at N/In = 50,000 and T = 500, 530, and 550 oC...........................................................105 5-9. Full Width Half Maximum (F WHM) of XRC for InN/LT-InN on Al2O3 (0001) at N/In = 50,000 and T = 500, 530, and 550 oC...............................................106 5-10. X-ray Diffraction (XRD) -2 scan for InN/LT-InN on GaN/Al2O3 (0001) at N/In = 50,000, T = 500, 530, and 550 oC.....................................................106 5-11. Full Width Half Maximum (FWH M) of XRC for InN/LT-InN on GaN/Al2O3 (0001) at N/In = 50,000, T = 500, 530, and 550 oC...................................107

PAGE 13

xiii 5-12. X-ray Diffraction (XRD) -2 scan for InN/LT-GaN on Si (111), at N/In = 50,000, T = 500, 530, 550, and 570 oC......................................................108 5-14. Growth rate of InN on variou s substrates (a) InN/LT-GaN on Al2O3 (0001) at N/In = 3000, (b) for InN/LT-GaN on Al2O3 (0001) at N/In = 50,000, (c) InN/LT-InN on GaN/Al2O3 (0001) at N/In = 50,000, and (d) InN/LT-InN on Si (111) at N/In = 50,000......................................................................109 5-15. Cross-sectional SEM micrographs of InN for 60 min growth at 530, 550, and 570 oC, and N/In = 50,000 with LT-GaN buffer........................................110 5-16. X-ray Diffraction (XRD) -2 scan for InN/LT-InN (TLT-InN = 450 oC) on Al2O3 (0001) at N/In = 50,000, and TLT-InN = 450 oC without and with nitridation...................................................................................................111 5-17. X-ray Diffraction (XRD) -2 scan for InN/LT-InN GaN/Al2O3 (0001) at N/In = 50,000, T = 530 oC, and TLT-InN = 450, 500 oC with and without nitridation...................................................................................................112 5-18. X-ray Diffraction (XRD) -2 scan for (a) InN/LT-InN on Si (111), at N/In = 50,000, T = 530 oC, and TLT-InN = 450, 500 oC with the nitridation and without the nitridation................................................................................113 5-19. Electron Spectroscopy of Chemi cal Analysis (ESCA) spectra of Si 2p3 peak for Si annealed at 850 oC in 1.0 slm N2 (a) with 100% NH3 at 1.0 slm (b) without NH3................................................................................................114 5-20. X-ray Diffraction (XRD) -2 scan at N/In=20,000, 30,000, and 50,000, T = 550 oC for InN/LT-GaN on Al2O3 (0001) at N/In of 50,000......................115 5-21. Full Width Half Maximum (F WHM) of XRC for InN/LT-GaN on Al2O3 (0001) at N/In of 50,000........................................................................................116 5-22. X-ray Diffraction (XRD) -2 scan for InN/LT-InN on GaN/Al2O3 (0001) at, T = 530 oC, TLT-InN = 400 oC, and N/In = 30,000 and 50,000........................116 5-23. X-ray Diffraction (XRD) -2 scan for InN/LT-GaN on Si (111) at N/In = 20,000, 30,000 and 50,000, T = 530 oC.....................................................117 5-24. Growth rate vs. N/In ratio for InN/LT-GaN on Al2O3 (0001), GaN/Al2O3 (0001), and Si (111) at N/In = 6000, 9000, 12,000, and 15,000 with T = 550 oC, TMI = 0.26 sccm and NH3 = 1600-4000 sccm...........................................118 5-25. X-ray Diffraction (XRD) -2 scan for InN/LT-InN on Al2O3 (0001) at N/In = 50,000, T = 530 oC, and TLT-InN = 400, 450, and 500 oC............................119 5-26. X-ray Diffraction (XRD) -2 scan for InN/LT-InN on GaN/Al2O3 (0001) at T = 530 oC, TLT-InN = 350, 400, 450, and 500 oC, and N/In = 50,000..............120

PAGE 14

xiv 5-27. Full Width Half Maximum (FWH M) of XRC for InN/LT-InN on GaN/Al2O3 (0001) at T = 530 oC, TLT-InN = 350, 400, 450, and 500 oC, and N/In = 50,000.........................................................................................................120 5-28. X-ray Diffraction (XRD) -2 scan for InN/LT-InN on Si (111) at N/In = 50,000, T = 530 oC, and TLT-InN = 400, 450, and 500 oC........................121 5-29. X-ray Diffraction (XRD) -2 scan for InN/LT-InN on Si (111) at N/In = 50,000, T = 530 oC, TLT-InN = 450 oC, and t = 5, 15, and 30 min...............121 5-30. X-ray Diffraction (XRD) -2 scan for InN/LT-InN (TLT-InN = 450 oC) and LT-GaN (TLT-GaN = 560 oC) on Al2O3 (0001) at N/In = 50,000, T = 530 oC (LT-InN buffer) and T = 550 oC (LT-GaN buffer)........................122 5-31. Full Width Half Maximum (FWHM) of XRC for InN/LT-InN (TLT-InN = 450 oC) and LT-GaN (TLT-GaN = 560 oC) on Al2O3 (0001) at N/In = 50,000, T = 530 oC (LT-InN buffer) and T = 550 oC (LT-GaN buffer).....123 5-32. Root Mean Square (RMS) roughness by ATM for (a) InN/LT-InN (T = 530 oC, TLT-InN = 450 oC), (b) InN/LT-GaN (T = 550 oC, TLT-GaN = 560 oC), (c) asgrown LT-InN (450 oC), and (d) as-grown LT-GaN (560 oC) on Al2O3 (0001) at N/In=50,000................................................................................124 5-33. X-ray Diffraction (XRD) -2 scan for InN/LT-InN and InN/LT-GaN on Si (111) at N/In = 50,000, T = 530 oC, TLT-InN = 450 oC, and TLT-InN = 560 oC................................................................................................................125 5-34. X-ray Diffraction (XRD) -2 scan for InN/LT-InN on GaN/Al2O3 (0001) at N/In = 50,000, TLT-InN = 450 and 500 oC and T = 530 oC with the different growth pressure of LT-InN..........................................................................126 5-35. X-ray Diffraction (XRD) -2 scan for InN/LT-InN on Si (111) at N/In = 50,000, TLT-InN = 450 and 500 oC and T = 530 oC with two different growth pressures for LT-InN......................................................................126 5-36. Photoluminescence for (a) InN grown on Al2O3 (0001) at T = 530 oC, TLT-InN = 500 oC, (b) InN grown on GaN/Al2O3 (0001) at T = 530 oC, TLT-InN = 400 oC, (c) InN grown on Si (111) substrate at TLT-GaN = 560 oC, T = 550 oC and N/In = 50,000.......................................................................................128 5-37. Carrier concentrations and mobilities of InN films grown with different growth conditions at different characte rization temperature using Hall measurement...............................................................................................128 5-38. Carrier concentration and mobility of InN on Si (111) at different characterization temperature using Hall measurement..............................129

PAGE 15

xv 5-39. Scanning Electron Microscopy (S EM) and EDS for the surface of InN/LT-GaN on Al2O3 (0001) at N/In = 3000, 6000, 9000, 20,000, 30,000, and 50,000.131 5-40. Number density of indium droplets vs. N/In ratio depending on different N/In, when InN was grown on Al2O3 (0001) at T = 550 oC and N/In = 3000, 6000, 9000..................................................................................................132 5-41. Percent (%) vs. indium droplet size depending on different N/In, when InN was grown on Al2O3 (0001) at T = 550 oC and N/In = 3000, 6000, 9000.........132 5-42. X-ray Diffraction (XRD) -2 scans for InN/LT-GaN on Al2O3 (0001) at N/In = 3000, T = 450, 550, 650, and 750 oC before HCl wet etching...................133 5-43. X-ray Diffraction (XRD) -2 scans for InN/LT-GaN on Al2O3 (0001) at N/In = 3000, T = 450, 550, 650, and 750 oC after HCl wet etching......................133 5-44. X-ray Diffraction (XRD) -2 scan for InN/LT-GaN on Si (111) at N/In = 3000, T = 450, 550, 650, and 750 oC before HCl wet................................134 5-45. X-ray Diffraction (XRD) -2 scan for InN/LT-GaN on Si (111) at N/In = 3000, T = 450, 550, 650, and 750 oC after HCl wet etching......................134 5-46. Characterization result by AES for In droplets formed during the growth of InN on Al2O3 (0001) with LT-GaN buffer layer before the HCl wet etching after the HCl wet etching at N/In = 3000...................................................135 5-47. Full Width Half Maximum (F WHM) of XRC of InN/LT-InN (450 oC) on Al2O3 (0001) at T = 450 oC in N2 flow with different annealing time (0, 10, 30, 60 and 90 min)............................................................................................137 5-48. Full Width Half Maximum (F WHM) of XRC of InN/LT-InN (400 oC) on GaN/Al2O3 (0001) at T = 450 oC in N2 flow with different annealing time (0, 10, 30, and 60 min).......................................................................137 5-49. Schematic for three types of inlet tubes used for the Fluent simulation................141 5-50. Flow of NH3 in the reactor with the current inlet tube..........................................142 5-51. Flow of NH3 1 mm above the surface of substrat e with the current inlet tube.....142 5-52. Flow of NH3 in the reactor with the horizontally extended inlet tube...................142 5-53. Flow in the reactor with the vertical inlet tube......................................................143 5-54. Flow of NH3 1 mm above the surface of substrate with the vertical inlet tube.....143 5-55. X-ray Diffraction (XRD) -2 scan for InN/LT-InN on Al2O3 (0001) at N/In = 50,000, T = 530 oC and TLT-InN = 450 oC with different inlet tubes...........146

PAGE 16

xvi 5-56. Full Width Half Maximum (F WHM) of XRC for InN/LT-InN on Al2O3 (0001) at N/In = 50,000, T = 530 oC and TLT-InN = 450 oC with different inlet tubes...........................................................................................................146 5-57. Cross-sectional SEM for InN/LT-InN on Al2O3 (0001) at N/In = 50,000, T = 530 oC, and TLT-InN = 450 oC with the horizontal and vertical inlet tubes..147 5-58. X-ray Diffraction (XRD) -2 scan for InN/LT-InN on GaN/Al2O3 (0001) at N/In = 50,000, T = 530 oC and TLT-InN = 450 oC with different inlet tubes.147 5-59. Full Width Half Maximum (FWH M) of XRC for InN/LT-InN on GaN/Al2O3 (0001) at N/In = 50,000, T = 530 oC and TLT-InN = 450 oC with different inlet tubes...................................................................................................148 5-60. Cross-sectional SEM for InN/LT-InN on GaN/Al2O3 (0001) at N/In = 50,000, T = 530 oC and TLT-InN = 400 oC with the horizontal and the vertical inlet tubes...........................................................................................................148 5-61. Grazing Angle Incident X-ray Diffraction (GIXD) for InN grown on Al2O3 (0002) with the incident angle of 1 de gree when the horizontal inlet tube was used.....................................................................................................149 5-62. Grazing Angle Incident X-ray Di ffraction (GIXD) for InN grown on GaN/Al2O3 (0002) with the incident angle of 1 de gree when the horizontal inlet tube was used.....................................................................................................150 5-63. Grazing Angle Incident X-ray Di ffraction (GIXD) for InN grown on (a) Al2O3 (0002) and (b) GaN/Al2O3 (0002) with the incident angle of 1 degree when the vertical inlet tube was used.........................................................151 5-64. X-ray Diffraction (XRD) -2 scan of InN/LT-InN/Ga N/LT-GaN on Si (111) at N/In = 50,000, T = 530 oC and TLT-InN = 400 oC for both horizontal and vertical inlet tubes......................................................................................151 5-65. Full Width Half Maximum (F WHM) of XRC of InN/LT-InN on Al2O3 (0001) at N/In = 50,000, T = 530 oC for both horizontal and vertical inlet tubes. The annealing test is performed at T = 450 oC for 30 min.........................152 5-66. Full Width Half Maximum (FWH M) of XRC of InN/LT-InN on GaN/Al2O3 (0001) at N/In = 50,000, T = 530 oC for both horizontal and vertical inlet tubes. The annealing test is performed at T = 450 oC for 30 min..............152

PAGE 17

xvii Abstract of Dissertation Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy INDIUM NITRIDE GROWTH BY META L-ORGANIC VAPOR PHASE EPITAXY By Taewoong Kim August 2006 Chair: Timothy J. Anderson Major Department: Chemical Engineering InN and In-rich compositions of InxGa1-xN, have potential for a variety of device applications including solar cells. This work addresses the growth of high quality InN by metalorganic vapor phase epitaxy. To bette r understand the material a thermodynamic assessment of the In-N-C-H sy stem was performed to yield the In-N P-T diagram. In addition, the InN critical thickness was calculated for several candidate substrates to guide substrate selection. Furthermore, com putational fluid dynamics was used to design an improved reactor. A vertical NH3 tube design produced the lowest reported -2 rocking curve FWHM value of (574 arcsec) for InN grown on GaN/Al2O3 (0001). The film surface was also mirror-like as judged by AFM (RMS roughness = 4.2 nm). The PL peak energy of 0.82 eV was obtained for InN grown on Si, consistent with recent reports of a considerably lower of bandgap energy.

PAGE 18

1 CHAPTER 1 INTRODUCTION During the last few years the interest in indium nitride (InN) has brought remarkable attention due to its highly attractiv e inherent properties, such as high mobility and high saturation velocity [Yam02 and Yam04b]. Epitaxial growth of InN films by meta l-organic vapor phase epitaxy (MOVPE), was first reported by Matsuoka et al. and Wakahara et al. in 1989 independently [Mat89, Wak89]. In the 1990s, epitaxial growth of In N films was performed by several scientists [Wak90, Yam94a, Yam94b, Guo95a, Guo95b, Uch96, Che97, Yam97a, Sat97a, Yam98a, Yan99, Tsu99, Pan99, Yam99a]. These st udies included the growth by MOVPE and MBE on different substrates su ch as Si, GaAs, GaAsB, Al2O3 and GaP over a wide range of growth conditions but had not shown any good results, however, no high quality films were produced. Because of the low decomposition temperature of InN (~ 650 oC), poor lattice matched substrate, high equilibrium pressure of nitrogen, and the low cracking efficiency of NH3 at the growth temperature, the growth of high quality InN film is challenging. Since the bandgap energy of InN has recently been discovered to be ~ 0.7 eV,[Wu02] the use of InN with GaN and AlN make it possible to extend the emission of nitrided-based LEDs from ultr aviolet to infrared regions. The bandgap energies of the semiconductor materials are shown in Fig.1. 1. Alloying InN with GaN creates an InxGa1xN active layer that is suitable for light emitting devices because InxGa1-xN considerably

PAGE 19

2 increases luminescence efficiency due to the localized energy states formed by alloy composition fluctuations of the InxGa1-xN [Nak92]. Figure 1-1. Bandgap energies Eg of the semiconductor materials. InN was predicted to have the lowest effective mass for electrons among all IIInitride semiconductors, which leads to high mo bility and high saturation (drift) velocity. The theoretical maximum mobility cal culated in InN at 300K is 4400 cm2 V-1S-1 (GaN ~ 1000 cm2 V-1S-1) [Chi94]. It was found that InN exhibits an extremely high peak drift velocity of 4.2107 cm/s [Bel99]. Thus, InN is promising as a highly potential material for the fabrication of high-speed high-performance heterojunction field-effect transistors (FETs). The use of wurtzite InN would permit photonic devices in infrared and much faster electronic devices, because it could induce higher mobility and high peak (saturation) velocity than most ot her III-nitride based materials. Single crystalline epitaxial InN films by MOVPE were first repor ted in [Mat98 and Wak89]. The typical FWHM of single crys talline InN grown by MOVPE is 4000 5500 arcsec [Che97]. Significant improvement in the growth of InN film has been made by

PAGE 20

3 MOVPE during the last few years. The aver age reported data on FWHM of the X-ray Rocking Curve (XRC) is ~2000 arcsec for si ngle crystal InN [Y am01b, Yan02a, and Yam04a] which may indicate that the highest crystalline qualit y of this material has not been achieved yet. The high crystalline quality InN can reduce the leakage current and extend the lifetime of the laser diodes (LDs) due to reduced disloca tions. Therefore, the growth of high quality crysta lline InN is essential to obta in high performance devices. The highest mobility and lowest background carrier concentration of InN by MOVPE are reported to be 900 cm2/Vs and 51018 cm-3, [Yam04b]. Better results were achieved using molecular beam epitaxy (MBE), and the highest reported data on mobility and lowest background carrier concentration of InN are 2050 cm2/Vs and 3.491017 cm-3, respectively [Lu02a]. In addition, InN has potential for highly efficient low cost solar cell. Yamamoto et al. proposed InN for a top cell material of a two-junction tandem solar cell [Yam94a]. In summary, InN is a very attractive materi al for semiconductor de vice applications and high structural quality, low defect density material with high mobility and low carrier concentration has not been achieved to date. Th erefore, more research is needed for the improvement of crystalline quality, transport an d optical properties of InN epitaxial films by MOVPE. CHAPTER 2 LITERATURE REVIEW

PAGE 21

4 CHAPTER 2 LITERATURE REVIEW 2.1 Indium Nitride (InN) and Indium Gallium Nitride (InxGa1-xN) Properties In this section, the fundamental properties of InN and InxGa1-xN such as structural, physical, electrical, and optical properties will be discussed. The understandings of these properties are important when selecting a suit able substrate and buffer layer for obtaining high quality InN and InxGa1-xN films. Also, possible applications of InN and InxGa1-xN will be assessed. 2.1.1 Structural Properties Lattice parameters of the wurtzite crystalline structure of InN was first reported as a = 3.53 and c = 5.69 [Juz38]. However, the latt ice parameter in the rf-sputtered InN film measured by Tansley and Folsey, a = 3.548 and c = 5.760 [Tan86a], showing a slight increase in the lattice para meter values, which also differs from the lattice parameter measured in th e rf-sputtered InN film by Kubota et al ., a = 3.540 and c = 5.705 [Kub89]. The crystalline quality of InN obtained by Kubota et al was higher than the other previously reported InN films a nd the lattice parameter is much closer with the lattice parameter measured in the sing le crystalline InN film. Based on the recent results reported by Davydov, the lattice para meter in the high qua lity single crystal hexagonal InN film was reported to be a = 3.5365 and c = 5.7039 [Dav02a]. The lattice parameters for polycrystalline and single crystalline InN reported by several groups are plotted in Fig. 2.1 [Bhu03b]. Probable reasons fo r the variation in lattice parameter are different crystalline qual ity and oxygen incorporation [Yam03].

PAGE 22

5 pattern [Lim99]. At 2002, Bhattacharya et al. reported the observation of zincblende phase in InN thin film grown by pulsed la ser deposition (PLD) a nd measured a lattice constant of 5.090.04 [Bha02]. The differe nt lattice constants of InN were summarized in Table 2-1. Figure 2-1. Lattice parameter for polycrystal line and single crystalline InN reported by different groups. Table 2-1. Lattice constants of InN. Structure a () c () References Wurtzite 3.53 5.69 Zuda and Hahn [Juz38] Wurtzite 3.548 5.760 Tansley and Foley [Tan86a] Wurtzite 3.540 5.705 Kubota [Kub89] Wurtzite 3.5365 5.7039 Davydov [Dav02a] Zincblende 4.98 Strite [Str93] Zincblende 4.98-5.04 Lima [Lim99] Zincblende 5.090.04 Bhattacharya [Bha02] InxGaxN films were usually deposited on GaN buffer layers, because the lattice constant of InxGaxN is closer to that of GaN than sapphire when the mole fraction of indium (x) in InxGaxN is less than 0.3. The indium mole fraction of InxGa1-xN films is estimated from the lattice constant along with the c axis measured by x-ray diffraction, assuming that the lattice constant

PAGE 23

6 changes linearly with the indium mole fract ion given (Eq. 2-1). In the calculation, aGaN = 3.189 , cGaN = 5.178 , aInN = 3.548 and cInN = 5.7034 [Nak02, Qia02, Yos91]. The fundamental properties of GaN and InN are listed below in Table 2-2. GaN InN N x Ga x Ina x a x a ) 1 (1 GaN InN N x Ga x Inc x c x c ) 1 (1 (2-1) Table 2-2. Properties of GaN and InN. Fundamental properties of GaN F undamental properties of InN Wurtzite type: Band gap energy Eg(300K)= 3.39 eV Eg(1.6K)= 3.50 eV Temperature coeff. dEg/dT=-6.010-4eV/K Pressure coefficient dEg/dT= 4.210-3 eV/kbar Lattice constants a=3.189 c=5.185 Thermal expansion a/a= 5.5910-6 K c/c= 3.1710-6 K Thermal conductivity =1.3 W/cmK Index of refraction n (1 eV) = 2.33 n (3.38 eV) = 2.67 Zincblende polytype: Band gap energy Eg(300K)= 3.2-3.3 eV Lattice constants a= 4.52 Index of refraction n(3 eV ?)= 2.9 Wurtzite type: Band gap energy Eg(300K)= 0.6-0.9 eV Temperature coeff. dEg/dT=-1.810-4 eV/K Lattice constantsa a=3.537 c=5.704 Thermal expansion a/a410-6K c/c310-6K Thermal conductivity =0.8 0.2 W/cm K Index of refraction n=2.9-3.05 Zincblende polytype: Band gap energy Eg(300K)= 2.2 eV Lattice constantb a=5.09 [bBha02, aDav02a, Mor94] In summary, the different lattice constant s of InN obtained by several scientists were discussed. The difference of lattice constants is thought to be caused by the difference in the crystalline quality of InN. The lattice constant of InxGaxN can be calculated by using the Vegard’s law with the lattice constants of InN and GaN.

PAGE 24

7 2.1.2 Physical Properties Directly measured density of wurtzite InN is 6.89103 kg m-3 at 25 C [Hah40]. A comparable value of 6.81103 kg m-3 has been estimated from X-ray data [Pea67]. The cell volume, taken in conjunction with a molar mass of 128.827 g mol-1, yields densities of (6.810.05)103 kg m-3 and 6.9710 kg m-3 for the wurtzite and zinc blende polytypes, respectively. Bulk modulus has been calcula ted from first principles by a local-density approximation [Cam90] and by a linear muffin -tin orbital method [Kub89], suggesting a value of B = 165 GPa. The five distinguishable second-order elastic moduli in a hexagonal crystal are c11, c12, c13, c33 and c44. Other researchers have utilized em pirical and theoretical approaches to calculate the thermoelasti c properties of the wurtzite structure InN [She91, Kim96a, Wri97, Mar98, Chi99]. Table 2-3 summarizes the room-temperature elastic constants from both experimental and th eoretical results. Estimates of the principal transverse and longitudinal elastic constants ct and cl are given in Table 2-4. Table 2-3. Elastic constants of wu rtzite InN at room temperature. Elastic constants Sheleg and Savastenko [She79] Kim et al. [Kim96a] Wright [Wri97] Marmalyuk et al. [Mar98] Chisholm et al. [Chi99] C11 (GPa) C12 (GPa) C13 (GPa) C33 (GPa) C44 (GPa) 190 104 121 182 9.9 271 124 94 200 46 223 115 92 224 48 257 92 70 278 68 297.5 107.4 108.7 25.05 89.4 [Wan01] Table 2-4. Physical properties of InN. Property Value Ref. Comments

PAGE 25

8 Density (wurtzite) Density (zinc blende) Molar mass Mol. Vol. (wurtzite) Mol. Vol. (zinc blende) ct cl Deformation potential TO LO 6.89103 kg m-3 (6.810.05)103 kg m-3 6.97103 kg m-3 128.827g mol-1 31.2 3 30.9 3 4.421011 dyn cm-2 2.651012 dyn cm-2 7.1 eV 59.3 meV (478 cm-1) 57.1 meV (460 cm-1) 86.2 meV (694 cm-1) 89.2 meV (719 cm-1) H. Hahn S. Strite V. W. Chin V. W. Chin V. W. Chin K. Osamura T. L. Tansley K. Osamura T. L. Tansley Meas. by displacement Various X-ray data X-ray data From lattice constants From lattice constants Estimate Estimate Estimate Reflectance meas. Transmission meas. Est.-Brout sum rule Est.-Brout sum rule [Edg94], (reprinted from the Institute of Electrical Engineers w ith the permission of INSPEC) The piezoelectric constant has not been re ported, but its dependence on the dielectric constants r and e14 [Wol89] allows values of about 50 % of those found in AIN to be inferred [Chi94]. Indium nitride has twelve phonon modes at the zone centre (symmetry group C6v), three acoustic and nine optical with the acousti c branches essentially zero at k = 0. The IR active modes are E1 (LO), E1(TO), A1(LO) and A1(TO). A transverse optical mode has been identified at 478 cm-1(59.3 meV) by reflectance and 460 cm-1 (57.1 meV) by transmission [Tan88]. In both reports the location of a longitudinal optical mode is inferred from the Brout sum rule, giving respective values of 694cm-1 (86.1 meV) and 719cm-1 (89.2 meV). In summary, the physical prop erties of InN films were br iefly discussed, especially the elastic constants used to calculate the strain energy and thus estimate the critical thickness of InN film.

PAGE 26

9 2.1.3 Electrical Properties of InN 2.1.3.1 Background Defects As-grown InN is always n-type with a ve ry high background carrier concentration. There has been much speculation as to what species is res ponsible for the high background donor concentration in InN. Po tential candidates for such high background donors are native defects, such as N vacancy or nitrogen antisite, and impurities, such as ON, SiIn, and possibly interstitial H. According to the oldest and most common view, the nitrogen vacancy is the most probable reason for n-type conductivity of InN. Tansley and Foley [Tan84b] had speculated that the n-type behavior is cause d by an antisite defect: N on an In site (NIn), which they had suggested mi ght be a double donor. Jenkin s and Dow [Jen89] showed that the native defect responsible for naturally occurring n -type InN is a nitrogen vacancy. Another defect po ssibly responsible for the n-type character of InN is oxygen on an N site, which is not a native defect but is nevertheless likely to be present in significant concentration. It is most likely that every nitrogen vacancy donates a single donor but possibly donates three electrons to the conduction band [Jen89]. Tansley and Egan [Tan92a, Tan92b] have also speculate d that the N vacancy might be the defect responsible for natural n-type character of InN. There is a simple approach to how the nitrogen vacancy contributes a donor in the as -grown InN film. The donor nature of the N vacancy is constructed as a missing N atom surrounded by four indium atoms that provide three valence electrons to complete the bonding octet w ith the five missing electrons of nitrogen. Two of these three electrons would be don ated to the conduction band. Therefore, it has been believed that ni trogen vacancy is the dominant donor in the as-grown InN film [Yam01, Yam02].

PAGE 27

10 In contrast with the above views, there ar e also some theoretical and experimental evidence, which argues against the nitr ogen vacancy being responsible for the background n-type conductivity. Stampfl et al. [Sta00] performed firs t-principles densityfunctional calculation to investigate the elec tronic and atomic structure and formation energies of native defects an d selected impurities (O, Si, and Mg) in InN. Their calculation showed that oxyge n and silicon impurities act as donors and that they can easily be incorporated during growth. At 2002, Look et al. [Loo02] presented a rule to determine donor and acceptor concentrations in degenerate InN. From a comparison with glow discharge mass spectroscopy measurement and the developed theory, they suggested that a potential candidate for the dominant donor in InN is H. However, the native de fects also cannot be completely ruled out. As discussed above both theoretical cal culation and experimental result give conflicting views and opinions regarding th e major reasons responsible for high n-type conductivity of as-grown InN film. However, on the basis of the data available in the literature, two major reasons can be concluded. One is native defects, mainly nitrogen vacancy, and one is impurities, mainly oxygen. 2.1.3.2 Hall mobility and Electron Co ncentration in Undoped InN The carrier concentrations and Hall mobilities reported for undoped InN films grown by a variety of techniques are plotted against the calendar year in Fig. 2.2 [Bhu03b]. The growth methods are divided into fi ve categories: molecular beam epitaxy (MBE), metal-organic chemical vapor pha se epitaxy (MOVPE), hydride vapor phase epitaxy (HVPE), sputtering, and others, includ ing electron beam plasma method, reactive

PAGE 28

11 evaporation and pulsed laser deposition. Un til the 1980s most of the InN films were deposited using sputtering. The grown film s were polycrystalline with a carrier concentration scattered from 1018 to 1021 cm-3 and Hall mobility from 20 to 250 cm2/Vs with the exception of the results obta ined by Tansley and Foley [Tan84a]. Tansley and Foley [Tan84a] attained a dramatic reduction of the carrier concentration with very high electron mobility A room temperature electron mobility of 2700 cm2/V s, which reached a maximum value of 5000 cm2/V s at 150 K, was measured. These are the best electrical proper ties ever reported in In N. It should be noted that the InN was a polycrystalline. Unfo rtunately, the InN film prepared by reactive sputtering in other laboratories has not met these results of Tansley and Foley and has universally high carrier concentration near 1020 cm-3 and constantly low electron mobility of less than 100 cm2/Vs. The InN film grown by differe nt techniques al so showed the high carrier concentrations and low electron mobility. Sato [Sat97b] achieved a car rier concentration of 41019 cm-3 in the InN epitaxial layer grown on sapphire subs trate by plasma-assisted MOVPE in 1997. However, there is no further improvement or report on the electr ical properties of InN by plasma-assisted MOVPE. The significant improvements in conventional MOVPE grown InN films started with the work of Yamamoto et al. and Pan et al. [Yam98a, Pan99] in which they reported an electron concentration of 51019 cm-3 with a Hall mobility of about 300 cm2/V s in the InN film grown on sapphire substrate.

PAGE 29

12 Figure 2-2. Carrier concentration and hall m obility reported for undoped InN film grown in a variety of technique is plot ted against the calendar year. Yamaguchi et al. [Yam99a] showed that using a GaN underlying layer increased InN film thickness and significantly improve the Hall mobility. A Hall mobility of about 700 cm2/V s was obtained in the InN film grown on GaN even at an electron concentration of 51019 cm-3. Yamamoto et al. [Yam98a, Yam01, Yam02] showed a high NH3/TMI molar ratio and enhanced NH3 decomposition (by growth temperature, atmospheric pressure growth, reduced flow velocity, etc.) significantly improved the electrical properties of MOVPE grown InN film. As a result, a carrier concentration in the order of 1018 cm-3 and the electron mobility of 730 cm2/Vs were reported. Recently, Yamamoto et al. [Yam04b] also reported a carrier concentration of 51018 cm-3 and the electron mobility of 900 cm2/V s for the MOVPE grown InN film. Laser-assisted MOVPE has the potential to decompose NH3 photolytically independent of the substrate temperature [Bhu02a].

PAGE 30

13 Lu et al. [Lu00] have obtained an electron concentration of 31018 cm-3 with a Hall mobility of 542 cm2/V s in the InN film grown by M EE (Migration Enhanced Epitaxy). They also showed that the Hall mobility for both growth methods, MEE and MBE, increases with film thickness. Similar thic kness dependence in Hall mobility was also observed in the MOVPE grown InN film [Y am99a]. The thickness dependence of the Hall mobility is presumed to be caused by the reduced defect density away from the lattice-mismatched substrate. Higashiwaki an d Matsui [Hig02] found that there was an immediate sharp increase in mobility up to a film thickness of 150 nm, beyond which it almost leveled out. The room-temperature Hall mobility as a function of InN thickness in the InN film grown by MBE, MPVPE, and MEE is shown in Fig. 2.3 [Hig02a]. Lu et al. [Lu02a] have achieved a carrier concentration in the order of 1017 cm-3 and a mobility of more than 2000 cm2/V s for the thick InN f ilm grown on HVPE grown on bulk GaN template. The use of a buffer layer of AIN, GaN or InN seems to contribute to the improvement of structural and electrica l properties of MBE grown InN. The better electrical properties in the MBE InN film compared with the MOVPE are believed to be because the active nitrogen can be supplied i ndependently of the growth temperature and reduced impurity incorporat ion in the MBE growth. Figure 2-3. Room-temperature Hall mobility as a function of InN thickness in InN films grown by MBE, MOVPE, and MEE.

PAGE 31

14 Table 2-5. Carrier concentration and Hall mobility for the different growth methods. Growth methods Carrier concentration (cm-3) Hall mobility (cm2/V s) References MOVPE ~ 51018 ~ 900 Yamamoto [Yam04b] PA-MOVPE ~ 41019 Sato [Sat97b] HVPE ~ 1017 ~ 2000 Lu [Lu02a] MBE 1017-1020 600-1200 Bhuiyan [Bhu02a] MEE ~ 31018 ~ 542 Lu [Lu00] Sputtering 1018-1021 20-250 Bhuiyan [Bhu02a] The typical range of carrier concentrations and mobilities for the different growth methods including MOVPE, PA-MOVPE, HV PE, MBE, MEE, and sputtering was discussed in detail and summarized in Table 2-5. 2.1.4 Optical Properties of InN Until 2001, the measured bandgap of 1.89 eV has been commonly accepted for InN [Tan86a]. However, a few groups recently sh owed by PL measurements that the band gap energy of InN is in between 0.65 and 0.90 eV, [D av02a, Dav02b, Dav02c, Wu02, Tat02, Hor02, Sai02, Miy02] which is much smaller than 1.89 eV. Evidence of a narrower band gap for InN was reported in 2001. Inushima et al. insisted that the fundamental absorptio n edge of MBE grown InN laye r lies around 1.1 eV, which is much lower than the previously reported values [Inu01]. Davydov et al. reported a band gap value of 0.9 eV for high quality MB E grown InN, studied by means of optical absorption, PL, photoluminescence excitation (PLE) spectroscopy, as well as by ab initio calculation [Dav02a]. Figure 2-4 shows phot oluminescence spectra for MBE grown InN sample which showed that the band gap of InN was much less than the previously reported value (around 1.9 eV) [Dav02a]. They further studied in detail with different high quality hexagonal InN films grown by diff erent epitaxy methods. Analysis of optical absorption, PL, PLE, and photoreflectivity da ta obtained on single crystalline hexagonal

PAGE 32

15 InN film leads to the conclusion that the true band gap of InN is Eg ~ 0.7 eV [Dav02b, Dav02c]. The larger band gap (~1.89 eV) cited in the literature may be due to the formation of oxynitrides, which have much larger band gaps th an that of InN. As can be seen in Fig. 2.5, the energy gap data less than 1 eV were obtained for single crystalline InN film with a relatively low carrier concentration, wh ile the larger values were mostly for polycrystalline InN film [Bhu03a]. It shou ld also be pointed out that the band gap obtained from epitaxial films shows a remark able dependence of carrier concentration, which is different from the larger one obtained from polycrystalline films. Polycrystalline films show a similar band gap (~ 2 eV) in spite of the wide range variation of carrier concentration 1016-1021 cm-3. Figure 2-4. Photoluminescence spectra for MBE grown InN. As Motlan et al. [Mol02] reported, oxygen incorporat ion is one of the causes for the large band gap energy. Therefore, the larger values may be related to oxygen incorporation into grown InN because polycr ystalline films can contain a high density of oxygen atoms at their grain boundaries.

PAGE 33

16 Figure 2-5. Band gap energy for InN film s as a function of carrier concentration. Davydov et al. [Dav02c] showed that the sample with band gap in the region of 1.82.1 eV contained up to 20 % of oxygen, much higher than for samples with narrow band gap. It can be assumed that oxygen is respon sible for a high concentration of defects. Therefore, this increase of the band gap energy can be caused by formation of oxynitrides, which have a much larger band gap than that of InN. 2.1.5. Indium Nitride (InN) andI indium Gallium Nitride (InxGa1-xN) Applications The latest progress in improving the InN f ilm quality indicates that the InN film almost meets the requirements for application to practical devices. Nowadays, the bandgap energy of InN is known as 0.7 eV and thus InxGa1-xN layers can be used as absorber layers in tandem solar cells where th e mole fraction of indium (x) is varied from 0 to 1 which tunes the bandgap from 0.7 to 3.4 eV. This energy range covers the majority of the solar spectrum, therefore improving efficiency. In addition to the tandem solar cells, In N can also be applied to LED and LD similar to other III-V nitride compounds. B ecause the reported band gap value of InN is about 0.7 eV, which is compatible with the wa velength of the optical fiber, another very

PAGE 34

17 important potential application of InN, fabr ication of high-speed LD and PD in the optical communication system, is expected. It is expected to be a highly promising material for the fabrication of high performance high electron mobility transistor (HEMT). InN as a HEMT channel requires a larger band gap barrier to induce and co nfine electrons. The significant lattice mismatch between InN and GaN or AIN can re sult in a large piezo-electric charge, which is very advantageous for HEMT applications. The strained InxGa1-xN or InxAl1-xN is also a good choice as a barrier layer. InxGa1-xN is a very important compound semiconductor among III-V nitride compounds because the InxGa1-xN active layer emits light by the recombination of the injected electrons and holes into this activ e layer. The addition of a small amount of indium into the GaN was very important in obtaining a strong band-to-band emission because GaN without the indium could not emit a strong band-to-band emission at RT. This reason is considered to be rela ted to deep localized energy states. Currently, InxGa1-xN is usually applied for the active layer in LEDs and LDs for this characteristic of the deep localized energy states, which can facilitate the efficiency of the band-to-band emission. For InxGa1-xN-based LDs, however, the TDs (threading dislocations) density had to be decreased to lengthen the lifetime by using the ELOG (Epitaxial Lateral Overgrowth). For InxGa1-xN-based LEDs, the lifetime of the LEDs is more than 100,000 hours in spite of the large number of dislocations. This difference in lifetime-behavior between LDs and LEDs is probably caused by the difference in the operating current density in the two devices The operating current density of LDs is about one order higher than that of LEDs. Nu merous studies have investigated the origin

PAGE 35

18 of these defects, and their effects on the stru ctural, optical, electr onic, and morphological properties of heteroepitaxial InxGa1-xN layers [Chen06, Cho04, Jin06, Lil06]. 2.2 Thermodynamic Analysis and Phase Separation in the InxGa1-xN System Thermodynamics give the guideline for the epitaxial growth process for all techniques, including MOVPE, since epitaxial growth is simply a highly controlled phase transition. A thermodynamic understanding of epitaxy allows the determination of alloy composition as well as the solid stoichiometry. The thermodynamics of mixing of semicondu ctor alloys (III/V, II/VI, and IV/IV) determines many characteristics of the growth process as well as the properties of the resultant materials. For example, Thermodynamic factors may limit the mutual solubility of the two (or more) components of an alloy. When the sizes of the constituent atoms are sufficiently different, miscibility gap exist. In addition to solid-phase immiscibility in important alloys systems such as GaInAsP and InxGa1-xN, this size difference also leads to microscopic structures far different than the random, totally disordered state normally expected for alloys. Both miscibility gaps an d deviations from a random distribution of the atoms constituting the lattice affect the electrical and opt ical properties of semiconductor alloys in ways that are extrem ely important for many types of devices. The thermodynamics of the surface must al so be considered in any effort to understand the growth processes as well as th e characteristics of the materials produced epitaxially. The basic goal of thermodynamics, as applied to epitaxy, is to define the relationship between the compositions of the va rious phases in an equilibrium system at constant temperature and pressure.

PAGE 36

19 2.2.1 Thermodynamic Models in Solid Solution 2.2.1.1 Regular Solution Model The term regular solution was first used by Hiderbrand to de scribe a class of solutions that are nonideal but consist of a random arrangement of the constituents. The term has since come to designate a more restricted, semiquantative model for the calculation of the free energy of mixing of multicomponent systems. Two additional assumptions are (1) interactions between the constituent atoms occur only pair-wisethat is, only between neares t neighbor pairs, and (2) the atoms reside on a lattice with each atom surrounded by Z neighbors. The bond energies are commonly thought of as being the sum of “chemical” energies, frequently relate d to charge transfer due to differences in electronegativity, and “strain” energies rela ted to distortions in the lattice due to differences in the sizes of the constituent at oms. The enthalpy of mixing is obtained by summing nearest-neighbor bond energies ) 1 (x x HM (2-2) where interaction parameter ( ) is CC AA AC oH H H ZN 2 1 (2-3) where No is Avogadro’s number. 2.2.1.2 Bonding in Semiconductor Solid Solutions Model Traditionally, semiconductor alloys have be en described in terms of the virtual crystal approximation (VCA), where the latt ice on which the atoms are situated is uniform; that is, the individual bonds are disl ocated to form a microscopically uniform solid solution. This was believed to be dictated by the accuracy with which Vegard’s law describes the linear dependence of lattice cons tant on solid solution. However, it has been

PAGE 37

20 recently found that the virtual crystal model fo r semiconductor solid solution is in fact not a good description of the solid. The bond lengths in the alloy more nearly resemble the bond lengths in the pure binary compounds than the average va lues anticipated from the virtual crystal model [Ega02]. The valence force field (VFF) model can be used to explain this behavior [Pos02]. The interactions between atoms are considered to be due to entirely to strain (i .e., the stretching and bending of th e bonds). The simplest form of the VFF calculation for an alloy AC-BC assumes that the lattice is composed of five types of hetrahedra shown in Fig.2.6 [Ich86, K ea66]. It is known that the bonding in a semiconductor is due to long-ra nge effects, particularly th e distributed el ectron energy stated in the solid. The same valence electr ons that determine the optical and electrical properties of the semiconductor also determ ine the bonding, as well as the elastic constant. This is contrary to the basic assu mptions of the regular solution model, which cannot be expected to provide a physically accurate, predictive description of the enthalpy of mixing in semiconductor alloys. Figure 2-6. Tetrahedral cells in a ternary III-V alloy semiconductor.

PAGE 38

21 2.2.1.3 Delta Lattice Parameter (DLP) Mo del for Enthalpy of Mixing Important information needed for the calc ulation of solid-solid solid-liquid, and solid-vapor phase equilibriums is the heat of mixing in the solid, HM. This coupled with the assumption of a random distribution of constituents on their respective sublattices allows the calculation of the free energy of mixing of the solid alloy. Several researchers have suggested that the bonding energy in se miconductors is linearly related to the bandgap [May90, Pan98, Sun94, and Ho96]. The work of Phillips and Van Vechten suggested that the average band ga p should be used in this rela tionship. Since it varies as 5 2 oain semiconductors that are nearly co valent such as the III-V compounds, Hat, which is used as a measure of bondi ng energy, might be written 5 2 o ata K H (2-4) Considering the zero of enthalpy to be infinitely separated atoms, the interaction parameter can be calculated from the en thalpy of mixing at x = 1/2, yielding 5 4 2 5 2 5 2 5 299 2 1 2 4B A B A B A B A sa a a a K a a a a K (2-5) Using Vegard’s law to obtain the latti ce constant at x=0.5. The value of K was obtained by making a leastsquare fit of Equation (Eqn.2-5) to available experimental values of s that are listed in Table 2-6. The DLP calculation also appears to be quite accurate for the III/V nitride alloys. A striki ng feature of the DLP model is that the interaction parameter, hence the enth alpy of mixing, is always positive.

PAGE 39

22 Table 2-6. Comparison of interaction paramete rs calculated using various models with experimental data. Alloy s (exp)a (kcal/mol) s (DLP) (kcal/mol) s (VFF) (kcal/mol) s (Mod VFF) (kcal/mol) Phase Separation AlGaN 1.19 1.34 0.87 AlInN 17.45 18.10 11.44 GaInN 9.60 9.62 5.98 Yes AlPN 19.68 60.79 36.56 AlAsN 57.93 85.33 53.42 GaPN 23 28.90 42.43 27.38 Yes GaAsN 42.78 59.09 36.84 Yes InPN 19.68 29.09 16.33 Yes InAsN 26.71 39.14 21.87 Yes [Str99] 2.2.1.4 Strain Energy Model In the traditional regular solution model, the uniformly positive values of enthalpy of mixing strongly suggest that the enthalpy of mixing is due to strain, rather than chemical factors. The mixing enthalpy can al so be estimated using the simplified VFF model. The solid is considered to be made up to identical tetrahedra (Fig.2.6) with the position of the central atoms, located on the s ub-lattice with no mixing, allowed to relax to the position giving the lowest strain ener gy, considering both st retching and bending distortions. The strain energy due to the stretching and bending of the bonds in each type of tetrahedron is summed over th e five types of tetrahedral weighted by the distribution probability (random arrangement was not assumed in reference [Ich86]). The two terms are coupled and must be solved simultaneously [Ich86]. This approach allows a calculation of the free energy of mixing. Th ere are two major drawbacks to the simple forms of the VFF model described here. First, when the lattice is assumed to be made up of tetrahedral where the corner atoms take the VCA positions, one of the sublattices is not relaxed. This causes a significant overestimati on of the total strain energy. Second, the

PAGE 40

23 difference in energy between the several te trahedral types is much greater than kT for many types of tetrahedrals. Taking into account the effects of the resulting short-range order (SRO) makes the calcula tion of the mixing enthalpy di fficult, since it couples the two factors [Ich86]. These problems can be surmounted by consideri ng a large ensemble of several hundred atoms with the positions of each allowed to relax while maintaining a relatively simple calculation by considering only the dilute limit, where the effect of the SRO is negligible [Sch91, Ho96]. This approa ch was developed specifically for dealing with systems with very low solubility limits, in particular for the solubility of the very small N atom in conventional III/V semiconductors such as GaAs, InP, GaP, and so forth. 2.2.1.5 First-Principal Models Advances in fundamental insight for the energy of a semiconductor lattice and the methodology of solving mathematical problems of large matrices have been achieved recently due to the availability of high-powered computers. These achievements can make possible the first-principle local density selfconsistent total energy minimization cal culations in semiconductor alloy systems [Zun94]. Using these quantum mechanical calculations, the thermodynamics of semiconductor solid solutions can be calculate d without any of the extreme simplifying approximations necessary to obtain simple analytic models. The total energy minimization calculations are based on the entire complex band structures. The results from such calculati ons are included in Table 2-4. The mixing enthalpies have also been calculated for InxGa1-xN, InAlN, and AlGaN alloys using a pseudopotential perturbation approach [Ito97].

PAGE 41

24 2.2.2 Thermodynamic Analysis of InN Koukitu and Seki have performed a thermodynamic analysis of the MBE growth of III-nitrides [Kou97a]. The equilibrium partia l pressure and the growth rate were calculated for input V/III ratio, input partia l pressure of group III elements and growth temperature. A summary of their calcula tion results as a phase diagram for the deposition, indicating etching, droplet form ation and growth regions are shown in the Fig. 2.6. The chemical reaction, which conn ects all species at the substrate surface, is In(g) + N(g) = InN(s). (2-6) The equilibrium equation for the reaction is as follows: K l =1/(PIn+PN). (2-7) From the conservation constraint we have 0InPPIn= 0NPPN (2-8) where 0InP and 0NP are the input partial pressures, which are obtained from the incident beam flux, and PIn and PN are the equilibrium partial pressures. Equation (2-8) expresses that the deposition occurs in the ratio of 1: 1 for In and N. The equilibrium partial pressures at the substrate surface can be obtained from the solution of the above simultaneous equations. The value of the equi librium constant was obtained from the literature [Kou97a]. The corresponding free energy to the chemical reaction (Eq.2-6) used in the analysis is as follows: Go(kcal/mol)= (-1.764102)+ 3.067102/T + (-1.45110-3)TIn(T)+7.90910-2T + 3.88310-11TT. (2-9)

PAGE 42

25 The calculation for the MBE gr owth technique using an NH3 source was performed in a similar manner, using atomic n itrogen. The chemical reaction is In(g)+ NH3(g)= InN(s) + (3/2)H2(g). (2-10) In NH3 case, they introduce the molar fraction of decomposed NH3, into the calculation as follows: NH3(g) (1-)NH3(g) + (1/2)N2(g) + 3(1/2)H2(g) (2-11) The value of is assumed appropriately as that of MOVPE growth [Kou96], because it is difficult to know the exact value. The equilibrium partial pressure and the growth rate were calculated for input V/III rati o, input partial pressure of In, and growth temperature. In the growth of InN, they conclude that three deposition modes, i.e., etching, droplet formation and growth regions appear in the temperature range from 500 to 900 C. The temperature suitable for the In N growth is predicted to be from 600 to 700 C with V/III 1, which is essential in the MBE growth. However, the experimental growth temperature is much lower than this theoretical prediction, and almost experiments have been done in the temperat ure range from 450 to 550 C. They also reported that there is a difference between the atomic nitrogen and the NH3 source as shown in the corner of diagram (Fig. 2.7) where the etching region appears [Kou97a]. In the case of the atomic nitrogen source, the et ching region appears cons tantly at the region where the input V/III ratio and the input 0InP are low value. On the other hand, in the case of the NH3 source, it appears at the region wher e the V/III ratio is high and the input 0InP is low. They concluded that this is due to the decomposition of NH3: when NH3 is decomposed, H2 gas is produced, and the produced H2 drives Eqn. (2-10) to the left hand. Consequently, the deposition moves into th e etching mode due to the increase in H2.

PAGE 43

26 Figure 2-7. Calculated phase diagram for th e MBE deposition of InN using atomic N and NH3 gases. There are three deposition m odes: etching, droplet and growth. 2.2.3 Phase Separation in InxGa1-xN The large positive enthalpy of mixing for sy stems with a large lattice mismatch can overwhelm the negative entr opy of mixing for temperat ures below the critical temperature. This results in a free energy versus composition curve shown schematically in Fig. 2.8, with an upward bowing in the cente r [Str99]. This dictates that at equilibrium, a random alloy with composition between points A and B will decompose into a mixture of two phases. Two other important points in the G versus energy curve shown in Fig. 2.8 are the inflection points lying between A a nd B. Between these two points the solid solution is unstable against an infinitesima l fluctuation of composition. The spinodal appears on the T-x phase diagram, as indicated in Fi g. 2.9 [Str99]. In the pseudobinary phase diagram, the boundary of the unstabl e region is defined by the locus of (d2G/dx2)T,P = 0 [25], called the spinode. Inside this region, the solid can decompose “spoinodally,” with no energy barrier.

PAGE 44

27 Figure 2-8. Free energy versus solid com position for a hypothetical semiconductor alloy having a large positive enthalpy of mixing. Point A and B are the bimodal points, and points C and D repr esent the spinodal points. The growth of InxGa1-xN alloys has proven to be extr emely challenging, mostly due to the trade-off between the ep ilayer quality and the amount of InN incorporation into the alloy as the growth temperature is change d. Growth using high temperatures of approximately 800oC, typically results in high crysta lline quality but the amount of InN in the solid is limited to low values because of the high volatibility of N over InN. Ho and Stringfellow performed a theore tical calculation of the en thalpy of mixing, the solid phase interaction parameter, and the extent of the miscibility gap for InxGa1-xN alloy system using a modified valence-force-field (VFF) model calculation where the lattice is allowed to relax beyond the fi rst nearest neighbor [Str97]. Figure 2-9. Schematic liquid-soli d pseudobinary phase diagram.

PAGE 45

28 The VFF model, itself, is found to overestim ate the total strain energy of a ternary system due to the constraint that only one of the two sublattices is allowed to relax. The calculation of the enthalpy of mixing or the interaction parameter in III-V system has been a topic of interest for nearly twenty-f ive years. In 1972 Stri ngfellow developed the semi-empirical delta-lattice-parameter (DLP) model, which is found to yield surprisingly accurate interaction parameters for a wide ra nge of III-V alloys knowing only the lattice constants of the binary constituents. The temperature dependence of the bimodal and spinodal lines in the InxGa1-xN system was calculated using a modified VFF model. The strain energy is found to decrease until appr oximately the sixth nearest neighbor, but this approximation is suitable only in the dilute limit. Assuming a symmetric, regular solution-like composition dependence of the enthalpy of mixing yields an interaction parameter of 5.98 kcal/mole and a critical temperature for the phase separation of 1250 oC (Fig. 2.10) [Ho96]. At a typi cal growth temperature of 800 oC, the solubility of indium in GaN calculated to be less than 6 %. The miscibility gap is expected to represent a significant problem for the epitaxial growth of these alloys [Ho96]. Figure 2-10. Binodal (solid) a nd spinodal (dashed) curves for the InxGa1-xN system, calculated assuming a constant average value for the solid phase interaction parameter.

PAGE 46

29 Singh et al. reported the growth of InGaN thick (0.3~0.4 m) films and InxGa1-xN /GaN double heterostructures by MBE at the substrate temperatures 700-800oC. X-ray diffraction and optical absorption studies show ed that the phase separation of InN of InxGa1-xN thick films occurred with x 0.3. On the other hand, InxGa1-xN/GaN double heterostructures showed no evidence of pha se separation. These observations were accounted for using Stringfellow’s model (DLP model) on phase separation, which gives a critical temperature for mi scibility of the GaN-InGaN system equal to 2457 K. The maximum value of the critical temperature (Tc) above which the InN-GaN system is completely miscible can be computed from Stringfellow’s equation for a binary system (Eq.2-12) [Sin97]. 5 4 24 75 8 a a R K Tc (2-12) Where a is the difference in the lattice constants of GaN and InN, a is the average lattice of GaN and InN, and R is the gas constant. K is the proportionality constant between atomization enthalpy (bondi ng energy). Phase separation in any alloy requires long-range diffusion and thus a correlation should ex ist between phase separation and a time required for the growth of the film. They believed this is one of the reasons for the non-observable phase separation in GaN/ InxGa1-xN /GaN double heterostructures with thin InxGa1-xN layers. Strain associated with thin InxGa1-xN quantum wells could also stabilize al loys against phase separation. Wakahara et al calculated the compositional imhomogeneity in InxGa1-xN by using a theoretical estimation of the interaction parameter based on DLP model [Wak97]. Table 2-7 summarize the lattice mismatching, the interaction parameter () and the critical

PAGE 47

30 temperature of spinodal decomposition, which is denoted as Tc = /2R, for the III-V ternary alloy system. It can clearly be seen that the nitride alloys including the In N have a very large interaction parameter thus, the critical temp erature of the spinodal decomposition also becomes very high. It is expected that the immiscibility of the InxGa1-xN alloy is very strong. The critical temperature of the spinodal decomposition defined at the composition x = 0.5 is much higher than the typically used growth temperature of 800 oC. The evidence of the phase separation in the InN containing nitride alloys was resulted in [Mor94]. Recently, Koukitsu and Seki [Z un94] reported compositional inhomogeneity based on a thermodynamic analysis of the vapo r-solid interface. They predicted that the compositional inhomogeneity of InxGa1-xN increases with an increase of the growth temperature and in the partial pressure of the hydrogen but decreases with V/III ratio. Table 2-7. Interaction parameters for various III-V ternary alloy systems. III-V ternary alloy system Lattice mismatch a/a (%) Interaction parameter (DLP model) (cal/mol) Critical temperature Tc (K) AlAs-GaAs 0.159 0 0 GaAs-InAs 6.92 2815 709 AlP-GaP 0.239 0 0 GaP-InP 7.39 3630 914 GaP-GaN 18.9 28900 7276 AlN-GaN 2.93 931 233 AlN-InN 13.38 17300 4338 GaN-InN 10.46 10300 2583 [Wak97] The resulting strain in the layers could le ad to deviations fro m homogeneity of the sublattice. Zunger and Mahajan have reviewed seve ral observations, which indicate that when the tetrahedral radii are different, two types of structural variations are observed: phase separation and atomic ordering. The difficulties in InxGa1-xN growth are mainly

PAGE 48

31 due to (a) very high equilibrium vapor pressure (EVPs) of nitrogen over InN and (b) a large lattice mismatch (11 %) between InN and GaN. The lattice mismatch between InN and GaN (due to the very different tetrah edral radii) results in highly strained InxGa1-xN alloys. Therefore, at relativel y low growth temperature (650-800 oC), phase separation is a major concern. The majorities of the III-V ternary and quaternary alloys are predicted to be thermodynamically unstable and show a tendency towards clustering and phase separation. The phase separation and ordering phenomena in InxGa1-xN alloys with MBE was studied by Doppalapudi et al [Kel98]. The ordering para meter was found to increase with the growth rates of the films, a result which is consistent with the notion that ordering is induced at the surface of the growing films where it is thermodynamically stable and is then subsequently “frozen in “ during further growth. Phase separation was found to be essential for films with high indium content ( 25 %), while ordering was noticeable for films with small indium content ( 10 %). This competition between the two phenomena is consistent with the proposal th at lattice strain is the driving force for both. The effect of elas tic strain in epitaxial InxGa1-xN layers coherently grown on GaN wafers on spinodal decomposition of the te rnary compound was examined. The effect results in considerable suppression of phase separation in the strained InxGa1-xN layers. The elastic strain effect is predicted to lowe r the critical temperat ure. This effect does work only if the relaxation of strain in the epitaxial layer is not yet started. In summary, the results obtained by the se veral models used in the thermodynamic analysis about InN and InxGa1-xN were reviewed. For InN, the thermodynamic result showed that the growth region is and the stab le InN growth can be achieved at high V/III ratio. For InxGa1-xN, it was found that phase separa tion commonly occurs and that the

PAGE 49

32 critical temperatures could be calculated through the interaction parameter and these critical values depend on the chosen mode l. The maximum mole fraction of indium incorporated into InxGa1-xN, which depends on the value of elastic strain of InxGa1-xN was studied. From this result, it can be s uggested that the maxi mum mole fraction of indium incorporated into InxGa1-xN can be increased by decreasing the strain energy of InxGa1-xN. 2.3 Indium Nitride (InN) and Indium Gallium Nitride (InxGa1-xN) Growth Challenges There are several problems to be over come for high crystalline InN and InxGa1-xN film growth. These problems are narrow regi on of growth temperature due to the low decomposition temperature, lo w cracking efficiency of NH3, no suitable nitrogen precursors to improve the decomposition efficiency of NH3, and carrier gas. These problems which occur during the growth of InN and InxGa1-xN are briefly discussed in this part. 2.3.1 Growth Temperature and V/III Ratio The growth of InN is the most difficult among the III-nitrides because the equilibrium vapor pressure of nitrogen over th e InN is several orders higher than AlN and GaN [Amb96]. Because of the low InN dissoc iation temperature and high equilibrium N2 vapor pressure over the InN film [McC70], the growth of InN requires a low growth temperature. Due to the low (~550 C) growth temperature, the MOVPE growth of InN is thought to be restricted by a low decomposition rate of NH3. Although a higher growth temperature is expected to result in a higher decom position rate of NH3, it can also result in thermal decomposition (thermal etching) of the grown InN. On the other hand, the growth at a low temperature (lower than 400 C) is dominated by the formation of

PAGE 50

33 metallic indium droplets due to the shortage of reactive nitrogen. Epitaxial growth at low temperature becomes impossible due to reduc ed migration of the deposited materials. Therefore, the region suitable for the deposition of InN is very narrow. Koukitu et al carried out a thermodynamic st udy on the MOVPE growth of IIInitrides [Kou97b]. They pointed out that a hi gh input V/III ratio, the use of an inert carrier gas, and a low mole fraction of the decomposed NH3 are required for the growth of InN. Experimental results also match well with the firs t two points (high input V/III ratio and the use of inert carri er gas) but are not clear on the last point (a low mole fraction of the decomposed NH3). High input V/III provides sufficient amount of reactive nitrogen, since the NH3 decomposition rate is low at low growth temperature. NH3 is decomposed thermodynamically into H2 and N2 with low decomposition efficiency at temperature higher than 300 oC, which results in the increase of H2 partial pressure [this sentence does not make sense, there is low decomposition efficiency at T < 300oC, and at higher temperatures the H2 partial pressure increa ses]. Thus too high ratio of V/III significantly prevents the growth of InN and leads to th e etching of InN due to the increase of H2 partial pressure (Eq. (2-10)). A suitable re gion of V/III ratio and growth temperature is required for the high quality InN growth w ithout indium droplets formation during the growth. The main problem in growing InxGa1-xN has been the phase separation at high indium incorporation (x (In) 0.3) due to the very high equilibrium vapor pressure of nitrogen over InN. The compositional contro l has only been achieved for relatively low growth temperature, up to 650 oC. However, crystal quality is not good because the

PAGE 51

34 decomposition rate of NH3 is very low below 1000 oC. Yoshimoto et al. [Yos91] obtained the relatively high-quality InxGa1-xN using a high temperature (800 oC) and a high indium flow rate. In that case of the temperature growth (650-800 oC), the phase separation may occur because of the large mismatch (11 %) between InN and GaN. The lattice mismatch between InN and GaN (due to the very different tetrahedral radii) results in highly strained InxGa1-xN. Therefore at the grow th temperature of 650-850 oC, phase separation is still a major concern. 2.3.2 Nitrogen Source InN and InxGa1-xN are typically grown by MOVP E using conventional group III precursors such as tri-methyl indi um and tri-ethyl indium with NH3 as the active nitrogen source. However, InN and InxGa1-xN are relatively difficult to produce with the high quality required for minority carrier devices due to high equilibrium N2 vapor pressure and the low decomposition efficiency of NH3 as discussed earlier. NH3 is almost stable even at 1000 C and decomposes only 15 % at 950 oC, even when catalyzed by GaN [Che91]. The combinati on of high growth temperatures and high nitrogen volatility leads to hi gh concentrations of N vacancies in GaN and InN. This is often cited as the reason that the GaN and InN epitaxial layers are n-type. Solution of this problem will probably requ ire N precursor to give the high active nitrogen reduction at the low gr owth temperature. Hydrazine (N2H4) is an attractive N precursor because it contains no carbon atoms to be incorporated into the solid, and the hydrogen atoms are potentially beneficial for removal of the alkyl radicals (from the group III precursors) from the surface. It decom poses at temperatures as low as 400 C, considerably lower than temperatures required for NH3 because of the weaker N-N bond [Gas86], thus making it suitable for growth at temperatures well at the current growth

PAGE 52

35 temperature (~550 oC) and at low V/III ratio as low as 10 which prevents the waste of N precursor. It also has a favorable vapor pres sure of approximately 10 Torr at 18C, as indicated in Table 2-8. Howe ver, hydrazine is a toxic ma terial, a rocket fuel, and explosive. When hydrazine is used in III-V nitride growth with the trimethyl-group III alkyls, the adduct formation between hydr azine and the group III precursors was observed. The layers were found to be c ontaminated with both oxygen and carbon. The hazard associated with the t oxicity and explosiveness of hydrazine makes its use in a production environment unlikely. Unsymmetrical dimethylhydrazine (H2N2 (CH3)2, 1,1 DMHy) is a considerably safer alternative to hydrazine. It has a vapor pressure of 157 Torr at 25 C and pyrolyzes at temperatures considerable lower than for NH3. However, relatively high (>1019 cm-3) levels of oxygen and carbon were observed, both of which are associated with the use of DMHy. A potentially less hazardous precursor, phenyl hydrazine, has al so been explored [Jon95]. However, the vapor pressure of 0.03 To rr at room temperature is far too low to be acceptable. Hydrogen azide, or hydrazoic acid (HN3) has also been successfully used for MOVPE growth in a low-pressure reactor [Cht 97].This precursor is attractive because it has a high vapor pressure (the boiling te mperature is 37 C) and decomposes at approximately 300 C to yield HN radicals with two dangling bonds a potentially good source of atomic nitrogen, and N. However, it is highly toxic and potentially explosive. The N precursor tertiary-butylamine ((C4H9)NH2 or TBAm) has a convenient vapor pressure of 340 Torr at 25 oC, a low toxicity, and is stable. However, the use of TBAm

PAGE 53

36 for the growth of GaN has proven unsuccessf ul [Rus96, Bea97]. They observed no GaN deposition but rather deposition of a la yer consisting mostly of carbon [Bea97]. In summary, several candidate s for nitrogen sources were reviewed and all of them have some problems such as toxicity, e xplosion, low decomposition efficiency, and contamination. NH3 has been still widely used as a nitrogen source in spite of low decomposition efficiency. The design of the optimum nitrogen source for the growth of III/V nitrides is st ill required even though it is tricky. Table 2-8. Properties of nitr ogen precursors for MOVPE. Precursor Melting Point (oC) Boiling Point (oC) Vapor Pressure a b, K P (Torr)/T(oC) NH3 N2H4 MMHy 1,1 DMHy N2H3(C6H5) (phenyl hydrazine) HN3 (hydrogen azide) TBAm -77.7 1.3 -67 113 37 45.2 9.9974 31.211 8.749 3.014 7.61 1,509.8 10/18 49.7/25 157/25 0.03/23 288/20 [Str99] 2.3.3 Carrier Gas Carrier gas is used as the medium to give the uniform flow pattern of precursors in MOVPE reactor. H2 and N2 carrier gas have been usually used in InN growth. Koukitu et al carried out a detailed thermodynamic study on the role of hydrogen during the MOVPE growth of III-nitrides [Kou99a]. They showed that increase of hydrogen in the growth system results in a decrease of InN deposition rate (called etching), which they suggested was due to the decrease of driving force for the deposition. Thus the reaction for the growth of InN pr oceeds more effectively in the inert gas system and is prevented with the increase of H2. Therefore it is necessary to use inert

PAGE 54

37 carrier gas in the growth of InN. These theo retical and experimental results confirm that using a N2 carrier gas is preferre d (and widely used) for su ccessful InN growth. 2.4 Indium Nitride (InN) Growth Techniques In this part, we review the several growth techniques commonly used for InN growth each of which deals briefly with the characteristics of the reaction system, the precursors, the chemistry, the applications the disadvantages and advantage of each growth technique. 2.4.1 Chemical Vapor Deposition (CVD) CVD involves the dissociation and/or chemical reactions of gaseous reactants in an activated (heated, plasma etc.) environment, followed by the formation of a solid film. The deposition involves homogene ous gas phase reactions, which occur in the gas phase, and heterogeneous chemical reactions which occur on a heated surface leading to the formation of epitaxial films. In general, the CVD process involves the fo llowing key steps as shown in Fig. 2.11 [Cho00b, Cho03]. (1) Generation of active ga seous reactant species. (2) Transport of the gaseous speci es into the reaction chamber. (3) Gaseous reactants undergo gas phase reactions forming intermediate species: (4) Absorption of gaseous reactants onto the heated substrate, and the heterogeneous reaction occurs at the gas—so lid interface (i.e. heated substrate) which produces the deposit and by-product species. (5) The deposits will diffuse along the heated substrate surface forming the crystallization centre and growth of the film.

PAGE 55

38 (6) Gaseous by-products are removed from the boundary layer through diffusion or convection, (7) The unreacted gaseous precursors and by-products will be transported away from the deposition chamber. Figure 2-11. Schematic illustration of the key CVD steps during deposition. 2.4.1.1 Metal-Organic Vapor Phase Epitaxy (MOVPE) Metalorganic Vapor Phase Epitaxy (MOVP E) is one growth method among CVD, which has been classified according to the use of metalorganics as precursors. Compounds containing metal atoms bonded to organic radicals are known as “Metalorganics”. MOVPE can be used to deposit a wide range of materials in the form of amorphous, epitaxial, and polycrystalline films. The schematic of MOVPE was shown in Fig. 2.12 where TMI is delivered by N2 carrier gas and NH3 is also delivered directly in to MOVPE reactor and the thermal environment for the decomposition and/or deposition reaction of the precursors can be supplied using resistance heati ng, radio-frequency or infrar ed lamp heating. MOVPE tend to involve endothermic reactions, thus cold-w all reactors with a single temperature zone can be used.

PAGE 56

39 Figure 2-12. Schematic of horiz ontal cold-wall MOVPE system. The metalorganic precursors genera lly undergo decomposition or pyrolysis reactions. In general, metalorganics precurs ors have lower decomposition or pyrolysis temperatures than halides, hydrides or ha lohydrides. Thus, metalorganic precursors enable MOVPE process to perform at a lowe r deposition temperature than conventional CVD, which generally uses halides or hydrides. The source materials generally used for the MOVPE growth of InN, are trimethylindium (TMI) as In source, and ammonia (NH3) as N source. The pyrolysis of TMI in MOVPE was firs t studied by Jacko and Price who founded that the decomposition occurred in three steps as each of the In-CH3 bonds were broken at the temperature above 400 oC [ Jac64]. The methyl radical s thus formed were then found to recombine to yield ethane (C2H6). This mechanism is given by Eq. (2-13) to (216). In(CH3)3 In(CH3)2 CH3 (2-13)

PAGE 57

40 In(CH3)2 In(CH3) CH3 (2-14) In(CH3)In CH3 (2-15) CH3 CH3 C2H6 (2-16) The indium reacts with NH3 at the substrate surface in high temperature ( 500oC) to form InN (Eq. (2-17)). In (g) NH3 (g) = InN (s) + 3/2 H2 (g) (2-17) The other method was suggested by Koukitu an d the reaction pro cedure is given as follows [Kou97b]. First, thermodynamically, almost all the NH3 is decomposed into N2 and H2 at temperatures higher than 300 oC. However, it is well known that the decomposition rate of NH3 under typical growth conditions is slow without a catalyst and the extent of the decomposition strongly de pends on the growth conditions. The mole fraction of decomposed NH3, into the calculation as follows (Eq. (2-18)). NH3 (g) (1) NH3 (g) + /2 N2 (g) + 3 /2 H2 (g) (2-18) The metal-organic precursors TMI are decomposed irreversibly, according to the following homogeneous reaction, near the vapor -solid interface (Eq. (2-19)). (CH3)3In (g) + 3/2 H2 (g) In (g) + 3CH4 (2-19) The chemical reaction which occurs at the s ubstrate surface to form InN is the same as the former method given in Eq.(2-17). MOVPE can be performed at atmospheric pressure and low pressure (about 2.726.7 kPa). For a typical MOVPE process, the de position is entirely kinetically controlled at very low deposition pressure ( 1 kPa), even though the deposition temperature is relatively high. At pressures above 1 kPa, the growth rate is predominantly controlled by diffusion-rate limited mechanism [Dup95].

PAGE 58

41 Despite the high cost of precursors, MOVP E have been developed especially for the growth of epitaxy of III-V as well as II-VI and IV-VI semic onducting material for optoelectronic applications (e.g. light-emitting diode, heterojunction bipolar transistor, solar cells, photocathode adva nced laser designs such as quantum well and double heterostructures, etc.). 2.4.1.2 Hydride Vapor Phase Epitaxy (HVPE) Hydride vapor phase epitaxy (HVPE) has been important in the development of a variety of semiconductors including the II I-arsenides, the III-phosphides and the III nitrides such as InN. HVPE occurs usually at atmospheric pres sure (horizontal or vertical). Generally, HVPE using chloride sources provides a high growth rate ( 30m) compared with that of MOVP E and MBE. Because of the gr oup III element is transported to the substrate as a volatile compound (usua lly a chloride), this technique is often referred to as chloride-tra nsported vapor phase epitaxy. The source material generally used for the HVPE growth of InN is liquid indium as indium source, which will react with HCl a nd form indium monochloride (InCl) or indium trichloride (InCl3) and ammonia (NH3) or monomethylhydrazine (MMHy) as nitrogen source. The source of InCl is fo rmed by the reaction between metallic In and HCl at 780 oC and the InCl3 is presynthesized and is evaporated from the source boat in the temperature range from 325 to 375 oC (Fig. 2.13) [Tak97a]. The reaction chemistry for InN growth was given InC1 (g) + NH3 (g) = InN (s) + HC1 + H2 (2-20)

PAGE 59

42 Figure 2-13. Schematics of horizontal hot-wa ll hydride vapor phase epitaxy chamber. Theoretically, the chlorine in the HVPE chemistry should reduce the amount of impurities from the system due to the formation of highly volatile species, producing films with low carrier concentrations. However, Si and O impurities from the quartz tube can cause highly n-type films. 2.4.1.3 Plasma Enhanced Chemical Vapor Deposition (PECVD) Plasma Enhanced Chemical Vapor Deposition (PECVD) is also known as glow discharge chemical vapor deposition. It uses electron energy (plasma) as the activation method to enable deposition to occur at a low temperature and at a reasonable rate. Supplying electrical power at a sufficiently hi gh voltage to a gas at reduced pressures (<1.3 kPa) results in the break ing down of the gas and generates a glow discharge plasma consisting of electrons, ions and electronically excited species. Uncracked trimethylindium (TEI) carried by Ar or H2 as In source and uncracked N2 as N source are ionized and dissociated by electron impact, and hence generating chemically active ions (radicals). These ra dicals undergo the heterogeneous chemical reaction at or near the heated substrate su rface and deposit the thin film (Fig.2.14) [Cho03].

PAGE 60

43 Figure 2-14. Schematics of PECVD. The disadvantage of PECVD is that it re quires the use of a vacuum system to generate the plasma, and a more sophisticated reactor to contain the plasma. PECVD is often more expensive and in general has diffi culty in depositing high purity films. This is mostly due to the incomplete desorption of by-product an d unreacted precursor at low temperatures, especially hydrogen which re mains incorporated into the films. However, PECVD can find applications wh ere technology will balance the cost of fabrication and also where low deposition temperatures are required on temperature sensitive substrates, which can not be met by the conventional CVD. The main advantage of PECVD over other CVD methods is that the deposition can occur at relatively low temperatures on large areas. It also offers flexibility for the microstructure of the film and deposition to be controlled separately. The ion bombardment can be substituted for deposition temperature to obtain the required film density. Such low temperature deposition is im portant for applications that involve the use of temperature sensitive substrates.

PAGE 61

44 2.4.2 Molecular Beam Epitaxy (MBE) and Me talorganic Molecular Beam Epitaxy (MOMBE) MBE is a common technique for thin epit axial growth of semiconductors, metals and insulators. MBE has the capability of growing device quality layers of semiconductors with atomic resolution. The low growth temperature (300-600 C) ensures negligible dopant diffusion. The apparatus consists of an UHV cold-wall chamber, independently controlled thermal or e-beam cells to supply the sources, in situ heating and cleaning, and in-situ monitors of growth and chemical analysis. The simplest method for generating molecula r beams is from heated Knudsen cells containing Ga, In, Al or dopant material. Shutters open to allow the molecular beams to leave the cells and the beams are directed at a heated substrate. Thermal beams of atoms or molecules react on a clean substrate su rface to grow an epitaxial film under UHV conditions. The atomic species undergo adso rption and migration on the surface. MBE can be also equipped with a number of in-situ probes that monitor the growth real time including RHEED where high-energy electrons are diffracted off the growing surface and imaged to describe the nature of the epitaxy. In the MBE growth of III-nitr ides, the solid sources of the group III elements such as Ga, In, and Al are used in general, but the nitrogen is supplied by the gas source such as N2 and NH3. In general, this type of MBE system is called gas source MBE [Dav97]. When organometallic sources replace the group-III elemental source, it is called metalorganic molecular beam epitaxy (MOMBE) or chemical beam epitaxy [Abe97]. In both types of MBE, the key issue in the grow th is the nitrogen source. The dissociation energy of N2 molecules is as high as 9.5 eV, therefore, the supply of N2 gas to the substrate surface with the group-III elementa l beams can not induce any growth of the

PAGE 62

45 nitrides. For obtaining atomic reactive nitrogen, the N2 molecules are dissociated by the radio-frequency (rf)-plasma or the electron cy clotron resonance (ECR) method. Although the rf-plasma source is the most popular and the rf-radical source produces considerably fewer ions than an ECR source due to the higher plasma pressures [Hug95], it is well known that ion damage can still induce duri ng the epitaxy [Pow93]. Some techniques to avoid such ion damage by an ion tapping system using the static electric field, have been tried [Bot95, Mol94, Iwa96]. The other seri ous problems induced by the plasma may be some contamination such as oxygen or carbon dioxide. MOMBE is also one of the potential grow th techniques where the advantages of both MOVPE and MBE can be utilized. Film can be grown relatively at low temperature by MOMBE and premature reaction of the prec ursors, a serious problem in MOVPE, is minimized due to the large mean free path of gaseous molecules. 2.4.3 Atomic Layer Deposition (ALD) Atomic Layer Deposition (ALD) can be considered as a special mode of CVD. It is a surface deposition process that can be used for the controlled growth of epitaxial films, and the fabrication of tailored molecular stru ctures on the surfaces of solid substrates. ‘Monatomic layers’ can be grown in sequence wh ich is a characteristic feature of ALD. Therefore, the desired coating thickness can be produced simply by counting the number of reaction sequences in the process. The su rface reconstruction of the monolayer formed in the reaction sequence will influence the saturation mechanism and the saturation density of the precursor. The ALD reaction sequences are normally perform in an ‘effective overdosing’ condition to ensure a complete saturation of the surface reaction to form the monoatomic layer. Furthermore, such effective overdos ing’ condition also provides good conformal

PAGE 63

46 coverage that allows unifor m coatings onto complex shaped substrates. The sequencing in ALD also eliminates the gas phase reacti ons, and enables a wide r choice of reactants (e.g. halides, metalorganics, elemental metal, etc.). The ALD process has the potential to be scaled up for the deposition of high quality thin films with excellent uniformity and reproducibility onto large area substrate [Nii96, Lau98]. The ALD process can be performed at atmos phere pressure or in a vacuum system as in molecular beam epitaxy. The use of vacuum enables a vari ety of in-situ surface analysis methods to be incorporated into th e ALD equipment for the in-situ analysis of the growth mechanism and the deposited su rface structures [Bac97, Kou97c, Her99]. The distinctive sequencing feature in AL D makes it an attractive method for the precise growth of crystalline compound laye rs, complex layered structures [Cha98] superlattices [Tor00, Har98] and laye red alloys with precise interfaces. Currently, a wide range of thin films ha ve been synthesized using ALD methods. These include semiconductor III-V, II-VI, oxides nitrides, phosphides, and metallic films [Gup98, Hsu98, Utr99, Mar99, Ish97b, Utr00]. The ALD process can produce films with good conformal coverage and it has the ability to control film thickness accurately at the sub-nanometer level. Such distinctive advantages have made it a potentially valuable tool for nanotechnology. 2.5 Substrate Materials The important properties of substrate are the lattice constant which causes the lattice mismatch for the epitaxial InN film and thermal expansion coefficient which also can create dislocations during cooling when th ere is a large thermal coefficient mismatch

PAGE 64

47 between the film and substrate. The propertie s of the substrate commonly used in III-V nitrided film growth are summarized based on InN film in Table 2-9. Table 2-9. Structural prop erties of substrates. Substrates Lattice constant a () Lattice constant c () Lattice mismatch (%) TEC (10-6 K-1) a c TEC mismatch (%) InN-Wurtzite GaN-Wurtzite AlN – wurtzite 6H-SiC – wurtzite Al2O3 – rhombohedral ZnO wurtzite LiAlO2 – tetragonal LiGaO2 – orthorhombic 0.537b 3.189a (5.524 when rotated 30 oC) 3.112b 3.08b 4.758b 3.252b 5.17b (3.134 as a wurzite) 5.40b (3.186 as a wurtzite) 5.704b 5.18a 4.98b 15.1b 12.991b 5.213 5.1687b 5.007b -10.9 -13.7 14.8 25.7 -8.8b -12.9 (wurtzite) -11 (wurtzite) 5.70b 5.59a 4.2b 4.2b 7.5b 2.9b 7.1b 6b 3.70b 3.17a 5.3b 4.7b 8.5b 4.8b 7.5b 7b 0 -30.9 -8.1 -8.1 -48.5 33.1 -45.6 -35.7 GaN – zincblende Si (111) – cubic 3C-SiC – zincblende GaAs zincblende 4.53b 5.43b ( 3.84 as a hexagonal) 4.36b 5.65b 21.9 7.9 (hexagonal) 18.9 37.4 5.2b 6.2b 2.7b 6.0b -25.8 -37.7 43 -35.7 [aMor94, bDav02a] 2.5.1 Sapphire Substrate (Al2O3) (0001) Sapphire is the most extensively used substr ate material for the epitaxial growth of III-V nitride materials. Large ar ea good quality crystals of sa pphire are easily available at relatively low cost. They are transparent and stable at high temperature. The large lattice mismatch (25.7%) and thermal expansion co efficient difference (48.5 %) for InN can result in an extremely high density of st ructural defects of InN film. However, researchers have revealed that the substr ate surface pretreatment and insert of an

PAGE 65

48 intermediate buffer layer between the substrat e and epilayer can significantly improve the film quality. Nitridation of the sapphire su bstrate surface significantly improves the crystalline quality of III-V nitride growth as a result of the formation of AlN which reduce the lattice mismatch from 25.7 % for InN/Al2O3 to 13.7 % for InN/AlN. The single crystal can be described by both rhombohedral unit cells with volume 84.929 3 and hexagonal unit cell with volume 254.792 3. The unreconstructed basal c plane perspective views for both unit cells ar e shown in Fig. 2.15 [Edg02]. The faceting of sapphire crystal is shown in Fig. 2.16 [Amb98]. Figure 2-15. Perspective views in (221) unit cell: (a) along [0001] direction in a rhombohedral unit cell; (b) along the (0001) direction in hexagonal unit cell. Figure 2-16. Common facets of sapp hire crystals: (a) view down c -axis; (b) surface planes.

PAGE 66

49 2.5.2 Silicon (Si) Substrate Si (111) substrate is usually spotlighted as an attractiv e substrate because of the high quality and low cost of Si. The availability of the either n or p -type substrate is advantageous. The doped substrates can si gnificantly simplify device structures. The bulk Si crystal is a diamond st ructure and has lattice constant a = 5.43 at room temperature. However, Si (111) surf ace has hexagonal surface and lattice parameter of a = 3.84 . Therefore, Si has the small lattice mismatch (7.9 %) for InN. The unit cell is outlined as a diamond sh ape with seven atoms along each edge, for two different orientations. Si has a diam ond-lattice structure with the space group of Fd3 m (no.227), which belongs to the cubic-crysta l family. It can be represented as two interpenetrating fcc sublattices with one sublattice displaced from the other by one quarter of the distance along a body diagonal of the cube (i.e. the displacement of a 3a/4. where a = 0.543 nm). Each atom in the latt ice is surrounded by foul equidistant nearest neighbors that lie at the comers of a tetrahedron. Figure 2-17 illustrates the perspective views along the [ 001], [011] and [111] directions of a Si unit cell [Liu02]. There are several methods for Si substrate pr eparation (Table 4) [Yan96, Gru91, Dad01a, Wat93]. Figure 2-17. Perspective views of Si along va rious directions: (a) [001]; (b) [011]; (c) [111].

PAGE 67

50 2.5.3 Gallium Nitride (GaN) and Al uminium Nitride (AlN) Substrate Gallium nitride substrate has a small lattice mismatch of 10.9 % with InN compared with sapphire substrate (25.7 %) and AlN substrat e (13.7 %) even if it is greater than that of silicon of 7.9 %. Gallium nitride normally has a wurtzite structure, with the space group of P 63mc (no.186). The wurtzite structure cons ists of alternating biatomic closepacked (0001) planes of Ga and N pairs st acked in an ABABAB sequence. Atoms in the first and third layers are directly aligned with each other. Figure 2-18 displays the perspective views of wurtzite GaN along [0001], [1120] and [1010] directions, where the large circles represent gallium atoms and th e small circles nitrogen [Liu02]. The closepacked planes are the (0001) planes. The gr oup III nitrides lack an inversion plane perpendicular to the c -axis, thus, crystals surfaces have either a group III element (Al, Ga, or In) polarity (designated (0001) or (0001)A) or a N-polar ity (designated (0001) or (0001)B). An excellent review on crystal pola rity is given by Hellman [Hel98]. The zincblende structure (space group F43 m ) of GaN can be stabilized in epitaxial films. The stacking sequence for the (111) close-packed planes in this structure is ABCABC. Perspective views of the zincblende stru cture are shown in Fig. 2.19 [Liu02]. Figure 2-18. Perspective views of wurtzite GaN along various directions: (a) [0001]; (b) [1120]; (c) [1010].

PAGE 68

51 Figure 2-19. Perspective views of zincble nde GaN along various di rections: (a) [100] (111 unit); (b) [110] (222 units); (c) [111] (222 units). AlN normally has the wurtzite structure, although epitaxial laye rs of zincblende structure AlN have been ma de [Oku98a, Oku98b]. Wurtzite AlN has the space group of P 63mc (no. 186) as same as wurtzite GaN. The (0001) surfaces of AlN are polar, which has an important effect on its etching, bulk crystal growth and GaN epitaxy. AlN has the properties such as high th ermal conductivity, low thermal expansion coefficient, high electrical resistivity, good dielectric prope rties, and excellent oxidation resistance. 2.5.4 Other Substrates GaAs has the same structure as zincblende GaN. GaAs is less stable than SiC or sapphire. Above 800 C its decomposition rate to liquid gallium and arsenic vapor is considerable. GaAs has the large lat tice mismatch of 37.4 % for InN film. Zinc oxide (ZnO) has a wurt zite structure and its stacki ng order match with lattice constants closely matched to GaN ( a =3.249 , c = 5.205 ). The small lattice mismatch of 8.8 % for InN makes ZnO attractive substrate for InN growth. Lithium gallate (LiGaO2) also has the small lattice mismatch of 11 % for InN film. Therefore, LiGaO2 is another candidate for the suitable substrate for InN growth.

PAGE 69

52 2.5.5 Buffer Layer There is no lattice matched substrate availa ble for InN so far. For example, the InN has a lattice mismatch of 25 % with sapphire 8% with Si (111), 37.4 % with GaAs, and 11 % with GaN. High quality single crystal line InN is very difficult to be obtained because of these problems. The two-step growth method or growth using buffer layer has now become a standard method for the heteroepitaxial growth of thin films. This method is commonly used to alleviate lattice mismatch and th ermal expansion coefficient difference the substrate and epilayer. In this method, a thin buffer layer is grown at a low temperature in the first step. The main epilayer is grown in the second step at a high temperature. The buffer layer provides the high density of nuc leation centers and promotes the lateral growth of the main epilayer. Th e two-step growth of InN is not well studied, especially in the MOVPE growth. There are very few studies about the M OVPE growth of InN using buffer layer such as GaN, AlN, and InN. There is no si gnificant report that use of low temperature InN buffer layers in the grow th InN gives improvement. Pan et al studied two-step growth of InN using conventional MOVP E [Pan99]. Based on their findings, they concluded that the two-step gr owth is not adequate for InN, which may correlate to the unstable nature of the InN film. Guo et al reported that if a single crystalline InN film is heated above 550 oC in a N2 flow, the surface undergoes a considerable change, owing to the decomposition and desorption of nitrogen [Guo93].

PAGE 70

53 2.6 Summary for Growth of InN on Different Substrate 2.6.1 Growth on Sapphire (Al2O3) Substrate The growth of InN in horizonta l MOVPE reactor has been studied using -Al2O3 (0001) substrate by Yamamoto [Yam94b]. A sing le-crystalline InN film was obtained on -Al2O3 substrate at 500 oC, in spite of the larger la ttice mismatch for InN (0001)/ -Al2O3 (0001) by the nitridation of the Al2O3 (0001) substrate prior to the growth. Nitridation of -Al2O3 surface occurs at the temperature region from 800 oC to 1000 oC. AlN is formed during the nitridation and the lattice mi smatch is reduced from 25 % for InN/ -Al2O3 to about 13 % for InN/AlN [Yam94b, Pan99]. Chen found that the InN film qualit y is strongly dependent on the growth temperature and V/III ratio [Che97]. He reporte d the best quality of InN film was grown at 375 oC under a high V/III ratio of 146000 and the flow rate of NH3 of 2000 sccm. InN film growth was carried out in the atmosphe ric-pressure horizontal MOVPE reactor with a cross-section of 30 14 mm2. The FWHM of the best quality of InN (0002) was 96 arcsec with InN (10-11) existing while th e typical FWHM of XRC of MOVPE-grown InN is from 4000 to 5500 arcsec [Che97]. Surface morphology study of InN grown in MOVPE was carried out by AFM with different growth condition by Yamamoto [Yam01a]. A continuous InN film with enhanced two-dimensional gr owth was obtained at 630-650 oC. It was reported that growth rate was increased with increasing gr owth temperature in the range of 500-630 oC, while it is independent of growth temperature at a temperature higher than 630 oC. It was suggested that when the growth is performed at 630-650 oC, growth rate is proportional to TMI supply. Th e increase in growth rate with increasing growth temperature at a temperature less than 630 oC can be explained by taking account that

PAGE 71

54 growth rate is limited by NH3 decomposition rate. Yamamoto studied the effect of GaN buffer layer on InN and found that uniformity for grown InN film is markedly improved by employing a GaN buffer layer and this improvement is due to the uniform nucleation of InN [Yam04a]. The growth of InN in vertical resistiv e heated MOVPE reactor was performed by Hwang [Hwa01] where InN was grown at 360-540 oC; high V/III ratio was used to prevent indium droplet formation. The best InN was obtained at 540 oC and there was no reported about the value of FWHM. Takahashi carried out the gr owth of InN by HVPE at V/ III = 10-100 using InCl and InCl3 as In sources and NH3 and MMHy as N sources where the source of InCl was formed by the reaction between metallic In and HCl at 780 oC and InCl3 was evaporated from the source boat in temperature range 325-375 oC. The InCl3-NH3 system showed an appreciable growth rate of InN (~0.3 m/hr) and the growth rate initially increases with increasing growth temperature up to 550 oC and then gradually decreases to 700 oC. The other systems such as InCl3-MMHy, InCl-NH3, and InCl-MMHy showed the very small growth rate ( 0.05 m) [Tak97a]. The hydride vapor phase epitaxy growth of InN was performed by Yuichi Sato where HCl (diluted with N2 to 1 %) gas was passed over the indium metal source, which was kept in a quartz boat and the i ndium source was maintained at 900 oC in order to form InCl. The growth rate of the film gradually increases with increasing growth temperature and reaches the maximum growth rate of 0.3 m/h at 510 oC [Sat94a].

PAGE 72

55 In addition to MOVPE and HVPE, atomic layer deposition (ALD) and molecular beam epitaxy (MBE) were also used for InN growth on sapphire substrate by several researchers [Bed 97, Hig20, and Mam99]. In summary, the growth conditions of In N on sapphire substrate for the different growth methods such as MOVPE, HVPE, a nd ALD were discussed. The MOVPE is the commonly used growth method for InN and the high quality single crystalline InN growth by MOVPE is still required because th e typical range of FWHM of XRC is higher than 1000 arcsec. 2.6.2 Growth on Silicon (Si) Substrate The growth of InN on Si in ho rizontal MOVPE reactor was carried out by Yamamoto [Yam94b]. For Si substrate, rela tively well oriented InN films are grown at about 400 oC. Polycrystalline InN films are grown both at 350 oC and at 500 oC on Si substrate. Polycrystalline InN growth below 350 oC is believed to be due to reduced migration of deposited materi als on Si or decomposition ra te of raw materials. The growth at a temperature higher than 450 oC results in serious dete rioration of InN films grown on Si substrates. It was shown that the ni tridation occurs at a temperature as low as 500 oC by exposing to NH3 [Yam94b]. The cause for poorly-oriented or polycrystalline InN film growth on Si at a temperature above 400 oC was due to the formation of amorphous silicon nitride (SiNx) on Si substrates before the growth. He suggested that epitaxial growth of In N on Si without a buffe r layer is found to be difficult owing to the nitridation of Si s ubstrate. The application of InN on Si to a tandem solar cell was suggested by Yamamoto [Yam94a, Yam94b]. Yang et al improved the growth rate of InN on Si with a double-zone MOVPE system consisting of a high temperature NH3 pre-cracking zone and a low temperature

PAGE 73

56 deposition zone [Yan02c]. A maximum growth rate of 6 m /h was achieved due to the high cracking efficiency of NH3. In this experiment, he used N2 as a carrier gas, the flow rate of NH3 at 800-1600 sccm, and V/III ratio of several hundred s. The optimal growth temperature was 450 oC [Yan02c]. In summary, single crystalline InN growth on Si was obtained but no reports on crystalline quality (FWHM of XRC) have been reported Therefore, the growth conditions for high quality single crystallin e InN should to be studied and optimized. 2.6.3 Growth on Gallium Arsenide (GaAs) Substrate InN films was obtained on GaAs(111) at 500 oC, 1.3 Torr, and N2 flow rate of 200 sccm, using microwave-excited MOVPE by Guo et al .[Guo95b]. Yamamoto et al studied thermal nitridation of GaAs (111) in flowing NH3 and horizontal MOVPE growth of InN on the nitrided GaAs (111) as a resu lt of the thermal nitridation [Yam97a]. In the case of GaAs(111) substrates, crys tal structure of a GaN layer formed by the nitridation before the InN growth was found to be dependent on nitridation temperature TN ; zincblende structure for TN 700 oC and wurtzite for TN 800 oC. For an InN film grown on a GaAs (111) s ubstrate with a zincbl ende GaN layer, its crystalline structure is changed from zincbl ende to wurtzite when the thickness exceeds about 0.2 m. On a GaAs (111) with a wurtzite Ga N layer, on the other hand, growth of zincblende InN is not found [Yam98a]. Using an atmospheric HVPE system, InN growth was carried out on a GaN layer which was formed on a GaAs (100) substrate inclined 10 o toward the 111 B direction of GaAs substrate. An importa nt requirement for growth was to keep low temperatures of less than 750 oC in the upstream region of the r eactor to raise the amount of InCl3, where

PAGE 74

57 indium chloride is formed at the temperature higher than 750 oC. Furthermore, it was necessary to exclude H2 from the reaction system for de position to occur because the high partial pressure of H2 increases the amount of InCl. These results indicate that the effective chemical substance of indium chlorides for the growth is InCl3. Growth rate of 1.5 m/h was obtained at 570 oC and single crystalline InN growth was confirmed by Xray diffraction measurement [Sun96]. The growth of InN using MOVPE and HVP E was discussed in terms of growth conditions. For MOVPE, the structure of InN depends on the nitridation temperature for GaAs(111)B substrates. For HVPE, InCl3 forms InN film more effectively than InCl does. 2.6.4 Growth on Gallium Phosphorus (GaP) Substrate Guo et al reported that InN films had been grown on GaP (111) substrate at 500 oC using microwave-excited MOVPE and TMI and nitrogen were used as the source materials. The epitaxial InN film was obtaine d on GaP (111) by exposing the substrate to the nitrogen plasma for 60min before gr owth [Guo95b]. InN film s have a wurtzite structure [Guo95b]. Bhuiyan et al obtained InN on GaP(111)B by th e horizontal MOVPE reactor where single crystalline InN films can be obtained on GaP(111)B only when the nitridation of the substrate is not made intentionally. InN films grown on nitrided GaP(111)B are found to be polycrystalline. XPS analysis shows the formation of PNx as well as GaN after the nitridation of Ga P (111)B substrate surfaces by flowing NH3 above 500 oC. Formation of PNx is responsible for the poor crystalline structure for InN. A single crystalline InN film with an excellent surface morphology can be grown on

PAGE 75

58 GaP(111)B at high te mperature (600-650 oC) using a low temper ature InN buffer layer [Bhu00a,Bhu01,Bhu02b]. The growth of InN on GaP substrate usin g MOVPE was briefly discussed. When the growth of InN is performed on GaP subs trate, the nitridation step should not be required in order to obtain the single crystalline InN. 2.6.5 Growth on Gallium Nitride (GaN) a nd Alumimum Nitride (AlN) Substrate Yamaguchi et al presented the result of the InN film grown on GaN substrate with AlN buffer layer using atmospheric M OVPE. Growth temperature was 450 oC and V/III ratio was 105. The FWHM of XRC decreases with in creasing the thickness of InN film [Yam99a]. The effects of reactant-gas velocity on the growth of InN on GaN/sapphire by MOVPE were studied by Yang et al With a high-speed reactant gas, the thickness of the stagnant layer is reduced so that the reacta nt species can reach the surface effectively. A layer like growth of InN was achieved, resultin g in a significant improvement of the film quality. In addition, significant enhan cement of the growth rate up to 2 m/h was obtained. The FWHM of XRC decreased with increasing gas velocity. FWHM of XRC for InN (0002) with 476 arcs ec was reported but there wa s no report about whether the InN is single or poly crystalline [Yan02a]. The possibility that high quality single crystalline InN can be grown on GaN/sapphire substrate using M OVPE is studied and it is found that the flow pattern of source materials can have an effect on the InN film quality.

PAGE 76

59 2.7 Overview The latest lattice constant of single crystal InN with wurtzite was reported to be a = 3.537 and c = 5.704 [Dav02a]. The band-ga p energy of InN is nowadays accepted to ~ 0.7 eV instead of 1.89 eV. Thermodynamic analysis helped us to understand at which growth condition the growth and etching happen and therefore help us to predict where InN film can be grown before the epitaxial growth is performed. The possible candidates as N precursor were reviewed due to low decomposition efficiency of NH3 at low growth temperature of InN (~ 550 oC). Because all of other candidates for nitrogen source have the severa l problems such as t oxicity, explosion and contamination, NH3 is still widely used N precursor and N2 carrier gas is better than H2 as carrier gas. MOVPE is still the most widely used grow th technique for InN for the industry and academy to date. Based on available published data, the typi cal range of FWHM of XRC for single crystalline InN grown by MOVPE is higher th an 2000 arcsec [Yam04a]. For Si substrate, it is still very difficult to have the high crystalline InN because of the bad coverage of InN on Si substrate despite of a small lattice mismatch [this topic was not reviewed in the InN growth on Si substrates section]. Therefore the study for the growth high quality single crystalline InN has been still required. It is found that some factors such as gr owth temperature, V/III ratio, substrate, nitridation treatment, buffer laye r, and flow pattern of source gases can have an effect on the structural quality of InN film. Based on the re sults of this review, these factors will be analyzed in detail to conduct our experiments of InN growth by MOVPE.

PAGE 77

60 CHAPTER 3 THERMODYNAMIC ANALYSIS OF InN AND InXGa1-XN MOVPE GROWTH 3.1 Thermodynamic Analysis of InN and InxGa1-xN The results of a study on the effect of pr essure and temperature on the equilibrium growth of InN and InxGa1-xN are presented in this chapter. Specifically, equilibrium in IN-Ga-N-C-H system is studied to clarify the impact of process variables on film composition and to estimate a suitable growth condition for InN and InxGa1-xN to support the experimental studies. For example, it is interesting to know the maximum content of indium that can be incorporated into the InxGa1-xN phase without the phase separation. The Gibbs energy functions for InN and GaN from the Scientific Group Thermodata Europe (SGTE) and the ThermoCalc software package were used for these calculations. 3.1.1 Reaction Mechanism and Kinetics of InN MOVPE Growth of InN by MOVPE typically uses trimethylindium (TMI) and NH3 precursors in a N2 carrier gas. The pyrolysis of TMI was studied by Jacko et al [Jac64, Tra78, and Lar85] and they proposed the se quential hemolytic fiss ion of the In-C bond along methyl radical recomb ination described by Eq. (3-1a) to (3-1d). In(CH3)3 In(CH3)2 CH3 (3-1a) In(CH3)2 In(CH3) CH3 (3-1b) In(CH3) In CH3 (3-1c) CH3 CH3 C2H6 (3-1d)

PAGE 78

61 It is reported that reactions 3.1a and 3.1c are sl ow steps, thus producing In(CH3) into the vapor phase. Stepwise hemolytic fission of the In-C bond in TMIn was first proposed as in Eq. 3-1 [Jac64] and recently mono-methyl indi um (MMIn) and atomic indiums were experimentally observed in the gas phase [Par02]. New reaction intermediates are proposed based on experimental evidence using in situ Raman and computational chemistry supports by Hwang [Hwa04]. It has been suggested from Hwang’s experimental results [Hwa04] that MMIn and/or dimethyl i ndium (DMIn) seem to hide from Raman detection by forming another inte rmediate, presence of which in contrast to MMIn is evident. A new intermediate (HInCH3) was found to exist during TMIn decomposition in a nitrogen carrier. It has some of its characteristi c vibrations at 416 [ (H-In-C)], 464 [ (In-C)], and 1560 cm-1 [ (H-In)]. The new intermediate was experimentally observed to decompose very qu ickly in a high temperature region of the reactor. In addition, it was consid ered highly probable that (DMIn)2 and DMIn-MMIn would form during TMIn decomposition, as shown in Eq. (3-2) [Hwa04]. 2DMIn (DMIn)2 (3-2a) DMIn + MMIn DMIn-MMIn (3-2b) DMIn-MMIn CH3InCH2 + HInCH3 (3-2c) TMIn + MMIn (DMIn)2 (3-2d) In terms of the kinetics of TMI decomposition (Eq. (3 -1)), several experimental results are summarized in Table 3-1. Reaction rate constant and activation energies of TMI decomposition are shown in Table 3-1.

PAGE 79

62 Table 3-1. Reported reaction rate constants for TMIn decomposition. k0 (s-1) Ea (kcal/mol) Carrier Hwang [Hwa04] 1017.9 56.1 N2 Jacko & Price [jac64] 1015.7 47.2 Toluene 1012.6 40.5 N2 Larsen & Stringfellow [Lar86] 1012.0 35.9 H2 1017.9 54.0 He 1013.4 39.8 D2 Buchan et at. [Buc88] 1015.0 42.6 H2 Ammonia is the most widespread precurs or for III-Nitrides growth by MOVPE. Complex chemical equilibrium analysis by Koukitu suggested that most of the NH3 should be decomposed into N2 and H2 at temperatures greater than 300 oC [Kou97b]. It is well known, however, that the decomposition rate of NH3 under typical growth conditions is slow without a catalyst and the extent of the decomposition strongly depends on the growth conditions. The decomposition reactions of NH3 are presented by Eq. (3a-3c). NH3 NH2 H (3-3a) NH2 NH H (3-3b) NH N H (3-3c) N N N2 (3-3d) H H H2 (3-3e) Based on the aforementioned consideration, the several possible reactions for InN formation are suggested (Eq. (3-4)). In N InN (3-4a) In(CH3) NH InN CH4 (3-4b) In(CH3)2 NH2 InN 2CH4 (3-4c)

PAGE 80

63 (DMIn)2 + 2H 2InN + 2CH4 (3-4d) This complex chemical equilibrium anal ysis of the growth of InN requires computing the equilibrium state in the In-N -C-H system. The growth conditions of InN were calculated as a function of deposition te mperature, pressure, and composition. It is assumed that the vapor phase follows an id eal gas mixture and the vapor species whose equilibrium mole fractions are negligible (below 10-30) are not taken into account in this calculation because the same result for P-T di agram was obtained in either case when the species with the mole fraction less than 10-30 (C4H10, C4H2, C4, and N3 etc .) were included or excluded in the calcul ation. Therefore, the species with the mole fraction less than 10-30 are thought to be insigni ficant in the calculation. The equilibrium mole fractions of each component can be obtaine d from the equilibrium result data of ThermoCalc. With this assumption, species, phase, and thermodynamic properties included in this analysis are summarized in Table 3-2. Diamond was not considered in the calculation as the phase of graphite is taken as a stable one. The equilibrium state for the growth of InN without indium formation of 2-phase (In ( l ) + InN) was computed in the range of P = 1 to 101.3 kPa (7.5 to 760 Torr), T = 400 to 1000 oC, and V/III (NH3/TMI) = 50,000. The P-T diagram is shown in Fig. 3.1 a nd the results indicate that the etching temperature (decomposition temperature) is 810 oC at P = 13.3 kPa (100 Torr) and V/III = 50,000. Pressure and V/III ratio were chosen based on our current operation conditions of the MOVPE system used in this study.

PAGE 81

64 Table 3-2. Species, phases, and thermodynami c properties included in the analysis of MOVPE of InN. Phase Species Parameter (J/mol) Solid (s) C, InN, In GC(s) = -17368.4408+170.730318T-24.3TLN(T)-4.723 10-4 T2+ 2562600T-1-2.643 108T-2+1.2 1010T-3 GInN(s) = -149963.181+215.110609T-38.0744TLN(T) -0.0060668T2 GIn(s) = -6978.89011+92.3381153T-21.8386TLN(T) -0.00572566T2 -2.12032167 10-6T3-22906T-1 Liquid (l) In, N GIn(l) = -6978.89011+92.3381153T-21.8386TLN(T) -0.00572566T2 -2.12032167 10-6T3-22906T-1 + 3283.7-7.6402121T GN(l) = -4461.675+60.74575T-12.7819TLN(T)-0.00176686T2 +2.680735 10-9T3-32374T-1 Gas (g) C, C1H4, H, H2, NH3, In, In2, N, N2 GC(g) = 710430.933-17.7062915T-20.97529TLN(T)+1.998237 10-4T2 -3.34617167 10-8T3+1680.6515T-1+RTLN(10-5P) GC1H4(g) = -77295.5632-147.095196T-2.234656TLN(T) -0.048463265T2 +4.33754333 10-6T3-305431.45T-1+RTLN(10-5P) GH(g) = 211801.621+24.4989821T-20.78611TLN(T)+ RTLN(10-5P) GH2(g) = 9522.9741+78.5273879T-31.35707TLN(T)+0.0027589925T2 -7.46390667 10-7T3+56582.3T-1+ RTLN(10-5P) GNH3(g) = -53688.8736-38.3667407T-21.21774TLN(T) -0.022871695T2 +1.80809167 10-6T3-76698.65T-1+RTLN(10-5P) GIn(g) = 236267.082-68.7705731*T-15.35206TLN(T)-0.00527185T2 -3.98269833 10-7T3-94519.9T-1+RTLN(10-5P) GIn2(g) = 407673.852-41.5349376T-35.82134TLN(T) -0.00654889T2 +2.03928167 10-8T3-20133.605T-1+RTLN(10-5P) GN(g) = 466446.153-13.3752574T-20.89393TLN(T)+8.45521 10-5T2 -1.0018685 10-8T3+2788.7865T-1+RTLN(10-5P) GN2(g) = -8000.12556-8.81620364T-27.22332TLN(T)-0.0012599175T2 -5.39381 10-7T3-38326.695T-1+RTLN(10-5P)

PAGE 82

65 Figure 3-1. Calculated P-T phase diagram for InN at X(In) = 5.31212 10-6, X(N) = 0.24998, X(H) = 0.75000, X(C) = 1.59364 10-5 and V/III = X(N)/X(In) = 50,000. In summary, the InN P-T diagram was computed and the maximum growth temperature was estimated at the condition used in this study (P =13.3 kPa (100 Torr) for V/III = 50,000) using ThermoCalc software. It is known that the typical MOVPE growth temperature of In N is in the range of 400 to 700oC depending on V/III ratio, pressure, residence time, and probing methods [Bhu03b]. In terms of the growth temperature re gion of InN, the difference between thermodynamic calculation and the experimental result is thought to be caused by the fact that the MOPVE growth of InN is non-equilibrium reaction. 3.1.2 Pressur-Temperature (P-T) Phase Diagram of InxGa1-xN and Phase Separation in InxGa1-xN For MOVPE growth of InxGa1-xN in this study, trimethylindium (TMI) was used as In precursor, triethylgallium (TEG) as Ga precursor, NH3 as the reactive N source, and

PAGE 83

66 N2 as a carrier gas. Thus chemical equilibr ium state was computed in the In-Ga-N-C-H system. The growth condition for the growth of InN was calculated as a function of deposition temperature, pressure, and compos ition. In the thermodynamic analysis, the calculation procedures for InxGa1-xN are the same as that of InN, which was already discussed 3.1.1. The species, phases and thermodynamic properties included in this analysis are summarized in Table 3-3. The Redlich-Kister equation was used and the interaction parameters (Lo and L1) of liquid indium and liqui d gallium are presented in Table 3-3. The input mole fraction ratio of TMI to TEG which leads to the growth of In0.3Ga0.7N was calculated to show the relation of the input ratio of indium with the indium content in InxGa1-xN, which is obtained experimentally (Fig 3.2) [Mat92]. The indium content of 0.3 in In0.3Ga0.7N is chosen in this calculation because the maximum indium mole fraction is reported to be about 0.3 [Elm 98, Sin97, Nak94, Shi94, Nak93, Mat92]. The flow rate of TEG of 0.44 sccm, the flow rate of H2 of 4 slm, the flow rate of TMI of 0.27 sccm and the flow rate of NH3 of 1.6 slm were chosen in this study. This condition corresponds to X(In) = 1.87328 10-5, X(Ga) = 3.05276 10-5, X(N) = 0.111, X(H) = 0.8887, X(C) = 2.39364 10-4. The P-T phase diagram (Fig. 3.3) shows that the stable In0.3Ga0.7N can be obtained below the growth temperature of 730 oC and at P = 13.3 kPa (100 Torr), which is the operating pressure of our MOVPE system [Mat92]. From theses results, it is also clear that In0.3Ga0.7N is decomposed at T > 800 oC. Experimentally, the growth of In0.3Ga0.7N is very difficult at the temperature above 800 oC [Mat92]. These calculated results are in th e good agreement with the data obtained by Matsuoka [Mat92].

PAGE 84

67 Table 3-3. Phases and species includ ed in the analysis of MOVPE of InxGa1-xN. Phase Species Parameter (J/mol) Solid (s) C, InN, GaN, InGaN GC(s) = -17368.4408+170.730318T-24.3TLN(T)-4.723 10-4 T2+ 2562600T-1-2.643 108T-2+1.2 1010T-3 GInN(s) = -149963.181+215.110609T-38.0744TLN(T) -0.0060668T2 GGaN(s) = -137112+272.388T-44.3769TLN(T)-0.0063011T2 +586387T-1 LInGaN(s) = 40320 Liquid (l) In, Ga, N GIn(l) = -6978.89011+92.3381153T-21.8386TLN(T) -0.00572566T2 -2.12032167 10-6T3-22906T-1 + 3283.7-7.6402121T GGa(l) = 5491.298-18.073995T-7.017 10-17T7 -7055.643+132.73019T -26.0692906TLN(T)+1.506 10-4T2 -4.0173 10-8T3-118332T-1 +1.645 1023T-9 GN(l) = -4461.675+60.74575T-12.7819TLN(T)-0.00176686T2 +2.680735 10-9T3-32374T-1 Lo(Ga, In) = 4450+1.19185T L1 (Ga, In) = 0.25943 T Gas (g) C, C1H4, H, H2, N1H3, Ga1, Ga2, In, In2, N, N2 GC(g) = 710430.933-17.7062915T-20.97529TLN(T)+1.998237 10-4T2 -3.34617167 10-8T3+1680.6515T-1+RTLN(10-5P) GC1H4(g) = -77295.5632-147.095196T-2.234656TLN(T) -0.048463265T2 +4.33754333 10-6T3-305431.45T-1+RTLN(10-5P) GH(g) = 211801.621+24.4989821T-20.78611TLN(T)+ RTLN(10-5P) GH2(g) = 9522.9741+78.5273879T-31.35707TLN(T)+0.0027589925T2 -7.46390667 10-7T3+56582.3T-1+ RTLN(10-5P) GNH3(g) = -53688.8736-38.3667407T-21.21774TLN(T) -0.022871695T2 +1.80809167 10-6T3-76698.65T-1+RTLN(10-5P) GGa(g) = 259072.279+88.0130701T-38.71057TLN(T)+0.01053784T2 -9.86907833 10-7T3 (298.13 K T 600 K) 263812.519+33.4871429T-30.75007TLN(T)+0.00537745T2 -5.46534 10-7T3-150942.65T-1 (600 K T 1400 K) GGa2(g) = 422882.385-36.0787973T-33.72863TLN(T)-0.009368525T2 +7.62775167 10-7T3-19520.385T-1 GIn(g) = 236267.082-68.7705731*T-15.35206TLN(T)-0.00527185T2 -3.98269833 10-7T3-94519.9T-1+RTLN(10-5P) GIn2(g) = 407673.852-41.5349376T-35.82134TLN(T) -0.00654889T2 +2.03928167 10-8T3-20133.605T-1+RTLN(10-5P) GN(g) = 466446.153-13.3752574T-20.89393TLN(T)+8.45521 10-5T2 -1.0018685 10-8T3+2788.7865T-1+RTLN(10-5P) GN2(g) = -8000.12556-8.81620364T-27.22332TLN(T)-0.0012599175T2 -5.39381 10-7T3-38326.695T-1+RTLN(10-5P)

PAGE 85

68 Figure 3-2. Relation between i ndium mole fraction (x) of InxGa1-xN and the flow rate ratio of the sum of group III source of TMI and TEG. Figure 3-3. Calculated P-T phase diagram for In0.3Ga0.7N at X(In)=1.87328 10-5, X(Ga)=3.05276 10-5, X(N)=0.111, X(H)=0.8887, X(C)=2.39364 10-4 and the data points () are from the measurements observed by Matsuoka.

PAGE 86

69 The main problem in growing InxGa1-xN has been the phase separation at high indium content (X(In) 0.3) due to the very high equi librium vapor pressure of nitrogen over InN. This phase separation leads to the miscibility gap in InxGa1-xN [Ho96]. In the calculation of the miscibility gap in InxGa1-xN, the 2-sublattice regular solution model was used for solid phase which consists of InN and GaN sub-lattices and the interaction parameter ( L ) of 5.98 kcal/mol [Ho96] and the Redlich-Kister model was used for liquid phase where the interaction parameter of L0 and L1 are given in Table 3-2. The regular solution model for the Gibbs excess energy of mixing for InxGa1-xN is given by L x x N Ga In GGaN InN x x xs mix ) (1 (3-5) The Redlich-Kister model for the Gibbs excess energy of mixing for In ( l ) and Ga ( l ) is given by ) ( ) ( 1 0 ) ( ) ()) ( ) ( (l Ga l In l Ga l In xs mixx x L L x x l Ga l In G (3-6) The existence of a miscibility gap and phase separation of InxGa1-xN grown by MOPVE was confirmed by using Thermo Calc Software (Fig.3.4) [Pin98]. The maximum calculated indium content in InxGa1-xN grown by MOVPE is ~ 0.1 at 780 oC, and the experimental data were used from the paper of E.L. Piner [Pin99]. Our calculations of the maximu m indium content in InxGa1-xN and the critical temperature are in the good agreement with results presented in [Str99]. However, based on experimental data the maximum indium content in InxGa1-xN is ~ 0.3 [Elm98].

PAGE 87

70 Figure 3-4. Thermodynamically calculated miscibility gap of InxGa1-xN grown by MOVPE and the data points () are from the measurements observed by Piner. This gap between the experimental result and theoretical one is caused by the fact that MOVPE growth of InxGa1-xN is non-equilibrium reaction and this thermodynamic results show the binodal curve which corresponds to the stable regi on but not show the spinodal curve corresponding to the meta-stable region which can be achieved experimentally. 3.2 Quantum Calculation of Phase Separation in InxGa1-xN 3.2.1 Boundary Passivation Method with Hydrogen The phase separation in InxGa1-xN was studied using Quantum calculation method (also called the first-principle method or ab-initio method). For this calculation, the Hatree-Fock Self-Consistent-Fie ld (HF-SCF) method was used and the total energy of the system was calculated by the Schrdinger equation (Eq.3-7). HP HPE H (3-7) where H is the Hamiltonian operator, E is the energy of system, and HP is the Hatree product (the wave function of the system);

PAGE 88

71 N HP 2 1 (3-8) i i i ih (3-9) Therefore, HP N i i N N i i HP N i i HPh h H 1 2 1 1 1 (3-10) j V r Z hi M k ik k i i 1 22 1 (3-11) 1 1 21j j ij j j ij j ir dr r j V (3-12) where Vi{ j } represents an interaction potential with all of the other electrons occupying orbitals { j } and i, j represent the electron and k the nucleus. In the first step of the SCF, one guesses the wave function for all of the occupied molecular orbitals (MOs) and uses these to construct the necessary one-electron operator, h Solution of each differential equati on (Eq. 3-11) provides a new set of presumably different from the initial guess. The one-electron Hamiltonians are formed anew using these presumably more accurate At some point, the difference between a newly determined set and the immediately pr eceding set falls some threshold criterion and the final set of is referred to as the ‘converged ’ SCF orbitals. The computational process is shown in Fig. 3.5.

PAGE 89

72 Figure 3-5. Flow chart of the HF-SCF procedure. For the calculation of phase separation in InxGa1-xN, the structure of the unit cell was set up, which contains In Ga, N, and H with different indium composition and all nitrogens at the wall sides were passivated by hydrogen to calculate th e total energy (see Fig. 3.6). As the indium content increases, the site of Ga is exchanged with In atom. Three sets of bond lengths for In-H, Ga-H and N-H were used for HF-SCF/3-21G calculation. The first one is derived from th e calculation of the software of Molden, the second one from that of Hiraoka [Hir94], and the third one is obtai ned using PM3 method, one of semi-empirical methods with Hype rchem software (Table 3-4). The other bond lengths were obtained from the data of In aba, who calculated these bond lengths using the CAmbridge serial Total Energy Package (CASTEP) code [Ina01]. The energies were

PAGE 90

73 calculated using three different calculation met hods with Gaussian software (Table 3-5) where Hatree energy is equal to 627.51kcal/mol. Figure 3-6. Structures used to compute the total energy for the InxGa1-xN vs. indium mole fraction.

PAGE 91

74 Table 3-4. Bond lengths for the calculation using HF-SCF. In-H Ga-H N-H Ga-N (x=0) In0.25Ga0.75N In-N Ga-N In0.38Ga0.62N In-N Ga-N In0.75Ga0.25N In-N Ga-N InN (x=1) 1 1.813 1.813 1.008 1.96 2.171.968 2.174 1.969 2.187 1.971 2.188 2 1.588 1.588 0.991 1.96 2.171.968 2.174 1.969 2.187 1.971 2.188 3 1.694 1.624 0.991 1.96 2.171.968 2.174 1.969 2.187 1.971 2.188 Table 3-5. Calculated total energy fo r three types of different bond length. Energy (Hatree) HF/3-21G (1) HF/3-21G (2) HF/3-21G (3) GaN -15758 -15763 -15762 In0.25Ga0.75N -23365 -23366 -23366 In0.38Ga0.62N -27166 -27167 -27167 In0.75Ga0.62N -38558 -38570 -38570 InN -46173 -46173 -46173 In the method of HF/3-21G, the 3 signifies that three core expanded Gaussian basis function are used as the core function and the 21G indicates that the valence functions are split into one basis function with two Gaussian type orbitals and one with only one Gaussian type orbital. When these energies were compared, all three showed a similar value for each mole fraction of In. After the first one was select ed, the energy in th e state without phaseseparation and the energy in the state with phase separation were compared. The energy in the state with phase sepa ration was calculated from th e summation of the energy of GaN and the energy of InN according to the le ver rule. It is found out that the energy with phase separation is less than the energy without phase separation when the indium mole fraction (X(In)) is changed from 0.25 to 0.38 (Table 3-6).

PAGE 92

75 Table 3-6. Calculated energies of InxGa1-xN with the phase separation and without phase separation. Energy (Hatree) Energy for InxGa1-xN (HF/3-21G (1)) Energy for (1-x)GaN + x InN (Phase separation) GaN -15758 -15758 In0.25Ga0.75N -23365 -23361 In0.38Ga0.62N -27166 -27315 In0.75Ga0.25N -38558 -38569 InN -46173 -46173 Based on presented data, the phase separation in InxGa1-xN started to occur at 0.25 X(In) 0.38. This result is in good agreement with the reported va lue of X(In) = 0.28 [Elm98]. In summary, the phase separation in InxGa1-xN was studied using quantum calculation method and 2-sublattice regular so lution model. Calculated values of the maximum indium mole fraction, X (In), in InxGa1-xN by quantum calculation method are in a good agreement with the re ported experimental data.

PAGE 93

76 CHAPTER 4 CALCULATION OF THE CRITICAL THICKN ESS OF InN ON GaN, AlN, Si, AND Al2O3 As the lattice mismatch between an epita xial film and substrate increases, highquality epitaxial growth can not continue in definitely because the strain energy of the layer is eventually completely or partiall y relieved (or relaxed) by the generation of dislocations at the interf ace (misfit dislocations). When a film with an unstrained lattice constant, af is deposited on a substrate with a different lattice constant, as (Fig. 4.1a), it initially grows with a lattice constant equal to that of the substrate. The mismatch is accomm odated by strain in th e layer. This is known as a pseudomorphic film. This continues until the film reaches some critical thickness hc (Fig. 4.1b). The critical thickness is the thic kness of the overgrown film at which misfit dislocation begins to occur. When the film thickness exceeds hc, the misfit is accommodated by the formation of misfit disloc ations which emanate from the interface between the film and substrate, and the lattice constant of the film relaxes toward the unstrained value (Fig. 4.1c) [Woo83]. The critical thickness fo r pseudomorphous growth is very important from a technological point of view. Fi rst, the misfit dislocations de teriorate the performance of the heterostructure devices due to the increa sed leakage current. Second, in uniformly strained epilayers, the interatomic spacing diffe rs from that in the unstrained (relaxed) ones, thus changing the bandgap energy. Sap phire traditionally has been the most commonly used substrate for III-Nitride growth Unfortunately, in case of InN the lattice

PAGE 94

77 mismatch is significant – 25.7%. Several altern ative substrates are considered, such as GaN, AlN, and Si with the lattice mismatch of 10.9 %, 13.7 %, a nd 7.9 % respectively. These values are much sm aller compare to In/Al2O3 case. The values of critical thickness of InN on grown on GaN, AlN, Al2O3, and Si substrates were calculated using different models and results are presented in this chapter. Figure 4-1. Schematic representation of the formation of misfit dislocations: (a) unstrained lattice; (b) thickness of the film is less than hc; (c) thickness of the film is greater than hc misfit dislocations are generated. 4.1 Calculation of Critical Thickness of InN by Matthews’ Method. In the simple model of Matthews (Mat75, Tu92), the critical thickness ( hc) for pseudomorphic epitaxy is derived by minimizing the total energy, as the sum of the stored energy as strain and the net energy whic h film releases from dislocation formation. It is assumed that film and substrate are cubic and prepared from elastically isotropic material. The elastic constants of th e two crystals are assu med equal. The misfit

PAGE 95

78 dislocations are considered to be arranged in a square network, to be in edge orientation, and to have Burgers vectors in the interface plane. The energy of a square grid made up of two perpendicular and non-inte racting arrays of edge dislo cation is therefore given by Eq. (4-1). The energy associated with the elasti c strain in the film is given by Eq. (4-2). 1 ln ) ( ) 1 ( 2 b h f b G Ef n dislocatio (4-1) h G Ef strain) 1 ( ) 1 ( 22 (4-2) 1 ln ) ( ) 1 ( 2 ) 1 ( ) 1 ( 22b h f b G h G Ef f total (4-3) where the in-plane strain is defined as f f flla a a (4-4) f s fa a a f (4-5) s f s fa a a a b ) ( 2 (4-6) where afll is the parallel-tothe-interface or in-plane lattice constant of the deposited film material, af is the lattice constant of film material in the bulk or unstrained state, as is the lattice constant of substrate, f is the misfit between film and substrate, Gf is the shear modulus of the film, b is the Burger vector of the dislocation, and is Poisson’s ratio for the film. The strain that minimizes the total energy is obtained by setting dEtotal/d = 0 to give the critical strain (*).

PAGE 96

79 1 ln ) 1 ( 8*b h h b (4-7) The critical thickness ( hc) is obtained by iteration with *= f 1 ln ) 1 ( 8 b h f b hc c (4-8) According to Eq. (4-7), the critical thic kness of the overgrown film depends on the misfit f Burger vector b and the Poisson’s ratio for the film. For the calculation of the critical thickne ss of InN film on GaN and AlN substrates, selected physical properties of these materials are needed including the lattice constants of InN, GaN, and AlN. These property values are summarized in Table 4-1, where the Burger vector is calculated using Eq. (4-6) and Misfit dislocation is calculated using Eq. (4-5). Table 4-1. Physical properties required for th e calculation of the cr itical thickness of InN on GaN, AlN, Al2O3, and Si substrates. [aBel04, bLiu02] Table 4-2. Calculated critical thickness of InN on GaN, AlN, Al2O3, and Si substrates using Mattews’ method. GaN (0001) substrate AlN (0001) substrate Al2O3 (0001) substrate Si (111) substrate Burger vector ( b ) 3.354 3.311 4.758 3.840 Poisson’s ratio of InN ( ) 0.3a (300K) 0.3a (300K) 0.3b (300K) 0.3b (300K) Misfit dislocation ( f ) -0.098 -0.120 0.345 0.086 GaN (0001) substrate AlN (0001) substrate Al2O3 (0001) substrate Si (111) substrate Critical thickness ( ) 0.658 0.599 0.440 0.760

PAGE 97

80 All these calculated values of critical thickness of InN lead to the conclusion that the misfit dislocations are formed during the growth of the fi rst monolayer on all considered substrates. 4.2 Calculation of Critical Thickness of InN by van der Merwe’s Method. When the square atomic meshes of the adjoining crystal planes are considered, the energy of the interface due to the lattice misfit will be equal to the energy of the homogeneous strain Ehs given by b b f hshf G E 1 1 22 (4-9) Beyond the critical thickness, misfit dislocations are intr oduced at the interface so that initially homogeneous strain and misfit dislocation energy coexis t. According to the theory of van der Merwe, the energy of the misfit dislocations (the homogeneous strain is absent) is naturally divided into two parts. The first is the energy of intersection between the atoms of the two crystal halves (Eq. 4-10) [Mar03]. 2 2 21 1 4 d b G Ei i (4-10) where d b is the separation of the atoms of the adjoining crystal planes. The second is the energy of the periodic elas tic strain energy which is distributed in the two crystal halves. This second energy is th e total strain energy pe r atom of the misfit dislocations (Eq. 4-11). 2 2 2 22 1 2 ln 4 d b G Ei e (4-11) The misfit dislocation energy Ed is then given by van der Merwe (Eq. (4-10)) through the sum of (Eq. 4-10) and (Eq. 4-11) [Mar03].

PAGE 98

81 2 2 2 22 1 2 ln 1 1 4 d b G Ei d (4-12) s f s f s f s f s f i b b a a ia a a a p a a a a b G G G G G G p b G G 2 1 1 1 2' where G’ is the shear modulus at the interface, Gf is the shear modulus of the film, Gs is the shear modulus of the substrate, b is Burger vector, and p is the vernier of misfit or the dislocation spacing as shown in Fig. 4.2 [Mar03]. Th e dashed lines located at a distance p /2 from the contact plane show the bou ndary beyond which th e periodic strains originating from the dislocations practica lly vanish. The physical properties for the calculation of critical thickness of InN on GaN, AlN, Al2O3, and Si substrates are summarized in Table 4-3. Figure 4-2. Model of epitaxial interface between two semi-infin ite crystals resolved in a sequence of misfit dislocations spaced at an average distance p

PAGE 99

82 Table 4-3. Physical properties required for th e calculation of critic al thickness of InN on GaN, AlN, Al2O3, and Si substrates. InN GaN (0001) substrate AlN (0001) substrate Al2O3 (0001) substrate Si (111) substrate Poisson’s ratio ( ) d0.3 d0.23 d0.2 a0.275 b0.218 Bulk Modulus (GPa) c165 c236 c248 a338 b98 [aBel80, bGeo99, cChi99, dBel04] In the isotropic materials, the shear modulus ( G ) can be calculated from a bulk modulus ( B ) or tensile (Young’s) modulus (E) by E B G 1 2 1 ) 1 ( 2 ) 2 1 ( 3 (4-13) The modulus ( G ) calculated from the given bulk modulus ( B ) for each material using equation (4-13) is shown in Table 4-4. Table 4-4. Calculated shear moduli for InN, GaN, AlN, Al2O3, and Si materials. InN GaN (0001) substrate AlN (0001) substrate Al2O3 (0001) substrate Si (111) substrate Shear modulus (GPa) 76 155 186 179 68 Equating the 2 Ed of a square grid of two perpendi cular and noninteracting arrays of misfit dislocations and the energy Ehs of the homogeneous strain gives ( a = b = ) 2 2) ( ) 1 ( ) 1 ( 4 f f G G b hf i c (4-14) f s fa a a f 2 2 22 1 2 ln 1 1 ) ( f (4-15)

PAGE 100

83 To illustrate the minimum energy considerations, the homogeneous strain energy Ehs is plotted for the numb er of InN epilayers and the dislocation energy Ed against the misfit for InN on a GaN substrate (see Fig. 4.3). 0 2 4 6 8 10 12 14 16 18 00.020.040.060.080.1 EhS(n=1) Misfit ((af-as)/af) E(J/m2) 2Ed 0.047 InN on GaN substrate Figure 4-3. Homogeneous strain energy, Ehs for the 1st InN epilayer and dislocation energy, Ed vs. the misfit, f on GaN substrate. Based on results presented in Fig.4.3, if the misfit f 0.047 (the critical value), Ehs is greater than Ed even for a monolayer, which means th at the misfit dislocation occurs during the growth of the 1st InN epilayer since the misfit f between the InN film and the GaN substrate is 0.098. The calculated criti cal thickness (using Eq. 4-14) of InN on GaN substrate is 3.927 . Identical calculation proce dure was performed for the growth of InN on an AlN substrate with the matching result that the misfit dislocation also occurs during the growth of the 1st InN epilayer since the misfit (f) between InN film and AlN substrate is 0.120, which is larger than 0.048 of the critical value of (Fig .4.4). The critical thickness of InN on AlN substrate is 3.044 (using Eq. 4-14).

PAGE 101

84 0 5 10 15 20 25 30 35 00.020.040.060.080.10.120.14 Misfit ((af-as)/af) E(J/m2) EhS(n=1) 2Ed 0.048 InN on AlN substrate Figure 4-4. Homogeneous strain energy, Ehs for the 1st InN epilayer and dislocation energy, Ed vs.the misfit, f on AlN substrate. Similar calculation procedure was perfor med for the growth of InN on an Al2O3 substrate with the result pointed to the conclu sion that the misfit dislocation also occurs during the growth of the 1st InN epilayer since the misfit (f) between InN film and Al2O3 substrate is 0.345, which is larg er than 0.061 of the critical valu e of (Fig.4.5). The critical thickness of InN on Al2O3 substrate is 0.574 from Eq. 4-14. 0 5 10 15 20 25 30 35 00.020.040.060.080.10.120.14 Misfit (laf-asl/af) E(J/m2) EhS(n=1) 2Ed 0.061 InN on Al2O3 substrate Figure 4-5. Homogeneous strain energy, Ehs for the 1st InN epilayer and dislocation energy, Ed vs. the misfit, f on Al2O3 substrate.

PAGE 102

85 Analogous calculation procedure was perf ormed for the growth of InN on Si substrate which led us to the conclusion that the misfit dislocation also occurs during the growth of the 1st InN epilayer since the misfit (f) between InN film and Si substrate is 0.086, which is larger than 0.039 of the critical value of (Fig .4.6). The critical thickness of InN on Si substrate is 3.450 from Eq. 4-14. All of the calculated critical thickness of InN on GaN, AlN, Al2O3, and Si substrate are given in Table 4-5. 0 2 4 6 8 10 12 14 16 00.010.020.030.040.050.060.070.080.09 Misfit (laf-asl/af) E(J/m2) EhS(n=1) 2Ed 0.039 InN on Si substrate Figure 4-6. Homogeneous strain energy, Ehs for the 1st InN epilayer and dislocation energy, Ed vs. the misfit, f on Si substrate. Table 4-5. Calculated critical thickne ss of InN using van der Merwe’s method. From these results, the smallest value of the critical thickness is assigned for the InN/Al2O3 system, although it has the biggest latt ice mismatch. Substrates with the GaN (0001) substrate AlN (0001) substrate Al2O3 (0001) substrate Si (111) substrate Critical thickness of InN ( ) 3.927 3.044 0.574 3.450

PAGE 103

86 lowest lattice mismatch, like Si or GaN, show ed significantly higher values of the critical thickness which may be a good indicator that the critical thickness depends not only on the lattice mismatch of InN film and substr ate material, but also the shear modulus and Poisson’s ratio should be taken into account. 4.3 Calculation of Critical Thickness of InN by the Methods of Shen, Jesser, and Wilsdorf. For the calculations of the critical thic kness by Matthews’ and van der Merwe’s methods, it was assumed in the calculation of stra in energy that the cr ystal structure is the cubic. Although the energy difference between the cubic and the hexagonal wurzite of the group III-nitride is small, the hexagonal structure is more stable one. Shen et al. proposed the equation for the calculations of the stra in energy for the hexa gonal structure [She02]. The compliances, s11 and s12 could be calculated from the elastic constants (a kind of modulus, stress per unit elastic strain) c11, c12, c13, c33, c44 (Eq. 4-16). These elastic constants are obtained using theoretical quant um calculation or the scattering and x-ray diffraction experimental method. The n is the number of layer and c is the lattice constant (5.704 ) in the vertical direction. Therefore n c is the thickness of overgrown InN film. c n b b b s s Ef f ll strain 2 12 111 (4-16) ) 2 )( ( ) 2 )( (2 13 33 12 33 11 12 11 33 12 2 13 12 2 13 33 12 33 11 12 11 2 13 33 11 11c c c c c c c c c c s c c c c c c c c c c s (4-17) In addition to the strain energy, the othe r energy such as dislocation energy should be considered in order to calculate the total energy. Jesser and Wilsdorf suggested the interfacial energy to include the dislocation energy. According to Jesser and Wilsdorf

PAGE 104

87 [Jes67], the energy per unit area of each of the two dislocation arrays in the absence of the other is given by xEd and yEd respectively, while the interfacial energy per unit area is given by the energy of the inte rfacial dislocations added to C N, the energy which the N overgrowth atoms per unit area possess when a ll of them are resting in the potential trough where C is the minimum interfacial pot ential energy per atom. Thus, the interfacial energy, EI is given by N C E E Ed y d x I (4-18) where it has been assumed that there is no interaction energy between the two dislocation arrays. At the same time, the in terfacial energy may be written as the sum of the overgrown film and substrate surface energies per unit area, E1 and E2, when the two crystals are separated, less EB (the binding energy between th e two crystals when joined), i.e. B IE E E E 2 1 (4-19) EB may be presented in terms of the dislocation energy and binding energy (Eb), which would exist when there is no misfit, as ) (d y d x b BE E E E (4-20) so that b d y d x IE E E E E E 2 1 (4-21) where the binding energy (Eb) is independent of misfit by virtue of definition. By summing the strain equation and the inte rfacial energy, the total energy is given by

PAGE 105

88 ) 2 ( 4 4 2 22 2 2 2 2 _ intc G a G b G c K f c G E E E E Ei a b i binding a surface b surface d erfacial (4-22) 4 2 1 1 1 2 2 1 2 ln 1 1 ) (' 2 2 2 c b a i b b a a iN K a b ab p b a ab c G G G G G G p c G G f 704 5int n E E Estrain erfacial total n =1, 2, 3 (4-23) where 5.704 is the lattice constant of c of InN (Table 4-6), n is the number of the layer of the overgrown film and Nc is the coordination number. As n increases, the total energy shows a minimum at the critical thickness. In other words, when 0 d dEtotal, misfit dislocations occur. The values of the elastic constants and compliances are summarized in Table 4-7. Table 4-6. Lattice constant () of InN, GaN, AlN. Material a () c () InN (wurtzite) b3.537 b5.704 GaN (wurtzite) a3.189 a5.185 AlN (wurtzite) a3.112 a4.98 [aMor94, bDav02a]

PAGE 106

89 Table 4-7. Elastic constants cij and compliances sij of InN. cij (GPa) c11 297.5 c12 107.4 c13 108.7 c33 250.5 c44 89.4 sij (1/GPa) s11 0.00424 s12 -0.00102 [Chi99] Results presented in Figure 4-7 indicate that for the growth of InN on GaN, the total energy shows a minimum at the 1st InN epilayer and as the result the misfit dislocation happen from the 1st monolayer of InN. The same outcomes were obtained for the growth of InN on AlN, Al2O3, and Si substrates (Fig. 4.8, Fig 4.9, and Fig.4.10). For the growth of InN on all named su bstrates, the critical thickness hc is less than 5.704 which points towards conclusion that the misfit dislocation o ccurs during the growth of the 1st monolayer of InN. 0 5 10 15 20 3.183.223.263.33.343.383.423.463.53.543.58 Es(1) Lattice constant () E(J/m2) EiEt(1) 3.523 (dEtot/d =0) InN on GaN substrate Figure 4-7. Total energy Et, strain energy Es, interfacial energy Ei vs. lattice constant of InN for 1st epilayer InN on GaN substrate.

PAGE 107

90 0 5 10 15 20 25 30 3.13.143.183.223.263.33.343.383.423.463.53.543.58 Es(1) Lattice constant () E(J/m2) EiEt(1) 3.530 (dEtot/d =0) InN on AlN substrate Figure 4-8. Total energy Et, strain energy Es, interfacial energy Ei vs. lattice constant of InN for 1st epilayer InN on AlN substrate. 0 20 40 60 80 100 120 140 160 180 200 220 3.53.63.73.83.944.14.24.34.44.54.64.74.8 Es(1) Lattice constant () E(J/m2) EiEt(1) 3.599 (dEtot/d =0) InN on Al2O3 substrate Figure 4-9. Total energy Et, strain energy Es, interfacial energy Ei vs. lattice constant of InN for 1st epilayer InN on Al2O3 substrate.

PAGE 108

91 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 3.533.563.593.623.653.683.713.743.773.83.83 Es(1) Lattice constant () E(J/m2) EiEt(1) 3.559 (dEtot/d =0) InN on Si substrate Figure 4-10. Total energy Et, strain energy Es, interfacial energy Ei vs. lattice constant of InN for 1st epilayer InN on Si substrate. The critical thickness hc of InN on GaN, AlN, Al2O3, and Si substrates was calculated using three different models and the results ar e summarized in Table 4-8. Compared with the Mattew’s model, the van der Merwe’s m odel shows the different value of critical thickness for all substrate but sa pphire since two models come from the different concept. The van der Merwe introduces the new paramete rs such as the vernier of misfit and shear modulus at the interface, and new concept of 2Ed=Ehs. It is considered that this may cause the different result. For sapphire, the simila r result happens to be obtained. Each independent calculation showed that the cr itical thickness of InN film grown on all considered substrates is less than a m onolayer width These results are in good agreement with the experimental data ba sed on HRTEM of InN on GaN [Bel04], which shows that the misfit dislocations were introduced during th e growth of the 1st epilayer of InN (thickness of the monolayer of InN = 5.704 ) for InN grown on GaN substrate.

PAGE 109

92 Table 4-8. Critical thickness ( hc) calculated of InN using three different models. hc () InN on GaN hc () InN on AlN hc () InN on Al2O3 hc () InN on Si Matthews 0.658 0.599 0.440 0.760 van der Merwe 3.927 3.044 0.574 3.450 Shen-JesserWilsdorf 5.704 5.704 5.704 5.704 In summary, the critical thickness is le ss than the monolayer width when three theoretical models are used. Th is result may suggest that the study of suitable substrate be necessary in order to improve the crystallinity of InN. In this chapter, the critical thickness of InN was calculated by taking into ac count the several energies that can exist in the InN epitaxial film and the physical properties of the epitaxial film and substrate using three different methods. As a result, it is concluded that our cal culated values are in good agreement with the experimental result s available for the growth of InN on GaN substrate [Bel04]. Considering the theoretical point of view the model of Shen, Jesser, and Kuhlmann-Wilsdorf is thought to be the most accurate one.

PAGE 110

93 CHAPTER 5 Indium Nitride (InN) GROWTH BY METAL-ORGANIC VAPOR PHASE EPITAXY (MOVPE) Unlike MOVPE growth of Ga and Al nitride, it is difficult to obtain high quality InN by MOVPE because the temperature range for successful growth is very narrow (500 to 650 oC) and for alloys not well matched to that for the other III-nitrides. Part of the problem is related to the low growth temper ature and the low amount of atomic nitrogen given the low decomposition efficiency of ammonia in this temperature range. Furthermore there is no obvious subs trate for growth of InN. In this chapter, re sults on the optimizati on of the growth conditions for InN are presented for several substrate and buffer laye r combinations. The process variables that were manipulated are growth temperature, N/In ratio, pressure, and the buffer layer growth conditions (temperature, GaN or InN buffer layer, and nitridation). The postgrowth annealing of InN film was also studi ed. The reactor design was modified from a horizontal type to an extende d horizontal and vertical ge ometry by changing the inlet tube. 5.1. Indium Nitride (InN) Growth Optimization Optimization of MOVPE growth condition is typically accomplished by empirical studies of the key process parameters. This study concentrates on substrate selection, growth temperature, N/In ratio, buffer laye r material (GaN or InN), and post-growth annealing. Several film propert ies are often important for a pa rticular application (e.g., background impurity concentration, defect de nsity, surface roughness). In general, the

PAGE 111

94 optimum conditions for achieving a particularly measure of film quality are different fro another film property. The qua lity factors included in this study are crystalline quality (XRD), surface roughness (AFM, SEM), background impurity level (PL, AES), and carrier type and mobility (Hall measurement). These factors are studied for each substrate and buffer layer. 5.1.1. Substrate Selection The selection of substrate is one of the most important considerations for epitaxial growth. In this study, thr ee substrates, Si (111), Al2O3 (0001), and GaN (5 m)/Al2O3 (0001), are explored for the epitaxial growth of InN. Since Al2O3 and Si have different crystal structures than InN, as well as di fferent lattice constants values and thermal expansion coefficients, heteroepitaxy of InN on these substrates is expected to be a challenge. The mismatch in la ttice constant important in heteroepitaxy is that which occurs at growth temperature, since growth mechanisms are influenced by the lattice spacing at growth temperature. Subseque nt adhesion and cracking issues upon cooling are determined in part by the mismatch in thermal expansion coefficients. Upon cooling, the mismatch in thermal expansion coefficients can lead to increasing strain, and possibl y producing cracking. The la ttice constants and thermal expansion coefficients of the substrates and InN are summarized in Table 5-1[aMor94, bDav02]. The lattice mismatch ( f ) between InN and substrate is defined by % 100 f s fa a a f (5-1) where af is the lattice constant of film material (InN) in the bulk or unstrained state, as is the lattice constant of substrate.

PAGE 112

95 The linear thermal expansion coeffici ent (TEC) mismatch between InN and substrate is calculated in the same way by % 100 f s fa a a mismatch TEC, (5-2) where af is the thermal expansion coefficient of film material in the bulk or unstrained state and as is the thermal expansion coefficient of substrate. Table 5-1. Structural prop erties of InN, GaN, Al2O3, Si, and AlN substrates. Substrates Lattice constant a () Lattice constant c () Lattice mismatch (%) TEC (10-6 K1) a c TEC mismatch (%) InN-wurtzite GaN-wurtzite Al2O3 – rhombohedral Si (111) – cubic AlN-wurtzite 3.537b 3.189a (5.524 when rotated 30 o) ‘4.758a 5.43a (3.84 as hexagonal) 3.112a 5.704b 5.18a 12.991a 4.98a -10.9 25.7 7.9 (hexagonal) -13.7 5.70b 5.59a 7.5a 6.2a 4.2a 3.70b 3.17a 8.5a 5.3a 0 -30.9 -48.5 -37.7 -8.1 When the GaN film is rotated by 30o with respect to the sapphire substrate, the minimum lattice mismatch in the GaN/Al2O3 system occurs so that the (0001)//(0001), [010 1]//[1210] GaN/Al2O3 interface is formed and the lattice constant of a = 5.524 at room temperature as shown in Fig. 5.1[She 02]. Bulk Si crystal has a diamond structure with lattice constant of a = 5.43 at room temperature. The Si (111) surface, however,

PAGE 113

96 presents an equivalent hexagonal surface with lattice parameter a = 3.84 as shown in Fig. 5.2 [Shu00]. Figure 5-1. Schematic for the de position of the ( 0001)//(0001), [010 1]//[ 10 2 1] GaN/Al2O3 system. Figure 5-2. Planes of Si (111) substrate. 5.1.1.1. Sapphire (c-Al2O3 (0001)) Sapphire is the most widely used substr ate for the epitaxial growth of InN. Relatively large area, good quali ty crystals of sapphire are commercially available at a reasonable cost. Sapphire is transparent and stable at high temperature. High quality epitaxial InN films can be grown easily on sapphire substrates by popular growth methods such as MOVPE and MBE. Sapphire, ho wever, has a large lattice mismatch of 25.7 % with InN. This large lattice mismatch and thermal expansion coefficient difference can result in an extr emely high density of structural defects. Nevertheless, this

PAGE 114

97 disadvantage may be partially overcome with substrate pretreatment and buffer layers [Bhu03b]. For example, nitridation of th e sapphire substrate surface significantly improves the crystalline quality of InN as a result of the formation of an AlN interfacial layer, which reduces the lattice mismatch from 25.7 % for InN/c-Al2O3 to 13 % for InN/AlN. Additionally, the InN/AlN has a comparatively small thermal expansion coefficient mismatch of 8.1 %. This study aims to improve the crystallin ity of InN on sapphire substrates through use of different buffer layers, different bu ffer layer temperature, different growth temperature, post-growth annealing, and modification of the inlet tube. 5.1.1.2. Gallium Nitride (GaN/c-Al2O3 (0001)) Gallium nitride has a small lattice mismatch of 10.9 % with InN compared with sapphire (25.7 %) and AlN (13 %) substrates, although it is greater than that of silicon (7.9 %). In addition, gallium nitride substrat es generally lead to good coverage of InN, which is very difficult to achieve with silic on substrates. It was re ported that the highest mobility of InN (~ 700 cm2/V s) was obtained from MOVPE on GaN substrates [Yam99a]. Therefore, this study explores the feasibility of th e gallium nitride on sapphire substrate for the epitaxial growth. The thickness of the GaN layer on Al2O3 is 5 m. 5.1.1.3. Silicon (Si (111)) Silicon is an excellent candidate as a s ubstrate for the epitaxial growth of InN because it has a smaller lattice mismatch, 7. 9 % for InN (0001)/Si (111), compared with the commonly used insulating sapphire substrates (25.7 % for InN (0001)/c-Al2O3 (0001), and 10.9 % for InN (0001)/GaN (0001)). Si has been used as a substrate for the ep itaxial growth of InN, but the film quality has been very poor to date and there has b een no report of the FWHM of XRD principal

PAGE 115

98 reflection to judge the crystal quality. Poor su rface coverage of InN on Si is observed. It is believed that MOVPE growth of InN di rectly onto Si was unsuccessful because of formation of a SiNx interfacial layer. The Si substr ate surface is nitrided during growth even at a low growth temperature (~ 400 oC) [Bhu03b]. Introduction of TMI into the reactor before flowing NH3 to prevent amorphous SiNx formation has been tried with the result Although the previous work has not be en encouraging, the high quality and low cost of silicon make it a very attractive substrate. The possibility of integrating optoelectronic InN devices with Si electronic de vices is also attractiv e. This study aims to achieve high quality single crystalline InN growth on Si substrates by adjusting the growth conditions. 5.1.2. Substrate Preparation Procedure Sapphire and GaN/sapphire substrates were degreased in boiling solvent in the following sequence, tri-chroloethylene, acetone and then methanol for 5 min each. In the case of silicon, an etch ing step was added after the degreasing step in which the silicon substrate was etched in ammonium bifl uoride (95 %) for 2 min to obtain an oxidefree and H-terminated silicon substrate. After degreasing and etching, all substrates were rinsed in de-ionized water and drie d under nitrogen flow [Etc01]. 5.1.3. Metal-Organic Vapor Phase Epitaxy (MOVPE) Reactor A horizontal, cold-wall MOVPE reactor (Ni ppon Sanso) with a RF-induced heated susceptor was used in this study. Trimethylindium (TMI) (99.9995 %, Shipley) and ammonia (NH3) (99.9999 %, Solktronics) were used as precursors and nitrogen was used as a carrier gas. TMI is kept in the bubbler at 20 oC and is transferred into the reactor via carrier gas. Ammonia is carried directly into the reactor without carrier gas. The outer quartz wall is kept at 25 oC by circulating cooling water in the quartz jacket. The pressure

PAGE 116

99 of the reactor is controlled with a Baratron gauge. A schematic of the MOVPE reactor is shown in Fig. 5.3 and a more detail ed description is given elsewhere. Figure 5-3. Image and schematic of hori zontal, cold-wall MOVPE reactor system. 5.1.4. Growth Chemistry and Conditions for InN Growth The epitaxial growth of InN by MOVPE is a non-equilibrium growth process that relies on vapor transport of precursors to the surface of a heated substrate with subsequent reaction of typically group III alkyls and group V hydrides The chemicals are transported as a dilute vapor to the surf ace of the heated substrate where pyrolysis reactions occur [Jac64, Tra78, and Lar85]. The reaction chem istry for deposition of InN

PAGE 117

100 was reviewed in detail in Chapter 3. Th e overall reaction invol ves trimethylindium reacting with NH3 to form InN and the reaction is given by In(CH4)3 (g) NH3 (g) = InN (s) + 3 CH4 (g) + 3/2H2 (3-5) The N/In ratio (volum etric flow ratio of NH3 to TMI) is calculated by assuming ideal gas and solution behavior. Theref ore, the volumetric flow ratio of NH3 /TMI at standard temperature and pr essure (STP) is equal to the pressure ratio of NH3 /TMI at constant volume and temperatur e, according the equation: n V RT P 4 22 (5-3) Given that the total TMI bubbler pressure is kept at 500 Torr and the vapor pressure of TMI is given by )) ( / 3204 ( 98 1010 ) (K T TMITorr P (5-4) the volumetric flow of TMI is calculated by ) 500 ( ) ( ) (2Torr P Torr P V sccm Vtotal TMI N TMI (5-5) where 2NVrepresents the volumetric flow rate of nitrogen introduced into the TMI bubbler. The N/In ratio is calculated by TMI NHV V In N3 (5-6) The pressure of MOVPE reactor was 100 To rr during the growth and two different buffer layers of GaN and InN were also studied to check which buffer layer gave better structural quality InN. The range of growth conditions st udied for depositing InN is summarized in Table 5-2. When the flow rate of TMI was kept at 0.26 sccm and the N/In ratio was varied from 3000 to 15,000 in the first growth condition set, indium droplets formed at the surface. To prevent indium dr oplet formation, a low flow rate of TMI (0.03~0.08 sccm) or a high N/In ratio (20, 000~50,000) was used in the second growth condition set. In the second growth condition set, the narrower growth temperature range 530 to 570 C was selected based on the results on the temperature influence on the InN

PAGE 118

101 structural quality obtai ned from the first growth conditio n set, which showed that the optimal growth temperature was around 550 C Table 5-2. Range of growth conditions examined for growth of InN. TMI Flow Rate (sccm) NH3 Flow Rate (sccm) N/In Growth Temperature (oC) 0.26 800-4000, 3000-15,000 450-750 0.03-0.08 1600 20,000-50,000 530-570 The growth sequence for each substrate is also shown in Fig. 5.4. For Al2O3 (0001), it is generally accepted that nitridation is re quired to obtain high quality InN by acting as a compliant layer. This effect will be discusse d later. For Si (111), nitridation should be avoided because SiNx leads to polycrystalline InN. For GaN/Al2O3 (0001), the effect of nitridation will be discussed later. Figure 5-4. Indium Nitride (InN) growth sequence for each of the three substrates.

PAGE 119

102 5.1.5. Indium Nitride (InN) Growth and Optimization 5.1.5.1. Influence of Growth Temperature Among all the factors, the growth temp erature has the largest impact on the epitaxial film quality in terms of the growth habit (single crystalline or polycrystalline) and the structural quality. The growth habit was determined by XRD -2 scan (XRD Philips APD 3720) and the structural quality was judged by evaluating the FWHM of Xray Rocking Curve (XRC) (Philips MRD X'Pert System). Substrate Studies For Al2O3 (0001), a low temperature GaN buffer layer (LT-GaN) grown at TLT-GaN = 560 oC, was employed after the nitridation at 850 oC for 15 min. The TMI flow rate was fixed at 0.26 sccm, NH3 at 800 sccm, and the N/ In ratio was 3000. In this growth condition, indium droplets (met al) formed on the surface in addition to InN (0002) as shown in the XRD spectrum in Fig. 5.5. Figure 5-5. X-ray Diffraction (XRD) -2 scans for InN/LT-GaN on (a) Al2O3 (0001) at N/In = 3000, T = 450, 550, 650, and 750 oC.

PAGE 120

103 With the reference to the peak position of indium droplets that solidified upon cooling, peaks are anticipated at 36.3 corresp onding to In (002), 39.2 In (110), 54.5 In (112), 67.0 In (103), and 69.1 In (202). As shown in Figure 5-5, it is clear that In formed. Indium droplet formation was found to occur when the N/In ratio was low, as will be confirmed by the results of subsequent th e study in which the N/In ratio was varied. Wet etching was used to remove the indium with a 15 % HCl solution. It is noted that the In N (10-11) peak at 33.1o is close to the In (101) reflection at 32.9o, thus the two peaks are likely to overlap. After the indium droplets were removed by etching with HCl, the underl ying InN film was characte rized again by taking a XRD -2 scan. For Al2O3 (0001), single crysta l InN (0002) film was obtained and the peak intensity of InN (0002) was the greatest at T = 550 oC. However, InN (10-11) occurred at T = 650 oC, which indicates that the growth di rection of InN depends on the growth temperature. A very broad and low intensity peak for InN was observed when growth was at T = 450 oC, indicative of poor quality InN. No growth of InN was observed at T = 750 oC (Fig. 5.6). It is generally accepted that In N can not be grown at a growth temperature above 700 oC due to the InN thermal decomposition, nor below a growth temperature of 400 oC due to the low decomposition efficiency of NH3. From these results, it was found that the optimum growth condition is 550 oC.

PAGE 121

104 Figure 5-6. X-ray Diffraction (XRD) -2 scan for InN/LT-GaN on (a) Al2O3 (0001) at N/In = 3000, T = 450, 550, 650, and 750 oC. Pure In was removed by etching with HCl. Growth Temperature The growth temperature effect was studied over a more limited set of temperatures: 530, 550, and 570 oC to fine tune the optimi zed growth temperature. A relatively small flow rate of TMI in the range 13 to 44 sccm, and high flow rate of NH3 (1600 sccm), gave a high N/In ratio of 50,000 to prevent the indium droplet formation during the growth. As expected single crystalline InN was grown without indium droplet formation at this growth condition. For Al2O3 (0001), the intensity of the peak of the InN (0002) reflection was lower at growth temper atures 530 and 570 oC while the InN (10-11) reflection was evident at T = 570 oC. Therefore, it is appears that the optimum growth temperature of InN film is in the vicinity of 550 oC on Al2O3 (0001) with a LT-GaN buffer layer.

PAGE 122

105 Figure 5-7. X-ray Diffraction (XRD) -2 scan for InN/LT-GaN on Al2O3 (0001) at N/In = 50,000, T = 530, 550, and 570 oC. The growth temperature of LT-InN buffer la yer was also examined in the range 500 to 550 oC. The result is an optim um growth temperature for the InN/LT-InN of 530 oC. Polycrystalline InN with the appearance of both InN (10-11) and InN (0002) reflections occurred at 500 and 550 oC, while a single reflection, In N (0002), was the strongest at 530 oC (see Fig. 5.8). Figure 5-8. X-ray Diffraction (XRD) -2 scan for InN/LT-InN on Al2O3 (0001) at N/In = 50,000 and T = 500, 530, and 550 oC.

PAGE 123

106 The FWHM of the XRC (X-ray rocking curve) for InN on Al2O3 (0001) at N/In = 50,000 and T = 530 oC was 4860 arcsec (see Fig. 5.9). Figure 5-9. Full Width Half Maximum (FWHM) of XRC for InN/LT-InN on Al2O3 (0001) at N/In = 50,000 and T = 500, 530, and 550 oC. A similar set of runs (T = 500, 530, and 550 oC, and N/In = 50,000 with LT-InN buffer layer) was conducted using GaN/ Al2O3 (0001) substrates. Again polycrystalline InN was observed at the 500 and 550 oC growth temperatures, while single crystalline InN (0002) appeared at 530 oC (see Fig. 5.10). Figure 5-10. X-ray Diffraction (XRD) -2 scan for InN/LT-InN on GaN/Al2O3 (0001) at N/In = 50,000, T = 500, 530, and 550 oC.

PAGE 124

107 The FWHM of the XRCs for the InN film grown at T = 530 and 550 oC are shown in Fig. 5.11, with the best crystallinity obtained at T = 530 oC with a FWHM of the XRC of 1039 arcsec. Therefore, the optimum grow th temperature of InN/LT-InN is near 530 oC. Figure 5-11. Full Width Half Maximu m (FWHM) of XRC for InN/LT-InN on GaN/Al2O3 (0001) at N/In = 50,000, T = 500, 530, and 550 oC. Finally for the third substrate examined, Si (111), InN was grown at T = 500, 530, 550, and 570 oC, and N/In = 50,000 with two buffe r layers: LT-GaN and LT-InN. A weak InN (0002) reflection appeared in the fi lms grown at all growth temperatures (see Fig. 5.12 and 5.13). With the LT-GaN bu ffer layer, the strongest peak of single crystalline InN (0002) occurred at T = 530 oC, while polycrystalline InN (InN (10-11) and InN (0002) reflections) occurred at T = 550 oC (see Fig. 5. 12). For a LT-InN buffer layer on Si, a strong single peak of was re corded, InN (0002), also at T = 530 oC (Fig. 5. 13). The peak intensity of InN grown on Si substr ate was smaller compared to those grown on either Al2O3 or GaN/Al2O3. This is believed to be due to difficulty in wetting InN on Si. From these results, the optimum grow th temperature is thought to be 530 oC on Si (111) with both LT-GaN and LT-InN.

PAGE 125

108 Figure 5-12. X-ray Diffraction (XRD) -2 scan for InN/LT-GaN on Si (111), at N/In = 50,000, T = 500, 530, 550, and 570 oC. Figure 5-13. X-ray Diffraction (XRD) -2 scan for InN/LTInN on Si (111), at N/In = 50,000, T = 500, 530, 550, and 570 oC. Growth Rate Studies The growth rate was determined as a function of the N/In ratio on Al2O3 (0001). In these studies In N was grown for 1 hr at select ed growth temperatures in the range 450 to 650 oC (N/In = 3000) and 500 to 550 oC (N/In = 50,000) to determine if the growth is chemical reaction-limited or transport-limited. The thickness was measured

PAGE 126

109 on cross-sections by SEM (FEG-SEM JEOL JSM 6335F) and the results are displayed in Figures 5.14 and 5.15. As shown in Fig. 5.14, the temperature dependence changed depending on the kind of substrate, presumably because of different texture of the InN. Figure 514 (a) and (b) show that the growth rate remained unchanged in the range 550 to 650 oC at N/In = 3000 and in the range 530 to 570 oC at N/In = 50,000 for Al2O3 (0001), which is expected for a mass transfer limited growth conditi on. For N/In = 3000 run, the low grow th rate (T = 450 oC) was a result of the lower efficiency of NH3 decomposition at the relatively low temperature (Fig. 5.14 (a)). For the N/In = 50,000 condition on Al2O3 (0001) the growth rate remain s unchanged in the range 530 to 570 oC (Fig. 5.14 (b)), and thus mass transport-limited. The same results is seen for GaN/Al2O3 (0001) and Si (111) (Fig. 5. 14 (c) and (d)). Furthermore the growth rate is independent of the substrate for the last 3 conditions. When the growth rate was compared between the N/In ratios of 3000 and 50,000, the rate at N/In = 3000 is higher than that at N/In = 50,000 because of the increased flow rate of TMI, the limiting reagent, from 0.03 to 0.26 sccm. Figure 5-14. Growth rate of InN on va rious substrates (a) InN/LT-GaN on Al2O3 (0001) at N/In = 3000, (b) for InN/LT-GaN on Al2O3 (0001) at N/In = 50,000, (c) InN/LT-InN on GaN/Al2O3 (0001) at N/In = 50,0 00, and (d) InN/LT-InN on Si (111) at N/In = 50,000

PAGE 127

110 Figure 5-15. Cross-sectional SEM micrographs of InN for 60 min growth at 530, 550, and 570 oC, and N/In = 50,000 with LT-GaN buffer. Table 5-3. Optimum growth temperature of InN on LT-GaN and LT -InN buffer layers on various substrates. Buffer layer Al2O3 (0001) GaN/ Al2O3 (0001) Si (111) LT-GaN T = 550 oC T = 530 oC LT-InN T = 530 oC T = 530 oC T = 530 oC In summary, the optimum growth temperature of InN was near 550 oC for LT-GaN and 530 oC for the LT-InN buffer layer on Al2O3 (0001), and 530 oC for both GaN/Al2O3 (0001) and Si (111). The grow th of InN was not chemical reaction-limited but mass transport-limited. The results for the optimiz ed growth temperature are summarized in Table 5-3. 5.1.5.2. Influence of Substrate Nitridation The nitridation of Al2O3 is an important step to enhance the quality of InN film. For Al2O3 (0001), it has been reported that the nitridation treatmen t results in the formation of an AlN or amorphous AlOxN1-x layer on the Al2O3 substrate, which was

PAGE 128

111 confirmed by several scient ists with XPS and EDS [Bry92a, Yam94b, Uch96, Pan99, Tsu99]. The results presented below are ba sed on growth of InN on sapphire prepared with the AlOxN1-x layer by the procedure previously described. The first run compared the InN quality w ith and without using nitridation of the sapphire on an InN/LT-InN (TLT-InN = 450 oC)/Al2O3 (0001) substrate with the conditions N/In = 50,000 and TLT-InN = 450 oC. The result of a XRD -2 scan showed that the InN (0002) is single crystal us ing nitridation (T = 850 oC in NH3 for 15 min), while without the nitridation, both InN (0002) and InN (1011) were evident (Fig. 5.16). For InN/LTInN GaN/Al2O3 (0001) at N/In = 50,000, T = 530 oC, and TLT-InN = 450, 500 oC Figure 5-16. X-ray Diffraction (XRD) -2 scan for InN/LT-InN (TLT-InN = 450 oC) on Al2O3 (0001) at N/In = 50,000, and TLT-InN = 450 oC without and with nitridation. The presence of the nitrided Al2O3 layer is believed to promote wetting of the successive InN over layer and therefore to improve the film quality. Th e nitridation of the Al2O3 surface significantly improves the crysta lline quality of InN as a result of the formation of AlN or AlOxN1-x layer through the reaction of nitrogen at the Al2O3 surface as discussed above. The marked improvement of InN film quality by the formation of this layer reduces the mismatch between the sapphire and III nitride layer. For example, AlN

PAGE 129

112 has the same lattice structure as InN, and th e lattice mismatch is reduced from 25.7 % for InN/Al2O3 to 13.7 % for InN/AlN. Another set of runs were made using LT-InN GaN/Al2O3 (0001) with and without nitridation of the GaN buffer (nitridation performed at 850 oC for 15 min in NH3). The growth of InN was performed at T = 530 oC, while two buffer layer temperatures were compared; TLT-InN = 450 and 500 oC at N/In = 50,000. The subsequent XRD patterns shown in Fig. 5.17 reveal that nitridation ag ain had an impact on the crystallinity. In contrast to previous results, the nitridation of sapphire produced pol ycrystalline InN with the InN (10-11) peak increasing in intensit y and the peak of InN (0002) broader. Figure 5-17. X-ray Diffraction (XRD) -2 scan for InN/LT-InN GaN/Al2O3 (0001) at N/In = 50,000, T = 530 oC, and TLT-InN = 450, 500 oC with and without nitridation.

PAGE 130

113 A final set of runs was performed with Si ( 111) with and without ni tridation at T = 850 oC for 15 min in NH3 ambient. For this study, the gr owth of InN was performed 530 oC using a TLT-InN = 450 or 500 oC, and N/In = 50,000. Although good results were note expected, for completeness this experiment was performed. The film with the nitridation treatment showed an InN (0002) peak while the one without ni tridation did not show this peak (see Fig. 5.18). It is known that th e nitridation of Si gives amorphous SiNx formation, which was reported to severely degrade the film quality [Yan02c]. However, the result of this study indicates that the nitridation process improved the structural properties of InN. Figure 5-18. X-ray Diffraction (XRD) -2 scan for (a) InN/LT-InN on Si (111), at N/In = 50,000, T = 530 oC, and TLT-InN = 450, 500 oC with the nitridation and without the nitridation. During the nitridation treatment in this system, the SiOxN formation on Si substrate was thought to exist on the surf ace of Si substrate as show n by the analysis using ESCA by a previous student in the group [Mas01]. SiOxN formation on Si (111), instead of SiNx formation, is believed to lead to the grow th of the single crystalline InN (Fig. 5.19)

PAGE 131

114 [Mas01]. To understand the initial stages of InN growth on Si, NH3 treatment on bare silicon was studied. Figures 5.19 shows ESCA spectra of the Si 2p3 peak for bare silicon treated with and without NH3 at T = 850 oC. The bulk Si-Si bonds have a binding energy at 100.1 eV. This peak shows a doublet splitting of the 2p subshell into 2p3/2 and 2p1/2 bands with an intensity ratio of 1:2. Emissi on of an electron with up or down spin from a p-type orbital creates a photoelectron w ith two possible energy levels. These two emission levels separated by 0.6 eV create an asymmetry in th e overall Si peak. Strained Si-Si bonds near the Si/SiOx interface are distorted from their typical tetragonal bonding configuration. This compressi ve distortion results in a silicon bonding peak shifted to 102.2 eV. A strong peak obser ved at 104.1 eV was assigned to the Si-O bond. For samples treated in ammonia, a strong peak from the Si-N bond was observed at 103.3 eV. By integrating the area of each peak, an es timation of the bonding configuration of the surface atoms is possible. Table 5-4 showed that nitrogen incorporates into the SiOx film for samples annealed in NH3 [Mas01]. Figure 5-19. Electron Spectroscopy of Chemi cal Analysis (ESCA) spectra of Si 2p3 peak for Si annealed at 850 oC in 1.0 slm N2 (a) with 100% NH3 at 1.0 slm (b) without NH3.

PAGE 132

115 Table 5-4. Comparison by ESCA of Si anneals. Temperature NH3 %Bulk Si Bonds %Si-O Bonds %Si-N Bonds High (850 oC) Yes 40.2 38.8 16.2 High (850 oC) No 64.6 31.5 Negligible In summary, it was confirmed that the nitridation of Al2O3 (0001) can improve the structural quality of single crystalline InN fi lm through the formation of an AlN layer, the nitridation of GaN/Al2O3 (0001) is not favorable for the growth of InN, and during the nitridation of Si (111), the SiOxN formation on Si ( 111), instead of SiNx formation, is led to the improved structural quality of InN. 5.1.5.3. Influence of N/In Ratio The N/In ratio is a key factor for InN grow th especially to prevent indium droplet formation and is also an important factor to influence the structural quality of InN. For the study of N/In effect on the indium droplet formation, the flow rate of TMI was varied from 0.03 to 0.08 sccm and the flow rate of NH3 was fixed at 1600 sccm to optimize the ratio of N/In, which leads to the growth of single crystalline InN wi thout indium droplet formation. From the results of XRD -2 (Fig. 5.20) scans and SEM images of the surface, the optimum ratio of N/In wa s found to be 50,000 for InN/LT-GaN on Al2O3 (0001). Figure 5-20. X-ray Diffraction (XRD) -2 scan at N/In=20,000, 30,000, and 50,000, T = 550 oC for InN/LT-GaN on Al2O3 (0001) at N/In of 50,000.

PAGE 133

116 Single crystalline InN ( 0002) was obtained at N/In = 50,000 and the indium droplet were observed at N/In = 20,000 and 30,000. The FWHM of XRC of InN was 14868 arcsec at N/In = 50,000 with LT-Ga N buffer layer (Fig. 5. 21). Figure 5-21. Full Width Ha lf Maximum (FWHM) of XRC for InN/LT-GaN on Al2O3 (0001) at N/In of 50,000. For GaN/ Al2O3 (0001), InN was grown at T = 530 oC, TLT-InN = 400 oC, and N/In = 30,000 and 50,000. At N/In = 30, 000, indium droplet formati on was observed but at N/In = 50,000 the single crystall ine InN (0002) was grown w ithout the indium droplet formation. From these results, the optimum N/ In ratio is for high values, e.g. 50,000 (Fig. 5.22). Figure 5-22. X-ray Diffraction (XRD) -2 scan for InN/LT-InN on GaN/Al2O3 (0001) at, T = 530 oC, TLT-InN = 400 oC, and N/In = 30,000 and 50,000.

PAGE 134

117 For Si (111), InN was grown at N/In ra tios of 20,000, 30,000, and 50,000, and T = 530 oC with a LT-GaN buffer layer. Indium dropl et formation occurred at N/In = 20,000 and 30,000, while the growth of single crys talline InN (0002) was achieved at N/In = 50,000, consistent with the results on other substrates (Fig. 5.23). Figure 5-23. X-ray Diffraction (XRD) -2 scan for InN/LT-GaN on Si (111) at N/In = 20,000, 30,000 and 50,000, T = 530 oC. The growth rate vs. N/In ratio was studied and the results are presented in Fig. 5.20. The N/In ratio was increased from 3000 to 15,000 by increasing th e flow rate of NH3 in the growth of InN/LT-GaN on Al2O3 (0001). After wet etching to remove the indium droplets, cross-section SEM images showed that the gr owth rate decreased with increasing amount of NH3 (i.e., increase flow rate of NH3). The growth rate vs. N/In ratio was same for each of the three substrates Al2O3 (0001), GaN/Al2O3 (0001), and Si (111) (see Fig. 5.24). The growth rate of InN was 0.27 m/hr at N/In = 6000, 0.21 m/hr at N/In = 9000, 0.17 m/hr at N/In = 12,000, and 0.13 m/hr at N/In = 15,000 (Fig. 5.24). This phenomenon can be also explained by thermodynamic reasoning. When the N/In ratio increases, the amount of H2 also increases due to the decomposition of NH3. The

PAGE 135

118 increase of H2 drives the reaction to the left, namely the etching process (Eq. (3-5)). In terms of the growth rate vs. N/In ratio, the results obtained for Al2O3 substrate, were the same as the results for GaN/Al2O3 and Si substrate because the growth rate is independent of the kind of s ubstrate but is dependent on the mass flow rate of the source materials. Figure 5-24. Growth rate vs. N/ In ratio for InN/LT-GaN on Al2O3 (0001), GaN/Al2O3 (0001), and Si (111) at N/In = 6000, 9000, 12,000, and 15,000 with T = 550 oC, TMI = 0.26 sccm and NH3 = 1600-4000 sccm. In summary, the growth rate decreases with increasing N/In ratio. However, indium droplet formation can be prev ented during the grow th of InN by growi ng at high N/In ratio. Therefore, the relatively large ratio of N/In is require d to avoid the indium droplets formation, although accompanied by a decrease of growth rate. It is also found that the growth rate is independent of the kind of substrate. 5.1.5.4. Influence of Buffer Layer and Morphological Study The growth temperature of th e buffer layer is also a significant factor because the buffer layer usually fails to act as a nuclea tion layer and stress-relie ving (compliant) layer in some temperature regions. Buffer layer can be used to improve the crystallinity of an InN film by first providing the nucleation site and thus leading to lateral growth and

PAGE 136

119 second reducing the lattice mismatch between the substrate and epitaxial film as a compliant layer. Therefore, it is also essential to find the optimized temperature for the InN buffer layer. A LT-InN buffer layer was used for this study because the optimized growth temperature of LT-Ga N was already found to be 560 oC by a previous student in our group [San04]. Without buffer layers, th e growth of single cr ystalline InN (0002) was difficult for Al2O3 (0001), and GaN/Al2O3 (0001), and Si (111). Therefore, the optimization of buffer layer growth temperature is required. For Al2O3 (0001), polycrystalline InN with peak s of InN (10-11) and/or InN (1120) appeared at TLT-InN = 400 and TLT-InN = 500 oC (Fig. 5.25). The InN buffer layer is thought to fail to relieve the st ress in the overgrown InN film because the InN with (1011) surface structure found is diffe rent from the (0001) found on the Al2O3. The optimized growth temperature of InN buffer layer was found to be 450 oC because the single crystalline InN (0002) was achieved at TLT-InN = 450 oC, thus pointing to the temperature sensitivity of the bu ffer layer growth temperature. Figure 5-25. X-ray Diffraction (XRD) -2 scan for InN/LT-InN on Al2O3 (0001) at N/In = 50,000, T = 530 oC, and TLT-InN = 400, 450, and 500 oC.

PAGE 137

120 For GaN/Al2O3 (0001) starting substrates, the grow th of InN was performed at T = 530 oC and N/In = 50,000, with several buffer layer growth temperatures: TLT-InN = 350, 400, 450, and 500 oC. Single crystalline InN (0002) was obtained at TLT-InN = 350 and 400 oC while polycrystalline InN with the pres ence of InN (10-11) and InN (0002) was found for TLT-InN = 450 and 500 oC (Fig. 5.26). When the crystalline quality of InN grown at TLT-InN = 350 and 400 oC were compared, the InN grown at TLT-InN = 400 oC showed smaller FWHM of XRC (1039 arcsec) than that (6386 arcsec) of InN grown at TLT-InN = 350 oC (Fig. 5.27). Figure 5-26. X-ray Diffraction (XRD) -2 scan for InN/LT-InN on GaN/Al2O3 (0001) at T = 530 oC, TLT-InN = 350, 400, 450, and 500 oC, and N/In = 50,000. Figure 5-27. Full Width Half Maximu m (FWHM) of XRC for InN/LT-InN on GaN/Al2O3 (0001) at T = 530 oC, TLT-InN = 350, 400, 450, and 500 oC, and N/In = 50,000.

PAGE 138

121 Finally for Si (111) starting substrates, the growth of InN was performed at N/In = 50,000 and530 oC, and three buffer layer temperatures: TLT-InN = 400, 450, and 500 oC. Single crystalline InN ( 0002) was obtained at TLT-InN = 400, 450, and 500 oC, but the intensity of InN (0002) was the strongest at TLT-InN = 450 oC (Fig. 5.28). The scan in Figure 529 shows that the peak intensit y of InN (0002) does not increase when the growth time for LT-InN buffer layer is longer than t = 15 min. Figure 5-28. X-ray Diffraction (XRD) -2 scan for InN/LT-InN on Si (111) at N/In = 50,000, T = 530 oC, and TLT-InN = 400, 450, and 500 oC. Figure 5-29. X-ray Diffraction (XRD) -2 scan for InN/LT-InN on Si (111) at N/In = 50,000, T = 530 oC, TLT-InN = 450 oC, and t = 5, 15, and 30 min.

PAGE 139

122 In summary, the optimum growth temper ature of LT-InN buffer layer was found to be 450 oC for Al2O3 (0001), 400 oC for GaN/Al2O3 (0001), and 450 oC for Si (111) to obtain single crystalline InN without indium droplet form ation. These results were summarized in Table 5-5. Table 5-5. Optimum growth temperatur e of LT-InN buffer layer depending on Al2O3 (0001). Buffer layer Al2O3 (0001) GaN/Al2O3 (0001) Si (111) LT-InN TLT-InN = 450 oC TLT-InN = 400 oC TLT-InN = 450 oC The choice of low temperature buffer laye r is also an important factor that influences the crystallin ity of InN. LT-GaN (560 oC) and LT-InN (450 oC) buffer layer were used to compare the effect of both bu ffer layers on the struct ural quality of InN grown on Al2O3 (0001). The LT-InN buffer layer enhanc ed the crystallinity of InN film more significantly than LT-GaN buffer layer as shown in Fig. 5.30, where the peak intensity was much stronger when using the LT-InN rather th an the LT-GaN. The FWHM (4860 arcsec) for LT-InN was much narrower than that (FWHM=14868 arcsec) of LT-GaN (Fig. 5.31). Figure 5-30. X-ray Diffraction (XRD) -2 scan for InN/LT-InN (TLT-InN = 450 oC) and LT-GaN (TLT-GaN = 560 oC) on Al2O3 (0001) at N/In = 50,000, T = 530 oC (LT-InN buffer) and T = 550 oC (LT-GaN buffer).

PAGE 140

123 Figure 5-31. Full Width Half Maximum (F WHM) of XRC for InN/LT-InN (TLT-InN = 450 oC) and LT-GaN (TLT-GaN = 560 oC) on Al2O3 (0001) at N/In = 50,000, T = 530 oC (LT-InN buffer) and T = 550 oC (LT-GaN buffer). Surface Morphology The surface morphology was also considered using Atomic Force Microscopy (AFM). In this study, InN was grown on Al2O3 (0001) at T = 530 oC using a LT-InN buffer layer (TLT-InN = 450 oC) and at T = 550 oC for LT-GaN buffer layer (TLTGaN = 560 oC), respectively, where the temperatures were selected according to Table 5-5. When the roughness of InN films obtained using the LT-InN and LT-GaN buffer layers was compared, the RMS roughness of InN film (4.2 nm) with LT-InN buffer layer was smaller than that (18.1 nm) with LT-GaN buffer layer (Fig. 5.32). The roughness of as-grown LT-InN buffer layer and LT-GaN buffer layer was compared to study the relation between the r oughness of InN film and that of as-grown buffer layer. The RMS roughness of as-grown LT-InN buffer layer (1.9 nm) was smaller than that (10.2 nm) of as-grown LT-GaN (Fig. 5.32).

PAGE 141

124 Figure 5-32. Root Mean S quare (RMS) roughness by ATM for (a) InN/LT-InN (T = 530 oC, TLT-InN = 450 oC), (b) InN/LT-GaN (T = 550 oC, TLT-GaN = 560 oC), (c) as-grown LT-InN (450 oC), and (d) as-grown LT-GaN (560 oC) on Al2O3 (0001) at N/In=50,000. Table 5-6. Root Mean Square (RMS) roughness for as-grown buffer layers and InN films. Material RMS (nm) Material RMS (nm) As-grown LT-InN 1.9 As-grown LT-GaN 10.2 InN/LT-InN 4.2 InN/LT-GaN 18.1 When the quality of the grown InN using either a LT-GaN or a LT-InN buffer layer on Si (111) was compared, the LT-InN buffer layer gave a higher intensity InN (0002) peak than that of LT-GaN (Fig. 5.33). This indicated that the LT-InN buffer layer is more favorable than LT-GaN buffer layer for the growth of InN.

PAGE 142

125 Figure 5-33. X-ray Diffraction (XRD) -2 scan for InN/LT-InN and InN/LT-GaN on Si (111) at N/In = 50,000, T = 530 oC, TLT-InN = 450 oC, and TLT-InN = 560 oC. In summary, it is conclude d that the roughness of the as -grown buffer layer affects the roughness of the InN film and the LT-InN buffer layer gives better surface morphology of InN film than the LT-GaN buffer layer (Table 5-6). Therefore, the LTInN buffer layer appears to be a more suita ble substrate than LT-GaN to improve the crystalline quality and surface morphology of InN. 5.1.5.5. Influence of Pressure The growth pressure effect on the crysta llinity of InN was st udied. To understand the growth pressure effect on InN growth two different pressures of 60 and 100 Torr were used during the growth of the LT-InN buffer layer. The growth pressure for the growth of InN film was 100 Torr after the gr owth of LT-InN buffer layer. The growth temperature was 530 oC and the InN buffer layer te mperatures were 450 and 500 oC, at N/In ratio = 50,000 on GaN/Al2O3 (0001). The a pplication of PLT-InN = 60 Torr in the growth of LT-InN buffer layer at TLT-InN = 450 and 500 oC made the peak intensity of InN (10-11) stronger (Fig. 5.34).

PAGE 143

126 Figure 5-34. X-ray Diffraction (XRD) -2 scan for InN/LT-InN on GaN/Al2O3 (0001) at N/In = 50,000, TLT-InN = 450 and 500 oC and T = 530 oC with the different growth pressure of LT-InN. For Si (111), the growth temperature was 530 oC, InN buffer layer temperatures were 450 and 500 oC, at N/In ratio = 50,000. The application of PLT-InN = 60 Torr in the growth of LT-InN buffer layer at TLT-InN = 450 and 500 oC made the peak intensity of InN (10-11) stronger (Fig. 5.35). Figure 5-35. X-ray Diffraction (XRD) -2 scan for InN/LT-InN on Si (111) at N/In = 50,000, TLT-InN = 450 and 500 oC and T = 530 oC with two different growth pressures for LT-InN.

PAGE 144

127 From these results, it was found that the low pressure (60 Torr) was not preferable for the growth of single crystalline InN, co mpared to the relativel y high pressure (100 Torr). High pressure at constant mass flow is usually thought to be favorable for the growth of InN since the high pressure can enhance the NH3 decomposition as a result of reduced flow velocity of the reactant gase s and thus, can suppress nitrogen evaporation during the growth. 5.1.5.6. Optical and Electrical Properties The band gap energy of InN has been known to be 1.89 eV for a long time [Tan86a]. However, the band gap energy of In N has been recently reported to be around 0.7 eV. Most recently the reported band gap energy of InN is in the range 0.7 to 1.0 eV [Dav02a, Dav02b, Dav02c, Wu02, Tat02, Hor02, Sai02, and Miy02]. The PL data of our InN films showed that the band gap energy is 0.84 eV for Al2O3 (0001), 0.94 eV for GaN/Al2O3 (0001), and 1.07 eV for Si (111), which are in good agreement with the recently reported data (Fig. 5.36). The band gap energy of InN grown on GaN/Al2O3 (0001) is 0.94 eV, which is higher than 0.84 eV for InN/Al2O3 (0001). This is related to the interfacial layer that exists between InN film and GaN substrate. The recently reported best electron density of InN is ~ 5.8 1018 cm3 and best mobility is ~ 900 cm2/Vs [Yam04b]. The carrier concentrations and mobilities of the several InN films grown with different conditions are shown in Fig. 5.37 and the growth conditions are summari zed in Table 5-7.

PAGE 145

128 Figure 5-36. Photoluminescence for (a) InN grown on Al2O3 (0001) at T = 530 oC, TLT-InN = 500 oC, (b) InN grown on GaN/Al2O3 (0001) at T = 530 oC, TLT-InN = 400 oC, (c) InN grown on Si (111) substrate at TLT-GaN = 560 oC, T = 550 oC and N/In = 50,000. Figure 5-37. Carrier concentrations and mob ilities of InN films grown with different growth conditions at different char acterization temperature using Hall measurement.

PAGE 146

129 Table 5-7. Growth conditions for InN. Sample Growth condition 1 Al2O3-LTGaN-InN, N/In = 50K 2 Al2O3-LTInN-InN, N/In = 50K 3 Si(111)-LTGaN-InN, N/In = 6K, after etching 4 GaN/Al2O3-LTInN-InN, N/In = 50K, vertical inlet tube 5 Al2O3-LTInN-InN,N/In = 50K, vertical inlet tube The best electrical propertie s of our InN film were obtained from InN on Si (111) and the net electron carrie r concentration is 7.0 1018 cm3 at T = 266 K and 4.5 1017 cm3 at T = 41 K, and the mobility is 623 cm2/Vs at T = 266 K and 9288 cm2/Vs at T = 45 K (Fig. 5.37 and Fig. 5.38). Figure 5-38. Carrier concentration and mobility of InN on Si (111) at different characterization temperature using Hall measurement. 5.1.5.7. Summary For Al2O3 (0001), the optimized growth condition of InN was found in terms of the growth temperature, N/In ratio, buffer layer growth temperature, substrate nitridation. The results are summarized in Table 5-8.

PAGE 147

130 Table 5-8. Optimum growth condition of InN for Al2O3 (0001), GaN/Al2O3 (0001), Si (111). Substrate Buffer layer Growth temperature Buffer layer temperature N/In ratio LT-InN T = 530 oC TLT-InN = 450 oC 50,000 Al2O3 (0001) LT-GaN T = 550 oC TLT-GaN = 560 oC 50,000 GaN/Al2O3 (0001) LT-InN T = 530 oC TLT-InN = 400 oC 50,000 Si (111) LT-InN T = 530 oC TLT-InN = 450 oC 50,000 The mirror-like surface of InN was obtained using a LT-InN buffer (RMS roughness = 4.2 nm). The best quality single crystalline InN grown on GaN/Al2O3 (0001) showed that the FWHM of XRC is 1039 arcsec with LT-InN buffer layer. The growth of InN was found to be the transport-limited under most conditions. The growth rate decreased with N/In ratio due to the in creased amount of H2 produced by thermal decomposition of NH3. The band gap energy of InN films on Al2O3 (0001) is 0.84 eV. The carrier concentration of 7.0 1018 cm3 at T = 266 K and 4.5 1017 cm3 at T = 41 K and the mobility of 623 cm2/Vs at T = 266 K and 9288 cm2/Vs at T = 45 K were obtained on InN grown on Si (111). 5.1.6. Indium Nitride (InN ) Droplet Formation Indium droplets normally occurred during the growth of InN at low N/In ratio. In this part, indium droplet formation was studi ed in more detail through XRD, Scanning Electron Microscopy (SEM), Ener gy Dispersive Microscopy (EDS), and Auger Electron Spectroscopy (AES) (AES Perkin-Elmer PHI 66 0 Scanning Auger Multiprobe). For this study, InN was grown on Al2O3 (0001) and Si (111) at T = 550 oC, TLT-GaN = 560 oC, and N/In = 3000 to 50,000 with LT-GaN buffer layer.

PAGE 148

131 At N/In = 3000, a high density of indium droplets formed at the surface. As the N/In ratio increases fr om 3000 to 50,000, the indium droplet density decreases and finally disappears. From N/In = 20,000, the size of the indium drop let is reduced significantly. Finally, at N/In = 50,000, i ndium droplets completely disappeared (Fig. 5.39). These results show that an increase of the N/In ra tio effectively reduces and then eliminates indium droplet formation. At N/In = 6000, the indium and InN phase s were characterized by EDS (Fig. 5.39). For the indium phase, the strong intensity of indium peak was obtained from thick indium precipitates. The weak Al peak was also obtained from Al2O3 substrate due to the depth penetration of EDS to ~1 m. For the InN phase, a weaker peak intensity was obtained compared to that of indium phase and a stronger Al peak was obtained from the Al2O3 substrate and thin InN film (Fig. 5.40). Figure 5-39. Scanning Electron Microscopy (SEM) and EDS for the surface of InN/LTGaN on Al2O3 (0001) at N/In = 3000, 6000, 9000, 20,000, 30,000, and 50,000.

PAGE 149

132 The relation between the number of indium droplets per unit area and N/In ratio was studied in the range of N/In = 3000 to 30,000. The percen t of the indium droplets with small size was increased w ith the N/In ratio (Fig. 5.40 and Fig. 5.41). The number density of indium droplets increases again from N/In = 9000 to 20,000 because the big indium droplets break into ma ny small indium droplets (Fig. 5.39). The density of indium droplets continued to be decr eased as the N/In ratio ch anged from 20,000 to 30,000. These results indicated that the indium dr oplets formation can be reduced by increasing the N/In ratio. Figure 5-40. Number density of indium droplets vs. N/In ratio depending on different N/In, when InN was grown on Al2O3 (0001) at T = 550 oC and N/In = 3000, 6000, 9000. Figure 5-41. Percent (%) vs. indium droplet size depending on different N/In, when InN was grown on Al2O3 (0001) at T = 550 oC and N/In = 3000, 6000, 9000.

PAGE 150

133 The possibility that the residual indium can be etched in the diluted HCl solution (15 %), was studied using XRD and AES. Fo r this study, the growth of InN was performed on Si and Al2O3 substrates at N/In = 3000, and T = 450, 550, 650, and 750 oC. The results of XRD -2 scans of InN grown on Al2O3 (0001) and Si (111) showed that after HCl etching all peaks of indium droplets di sappeared on both subs trates (Fig. 5.42 to Fig. 5.45). Because the InN (10-11) peak is at 33.1o and In (101) peak at 32.9o, these two peaks overlapped before the HCl wet etching. Figure 5-42. X-ray Diffraction (XRD) -2 scans for InN/LT-GaN on Al2O3 (0001) at N/In = 3000, T = 450, 550, 650, and 750 oC before HCl wet etching. Figure 5-43. X-ray Diffraction (XRD) -2 scans for InN/LT-GaN on Al2O3 (0001) at N/In = 3000, T = 450, 550, 650, and 750 oC after HCl wet etching.

PAGE 151

134 Figure 5-44. X-ray Diffraction (XRD) -2 scan for InN/LT-GaN on Si (111) at N/In = 3000, T = 450, 550, 650, and 750 oC before HCl wet. Figure 5-45. X-ray Diffraction (XRD) -2 scan for InN/LT-GaN on Si (111) at N/In = 3000, T = 450, 550, 650, and 750 oC after HCl wet etching. AES characterization also showed that th e indium droplets could be removed in HCl solution (Fig. 5.46). Before the HCl wet etching, only the peak due to indium droplets was detected with the indium atom ic concentration 37.2 % and the peak of

PAGE 152

135 nitrogen was not detected. After the HCl wet etching, the indium atomic concentration was decreased from 37.2 to 18 % and the peak s of In and N were detected together, which came from the InN film obtained after the HCl wet etching. Figure 5-46. Characterization re sult by AES for In droplets fo rmed during the growth of InN on Al2O3 (0001) with LT-GaN buffer laye r before the HCl wet etching after the HCl wet etching at N/In = 3000. The indium atomic concentration obtained from InN is 18 % and the nitrogen atomic concentration obtained from InN film and LT-GaN buffer layer was 29.1 %. Through the characterization results of XR D, SEM, EDS and AES, it is concluded that the increase of N/In ratio can reduce i ndium droplet formation effectively during the growth of InN.

PAGE 153

136 5.1.7. Annealing Effect The crystalline quality of an epitaxial film can often be improved further by postgrowth annealing. The post-growth anneali ng is thought to cause the rearrangement of crystallites. The as-grown single crystal InN films consist of mosaic crystallites with slightly different orientations nearly along the (0002) direc tion. Many disloc ations with an edge component must be arranged at the boundary between the crystallites. This climbing motion (nonconservati ve motion) of the boundary dislocations is strongly temperature dependent and is accommodated onl y through a transfer of matter by atomic diffusion. Therefore, thermal annealing ma y cause climbing motion of dislocations [Guo94c]. The movement of boundaries and some of the small crystallites begin to rotate, so that the free energy of the system may be reduced. This behavior will result in an improvement in crystalline quality of a mosaic crystal and the crystalline quality of InN film. Therefore, the annealing effect was st udied for improving the crystalline quality of InN on Al2O3 (0001) and GaN/Al2O3 (0001). The annealing test was performed at T = 450 oC in N2 flow with different annealing time (0, 10, 30, 60, and 90 min). For InN on Al2O3 (0001), the average FWHM of XRC of InN film is 7488 arcsec before the annealing test and FWHM of X RC was reduced with a nnealing (Fig. 5.47). The instrument error for FWHM of XRC is 1.8 arcsec ( 0.0005). Above 60 min of annealing time, the FWHM of XRC remained unchanged. For GaN/Al2O3 (0001), the average FWHM of XRC of the InN film was 1779 arcsec before the annealing test and FWHM of XRC was reduced after annealing (F ig. 5.48). Above 30 min of annealing time, the FWHM of XRC remained same.

PAGE 154

137 Figure 5-47. Full Width Ha lf Maximum (FWHM) of XRC of InN/LT-InN (450 oC) on Al2O3 (0001) at T = 450 oC in N2 flow with different annealing time (0, 10, 30, 60 and 90 min). Figure 5-48. Full Width Ha lf Maximum (FWHM) of XRC of InN/LT-InN (400 oC) on GaN/Al2O3 (0001) at T = 450 oC in N2 flow with differe nt annealing time (0, 10, 30, and 60 min). Based on 30 min of the annealing time, the FWHM of XRC was reduced from 7488 arcsec to 7264 arcsec by 4 % for InN on Al2O3 (0001) and the FWHM of XRC was reduced from 1779 arcsec to 1388 arcsec by 22 % for InN on GaN/Al2O3 (0001). Therefore, it was concluded that the ann ealing effect was larger for InN on GaN/Al2O3 (0001) rather than for InN on Al2O3 (0001). This difference in the annealing efficiency for the two substrates was thought to come fr om the different degree of rearrangement of

PAGE 155

138 mosaic crystallite with differe nt orientation along the (0002) direction and the different degree of the misorientation along the (0002) direction. It is concluded that the annealing method can improve the crystalline quality of InN film and the most effective annealing was achieved within approximately 30 min. 5.2. Computational Fluid Dynamic Analysis of the Flow of NH3 and Proposed Inlet Tube Modification to Im prove Flow Pattern of NH3 Ammonia (NH3) is a widely used n itrogen precursor for In N growth. It is very difficult to produce active ni trogen due to the low decom position efficiency at the relatively low growth temperature of 530 oC. This causes poor crystallinity of InN and low growth rate. Therefore, th e modification of inlet tube was proposed to figure out the flow pattern of NH3 in the MOVPE reactor with the curre nt horizontal inlet tube and two modified inlet tubes. Thus, the approach to optimize the flow pattern through the inlet tube to induce the uniform flow and enhance the amount of NH3 on the substrate was studied. The inlet velocity inlet of NH3 is 0.94 m/s obtained when the flow rate (1.6 slm) of NH3 and the radius of inlet t ube is 0.3 cm (Eq. (5-7)). s cm cm s cminlet/ 94 ) 3 0 ( 1 60 16002 3 (5-7) It was assumed that the temperature in the inlet tube is 25 oC and the temperature in the reactor changes from 25 oC (at the wall of the reactor) to 530 oC (at the substrate). The density and viscosity of NH3 change depending on the temp erature (Eq. (5-8) and (59)), and the velocity also changes due to the change of the diameter in inlet tube and the reactor (Eq. (1)). The ope rating pressure is 100 Torr.

PAGE 156

139 For calculation of the density of NH3, it was assumed that NH3 follows the ideal gas law. The density of NH3 can be calculated as RT PM (5-8) where the unit of viscosity is micropoise (10-7 kg/ms) and the unit of temperature, Kelvin (Table 5-9). Table 5-9. Density and velocity of NH3 at reactor wall and substrate. Property Inlet tube (25oC) Reactor wall (25 oC, P = 100 Torr) Substrate (530 oC, P = 100 Torr) D 0.006 m 0.081 m 0.081 m 0.940 m/s 0.005 m/s 0.005 m/s 0.092 kg/m3 0.092 kg/m3 0.033 kg/m3 Reynolds number can be calculated as D Re (5-9) where D is the diameter of the reactor, the average velocity in the reactor of NH3 is the density of NH3, and is the viscosity of NH3. The Re in the inlet tube is 49.8 at T = 25 oC and Re in the reactor is 3.72 in the position at T = 25 oC (wall of rector) and 0.47 in the position at T = 550 oC (substrate), respectively (Eq. 5.9, Table 5-10). Table 5-10. Reynolds number (Re) calculated in the inlet tube and in the reactor depending on temperature. Inlet tube (T = 25 oC) Reactor Wall (T = 25 oC) Substrate in reactor (T = 550 oC) Re 49.83 3.72 0.47 Considering the Fanning friction factor Re 16 the values Re indicate laminar flow

PAGE 157

140 because the critical number is 2000, at wh ich the flow changes from laminar flow to turbulent flow. The thermal expansion coefficient ( ) of NH3 was calculated by T (5-10) substrate wall 2substrate wallT T T 2substrate wallT T T The thermal expansion coefficient of NH3 was calculated to be 0.0025 K-1 at the average temperature. The importance of buoyancy for ces in a mixed convection flow can be measured by the ratio of the Grashof and Reynolds numbers; 2 1 2 2) ( Re L T T g forvce inertia force buoyancy Gr (5-11) 2substrate wallT T T where g is the gravity acceleration and 1 2T T is the temperature difference between the substrate and the wall of the reactor and L is the length between two temperature zones (0.0393 m). When this number approaches or ex ceeds unity, strong buoyancy is expected. Conversely, if it is very small, buoyancy forces may be ignored in our simulation. For the operating condition of the reactor at P = 100 Torr, the ratio of 2Re Gris 19450 near the substrate (T = 530 oC). From this result, it was found that the e ffect of convective flow should be taken into account because the ratio of 2Re Gris much greater than unity. Therefore, our calculation includes the gravity term to ta ke into account the effect of the buoyancy

PAGE 158

141 forces. The schematic for three types of inlet tubes used in the Flue nt simulation is shown in Fig. 5.49. The first one is the currently used conventiona l horizontal inlet tube and the second and third ones are proposed inlet tube based on the results of the computational fluid dynamic analysis. Figure 5-49. Schematic for three types of in let tubes used for the Fluent simulation. The physical prop erties of NH3 such as the density, th ermal conductivity, viscosity and heat capacity of NH3 were obtained from the data set of Fluent. The flow of NH3 in the reactor using the current inlet tube is show n in Fig. 5.50. Figur e 5-51 shows the flow of NH3 1 mm above the substrate. These results demonstrated that most of the NH3 passed along the wall of the r eactor due to the structure of the inlet tube and the low density of NH3 and therefore the flow of NH3 was very small near the substrate and not uniform. This was thought to deteriorate the structural quality of InN and lead to low growth rate. The first modified inlet tube is the horizo ntally extended one with the outlet located at the front side of the end of the inlet tube (Fig. 5.52). The si mulation results also

PAGE 159

142 indicated that most of the NH3 moved up to the top of the reactor due to the low density and therefore most of the NH3 still did not reach the surface of the substrate. Figure 5-50. Flow of NH3 in the reactor with the current inlet tube. Figure 5-51. Flow of NH3 1 mm above the surface of substrate with the current inlet tube. Figure 5-52. Flow of NH3 in the reactor with the horizontally extended inlet tube.

PAGE 160

143 The second proposed inlet modi fication is the vert ical inlet tube as shown in Fig. 5.53. These results showed that more of the NH3 reached the surface of the substrate and a uniform flow could be obtained on the surface of substrate with this vertical inlet tube design (Fig. 5.54). Figure 5-53. Flow in the reactor with the vertical inlet tube. (c) Figure 5-54. Flow of NH3 1 mm above the surface of substrate with the vertical inlet tube. It was thought that fast flow rate of NH3 on the substrate can reduce the boundary layer and improve the diffusion of the NH3 into the substrate due to the decrease of the mass-transport boundary layer thickness, and fi nally leads to high growth rate according to Eq. (5-13). The MOVPE reactor is usua lly modeled by a boundary-l ayer flow over a

PAGE 161

144 flat plate using the empirical expression [Y an99]. The following explanation showed the application of the scal ing concept to the Navi er-Stokes equation. Equation Stokes Navier g V v P V V t V :2 (5-12) Where the second term on the left side corr esponds to momentum by convection and the second term on the right side corres ponds to the momentum by diffusion. When the scaling concept was used in Navier-Stokes equation, the velocity was scaled by V and the length is scaled by L. Then the velocity, the longitudinal length (x), and transverse length (y) is expressed as Eq. (5-13). g L V g P V P y y x L x V V V2 2 (5-13) where g P y x V , ,is the dimensionless velocity, di mensionless longitudinal length, dimensionless transverse length, and dimensionless gravity ac celeration, respectively. All terms in Eqn. (5-12) correspond to V2/L. Therefore, in the boundary layer, 2 2 V L V (5-14) Therefore, the boundary laye r thickness is given by 2 1 V L (5-15) where is the boundary layer thic kness for mass transport, is the dynamic viscosity, x is the axial distance, ) / ( is the kinematic viscosity, is density, and V is the free stream velocity. The uniform flow of NH3 is supposed to enhance th e structural quality of

PAGE 162

145 InN. Equation (5-15) indicates that the increa sed velocity leads to the reduced thickness of boundary layer and therefore increased diffusion into the f ilm. The growth rate can be increased. In summary, based on the simulation results using the Fluent software, it was proposed that the vertical inle t tube would deliver more NH3 flow and TMI to the substrate and thus improve the growth rate and perhaps structural quality of InN. 5.3. Inlet Tube Modification and Growth Results From the result of the simulation obtained wi th Fluent software, the possibility to improve the structural quality of InN film by using the vertical inlet tube was studied. Two types of inlet tubes, an ex tended horizontal and a vertical design, were used for this experiment. The vertical inlet tube is expect ed to improve the crystallinity of InN film through the more uniform flow of NH3 and is also expected to increase the growth rate through the enhanced amount of th e precursors such as TMI and NH3 on the substrate. Using Al2O3 (0001) as the substrate, Figure 5. 55 and 5.56 shows that the single crystalline InN was obtained and a significan tly reduced FWHM of the XRC was also found, from 4860 arcsec (the previous horiz ontal inlet tube) to 1339 arcsec by using the vertical inlet tube. However the extended ho rizontal inlet tube caused In droplet formation since most of the NH3 circulated away from the substrate due to the low density, and because a relatively large amount of TMI was delivered to the Al2O3 (0001) due to the high density of TMI, which leads to a small N/In ratio at the surface. These results show that the structural quality of InN was significantly improved by using the vertical inlet tube for Al2O3 (0001).

PAGE 163

146 Figure 5-55. X-ray Diffraction (XRD) -2 scan for InN/LT-InN on Al2O3 (0001) at N/In = 50,000, T = 530 oC and TLT-InN = 450 oC with different inlet tubes. (b) FWHM of XRC Figure 5-56. Full Width Half Maximum (FWHM) of XRC for InN/LT-InN on Al2O3 (0001) at N/In = 50,000, T = 530 oC and TLT-InN = 450 oC with different inlet tubes. The growth rate of InN with the vert ical inlet tube increased from 0.1 m/hr (the previous horizontal inlet tube) to 0.3 m/hr due to the increased flow rate of TMI and NH3 on Al2O3 (0001), where InN was grown for 3 hrs (Fig. 5.57).

PAGE 164

147 Figure 5-57. Cross-sectional SEM for InN/LT-InN on Al2O3 (0001) at N/In = 50,000, T = 530 oC, and TLT-InN = 450 oC with the horizontal and vertical inlet tubes. For GaN/ Al2O3 (0001), the results shown in Figure 5. 58 and 5.59 i ndicate that single crystalline InN was obtained and th e FWHM of XRC of InN film was also significantly reduced from 1039 arcsec (the prev ious horizontal inlet tube) to 611 arcsec by using the vertical inlet tube These results suggest that th e structural quality of InN was also significantly improved by using the vert ical inlet tube for GaN/Al2O3 (0001). Figure 5-58. X-ray Diffraction (XRD) -2 scan for InN/LT-InN on GaN/Al2O3 (0001) at N/In = 50,000, T = 530 oC and TLT-InN = 450 oC with different inlet tubes.

PAGE 165

148 Figure 5-59. Full Width Half Maximu m (FWHM) of XRC for InN/LT-InN on GaN/Al2O3 (0001) at N/In = 50,000, T = 530 oC and TLT-InN = 450 oC with different inlet tubes. The growth rate of InN with the vert ical inlet tube increased from 0.1 m/hr (the previous horizontal inlet tube) to 0.3 m/hr due to the increased flow rate of TMI and NH3 on GaN/Al2O3 (0001), where InN wa s grown for 3 hrs (Fig. 5.60). Figure 5-60. Cross-sectional SEM for InN/LT-InN on GaN/Al2O3 (0001) at N/In = 50,000, T = 530 oC and TLT-InN = 400 oC with the horizontal and the vertical inlet tubes.

PAGE 166

149 The crystallinity of InN on the surface was characterized for InN grown on Al2O3 (0001) and GaN/Al2O3 (0001) with GIXD (Grazing Angl e Incident X-ra y Diffraction). GIXD is used for the characterization of th e crystallinity of the film on the surface because the incident angle of X-ray is very small. For GI XD characterization, the film with the better crystallinity does not show any peak because the incident angle is fixed at the angle less than 3 degrees. When any peak ex ists, this represents that the film with the growth direction co rresponding to the 2 was grown with the tilt. For this GIXD characterization, the incide nt angle was fixed at 1 degree. For InN grown with the horizontal inlet tube, InN on GaN/Al2O3 (0001) showed the better crystallinity compared with that of InN on Al2O3 (0001) as shown in Fig. 5.61 and 5.62. When InN was grown on GaN/Al2O3 (0001), most of InN (0002) was grown perpendicular to the surface of GaN/Al2O3 (0001) without the ti lting. When InN was grown on Al2O3 (0001), some of InN (0002) was grow n in the different direction from the perpendicular direct ion to the surface of Al2O3 (0001) with the tilting. The same explanation was applied to the InN (10-13) and InN (20-21). Figure 5-61. Grazing Angle Incident X-ra y Diffraction (GIXD) for InN grown on Al2O3 (0002) with the incident angle of 1 degree when th e horizontal inlet tube was used.

PAGE 167

150 (b) Figure 5-62. Grazing Angle In cident X-ray Di ffraction (GIXD) for InN grown on GaN/Al2O3 (0002) with the incident angle of 1 degree when the horizontal inlet tube was used. When the vertical inlet tube was used for the growth of InN on Al2O3 (0001) and GaN/Al2O3 (0001), the crystallinity of InN on th e surface was also ch aracterized with GIXD and compared with that of the horizontal inlet tube (Fig. 5.63). The intensity of InN (0002) peak was reduced sign ificantly on InN grown on both Al2O3 (0001) and GaN/ Al2O3 (0001) using the vertical inlet tube. Theref ore, the vertical inle t tube was found to improve the crystallinity on InN on the surface compared to that of the horizontal inlet tube. The best crystallinity of InN on the surface was obtained for InN on GaN/Al2O3 (0001) with the vertical inlet tube. For growth of InN on Si (111), the f ilms grown with both the vertical and the extended horizontal inlet tube designs showed single crystalline InN (002) growth, but th e intensity of InN (002) peak wa s not significantly increased (Fig. 5.64). No XRC was taken for InN grown on Si (111). From this resu lt, it is concluded that for Si (111), the quality InN film is believed not to be improved through the induced increased mass flow of TMI and NH3. There needs to be studied further in terms of the reaction at the surface of silicon substrate.

PAGE 168

151 Figure 5-63. Grazing Angle Incident X-ray Diffraction (GIXD) for InN grown on (a) Al2O3 (0002) and (b) GaN/Al2O3 (0002) with the incident angle of 1 degree when the vertical inlet tube was used. Figure 5-64. X-ray Diffraction (XRD) -2 scan of InN/LT-InN/GaN/LT-GaN on Si (111) at N/In = 50,000, T = 530 oC and TLT-InN = 400 oC for both horizontal and vertical inlet tubes.

PAGE 169

152 The crystalline quality of InN can be improved further by post-growth annealing presumably through rearrangement of the cr ystallites. Therefor e, the post-growth annealing for 30 min was applied for the furthe r improvement of the structural quality of InN on Al2O3 (0001) and GaN/Al2O3 (0001). This annealing te st was performed at T = 450 oC in N2 flow. The annealing found not to be effective for InN grown on Al2O3 (0001), but effective for InN grown on GaN/Al2O3 (0001) (Fig. 5.65 and 5.66). The FWHM of 574 arcsec wa s obtained for GaN/Al2O3 (0001) after the annealing for 30 min. This FWHM of 574 arcsec is the smallest one known so far among the single crystalline InN film. This result showed a similar trend to the result of the annealing test previously done for Al2O3 (0001) and GaN/Al2O3 (0001), where the annealing effect was much larger for GaN/Al2O3 (0001) than forAl2O3 (0001) (Fig. 5.65 and Fig. 5.66). Figure 5-65. Full Width Half Maximu m (FWHM) of XRC of InN/LT-InN on Al2O3 (0001) at N/In = 50,000, T = 530 oC for both horizontal and vertical inlet tubes. The annealing test is performed at T = 450 oC for 30 min. Figure 5-66. Full Width Half Maximum (F WHM) of XRC of InN/LT-InN on GaN/Al2O3 (0001) at N/In = 50,000, T = 530 oC for both horizontal and vertical inlet tubes. The annealing test is performed at T = 450 oC for 30 min.

PAGE 170

153 Table 5-11. Typical values of FWHM depending on different reactor systems. Reactor FWHM (arcsec) MOVPE ~ 4000-5500 ME-MOVPE ~ 420-2000 MEE (Lu,2000) ~ 3120 [Che97, Lu00, Oian02b] Table 5-12. Reference data available for FWHM for MOVPE reactor. Researcher FWHM (arcsec) Single/ Polycrystal Growth method Reactor Y.C. Pan(1999) ~ 700 Polycrystal InN/sapphire MOVPE S. Yamaguchi(1999) ~ 1700 Polycr ystal InN/LT-AlN/GaN MOVPE A. Yamamoto(2001) ~ 1500 Polycrystal InN/GaNSap MOVPE F.H. Yang(2002) 476 Polycrystal InN/GaNSap MOVPE A. Yamamoto(2004) ~ 2000 Si ngle InN/LT-AlN/Sap MOVPE [Pan99, Yam99a, Yam01b, Yan02a, Yam04a] In summary, it is found that the vertical inlet tube design enha nces the crystalline quality of InN through the increase d mass flow rates of TMI and NH3. These results show that the smallest FWHM of InN/LT-InN grown on Al2O3 (0001) and GaN/Al2O3 (0001) is 1339 and 574 arcsec, respectively. For Al2O3 (0001), the optimized growth temperature is 530 oC with N/In = 50,000 and TLT-InN = 450 oC and for GaN/Al2O3 (0001), the optimized growth temperature is 530 oC with N/In = 50,000 and TLT-InN = 400 oC. When these values of FWHM are compared to other ones (Table 5-11 and 5.12), they show that the FWHM of XRC for InN film grown on GaN/Al2O3 (0001) (FWHM of 574 arcsec) corresponds to the very high qua lity single crystalline InN.

PAGE 171

154 CHAPTER 6 CONCLUSIONS The suitable growth region of InN and InxGa1-xN was calculated thermodynamically in terms of temperature and pressure using ThermoCalc software. Based on the result of the calculation, the grow th of InN occurs at temperatures below T = 800 oC at V/III ratio = 50,000 and P = 100 Torr. Th ese theoretical results are in a good agreement with the experimental data (Tgrowth, 450-700 oC). For In0.3Ga0.7N, a maximum growth temperature of 780 oC was estimated, which is in a good agreement with the experimental data (730 ~ 780 oC). Some disagreement between the calculated values and experimental data may be attributed to the fact that the epitaxial growth of InN and InxGa1-xN by MOVPE is a non-equilibrium r eaction and the calculation assumes equilibrium conditions. The growth temperatur e was almost independe nt of the operation pressure for both InN and InxGa1-xN. For InxGa1-xN, the phase separation diagram was estimated using a 2-sublattice regular solution model and a quantum cal culation method. The 2-sublattice model showed that the phase separation occurred at xIn 0.1 when T = 730 oC and P = 100 Torr. The quantum calculation predicted that the onset of phase separation occurs at 0.25 xIn 0.38. The phase separation experimentally o ccurs from the indium mole fraction of 0.25-0.3 depending on the growth condition. Since the quantum calculation is the theoretical method with the least assump tions, it shows bett er agreement with experimental data.

PAGE 172

155 The critical thickness of In N on GaN (0001), AlN (0001), Al2O3 (0001), and Si (111) was calculated using th ree type of models, and all showed that the dislocation occurs at 1st monolayer of InN. Calcul ated values were consiste nt with the experimental result obtained by TEM. Based on the calcula ted and available experimental data, we could conclude that there is no suitable s ubstrate for the growth of InN and that a LTbuffer layer is necessary for th e growth of high quality InN. The growth conditions of InN on substrates such as Al2O3 (0001), GaN/Al2O3 (0001), and Si (111) substrate were also op timized with growth temperature, growth pressure, buffer layer materials (InN and GaN), growth condition of buffer, V/III ratio and annealing. The InN buffer was first intr oduced in InN growth by MOVPE. It was clearly shown that the structural quality of InN film was improved dramatically. The effect of SiOxN compliant layer was also studied for the growth of InN film. From this study, the optimum V/III ra tio was 50,000 and the optimized growth temperature of InN was 550 oC for LT-GaN buffer layer and 530 oC for LT-InN buffer layer. High V/III ratio could prevent the indi um droplets formation during the InN growth by MOVPE. The mirror-like surf ace and the improved structural quality of InN film was obtained with LT-InN buffer layer (FWH M of XRC ~ 4860 arcsec for InN on Al2O3) rather than with LT-GaN buffer layer (F WHM of XRC ~ 14868 arcsec for InN on Al2O3). The SiOxN compliant layer improved the stru ctural quality of InN film. Using the Fluent software, the flow pattern of NH3 in the MOVPE reactor was studied, for the three types of inlet tube such as the c onventional horizontal, extended horizontal and vertical inlet tube. From the results of this simulation, it was suggested that the vertical inlet tube could increase amount of NH3 and TMI on the substrate and

PAGE 173

156 therefore, introduce more amount of active ni trogen over the substr ate. The results of simulation also indicated that the uniform flow of NH3 could be obtained on the substrate with the vertical inlet tube. Experimentally, using vertical inlet tube the crystalline quality of InN was improved significantly and the growth rate of InN was increased from0.1 to 0.3 m/hr. For Al2O3 (0001), FWHM of XRC of InN was decreased from 4860 to 1339 arcsec. For GaN/Al2O3 (0001), FWHM of XRC of InN was d ecreased from 1039 to 611 arcsec. The characterization of GIXD also showed that th e InN film was grown with much smaller tilt along the (0002) direction with the vertical inlet. These studies of the inlet tube modification suggested that the change of the flow pattern can be one of key factor to influence the structural quality of InN. The effect of post-growth annealing was studied and further improvement in InN film quality was achieved. FWHM of InN wa s decreased further from 611 to 574 arcsec on GaN/ Al2O3 (0001) after annealing at T= 450 oC for 30 min in N2 environment. Optical and electrical propert ies of the InN film on differe nt substrate were studied using Hall measurement and PL. The ba nd-gap energy of InN on Si (111), Al2O3 (0001), and GaN/Al2O3 (0001) was 0.82, 0.84, and 0.94 eV re spectively. The mobility of InN on Si (111), Al2O3 (0001), and GaN/Al2O3 (0001) was 623, 35, and 115 cm2/Vs respectively. The carrier concentratio n of InN on Si (111), Al2O3 (0001), and GaN/Al2O3 (0001) was 7.051018, 8.671019, and 41019 cm-3 respectively. In future work, understanding why the electr ical and optical properties of InN films differ with substrate will be studied in detail. These future studies are important to produce the high mobility and low carrier con centrations that are necessary for InN

PAGE 174

157 devices. It is suggested that the use of othe r substrates be explor ed and the effect of different growth conditions such as usi ng double buffer layer and pressure are also investigated. The structural quality of InN film has been found to be dependent of the flow pattern and rate. When the vertical inlet tube is studied further, the effect of the position of the outlet and the fl ow rate are also suggested for the future work.

PAGE 175

158 LIST OF REFERENCES Abe93 C. R. Abernathy, S. J. Pearton, F. Ren, and P. W. Wisk, J. Vac. Sci. Technol. B 11, 179 (1993). Abe97 C. R. Abernathy, J. D. MacKenzie, and S. M. Donovan, J. Cryst. Growth 178,74 (1997). Ada00 M. Adachi, Y. Murakami, A. Hashimoto, and A. Yamarnoto, Proceedings of theInternational Workshop on Nitride Semiconductors (IWN' 2000), Nagoya, Japan, September 24-27, 2000, IPAP conference series 1, p. 339. Ade01 J. Aderhold, V. Yu. Davydov, F. Fedl er, H. Klausing, D. Mistele, T. Rotter, O. Semchinova, J. Stemmer, and J. Graul, J. Cryst. Growth 222, 701 (2001). Aka93 I. Akasaki, H. Amano, N. Koide, M. Kotaki, and K. Manabe, Physica B 185,428 (1993). Aka94 I. Akasaki, and H. Amano, J. Electrochem. Soc. 141, 2266 (1994). Amb96 O. Ambacher, M. S. Brandt, R. Di mitrov, T. Mctzger, M. Stutzmann, R. A.Fischer, A. Miehr, A. Bergmajer, and G. Dollinger, J. Vac. Sci. Technol. B 14, 3532 (1996). Amb98 O. Ambacher, J. Phys. D 31, 2653 (1998). And74 A. F. Andreeyva and O. J. Eliseejva, Z. Neorg. Chim. 13, 185 (1974). Art67 J. R. Arthur, J. Phys. Chem. Solids 28, 2557 (1967). Bac97 K. J. Bachmann, C. Hoepfner, N. Sukidi, A. E. Miller, C. Harris, D. E. Aspnes et al. Appl. Surf. Sci. 112, 38 (1997). Bea97 B. Beaumont, M. Vaille, T. Boufaden, B. E. Jani, and P. Gibart, J. Cryst. Growth 170 316 (1997). Bed97 S. M. Bedair, F. G. McIntosh, J. C. Roberts, E. L. Piner, K. S. Boutros, and N. A.El-Masry, J. Cryst. Growth 178, 32 (1997). Bel80 L. M. Belyaev, Rubby and Sapphire Amerind Publishing Co., New Delhi 1980, p.1.

PAGE 176

159 Bel99 E. Bellotti, B.K. Doshi, K.F. Brennan, J.D. Albrecht, and P.P. Ruden, J. Appl.Phys. 85, 916 (1999). Bel04 E. Bellet-Amalric, C. Adelmann, E. Sarigiannidou, J. L. Rouvi re, G. Feuillet, E. Monroy, and B. Daudin, J. Appl. Phys. 95, 1127 (2004). Bha02 P. Bhattacharya, T. K. Sharrn a, S. Singh, A.1ngale, and L. M. Kukreja, J. Cryst. Growth 236, 5 (2002). Bhu00a A. G. Bhuiyan, A. Hashimoto, A. Yamamoto, and R. Ishignmi, J. Cryst. Growth 212, 379 (2000). Bhu00b A. G. Bhuiyan. A. Yamamoto, A. Hashimoto and R. Ishignmi, Proceedings of the International Workshop on Nitride Semiconductors (IWN' 2000). Nagoya, Japan. September 24-27, 2000. IPAP conference series 1, p. 343. Bhu01 A. G. Bhuiyan, A. Yamamoto. and A. Hashimoto, Phys. Status Solidi B 228. 27(2001). Bhu02a A. G. Bhuiyan, T. Tanaka A. Yamamoto, and A. Hashimoto, Phys. StatusSolidi A 194, 502 (2002). Bhu02b A. G. Bhuiyan, A. Yamamoto, A. Hashimoto, and Y. Ito, J. Cryst. Growth 236.59 (2002). Bhu03a A. G. Bhuiyan, T. Tanaka, K. Kasashima, A. Hadhimoto, and A. Yamamoto, 5th International Conferen ce on Nitride Semiconductors (ICNS-5), Nara, Japan,May 25-30, 2003. Bhu03b A. G. Bhuiyan, A. Hadhimoto, and A. Yamamoto, J. Appl. Phys. 94, 2779 (2003). Bot95 A. Botchkarev, A. Salvador, B. Sverdlov, J. Myoung, and H. Morkoc, J. Appl. Phys. 77,4455 (1995). Bou94 D. P. Bour, R. S. Geels, D. W. Treat, T. L. Paoli, F. Ponce, R. L. Thornton, B. S. Kunsor, R. D. Bringans, D. F. Welch. IEEE J Quantum Electron. QE-30, 593(1994). Bro98 A. Brown et al., J. Vac. Sci. Technol. B 16, 1300 (1998). Bry90 W. A. Bryden, J. S. Morgan, T. J. Kistenmacher, D. Dayan, R. Fainchtein, and T. O. Poehler, Mater. Res. Soc. Symp. Proc. 162, 567 (1990). Bry92 T. L. Tansley and R. J. Egan, Phys. Rev. B 45, 10942 (1992). Bry92a W. A. Bryden, Y. H. Lee, O. Miki, R. Fainchtein and T. J. Kistenmacher, Thin Solid Films 213 (1992) 86.

PAGE 177

160 Bry94 W. A. Bryden, S. A. E eelberger, and T. J. Kistenmacher, Appl. Phys. Lett. 64, 2864 (1994). Buc88 Buchan, N.I., C.A. Larsen and G.B. Stringfellow, J. Cryst. Growth, 92, 591-604 (1988). But02a K. S. A. Butcher, M. Wintrebert-F ouquet, P. P.T. Chen, T. L. Tansley, and S. Srikeaw, Mater. Res. Soc. Symp. Proc. 693, 341 (2002). But02b K. S. A. Butcher, H. Dou, E. M. Goldys, T. L. Tansley, and S. Srikeaw, Phys. Status Solidi C 0, 373 (2002). Cam90 P.E. van Camp, V. E. van Doren, and J. T. Devreese, Phys. Rev. B (USA). 41.1598 (1990). Che91 C. H. Chen, Z. M. Fang, G. B. Stringfellow, a nd R. W. Gedridge, J. Appl. Phys. 69, 7605 (1991). Che97 W. K. Chen, Y. C. Pan, H. C. Lin, J. Ou, W. H. Chen, and M. C. Lee, Jpn. J. Appl Phys., Part 2 36, L1625 (1997). Che98 W. K. Chen, H. C. Lin, Y. C. Pan, J. Ou, C. K. Shu, W. H. Chen, and M. C. Lee, Jpn. J. Appl. Phys., Part 1 37,4870 (1998). Che00 H. Chen, R. M. Feenstra, J. E. Nort hrup, T. Zywietz, J. Neugebauer, and D. W. Greve, J. Vac. Sci. Technol. B 18(4), 2284 (2000). Chen06 Jeng-Hung Chen, Zhe-Chuan Feng, H ung-Ling Tsai, Jer-Ren Yang, P. Li, C. Wetzel, T. Detchprohm and J. Nelson, Thin Solid Films, Volume 498, Issues 1-2, 1 March 2006, Pages 123-127 Chi94 V. W. Chin, T.L. Transley, and T. Osotchan, J. Appl. Phys. 75, 7365 (1994). Chi96 S. Chichibu, T. Azuhata, T. Sota, and S. Nakamura: Appl. Phys. Lett. 69, 4188 (1996). Chi99 J. A. Chisholm, D. W. Lewis, and P. D. Bristowe, J. Phys. Condens. Matter. 11. L235 (1999). Cho00a H. K. Cho, J. Y. Lee, K. S. Kim, and G. M. Yang: Appl. Phys. Lett. 77, 247 (2000). Cho00b K. L. Choy, Handbook of nanostructured materials and nanotechnology, vol.1, synthesis and processing, Academic Press, San Diego, CA, P533 (2000). Cho03 K. L. Choy, Progress in Materials Science 48, 57 (2003). Cho04 H. K. Cho, J. Y. Lee and J. Y. Leem, Applied Surface Science, Volume 221, Issues 1-4, 15 January 2004, Pages 288-292

PAGE 178

161 Cht97 D. G. Chtchekine, L. P. Fu, G. D. Gilliland, Y. Chen, S. E. Ralph, K. K. Bajaj, Y. Bu, M. C. Lin, F. T. Bacalzo, and S. R. Stock, J. Appl. Phys. 81 2197 (1997). CVD93 CVD metalorganics for vapor phase ep itaxy, product guide and literature review, Morton International, Advanced Materials, Danvers, MA, (1993). Dad00 A. Dadgar, J. Blasing, A. Diez, A. Alam, M. Heuken, and A. Krost, Jpn. J. Appl. Phys. Lett. 39, L1183 (2000). Dad01a A. Dadgar, J. Christen, T. Riemann, S. Richter, J. Blading, A. Diez, and A. Krost, Appl. Phys. Lett. 78, 2211 (2001). Dad01b A. Dadgar A. Alam, T. Riemann, J. Blasing, A. Diez, M. Poschenrieder, M. Stressburg, M. Heuken, J. Christen, and A. Krost, Phys. Stat. Sol. (a) 188, 155 (2001). Dad02 A. Dadgar A. M. Poschenrieder, O. Cont reras, J. Christen, K. Fense, J. Blasing, A. Diez, F. Schulze, T. Riemann, F. A. Ponce, and A. Krost, Phys. Stat. Sol. (a) 192, 308 (2002). Dav97 R. F. Davis, M. J. Paisley, Z. Sitar, D. J. Kester, K. S. A iley, K. Linthicurn, L. B. Rowland, S. Tanaka, and R. S. Kern, J. Cryst. Growth 178, 87 (1997). Dav99a V. Y. Davydov, Appl. Phys. Lett. 75, 3297 (1999). Dav99b V. Y. Davydov, Phys. Status Solidi B 216, 779 (1999). Dav02a V.Y. Davydov, Phys. Status Solidi B 229, R1 (2002). Dav02b V.Y. Davydov, Phys. Status Solidi B 230, R4 (2002). Dav02c V.Y. Davydov., Phys. Status Solidi B 234, 787 (2002). Duc78 Duchemin JP, Bonnet M, Koelsch F, Huyghe D. J Cryst Growth 45. 181 (1978). Dup95 Dupuis RD. In: Glocker DA, Shah SI, editors. Handbook of thin film process technology, B1.1:1. Institute of Physics, Bristol,UK, (1995) Dyc98 J. S. Dyck, K. Kash, K. Kim, W. R. L. Lambrecht, C. C. Hayman, A. Argoitia, M. T. Grossner, W. L. Zhou, and J. C. Angus, Mater. Res. Soc. Symp. Proc. 482, 549 (1998). Edg94 J. H. Edgar, Properties of Group III Nitrides, INSPC, the Institution of Electrical Engineers, London, UK (1994). Ega00 T. Egawa, H. Ishikawa, T. Jimbo, M. Unemo. Bul.l Mate.r Sci. 22, 363 (2000).

PAGE 179

162 Ega02 T. Egawa, B. Zhang, N. Nishikawa, H. Ishikawa, T. Jimbo, and M. Umeno, J. App. Phys. 91 528 (2002). Elm98 N. A. El-Mastry, E. L. Pi ner, S. X. Liu, and S. M. Bedair, Appl. Phys. Lett. 72, 40 (1998). Erb92 A. Erbil, W. Braun, B, S, Zwak, B. J. Wilkens, L. A. Boatner, J, D. Budai. J. Cryst Growth. 24, 84 (1992). Etc01 Etching Bulk GaAs wafer for use as cathodes, www.jlab.org/accel/inj_group/docs/etch.htm, June 2, 2005. Far00 Q. Farecd, J. Yang, J. Zhang, V. Adivarahan, and M. A. Khan, Proceedings of the International Workshop on Nitride Semiconductors (IWN' 2000), Nagoya, Japan, September 24-27, 2000, IPAP conference series 1, p. 237. Fei97 D. Feiler, R. S. Williams, A. A. Talin, H. Yoon, and M. S. Goorsky, J. Cryst. Growth 171, 12 (1997). Fol85 C.P. Foley and T. L. Tansley, App1. Surf. Sci. 22/23, 663 (1985). Fol86 C.P. Folcy and T. L. Tanslcy, Phys. Rev. B 33, 1430 (1986). Fol87 C.P. Folcy and J. Lyngdal, J. Vac. Sci. Technol. A 5,1708 (1987). Fou99 B.E. Foutz, S.K. O’Leary, M.S. Shur, and L.F. Eastman, J. Appl. Phys. 85, 7727 (1999). Gas86 D. K. Gaskill, N. Bottka, and M. C. Lin, J. Cryst. Growth 77, 418 (1986). Geo99 A. George, in : Robert Hull (Ed.), Properties of Crystalline Silicon, IEEE, London, 1999, p98. Gha88 S. K. Ghandhi, I. B. Bhat. MRS Bulletin, 11, 37 (1988). Gra96 N. Grandjean, J. Massies, and M. Leroux, Appl. Phys. Lett. 69, 2071 (1996). Gru91 M. Grundmann, A. Krost, and D. Bimberg, Appl. Phys. Lett. 58, 284 (1991). Guo92 Q. Guo, 0. Kato, andA. Yoshida, J. Electrochem. Soc.139, 2008 (1992). Guo93 Q. Guo, 0. Kato, and A. Yoshida, J. Appl. Phys. 73, 7969 (1993). Guo94a Q. Guo and A. Yoshida, Jpn. J. Appl Phys., Part 1 33, 90 (1994). Guo94b Q. Guo and A. Yoshida, Jpn. J. Apph Phys., Part 1 33, 2453 (1994). Guo94c Q.,Guo, T.Yamamura, A. Yoshida,and N. Itoh, J. Appl. Phys. 75, 4927 (1994).

PAGE 180

163 Guo95a Q. Guo, H. Ogawa, and A. Yoshida, J. Cryst. Growth 146, 462 (1995). Guo95b Q. Guo, H. Ogawa, H. Yamano, and A. Yoshida, Appl. Phys. Lett. 66, 715 (1995). Guo98a Q. X. Guo, M. Nishio, H. Og awa, A. Wakahara, and A. Yoshida, Phys. Rev. B 58, 15304 (1998). Guo98b Q. X. Guo, N. Shinga i, M. Nishio, and H. Ogawa, J. Cryst. Growth 189/190, 466 (1998). Guo99 Q. X. Guo, M. Nishio, H. Ogawa, and A. Yoshida, Jpn. J. Appl. Phys., Part 238, L490 (1999). Guo00 Q. Guo, M. Matsuse, M. Nishio, and H. Ogawa, Jpn. J. Appl. Phys., Part 1 39, 5048 (2000). Guo02 Q. X. Guo, A. Okade, H. Kide ra, T. Tanaka, M. Nishio, and H. Ogawa, J. Cryst. Growth 237-239, 1032 (2002). Gup98 J. A. Gupta, J. C. Woicik, S. P. Watkins, K. E. Miyano, J. G. Pellegrino, E. D. Crozier, J. Cryst. Growth 34, 195 (1998). Hah40 H. Hahn, R. Juza, Z, Anorg. Allg. Chem. (Germany). 244. 111 (1940). Har98 P. Hartmann, R. Haubner, B. Lux. Int. J. Refract. Met. Hard. Mater. 16, 207 (1998). Hel98 E. S. Hellman, MRS Intemet J. Nitride Semicond. Res. 3, 11 (1998). Her99 M. A. Herman, J. T. Sadowski. Crystal Res. Technol. 34, 153 (1999). Hig02 M. Higashiwaki and T. Matsui, Jpn. J. Appl. Phys., Part 2 41, L540 (2002). Hir90 B. Hirsch, Proceedings of the 2nd International Conference on Polycrysta lline Semiconductors, 470 (1990). Hir94 Y. S. Hiraoka and M. Mashita, J. Cryst. Growth 136, 94 (1994). Hit93 M. L. Hitchman, Jensen KF, editors. CVD principles and applications, Academic Press San Diego (1993). Ho96 I. Ho and G. B. Stringfellow, Appl. Phys. Lett. 69, 2701 (1996). Hoc89 M. G. Hocking V. Vasantasree, P. S. Sidky. Metallic and ceramic coatings: production, high temperatur e properties and applications, John Wiley & Sons, New York (1989). Hok91 W. E. Hoke, P. J. Lemonias and D. G. Weir, J. Cryst. Growth 111, 1024 (1991).

PAGE 181

164 Hol80 N. Holonyak, R. M. Kolbas, R. D. Dupuis, P. D. Dapkus. IEEE J Quant Electron. QE-16, 170 (1980). Hor02 M. Hori, K. Kano, T. Yamaguchi, Y. Saito, T. Araki, Y. Nanishi, N. Teraguchi, and A. Suzuki, Phys. Status. Solidi B 234. 750 (2002). Hov72 H. J. Hovel and J. J. Cuomo, Appl. Phys. Lett. 20, 71 1972. 31 J. W. Trainor and K. Rose, J. Electron. Matcr. 3,821 (1974). Hsu98 C. T. Hsu,. Thin Solid Films 335, 284 (1998). Hug95 W. C. Hughes, W. H. Rowland, Jr., M. A. L. Johnson, S. Fujita, J. w. Cook, Jr., J. F. Schetzina, J. Ren, and J. A. Edmond, J. Vac. Sci. Technol. B 13, 1571 (1995). Hwa01 J. S. Hwang, C. H. Lee, F. H. Yang, K. H. Chen, L. G. Hwa, Y. J. Yang, and L. C. Chen, Mat. Chem. and Phys. 72, 290 (2001). Hwa04 J.Y. Hwang Dissertation (2004). Ich86 M. Ichimura and A. Sasaki, J. Appl. Phys. 60, 3850 (1986). Iga92 O. Igarashi, Jpn. J. Appl. Phys., Part 1 31, 2665 (1992). Ina01 Y. Inaba, T. Onozu, S. Takami, M. Kubo, A. Miyamoto, and A. Imamura, Jpn. J. Appl. Phys. 140, 2991 (2001). Inu99 T. Inushima, T. Shiraishi, V. Y. Davydov, Solid State Commun. 110, 491 (1999). Inu01 T. Inushima, V. V. Mamutin, V. A. Vekshin, S. V. Ivanov, T. Sakon, M. Motokawa, and S. Ohoya, J. Cryst. Growth 227-228, 481 (2001). Ish97a T. Ishi, Y. Tazoh, and S. Miyazawa, Mater. Res. Soc. Symp. Proc. 468, 155 (1997). Ish97b M. Ishii, S. Iwai, H. Kawata, T. Ueki, Y. Aoyagi, J. Cryst. Growth 180, 15 (1997). Ish98 H. Ishikawa, K. Yamamoto, T. Eg awa, T. Soga, T. Jimbo, and M. Umeno: J. Crystal Growth 189/190, 178 (1998). Isl95 M. R. Islam, R. V. Chelakara, J. G. Neff, K. G. Fertitta, P. A. Grudowski, A. L. Holmes, J Electron Mater 24, 181 (1995). Ito97 T. Ito, Jpn. J. Appl. Phys. 36, L1065 (1997). Iwa95 K. Iwata, H. Asahi, S. J. Yu, K. Asami, H. Fujita, M. Fushida, and S. Gonda, Jpn. J. Appl. Phys., Part 2 35, L289 (1996).

PAGE 182

165 Jac64 M. G. Jacko and S. J. W. Price, Can. J. Chem. 42, 1198 (1964). Jan00 S.C. Jain, M. Willander, J. Naranyan and R.V. Overstracten, J. Appl. Phys. 87. 965 (2000). Jen89 D. W. Jenkins and J. D. Dow, Phys. Rev. B 39, 3317 (1989). Jes67 W. A. Jesser and D. Kuhlmann-Wilsdorf, Phys. Stat. Sol. 19, 95 (1967). Jin06 J.R. Jinschek and C. Kisielowski, Physica B: Condensed Matter, Volumes 376377, 1 April 2006, Pages 536-539 Joh94 A. C. Jones, J. Alud, S. A. Rushworth, G. W. Critchow. J Cryst Growth 135, 285(1994). Juz38 R. Juza and H. Jahn, Z. Anorg. Allg. Chem. 239, 282 (1938). Juz56 R. Juza and A. Rabenau, Z. Anorg. Allg. Chem. 285, 212 (1938). Kac00 G. Kaczmarczyk, A. Kaschner, S. Reich, A. Hoffman, C. Thomsen., Appl. Phys. Lett. 76, 2122 (2000). Kan04. Sangwon Kang Dissertation 2004, University of Florida. Kap97 S. Kapov: MRS Internet J. Ntride Semicond. Res. 3, 16 (1997). Kea66 P. N. Keating: Phys. Rev. 145 (1966) 637; R. M. Martin: Phys Rev. 1B, 4005 (1970). Kel98 S. Keller, U. K. Misura S. P. Denbaars, and W. Seifert: Jpn. J. Appl. Phys. 37, L431 (1998). Kel00 S. Kellcr, I. Ben-yaacov, S. P. Denvers, and U. K. Mishra, Procecdings of the International Workshop on Nitride Semiconductors (IWN' 2000), Nagoya, Japan, September 24-27, 2000, IPAP conference series 1, p. 233. Kam97 M. Kamp, M. Mayer, A. Pelzmann, and K. J. Ebeling, Mater. Res. Soc. Symp. Proc. 449, 161 (1997). Kan05 S. W. Kang, H. J. Park, T. W. Kim, T. Dann, O. Kryliouk, T. Anderson, Phys. stat. sol. (c) 2, 2420 (2005). Kim96a K. Kim, W. R. L. Lambrecht and B. Segall, Phys. Rev. B 53, 16310 (1996). Kim96b W. Kim, A. Salvador, A. E. Botchka rev, 0. Aktas, S. N. Mohammad, and H. Morkoc, Appl. Phys. Lett. 69, 559 (1996).

PAGE 183

166 Kis90a T. J. Kistenmacher, D. Dayan, R. Fainchtein, W. A. Bryden, J. S. Morgan, andT. O. Poehler, Mater. Res. Soc. Symp. Proc. 162, 573 (1990). Kis90b T. J. Kistenmacher, W. A. Bryde n, J. S. Morgan, and T. O. Poeh1er, J. Appl. Phys. 68, 1541 (1990). Kis91 T. J. Kistenmacher and W. A. Bryden, Appl. Phys. Lett. 59, 1844 (1991). Kis92 T. J. Kistenmacher, S. A. Ecelberger, and W. A. Bryden, Mater. Res. Soc. Symp. Proc. 242, 441 (1992). Kis93a T. J. Kisternn acher and W. A. Bryden, Appl. Phys. Lett. 62, 1221 (1993). Kis93b T. J. Kistenmacher, S. A. Ecelberger, and W. A. Bryden, J. Appl. Phys. 74, 1684 (1993). Kou96 A. Koukitu, N. Takahashi, T. Taki, and H. Seki, Jpn. J. Appl. Phys., Part 2 35, L673 (1996). Kou97a A. Koukitu and H. Seki, Jpn. J. Appl. Phys., Part 2 36, L750 (1997). Kou97b A. Koukitu, N. Takahashi, and H. Scki, Jpn. J. Apph Phys., Part 2 36, L1136 (1997). Kou97c A. Koukitu, T. Taki. Appl. Surf. Sci. 112, 63 (1997). Kou99a A. Koukitu, T. Taki, N. Takahashi, and H. Seki, J. Cryst. Growth 197, 99 (1999). Kou99b A. Koukitu, y: Kumagai, N. Kubota, and H. Scki, Phys. Status Solidi B 216, 707 (1999). Kwo96 H. J. Kwon, Y.H. Lee, O. Miki, H. Yamano, and A. Yoshida, Appl. Phys. Lett. 69,937 (1996). Kub89 K. Kubota, Y. Kobayashi, and K. Fujimoto, J. Appl. Phys. 66, 2984 (1989). Kum94 N. D. Kumar, M. N. Kamalasanan, S. Chandra, Appl. Phys. Lett. 65, 1373 (1994). Kun96 P. Kung, A. Saxlcr, X. Zhang, D. Walkcr, R. Lavado, and M. Razcghi, Appl. Phys. Lett. 69, 2116 (1996). Kur94 K. Kurishima, H. Nakajima, S. Yamahata, T. Kobayashi, Y. Matsuoka, Appl. Phys. Lett. 64, 1111 (1994). Kur01 E. Kurimoto, H. Harima, A. Hashimoto, and A. Yamamoto, Phys. Status Solidi B 228, 1 (2001).

PAGE 184

167 Lah00 H. Lahreche, V. Bousquet, M. Lahgt, P. Vennegues, B. Beaumont, and P. Gibert, Proc. Int. Conf. Sic & Relat. Mater. 99, North Carolina, 273 (2000). Lar86 C. A. Larsen, G. B. Stringfellow, J. Cryst. Growth 75, 247 (1986). Lau98 J. E. Lau, K. W. Barth, G. G. Pete rson, D. Endisch, A. Topol, A. E. Kaloyeros, J. Electrochem. Soc. 145, 4271 (1998). Lee85 W. E. Lee, K. P. D. Lagerlof, J. Electron. Microsc. Technol. 2, 247 (1985). Lee91 R. R. Lee, J. Am. Cerem. Soc. 74 (9), 2242 (1991). Lee95 N. E. Lee, R. C. Powell, Y. W. Kim, and J. E. Greene, J. Vac. Sci. Technol. A 13,2293 (1995). Lee02a I. J. Lee, J. W. Kjm, T. B. Hur, Y.H. Hwang, and H.K. Kim, Appl. Phys. Lett. 81, 475 (2002). Lee02b I. J. Lee, J. W. Kim, Y.H. Hwang, and H.K. Kim, J. Appl. Phys. 92, 5814 (2002). Les95 S.D. Lester, F. A. Ponce, M. G. Craford, and D.A. Steigerwald, Appl. Phys. Lett. 67, 1249 (1995). Li94 X. Li, B. Zhou, K. S. A. Butcher, E. Florido, N. Syakir, and T. L. Tansley, Proceedings of the Australian Compound Optoelectronic Materials Devices Conference, Sydney, Australia, December 12-14, 1994, p. 43. Lil06 Z. Liliental-Weber, D.N. Zakharov, K.M. Yu, J.W. Ager III, W. Walukiewicz, E.E. Haller, H. Lu and W.J. Schaff, Physica B: Condensed Matter, Volumes 376-377, 1 April 2006, Pages 468-472 Lim99 A. P. Lirna, A. Tabata, J. R. Lcitc, S. Kaiser, D. Schikora, B. Schottker, T. Frey, D. J. As, and K. Lischka, J. Cryst. Growth 201/202, 396 (1999). Liu02 L. Liu and J. H. Edgar, Mat. Sci. and Eng., R37, 61 (2002). Liu02a C. H. Liu, R. W. Chuang, S. J. Chang, Y. K. Su, C. H. Kuo, J. M. Tsai and C. C. Lin, Materials Science and Engineering B, Volume 111, Issues 2-3, 25 August 2004, Pages 214-217 Loo02 D. C. Look, H. Lu, W. J. Schaff, J. Jasinski, and Z. Liliental-Wcber, Appl. Phys. Lett. 80, 258 (2002). Lu97 H. Lu, M. Thothathiri, Z. Wu, and I. Bhat: J. Electron. Mater. 26, 281 (1997). Lu00 H. Lu, W. J. Schaff, J. Hwa ng, H. Wu, W. Yeo, A. Pharkya, and L. Eastman,Appl. Phys. Lett. 77, 2548 (2000).

PAGE 185

168 Lu01a H. Lu, W. J. Schaff, J. Hwa ng, H. Wu, G. Koley, and L Eastman, Appl. Phys. Lett. 79,1489 (2001). Lu01b H. Lu, W. J. Schaff, J. Hwang, and L. F. Eastman, Mater. Res. Soc. Symp. Proc. 680E, E3.2 (200). Lu02a H. Lu, W. J. Schaff, L. F. East man, J. Wu, W. Walukiewicz, K. M. Yu, J. W. Auger III, E. E. HaIler, and O. Ambacher, Abstract of the 44th ElectronicMaterial Conference, Santa Barbara, CA, June 26-28, 2002. Lu02b H. Lu, W. J. Schaff, L. F. Eastman, and C. Wood, Mater. Res. Soc. Symp. Proc. 693, 9 (2002). Mam99a V. V. Mamutin, T. Renner, R. R. Parrsons. Phys. Status Solidi A 176,247 (1999). Mam99b V. V. Mamutin, V. A. Vekshin, V. Y. Davydov, V. V. Ratnikov, Y. A. Kudriavtsev, B. Y. Ber, V. v: Emtsev, and S. V. Ivanov, Phys. Status Solidi A176, 373 (1999). Man75 H. M. Manasevit, W. I. Simpson, J. Electrochem. Soc. 122, 144 (1975). Mar77 L. A. Marasina, 1. G. Pichugin, and M. Tlaczala, Krist. Tech. 12, 541 (1977). Mar98 A. A. Marmalyuk, R. K. Akchurin, and V. A. Gorbylev, Inorg. Mater. (Transl. of Neorg. Mater.). 34. 691 (1998). Mar99 P. Martensson, M. Juppo, M. Ritala, M. Leskela, J. O. Carlsson, J. Vac. Sci. Technol. B, 17, 2122 (1999). Mar03 I. V. Markov, Crystal Growth for Beginners, World Scientific, 360 (2003). Mas01 Mike Mastro “GROWTH AND CHARACTERIZATON OF THIN AND THICK GALLIUM NITRIDE” University of Florida, Gainesville, Florida, 2001. Mat75 J. W. Mattews, J. Vac. Sci. Techol. 12, 126(1975). Mat89 T. Matuoka, H. Tana ka, T. Sasaki, and A. Katsui, Inst. Phys. Conf. Ser. 106, 141(1989). Mat90 T. Matsuoka, H. Tana ka, T. Sasaki, and A. Katsui, Proceedings of the Sixteenth International Symposium on GaAs and Related Compounds, Karuizawa, Japan, September 25-29,1989 (Institute of Physics, Bristol, 1990), p. 141. Mat92 T. Matsuoka, N Yoshimot o, T. Sasaki, and A. Katsui, J. Electron. Mater. 21, 157 (1992).

PAGE 186

169 Mat97 T. Matsuoka, in GaN and Related Materials, edited by S. J. Pearton (Gordon and Breach, New York, 1997), pp. 53-59. Mat02 T. Matsuoka, H. Okamoto, M. Nakao, H. Harima, and E. Kurimoto, Appl. Phys. Lett. 81, 1246 (2002). May90 James W. Mayer, Electronic Materials Science for Intergrated Circuits in Si and GaAs, Macmillan Publishing Company, New York (1990). McC70 J. B. McChesney, P. M. Bridenbaugh, and P. B. O'Connor, Mater. Res. Bull. 5, 783 (1970). Mil96 J. R. Mileham, S. J. Pearton, C. R. Abernathy, J. D. Mackenzie, R. J. Shul, and S. P. Kilcoyne, J. Vac. Sci. Technol. A 14, 836 (1996). Miy02 T. Miyajima, Y. Kudo, K. L. Liu, T. Honma, Y. Sato., Phys. Status Solidi B 234, 801 (2002). Moh96 S. N. Mohammad and H. Morkoc, Prog. Quantum Electron.20, 361 (1996). Mol02 Motlan, E. M. Goldys, and T. L. Tansley, J. Cryst. Growth 241, 165 (2002). Mol94 R. J. Molnar and T. D. Moustakas, J. Appl. Phys. 76, 4587 (1994). Mor90 J. S. Morgan, T. J. Kistenmaeher, W. A. Bryden, and T. O. Pochler, Mater. Res. Soc. Symp. Proc. 162, 579 (1990). Mor91 J. S. Morgan, T. J. Kistenmach er, W. A. Bryden, and S. A. Ecelberger, Mater. Res. Soc. Symp. Proc. 202, 383 (1991). Mor94 H. Morkoc, S. Strite, G. B. Gao, M.E. Lin, B. Sver1ov, and M. Burns, J. App1. Phys. 76, 1363 (1994). Mot94 T. Motoda, M. Kato, K. Kadoiwa, A. Shima, M. Tsugami, T. Sonoda, et al. J. Cryst Growth 145, 650 (1994). Nag89 T. Nagatomo, T. Kuboyama H. Minamino, and O. Omoto: Jpn. J. Appl. Phys. Lett. 28, L1334 (1989). Nag02 B.R. Nag, Phys. Status Solidi B 233, R8-R9 (2002). Nak91 S. Nakamura, Y. Harada, and M. Seno, Appl. Phys. Lett. 58, 2021 (1991). Nak92 Shuji Nakamura and Takashi Mukai, Jpn. J. Appl. Phys. 31, L1457 (1992). Nak93 S. Nakamura, M. Senoh, and T. Mukai, S. Nagahama, and N. Iwasa, J. Appl. Phys., 74. 3911 (1993). Nak94 S. Nakamura, Microelectron. J. 25. 651 (1994).

PAGE 187

170 Nak96 S. Nakamura, M. Senoh, S. Nagahama, N. Iwasa, T. Yamada, T. Matushita, H. Kiyoku, and Y. Sugimoto, Jpn. J. Appl. Phys., Part 2 35, L74 (1996). Nak00 Shuji Nakamura: IEICE Trans. Electronics. E83-C 529 (2000). Nat80 B.R.Natarajan, A.H. Eltoukhy, J.E. Greene, T.L. Barr, Thin Solid Films 69, 201 (1980). Nat86 B.R. Natara.jan, A. H. Eltoukhy, J. E. Greenc, and T. L. Barr, Thin Solid Films 69, 201 (1980). Ng02 Y. F. Ng, Y. G. Cao, M. H. Xie, X. L. Wang, and S. Y. Tong, Appl. Phys. Lett. 81, 3960 (2002). Nic96 J. H. M. Nicholls, H. GalIaghc r, B. Hcndcrson, C. Tragcr.Cowan. P. G. Middleton, K. P. O'Donnel. T. S. Che ng, C. T. Foxon, and B. H. T. Chai, Mater.Res. Soc. Symp. Proc. 395, 535 (1996). Nii96 L. Niinisto, M. Ritala, M. Leskela, Mat. Sci. Enging. B. Solid-State Mat. Adv. Technol. 41, 23 (1996). Nil99. S. A. Nilishin, N. N. Faleev, V. G. Antopov, S. Francoeur and L. Grave de Peralta, Appl. Phys. Lett. 75, 736 (1999). Nis97 K. Nishida, S. Haneda, K. Hara, H. Munekata, and H. Kukimoto, J. Cryst. Growth 170, 312 (1997). Nr02a C. Nrenberg, M.G. Martin, R. A. oliver, M. R. Castell, and G. A. D. Briggs, J. Phys. D 35, 615 (2002). Nr02b C. Norenberg, R. A. Oliver, M. G. Mart in, L. Allers, M. R. Castell, and G.A.D. Briggs, Phys. Status Solidi A 194, 536 (2002). Ohk98 M. Ohkubo, T. Ijichi, K. Iketani, T. Kikuta. IEEE J Quantum Electron. 30, 408 (1998). Ohk00 M. Ohkubo and O. Takai, Proceedings ofthe International Workshop on Nitride Semiconductors (IWN' 2000), Nagoya, Japan, September 24-27, 2000, IPAP conference series 1, p. 770. Oku91 H. Okumura, S. Yoshida, and E. Sakuma, J. Cryst. Growth 120, 114 (1991). Oku98a H. Okumara, H. Hamaguchi, T. Koizumi, K. Balakrihnan, Y. Ishida, M. Arita, S. Chichubu, H. Nakanishi, T. Nagamoto, and S. Yoshida, J. Cryst. Growth 189/190, 390 (1998). Oku98b S. Okubo, N. Shibata, T. Saito, and Y. Ikuhara, J. Cryst. Growth 189/190, 452 (1998).

PAGE 188

171 Ono99 T. Onozu, I. Gunji, R. Miura, S. S. C. Ammal, M. Kubo, K. Teraishi, A. Miyamoto, Y. Iyechika, and T. Maeda, Jpn. J. Appl. Phys., Part 1 38, 2544 (1999). Osa72 K. Osamura, K. Nakajima, Y. Murakami, 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). Qia02 Z. G. Qian, W. Z. Sh en, H. Ogawa, and Q. X. Guo, J. App1. Phys. 92, 3683 (2002). Qian02b Z. G. Qian, W. Z. Shen H. Ogawa, and Q. X. Guo, Physica B. 318, 180 (2002). Qin02 Z.Qin, Z. Chen, Y. Tong, S. Lu, and G. Zang, Appl. Phys. A 74 655 (2002). Ou98 J.Ou, W. Chen, H. Lin, Y. Pan, and M. Lee: Jpn. J. Appl. Phys. 37, L633 (1998). Pan96 J. S. Pan, A. T. S. Wee, C. H. A. Hual1, H. S. Tan, and K. L. Tan, J. Phys. D 29, 2997 (1996). Pan98 J. I. Pankove, Gallium Nitride (GaN) I, Academic Press (1998). Pan99 Y. C. Pan, W. H. Lee, C. K. Shu, H. C. Lin, C. I. Chjang, H. Chang, D. S. Lin, M. C. Lee, and W. K. Chen, Jpn. J.Appl. Phys., Part 1 38, 645 (1999). Par02 Park, C., W.-S. Jung, Z. Huang and T.J. Anderson, J. Mater. Chem., 12, 356 (2002). Pas63 J. Pastrnak and L. Souckova, Phys. Status Solidi 3, K71 (1963). Pea67 W. B. Pearson, A Handbook of Lattice Spacing and Structures of Metals and Alloys, Pergamon Press, Oxford (1967) Pin98 E. L. Piner, N. A. El-Mas try, S. X. Liu, and S. M. Bedair, Materials Research Society Proceedings, Vol.482 (1998), Appl. Phys. Lett. 72, 40 (1998). Pos02 M. Poschenrieder, F. Schulze, J. Blas ing, A. Dadgar, A. Diez, J. Christen, and A. Krost, Appl. Phys. Lett. 81, 1591 (2002). Pow93 R. C. Powell, N. E. Lee, Y. W. Kim, and J. E. Greene, J. Appl. Phys.73, 189 (1993). Put86 N. Putz, H. Heinecke, M. Heyen, P. Balk, M. Wayers, H. Luth,. J. Crystal Growth 74, 292 (1986). Puy76 N. Puychcvricr and M. Mcnoret, Thin Solid Films 36, 141 (1976).

PAGE 189

172 Qia02 Z.G. Qian, W. Z. Shen, H. Ogawa, and Q.X. Guo, J. Appl. Phys. 92., 3683 (2002). Pea93 S. J. Pcarton, C. R. Abemathy, F. Ren, J. R. Lothian, P. W. Wisk, and A. Katz, J. Vac. Sci. Technol. A 11, 1772 (1993). Rai98 Raina N. Smith, Virginia Commonwealth University, May 5 (1998) http://www. Geocities.com/Capecanaveral/5702/Fe_Si.html. Ree02 Mike Reed Dissertation 2002, University of Florida. Ren58 T. Renner, Z. Anorg. Allg. Chem. 298, 28 (1958). Rom05 Electronic Materials, http://elect ronicmaterials.rohmhaas.com/, June 2, 2005. Ros97 S. J. Rosner, E. C. Carr, M. J. Ludowise, G. Girolami, and H. I. Erikson, Appl. Phys. Lett. 70, 420 (1997). Rus96 S. A. Rushworth, J. R. Brown, D. J. Houlton, A. C. Jones, J. S. Roberts, and G. W. Critchlow, Advanced Materials for Optics and Electronics 6, 119 (1996). Sai01 Y. Saito, Y. Tanabe, T. Yamaguchi, N. Teraguchi, A. Suzuki, T. Araki, and J Y. Nanishi, Phys. Status Solidi B 228, 13 (2001). Sai02a Y. Saito, H. Harima, E. Kurimoto T. Yamaguchi, N. Teraguchi. A. Suzuki, T. Araki, and Y. Nanishi, Phys. Status Solidi B 234, 796 (2002). Sai02b Y. Saito, T. Yamaguchi, H. Kanazaw a, K. Kano, T. Araki, Y. Nanishi, A. Suzuki, and N. Teraguchi, J. Cryst. Growth 237-239, 1017 (2002). Sam69 G.V. Samsonov. Nitridy Kiev, 1969. San92 C. Sant, P. Gibart, P. Genou, C. Verie, J Cryst Growth 124, 690 (1992). Sat89 Y. Sato and S. Sato, Jpn. J. Appl. Phys. Part 228, L1641 (1989). Sat94a Y. Sato and S. Sato, J. Cryst. Growth 144, 15 (1994). Sat94b Y. Sato, S. Kakinuma, and S. Sato, Jpn. J. Appl. Phys., Part 1 33, 4377 (1994). Sat95 Y. Sato and S. Sato, J. Cryst. Growth 146, 262 (1995). Sat96 H. Sato, H. Takahashi, A. Watanabe, and H. Ota, Appl. Phys. Lett. 68, 3617 (1996). Sat97a M, Sato, Jpn. J. Appl. Phys., Part 2 36, L595 (1997). Sat97b M. Sato, Jpn. J. Appl. Phys., Part 2 36, L658 (1997).

PAGE 190

173 Sat98 H. Sato, T. Sugahara, Y. Naoi, and S. Sakai: Jpn. J. Appl. Phys. 37, 2013 (1998). Sch91 M.C. Schabel and J. L. Martins: Phys. Rev. B 43, 11873 (1991). Sch00 W. J. Schaff, H. Lu, J. Hwang, and H. Wu, Proceedings of the Seventeenth Biennial IEEE/Cornell Conference on Advanced Concepts in High PerformanceDevices, August 7-9, 2000, p. 225. See97 M. Seelmann-Eggebert, J. L. Weher, H. Obloh, H. Zimmermann, A. Rar, and S. Porowski, Appl. Phys. Lett. 71, 2635 (1997). She79 A. U. Sheleg and V. A. Savastenko, Inorg. Mater. Transl. of Neorg. Mater. 15. 1257 (1979). She91 M. E. Sherwin, and T. J. Drummond, J. Appl. Phys. 69. 8423 (1991). She02 Jianyun Shen, Steven Johnston, Shunli Shang, Timothy Anderson, J. Cryst. Growth 240, 6 (2002). Shi94 M. Shimizu, K. Hiramatsu, and N. Sawaki, J. Crystal Growth, 145, 209 (1994). Shu98 R. J. Shul, C. G. Willison, M. M. Bridges, J. Han, J. W. Lec, S. J. Pearton, C. R. Abcmathy, J. D. Mackenzie, and S. M. Donovan, Solid-State Electron. 42, 2269 (1998). Shu00 Shuji Nakamura: IEICE Trans. Electrons. E83-C, 529 (2000). Sin97 R. Singh, D. Doppalapudi, T. D. Moustakas, L. T. Romano, Appl. Phys. Lett. 70, 1089 (1997). Sla98 G. A. Slack, Mater. Res. Soc. Symp. Proc. 512, 35 (1998). Smi98 Raina N. Smith, Virginia Commonwealth University, May 5, 1998 http;//www.geocities.xom/CapeCanaveral/5702/Fe_Si.html. Sta92 R. A. Stall, E. Wolak, P. Zawadski. Mat Res Soc Symp Proc. 282, 115 (1992). Sta00 C. Stampfl, C. G. Van de Wa1l e, D. Vogel, P. Kruger, and J. Pollmann, Phys. Rev. B 61, R7846 (2000). Str92 S. Strite and H. Morkoc, J. Vac. Sci. Technol. B 10,1237 (1992). Str93 S. Strite, D. Chandrasekhar, D. J. Smith, J. Sariel, H. Chen, N. Teraguchi, and H. Morkoc, J. Cryst. Growth 127, 204 (1993). Str97 W. Van der Stricht, I. Moerman, P. Demeester, L. Considine, E. J. Thrush, and J.A. Crawley, MRS Internet J. Ntride Semicond. Res. 2, 16 (1997).

PAGE 191

174 Str98 A. Strittmatter, A. Krost, K. Schatke, D. Bimberg, J. Bl..asing, and J. Christen, Proceedings of the Seventh Internatio nal Conference on SiC, III-Nitrides and Related Material, Trans Tech Publ., Part 2, 1145 (1998). Str99 G. B. Stringfellow, Organometallic Vapor-Phase Epitaxy, Academic Press, New York (1999). Sug91 T. Sugimoto, M. Yoshida, K. Ya maguchi, Y. Yamada, K. Sugawara, Y. Shirohara, J Crystal Growth 107, 692 (1991). Sul88 B. T. SuIlivan, R. R. Parsons, K. L. Westra, and M. J. Brett, J. Appl. Phys. 64,4144 (1988). Sun89 T. Suntola. Mat. Sci. Rep. 4, 261 (1989). Sun94 C. J. Sun, P. Kung, A. Saxier, H. Ohsato, K. Haritos, and M. Razeghi, J. Appl. Phys, 75, 3964 (1994). Sun96 H. Sunakawa, A. Yamaguchi, A. Kimura, and A. Usui, Jpn. J. Appl. Phys., Part 2 35, L 1395 (1996). Tab99 A. Tabata, A. P. Lima, L. K. Teles, L. M. R. Scolfaro, J. R. Leite, Appl. Phys. Lett. 74, 362 (1999). Tak97a N. Takahashi, J. Ogasawara, and A. Koukitu, J. Cryst. Growth 172, 298 (1997). Tak97b N. Takahashi, R. Matsumoto, A. Koukitu, and H. Seki, Jpn. J. Appl. Phys., Part 2 36, L743 (1997). Tak99 Takashi Mukai and Shuji Nakamura, Jpn. J. Appl. Phys. 38, 5735 (1999). Tan84a T. L. Tansley and C. P. Foley, Electro. Lett. 20, 1066 (1984). Tan84b T. L. Tansley and C. P. Foley, in Proceedings of the 3rd International Conference on Semi-Insulating III-V Materials, Warm Springs, Oregon, 1984,p. 497. Tan86a T. L. Tansley, C.P. Foley, J. Appl. Phys. 59, 3241 (1986). Tan86b T. L. Tansley and C. P. Foley, J. App1. Phys. 60, 2092 (1986). Tan88 T. L. Tansley, R. J. Egan, and E. C. Horrigan, Thin Solid Films. 164. 441 (1988). Tan92a T. L. Tansley and R. J. Egan, Phys. Rev. B 45,10942 (1992). Tan92b T. L. Tansley and R. J. Egan, Mater. Res. Soc. Symp. Proc. 242, 395 (1992).

PAGE 192

175 Tod95 A. Toda, T. Kawasaki D. Zmaniski, A. Ishibashi. Electron. Lett. 31, 235 (1995). Tom92 G. S. Tompa, E. Wolak, R. A. St all, M. A. George, M. Lippitt Norman JAT. Mat. Res. Soc. Symp. Proc. 282, 323 (1992). Tom94 K. Tominaga, Y. Sa kashita, H. Nakashima, M. Ada, J Cryst Growth. 145, 219 (1994). Tra74 J. W. Trainor and K. Rose, J. Electron. Mater. 3, 821 (1974). Tra78 N. N. Travkin, P. K. Stachov, I. G. Tonoyan, and B. I. Kozykin, J. Gen. Chem. USSR 48, 2428 (1978). Tra99 C. A. Tran, A. Osinski, and R. F. Kalicek, Jr., Appl. Phys. Lett. 75, 1494 (1999). Tra00 A. Trampert and K. H. Ploog, Cryst. Res. Technol., 35, 793 (2000). Tsu99 T. Tsuchiya, H. Yamano, 0. Mjki, A. Wakahara, and A. Yoshida, Jpn. J. Appl. Phys., Part 1 38, 1884 (1999). Tsu00a T. Tsuchiya, 0. Miki, K. Shimada, M. Ohnishi, A. Wakahara, and A. Yoshida, J. Cryst. Growth 220, 185 (2000). Tsu00b T. Tsuchiya, M. Ohnjshi, A. Wakahara, and A. Yoshida, J. Cryst. Growth 220, 191 (2000). Tu92 King Ning Tu, James W. Mayer, Leonard C. Feldman, Electronic Thin Film Science, Macmillian Publishing Company, (1992). Uch96 K. Uchida, A. Watanabe. F. Ya no, M. Kouguchi, T.Tanka and S. Hinakawa, J. Appl. Phys. 79 (1996) 3487. Utr99 M. Utriainn, C. Kovacs, J. M. Campbell, L. Niinistoe, F. Reti. J. Electrochem. Soc. 146, 189 (1999). Utr00 M. Utriainen, Kroeger, M. Laukka nen, L. S. Johansson, L. Niinistoe, Appl. Surf. Sci. 157, 151 (2000). Vis95a R.D. Vispute, J. Na rayan, H. Wu, and K. Jagannadham, J. Appl. Phys.77, 4724(1995). Vis95b R. D. Vispute, H. Wu, and J. Narayan, Appl. Phys. Lett. 67, 1549 (1995). Wak89 A. Wakahara and A. Yoshida, Appl. Phys. Lett. 54, 709 (1989). Wak90 A.Wakahara, T. Ts uchiya, and A. Yoshida, J. Cryst. Growth 99, 385 (1990). Wak97 A. Wakahara, T. Tokuda, X. Dang, S. Noda, and A. Sasaki, Appl. Phys. Lett. 71, 906 (1997).

PAGE 193

176 Wan01 K. Wang and R. R. Reeber, Appl. Phys. Lett. 79, 1602 (2001). Wat93 A. Watanabe, T. Takeuchi, K. Hirosawa, H. Amano, K. Hiramatsu, I. Akasaki, J. Crystal Growth, 128. 391 (1993). Wes88 K. L. Westra, R. P. W. Lawson, and M. J. Brett, J. Vac. Sci. Techno1. A 6, 1730 (1988). Wri97 A. F. Wright, J. Appl. Phys. 82. 2883 (1997). Wu02 J. Wu, W. Walukiewicz, K.M. Yu, J. W. Ager III, E.E. Haller, H.Lu, W.J. Schaff, Y. Saito, and Y. Nanishi, Appl. Phys. Lett. 80, 3967 (2002). Xu02 K. Xu, W. Terashima, T. Hata, N. Hashimoto, Y. Ishitani, and A. Yoshikawa, Phys. Status Solidi C 0, 377 (2002). Yam94a A. Yamamoto, M. Tsujino, M. Ohkubo, and A. Hashimoto, Sol. Energy. Mater. Sol. Cells 35, 53 (1994). Yam94b A. Yamamoto, M. Tsujino, M. Ohkubo, and A. Hashimoto, J. Cryst. Growth 137, 415 (1994). Yam96 A. Yamamoto, Y. Yamauchi,T. Ogawa, M. Ohkubo, and A. Hashimoto, Inst. Phys. Conf. Ser. 142, 879 (1996). Yam97a A. Yamamoto, Y. Yamauchi, M. Ohkubo, A. Hashimoto, and T. Saitoh, SolidState Electron. 41, 149 (1997). Yam97b A. Ynmamoto, Y. Yamauchi, M. Ohkubo, and A. Hashimoto, J. Cryst. Growth 174, 641 (1997). Yam98a A. Yarnamoto, T. Shin-y a, T. Sugiura, and A. Hashimoto, J. Cryst. Growth 189/190, 461 (1998). Yam98b A. Yamamoto, T. Shin-ya, T. Sugiura, M. Ohkubo, and A. Hashimoto, J. Cryst. Growth 189/190, 476 (1998). Yam99a S. Yamaguchi, M. Kari ya, S. Nitta, T. Takcuchi. C. Wctzcl, H. Amano, I. Akasaki, J. Appl. Phys. 85, 7682 (1999). Yam99b A. Yamamoto, M. Adachi, T. Arita, T. Sugiura, and A. Hashimoto, Phys. Status Solidi A 176,595 (1999). Yam01a A. Yamamoto, Y. Murakarni, K. Koide, M. Adachi, and A. Hashimoto, Phys. Status Solidi B 228. 5 (2001). Yam01b A. Yamamoto, M. Adachi, and A. Hashimoto, J. Cryst. Growth 230, 351(2001).

PAGE 194

177 Yam01c T. Yamaguchi, Y. Saito, K. Kano, T. Araki, N. Teraguchi, A. Suzuki, and Y. Nanishi, Phys. Status Solidi B 228, 17 (2001). Yam02 A. Yamamoto, T. Tanaka, K. Koide, and A. Hashjmoto, Phys. Status Solidi A 194,510 (2002). Yam03 A. Yamamoto, T. Tanaka, A. G. Bhui yan, K. Sugita, K. Kasashima, Y. Kimura, A. Hashimoto, and V. Yu. Davydov, 5th Intcmational Conference on Nitride Semiconductors (ICNS-5), Nara, Japan, May 25-30, 2003. Yam04a A. Yamamoto, N. Imai, K.Sugita, and A. Hashimoto, J. Cryst. Growth 261, 271 (2004). Yam04b A. Yamamoto, K. Sugita, H. Takatsuka, Y. Hamano, N. Imai, and A. Hashimoto, Proceedings of the International Workshop on NitrideSemiconductors (IWN' 2004), Pittsburgh, USA, July 19-23, p113 (2004). Yan95 Z. Yang, L. K. Li, and W. I. Wang, Appl. Phys. Lett. 67, 1686 (1995). Yan96 J.W. Yang, C. J. Sun. Q. Chen, M.Z. Anwar, M. A. Khan, S. A. Nikishin, G. A. Seryogin, A. V. Osinsky, L. Chernyak, H. Temkin, C. Hu, and S, Mahajan, Appl.Phys. Lett. 69, 3566 (1996). Yan99 C.C. Yang, G.C. Chi, C.K. Huang, and M.C. Wu, J. Cryst. Growth 200, 32 (1999). Yan00 J. W. Yang, A. Lunev, G. Simin, A. Chitnis, M. Shatalov, and M. Asif Khan, Joseph E. Van Notrand, and R. Gaska, Appl. Phys. Lett. 76, 273 (2000). Yan02a F. H. Yang, J.S. Hwang, Y. J. Yang, K.H. Chen, and J.H. Wang, Jpn. J. Appl. Phys., Part 2 41, L1321 (2002). Yan02b H. F. Yang, W. Z. Shen, Z. G. Qian Q. J. Pan, H. Ogawa, and Q. X. Guo, J. Appl. Phys. 91, 9803 (2002). Yan02c 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, 194 (2002). Yaw99 C. L. Yaws, Chemical Properties Handbooks, McGraw-Hill, 475 (1999). Yod01 T. Yodo, H. Ando, D. Nosei, and Y. Harada, Phys. Status Solidi B 228, 21 (2001). Yod02 T. Yodo, H. Yona, H. Ando, D. Nosei, and Y. Harada, Appl. Phys. Lett. 80, 968 (2002). Yos83 S. Yoshida, S. Misawa and S. Gonda, Appl. Phys. Lett. 42 427 (1983).

PAGE 195

178 Yos91 N Yoshimoto, T. Mats uoka, T. Sasaki, and A. Katsui, Appl. Phys. Lett. 59 2251 (1991). Zim01 S. Zimmer, B. Meyler, and J. Salzman: Appl. Phys. Lett. 78, 288 (2001). Zun94 A. Z. Zunger and S. Mahajan, in Handbook on Semiconductors, ed. T. S. Moss, Elsvier Science, Amsterdam, (1994) p.1402.

PAGE 196

179 BIOGRAPHICAL SKETCH Taewoong Kim was born to Doohoi Kim a nd Byungrae Yoo on July 31, 1968, in Seoul, Korea. He grew up in Seoul, Korea. He graduate d from Youngil High School in 1986. He then went on to Inha University in Inchon City, where he graduated in 1993 with a Bachelor of Science in polymer scien ce and engineering. He went on to graduate school in Inha University where he gradua ted in 1995 with a Master of Science in polymer science and engineering. In graduate school, he studied the ER-fluid (electrorheological fluid). After gra duating, he worked for LG Ch emical Company from 1995 to 1999, where he performed research on polymer processing. He changed his major and joined the Chemical Engineering Department at University of Florida in August of 1999. There he researched semiconductor epitaxial InN film growth in Dr. Timothy Anderson’s group.


Permanent Link: http://ufdc.ufl.edu/UFE0010058/00001

Material Information

Title: Indium Nitride Growth by Metal-Organic Vapor Phase Epitaxy (MOVPE)
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: UFE0010058:00001

Permanent Link: http://ufdc.ufl.edu/UFE0010058/00001

Material Information

Title: Indium Nitride Growth by Metal-Organic Vapor Phase Epitaxy (MOVPE)
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: UFE0010058:00001


This item has the following downloads:


Full Text












INDIUM NITRIDE GROWTH BY METAL-ORGANIC VAPOR PHASE EPITAXY


By

TAEWOONG KIM
















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

Taewoong Kim















ACKNOWLEDGMENTS

The author wishes first to thank his advisor, Dr. Timothy J. Anderson, for

providing five years of valuable advice and guidance. Dr. Anderson always encouraged

the author to approach his research from the highest scientific level. He is deeply thankful

to his co-advisor, Dr. Olga Kryliouk, for her valuable guidance, sincere advice, and

consistent support for the past five years.

Secondly, the author wishes to thank the remaining committee members of Dr.

Steve Pearton and Dr. Fan Ren for their advice and guidance

The author is grateful to Scott Gapinski, the staff at Microfabritech, and Eric

Lambers, the staff at the Major Analytical Instrumentation Center, especially for Auger

characterization.

Acknowledgement needs to be given to Sangwon Kang who worked with the

author for the past year and provided valuable assistance.

Thanks go to Youngsun Won for his useful discussion of quantum calculation and

SEM characterization, and to Dr. Jianyun Shen for her assistance about how to use the

ThermoCalc.

The author wishes to thank Hyunjong Park for useful discussion and Youngseok

Kim for his kindness and friendship.

Most importantly, the author is grateful to Moonhee Choi, his beloved wife, for her

endless support, trust, love, sacrifice and encouragement. Without her help, he would

have not finished the Ph.D. course.









The author is grateful to his mother, father, mother-in-law, father-in-law, sisters,

and brother for providing love, support and guidance throughout his life.
















TABLE OF CONTENTS


page

A CK N O W LED G M EN T S ......... ......... ............... ........................... ......................... iii

LIST OF TABLES ............. .... ...... .................................. viii

LIST OF FIGURES ............................... .... ...... ... ................. .x

ABSTRACT. ....................................... xvii

CHAPTER

1 INTRODUCTION ............... .................................. ................... 1

2 LITER A TU RE REV IEW .................................................. ............................... 3

2.1 Indium Nitride (InN) and Indium Gallium Nitride (InxGal-xN) Properties ............4
2.1.1 Structural Properties .............................................................................. 4
2.1.2 Physical P properties ............................................... ............................ 7
2.1.3 Electrical Properties of InN ................................... .................. ... ......... 9
2.1.3.1 B background D effects ............................. ....... ... ........................ 9
2.1.3.2 Hall mobility and Electron Concentration in Undoped InN ............10
2.1.4 O ptical Properties of InN ....................................................... .................. 14
2.1.5. Indium Nitride (InN) andlindium Gallium Nitride (InxGal-xN)
A application s ....................... ........ .................... ........................... 16
2.2 Thermodynamic Analysis and Phase Separation in the InxGal-xN System..........18
2.2.1 Thermodynamic M odels in Solid Solution...............................................19
2.2.1.1 Regular Solution M odel ....................... ..............................19
2.2.1.2 Bonding in Semiconductor Solid Solutions Model..........................19
2.2.1.3 Delta Lattice Parameter (DLP) Model for Enthalpy of Mixing .......21
2.2.1.4 Strain Energy M odel ............................................. ............... 22
2.2.1.5 First-Principal M odels.................................. ........................ 23
2.2.2 Thermodynamic Analysis of InN............................................................24
2.2.3 Phase Separation in InxGalxN .............................. ..................... ............... 26
2.3 Indium Nitride (InN) and Indium Gallium Nitride (InxGal-xN) Growth
Challenges ........................................................................ ..........32
2.3.1 Growth Temperature and V/III Ratio.................................... ...................32
2.3.2 N itrogen Source .................. .................. ..... ...................... 34
2.3.3 Carrier Gas.................... ................. ..... ............ 36









2.4 Indium Nitride (InN) Growth Techniques ......................................................37
2.4.1 Chem ical V apor D position (CVD ) ................................ .....................37
2.4.1.1 Metal-Organic Vapor Phase Epitaxy (MOVPE).............................38
2.4.1.2 Hydride Vapor Phase Epitaxy (HVPE) ............................................41
2.4.1.3 Plasma Enhanced Chemical Vapor Deposition (PECVD) ..............42
2.4.2 Molecular Beam Epitaxy (MBE) and Metalorganic Molecular Beam
E pitaxy (M O M B E ) ............................................... ......................... 44
2.4.3 Atomic Layer Deposition (ALD) ............................. ............... 45
2 .5 Sub state M materials .............................. .................. ... ........... ................... 46
2.5.1 Sapphire Substrate (Al203) (0001) .................................. ............... ..47
2.5.2 Silicon (Si) Substrate ......................................................................... ...... 49
2.5.3 Gallium Nitride (GaN) and Aluminium Nitride (A1N) Substrate .............50
2 .5.4 O their Sub strates......... .............................................. ........ .... .......... 5 1
2.5.5 Buffer Layer .......................... ........ ......... ............... .. 52
2.6 Summary for Growth of InN on Different Substrate................ .............. ....53
2.6.1 Growth on Sapphire (A1203) Substrate......................................................53
2.6.2 Growth on Silicon (Si) Substrate........................................................... 55
2.6.3 Growth on Gallium Arsenide (GaAs) Substrate.......................................56
2.6.4 Growth on Gallium Phosphorus (GaP) Substrate.......................................57
2.6.5 Growth on Gallium Nitride (GaN) and Alumimum Nitride (A1N)
Substrate................................... ......................................... .. 58
2 .7 O v erv iew .................................................... ................ 5 9

3 THERMODYNAMIC ANALYSIS OF InN AND InxGal-xN MOVPE GROWTH..60

3.1 Thermodynamic Analysis of InN and InxGal-xN ..................................................60
3.1.1 Reaction Mechanism and Kinetics of InN MOVPE ...............................60
3.1.2 Pressur-Temperature (P-T) Phase Diagram of InxGal-xN and Phase
Separation in InGaxN........................................ ..................................... 65
3.2 Quantum Calculation of Phase Separation in InxGal-xN ................................70
3.2.1 Boundary Passivation Method with Hydrogen .........................................70

4 CALCULATION OF THE CRITICAL THICKNESS OF InN ON GaN, A1N, Si,
A N D A 120 3 ......................................................... ................ 7 6

4.1 Calculation of Critical Thickness of InN by Matthews' Method. ........................77
4.2 Calculation of Critical Thickness of InN by van der Merwe's Method. ..............80
4.3 Calculation of Critical Thickness of InN by the Methods of Shen, Jesser, and
W ilsd o rf. ...............................................................................................................8 6

5 Indium Nitride (InN) GROWTH BY METAL-ORGANIC VAPOR PHASE
E P IT A X Y (M O V P E )......................................................................... ...................93

5.1. Indium Nitride (InN) Growth Optimization ................................ ............... 93
5.1.1. Substrate Selection ............................................. .......... .. ...... .... 94
5.1.1.1. Sapphire (c-A 1203 (0001)).................... ........................................ 96
5.1.1.2. Gallium Nitride (GaN/c-A1203 (0001))................ .............. ....97









5.1.1.3. Silicon (Si (111)) .................................................... 97
5.1.2. Substrate Preparation Procedure................................... .................. .....98
5.1.3. Metal-Organic Vapor Phase Epitaxy (MOVPE) Reactor........................98
5.1.4. Growth Chemistry and Conditions for InN Growth...............................99
5.1.5. Indium Nitride (InN) Growth and Optimization ...................................102
5.1.5.1. Influence of Growth Temperature.............................................. 102
5.1.5.2. Influence of Substrate N itridation................................................110
5.1.5.3. Influence of N /In R atio ........................................... ................. 115
5.1.5.4. Influence of Buffer Layer and Morphological Study.................... 118
5.1.5.5. Influence of Pressure .................................................................. 125
5.1.5.6. Optical and Electrical Properties............................127
5.1.5.7. Sum m ary ............................... .... ........ .. .............. .129
5.1.6. Indium Nitride (InN) Droplet Formation .............................................130
5.1.7. A nnealing Effect................. ................ ...... .. ....... .... .................136
5.2. Computational Fluid Dynamic Analysis of the Flow of NH3 and Proposed
Inlet Tube Modification to Improve Flow Pattern of NH3 .............................138
5.3. Inlet Tube M odification and Growth Results .......... .....................................145

6 CONCLU SION S ................................... .. .. ......... .. ............ 54

L IST O F R E FE R E N C E S ........................................................................ ................... 158

BIOGRAPHICAL SKETCH ............................................................. ............... 179
















LIST OF TABLES


Table pge

2-1. L attice constants of InN ............................................................................... ..... ..5

2-2. Properties of G aN and InN ........................................ .......................................6

2-3. Elastic constants of wurtzite InN at room temperature. .....................................7

2-4. Physical properties of InN ...... ........................... ........................................7

2-5. Carrier concentration and Hall mobility for the different growth methods ..............14

2-6. Comparison of interaction parameters calculated using various models with
experim ental data. .......................... ...................... .. .. .. .... ........... 22

2-7. Interaction parameters for various III-V ternary alloy systems.............................30

2-8. Properties of nitrogen precursors for MOVPE. ................... ................36

2-9. Structural properties of substrates. ........................................ ........................ 47

3-1. Reported reaction rate constants for TMIn decomposition. .....................................62

3-2. Species, phases, and thermodynamic properties included in the analysis of
M O V PE of InN ................ ..................................................... ........ .................64

3-3. Phases and species included in the analysis of MOVPE of InxGal-xN ....... ........ 67

3-4. Bond lengths for the calculation using HF-SCF.............. ..... .................74

3-5. Calculated total energy for three types of different bond length............................. 74

3-6. Calculated energies of InxGal-xN with the phase separation and without phase
separation ....................................................................... 75

4-1. Physical properties required for the calculation of the critical thickness of InN on
GaN, A1N, A1203, and Si substrates. ...................................... ............... 79

4-2. Calculated critical thickness of InN on GaN, A1N, A1203, and Si substrates using
M attew s' m eth o d .................................................................... ................ .. 7 9









4-3. Physical properties required for the calculation of critical thickness of InN on
GaN, A1N, A1203, and Si substrates. ...................................... ............... 82

4-4. Calculated shear moduli for InN, GaN, A1N, A1203, and Si materials ......................82

4-5. Calculated critical thickness of InN using van der Merwe's method ........................85

4-6. Lattice constant (A) of InN, GaN, A1N. ........................................... ...................88

4-7. Elastic constants c, and compliances s, of InN ......................................................89

4-8. Critical thickness (hc) calculated of InN using three different models....................92

5-1. Structural properties of InN, GaN, A1203, Si, and A1N substrates............................95

5-2. Range of growth conditions examined for growth of InN..................................... 101

5-3. Optimum growth temperature of InN on LT-GaN and LT-InN buffer layers on
various substrates. ....................................................... .. ........ .... 110

5-4. Comparison by ESCA of Si anneals..................................................... .. .......... 115

5-5. Optimum growth temperature of LT-InN buffer layer depending on A1203 (0001).
.............................................................................................. . 1 2 2

5-6. Root Mean Square (RMS) roughness for as-grown buffer layers and InN films....124

5-7. Grow th conditions for InN ............................................ ............................. 129

5-8. Optimum growth condition of InN for A1203 (0001), GaN/A1203 (0001), Si
(1 1 1). ........................................................................... 13 0

5-9. Density and velocity of NH3 at reactor wall and substrate.............. .................139

5-10. Reynolds number (Re) calculated in the inlet tube and in the reactor depending
on tem perature .................. ............................ .... ........ .. ........ .... 139

5-11. Typical values of FWHM depending on different reactor systems....................... 153

5-12. Reference data available for FWHM for MOVPE reactor...............................153
















LIST OF FIGURES


Figure p

1-1. Bandgap energies Eg of the semiconductor materials ......................................2

2-1. Lattice parameter for polycrystalline and single crystalline InN reported by
different groups. .................... ................ ................................ 5

2-2. Carrier concentration and hall mobility reported for undoped InN film grown in a
variety of technique is plotted against the calendar year............................. 12

2-3. Room-temperature Hall mobility as a function of InN thickness in InN films
grown by MBE, MOVPE, and MEE................................ ................13

2-4. Photoluminescence spectra for MBE grown InN...................................................15

2-5. Band gap energy for InN films as a function of carrier concentration....................16

2-6. Tetrahedral cells in a ternary III-V alloy semiconductor. .......................................20

2-7. Calculated phase diagram for the MBE deposition of InN using atomic N and
NH3 gases. There are three deposition modes: etching, droplet and growth..26

2-8. Free energy versus solid composition for a hypothetical semiconductor alloy
having a large positive enthalpy of mixing. Point A and B are the bimodal
points, and points C and D represent the spinodal points. ..........................27

2-9. Schematic liquid-solid pseudobinary phase diagram. .............................................27

2-10. Binodal (solid) and spinodal (dashed) curves for the InxGal-xN system,
calculated assuming a constant average value for the solid phase
interaction param eter ......................................................... ............. 28

2-11. Schematic illustration of the key CVD steps during deposition.............................38

2-12. Schematic of horizontal cold-wall MOVPE system.................... .................39

2-13. Schematics of horizontal hot-wall hydride vapor phase epitaxy chamber .............42

2-14. Schem atics of PE CV D .................................................. ............................... 43









2-15. Perspective views in (2 x2 x 1) unit cell: (a) along [0001] direction in a
rhombohedral unit cell; (b) along the (0001) direction in hexagonal unit
c ell. ........................................................ ................ 4 8

2-16. Common facets of sapphire crystals: (a) view down c-axis; (b) surface planes......48

2-17. Perspective views of Si along various directions: (a) [001]; (b) [011]; (c) [111]. ..49

2-18. Perspective views of wurtzite GaN along various directions: (a) [0001]; (b)
[112 0]; (c) [10 10] ................................................ ................... ... 50

2-19. Perspective views of zincblende GaN along various directions: (a) [100] (1 x 1 x 1
unit); (b) [110] (2 x 2 x 2 units); (c) [111] (2 x 2 x 2 units). ..........................51

3-1. Calculated P-T phase diagram for InN at X(In) = 5.31212x10-6, X(N) = 0.24998,
X(H) = 0.75000, X(C) = 1.59364x10-5 and V/III = X(N)/X(In) = 50,000..65

3-2. Relation between indium mole fraction (x) of InxGal-xN and the flow rate ratio of
the sum of group III source of TMI and TEG..............................................68

3-3. Calculated P-T phase diagram for In0.3Gao.7N at X(In)=1.87328x10-5,
X(Ga)=3.05276x10-5, X(N)=0.111, X(H)=0.8887, X(C)=2.39364x10-4
and the data points (0) are from the measurements observed by Matsuoka..68

3-4. Thermodynamically calculated miscibility gap of InxGal-xN grown by MOVPE
and the data points (AM) are from the measurements observed by Piner et
al..................... ....................................70

3-5. Flow chart of the HF-SCF procedure. .................................. ..................72

3-6. Structures used to compute the total energy for the InGal-xN vs. indium mole
fraction. ................................................................73

4-1. Schematic representation of the formation of misfit dislocations: (a) unstrained
lattice; (b) thickness of the film is less than hc; (c) thickness of the film is
greater than hc misfit dislocations are generated............. .................77

4-2. Model of epitaxial interface between two semi-infinite crystals resolved in a
sequence of misfit dislocations spaced at an average distance. ...............81

4-3. Homogeneous strain energy, Eh, for the 1st InN epilayer and dislocation energy,
Ed vs. the misfit, f on GaN substrate. ............ .............................................83

4-4. Homogeneous strain energy, Eh, for the 1st InN epilayer and dislocation energy,
Ed vs.the misfit,f on A1N substrate. .................................... .................84









4-5. Homogeneous strain energy, Eh, for the 1st InN epilayer and dislocation energy,
Ed VS. the misfit, fon A1203 substrate. .........................................................84

4-6. Homogeneous strain energy, Eh, for the 1st InN epilayer and dislocation energy,
Ed vs. the misfit, fon Si substrate........................................................ 85

4-7. Total energy Et, strain energy E,, interfacial energy E, vs. lattice constant of InN
for 1st epilayer InN on GaN substrate .......................................................89

4-8. Total energy Et, strain energy Es, interfacial energy E, vs. lattice constant of InN
for 1st epilayer InN on A1N substrate. ................................. .................90

4-9. Total energy Et, strain energy Es, interfacial energy E, vs. lattice constant of InN
for 1st epilayer InN on A1203 substrate ............. ..........................................90

4-10. Total energy Et, strain energy Es, interfacial energy E, vs. lattice constant of
InN for 1st epilayer InN on Si substrate............................................ 91

5-1. Schematic for the deposition of the (0001)//(0001), [1010 ]//[ 1210 ] GaN/A1203
sy stem ..................................................................... .. ......96

5-2. P lanes of Si (111) substrate. ................................................................................ 96

5-3. Image and schematic of horizontal, cold-wall MOVPE reactor system....................99

5-4. Indium Nitride (InN) growth sequence for each of the three substrates. ..............101

5-5. X-ray Diffraction (XRD) 0-20 scans for InN/LT-GaN on (a) A1203 (0001) at
N/In = 3000, T = 450, 550, 650, and 750 C........................................102

5-6. X-ray Diffraction (XRD) 0-20 scan for InN/LT-GaN on (a) A1203 (0001) at N/In
= 3000, T = 450, 550, 650, and 750 C. Pure In was removed by etching
w ith H C 1. .................................................................... 104

5-7. X-ray Diffraction (XRD) 0-20 scan for InN/LT-GaN on A1203 (0001) at N/In =
50,000, T = 530, 550, and 570 C. ........................................ ............... 105

5-8. X-ray Diffraction (XRD) 0-20 scan for InN/LT-InN on A1203 (0001) at N/In =
50,000 and T = 500, 530, and 550 C ...................................... .................105

5-9. Full Width Half Maximum (FWHM) of XRC for InN/LT-InN on A1203 (0001) at
N/In = 50,000 and T = 500, 530, and 550 C....................... ...............106

5-10. X-ray Diffraction (XRD) 0-20 scan for InN/LT-InN on GaN/A1203 (0001) at
N/In = 50,000, T = 500, 530, and 550 C............................. ... ................ 106

5-11. Full Width Half Maximum (FWHM) of XRC for InN/LT-InN on GaN/A1203
(0001) at N/In = 50,000, T = 500, 530, and 550 C. ................................. 107









5-12. X-ray Diffraction (XRD) 0-20 scan for InN/LT-GaN on Si (111), at N/In =
50,000, T = 500, 530, 550, and 570 C. ............. ................ ........ ....... 108

5-14. Growth rate of InN on various substrates (a) InN/LT-GaN on A1203 (0001) at
N/In = 3000, (b) for InN/LT-GaN on A1203 (0001) at N/In = 50,000, (c)
InN/LT-InN on GaN/A1203 (0001) at N/In = 50,000, and (d) InN/LT-InN
on Si (111) at N /In = 50,000 .............. ................................................ 109

5-15. Cross-sectional SEM micrographs of InN for 60 min growth at 530, 550, and
570 C, and N/In = 50,000 with LT-GaN buffer ................................110

5-16. X-ray Diffraction (XRD) 0-20 scan for InN/LT-InN (TLT-InN = 450 oC) on A1203
(0001) at N/In = 50,000, and TLT-InN = 450 C without and with
nitridation ........................................ ................ ............ ...... 111

5-17. X-ray Diffraction (XRD) 0-20 scan for InN/LT-InN GaN/A1203 (0001) at N/In
= 50,000, T = 530 C, and TLT-InN = 450, 500 C with and without
nitridation ........... .................... ........... .................. ............ 112

5-18. X-ray Diffraction (XRD) 0-20 scan for (a) InN/LT-InN on Si (111), at N/In =
50,000, T = 530 C, and TLT-InN = 450, 500 C with the nitridation and
without the nitridation. .......... ...... ...... .... ............. ............... 113

5-19. Electron Spectroscopy of Chemical Analysis (ESCA) spectra of Si 2p3 peak for
Si annealed at 850 C in 1.0 slm N2 (a) with 100% NH3 at 1.0 slm (b)
w without NH 3 ........ .......... ........................... ..... .......... 114

5-20. X-ray Diffraction (XRD) 0-20 scan at N/In=20,000, 30,000, and 50,000, T =
550 C for InN/LT-GaN on A1203 (0001) at N/In of 50,000..................115

5-21. Full Width Half Maximum (FWHM) of XRC for InN/LT-GaN on A1203 (0001)
at N /In of 50,000. ............. ... .. ....... .... ................... ...... ..... 116

5-22. X-ray Diffraction (XRD) 0-20 scan for InN/LT-InN on GaN/A1203 (0001) at, T
= 530 C, TLT-InN = 400 oC, and N/In = 30,000 and 50,000...................16

5-23. X-ray Diffraction (XRD) 0-20 scan for InN/LT-GaN on Si (111) at N/In =
20,000, 30,000 and 50,000, T = 530 C. ............................... ............... 117

5-24. Growth rate vs. N/In ratio for InN/LT-GaN on A1203 (0001), GaN/A1203 (0001),
and Si (111) at N/In = 6000, 9000, 12,000, and 15,000 with T = 550 C,
TMI = 0.26 sccm and NH3 = 1600-4000 sccm................................. ....118

5-25. X-ray Diffraction (XRD) 0-20 scan for InN/LT-InN on A1203 (0001) at N/In =
50,000, T = 530 C, and TLT-InN= 400, 450, and 500 C. ......................... 19

5-26. X-ray Diffraction (XRD) 0-20 scan for InN/LT-InN on GaN/A1203 (0001) at T
= 530 C, TLT-InN= 350, 400, 450, and 500 C, and N/In = 50,000..............120









5-27. Full Width Half Maximum (FWHM) of XRC for InN/LT-InN on GaN/A1203
(0001) at T = 530 C, TLT-InN= 350, 400, 450, and 500 C, and N/In =
50,000 .............. ...... .... ............... ............ ........ 120

5-28. X-ray Diffraction (XRD) 0-20 scan for InN/LT-InN on Si (111) at N/In =
50,000, T = 530 C, and TLT-InN = 400, 450, and 500 C......................121

5-29. X-ray Diffraction (XRD) 0-20 scan for InN/LT-InN on Si (111) at N/In =
50,000, T = 530 C, TLT-InN= 450 C, and t = 5, 15, and 30 min. ............121

5-30. X-ray Diffraction (XRD) 0-20 scan for InN/LT-InN (TLT-InN= 450 C) and
LT-GaN (TLT-GaN = 560 C) on A1203 (0001) at N/In = 50,000, T =
530 C (LT-InN buffer) and T = 550 C (LT-GaN buffer). .......................122

5-31. Full Width Half Maximum (FWHM) of XRC for InN/LT-InN (TLT-InN = 450
C) and LT-GaN (TLT-GaN = 560 C) on A1203 (0001) at N/In =
50,000, T = 530 C (LT-InN buffer) and T = 550 C (LT-GaN buffer).....123

5-32. Root Mean Square (RMS) roughness by ATM for (a) InN/LT-InN (T = 530 C,
TLT-InN = 450 oC), (b) InN/LT-GaN (T = 550 C, TLT-GaN = 560 oC), (c) as-
grown LT-InN (450 C), and (d) as-grown LT-GaN (560 C) on A1203
(0001) at N /In=50,000 .............................................................................124

5-33. X-ray Diffraction (XRD) 0-20 scan for InN/LT-InN and InN/LT-GaN on Si
(111) at N/In = 50,000, T = 530 C, TLT-InN = 450 C, and TLT-InN = 560
C ..............................................................................12 5

5-34. X-ray Diffraction (XRD) 0-20 scan for InN/LT-InN on GaN/A1203 (0001) at
N/In = 50,000, TLT-InN = 450 and 500 oC and T = 530 oC with the different
growth pressure of LT-InN .............................................. ............... 126

5-35. X-ray Diffraction (XRD) 0-20 scan for InN/LT-InN on Si (111) at N/In =
50,000, TLT-InN = 450 and 500 oC and T = 530 oC with two different
growth pressures for LT-InN .... ............ ...... ............................ 126

5-36. Photoluminescence for (a) InN grown on A1203 (0001) at T = 530 oC, TLT-InN =
500 oC, (b) InN grown on GaN/A1203 (0001) at T = 530 oC, TLT-InN = 400
oC, (c) InN grown on Si (111) substrate at TLT-GaN = 560 oC, T = 550 C
and N /In = 50,000............. .................................. .... ........ ............. 128

5-37. Carrier concentrations and mobilities of InN films grown with different growth
conditions at different characterization temperature using Hall
m easurem ent............................................................................................ 128

5-38. Carrier concentration and mobility of InN on Si (111) at different
characterization temperature using Hall measurement. ..........................129









5-39. Scanning Electron Microscopy (SEM) and EDS for the surface of InN/LT-GaN
on A1203 (0001) at N/In = 3000, 6000, 9000, 20,000, 30,000, and 50,000.131

5-40. Number density of indium droplets vs. N/In ratio depending on different N/In,
when InN was grown on A1203 (0001) at T = 550 C and N/In = 3000,
6000, 9000. .................................................................... 132

5-41. Percent (%) vs. indium droplet size depending on different N/In, when InN was
grown on A1203 (0001) at T = 550 C and N/In = 3000, 6000, 9000.........132

5-42. X-ray Diffraction (XRD) 0-20 scans for InN/LT-GaN on A1203 (0001) at N/In =
3000, T = 450, 550, 650, and 750 C before HC1 wet etching.................133

5-43. X-ray Diffraction (XRD) 0-20 scans for InN/LT-GaN on A1203 (0001) at N/In =
3000, T = 450, 550, 650, and 750 C after HC1 wet etching......................133

5-44. X-ray Diffraction (XRD) 0-20 scan for InN/LT-GaN on Si (111) at N/In =
3000, T = 450, 550, 650, and 750 C before HC1 wet .............................134

5-45. X-ray Diffraction (XRD) 0-20 scan for InN/LT-GaN on Si (111) at N/In =
3000, T = 450, 550, 650, and 750 C after HC1 wet etching..................134

5-46. Characterization result by AES for In droplets formed during the growth of InN
on A1203 (0001) with LT-GaN buffer layer before the HC1 wet etching
after the HC1 wet etching at N/In = 3000. ............................................. 135

5-47. Full Width Half Maximum (FWHM) of XRC of InN/LT-InN (450 C) on A1203
(0001) at T = 450 C in N2 flow with different annealing time (0, 10, 30,
60 and 90 m in)................................................................... ... ........ 137

5-48. Full Width Half Maximum (FWHM) of XRC of InN/LT-InN (400 C) on
GaN/A1203 (0001) at T = 450 C in N2 flow with different annealing
tim e (0, 10, 30, and 60 m in). ........................................... ............... 137

5-49. Schematic for three types of inlet tubes used for the Fluent simulation..............141

5-50. Flow of NH3 in the reactor with the current inlet tube. .......................................142

5-51. Flow of NH3 1 mm above the surface of substrate with the current inlet tube. ....142

5-52. Flow of NH3 in the reactor with the horizontally extended inlet tube...................142

5-53. Flow in the reactor with the vertical inlet tube................................................143

5-54. Flow ofNH3 1 mm above the surface of substrate with the vertical inlet tube.....143

5-55. X-ray Diffraction (XRD) 0-20 scan for InN/LT-InN on A1203 (0001) at N/In =
50,000, T = 530 C and TLT-InN = 450 C with different inlet tubes...........146









5-56. Full Width Half Maximum (FWHM) of XRC for InN/LT-InN on A1203 (0001)
at N/In = 50,000, T = 530 C and TLT-InN = 450 C with different inlet
tu b e s. ..................................................................... 14 6

5-57. Cross-sectional SEM for InN/LT-InN on A1203 (0001) at N/In = 50,000, T =
530 C, and TLT-InN = 450 C with the horizontal and vertical inlet tubes.. 147

5-58. X-ray Diffraction (XRD) 0-20 scan for InN/LT-InN on GaN/A1203 (0001) at
N/In = 50,000, T = 530 C and TLT-InN = 450 C with different inlet tubes. 147

5-59. Full Width Half Maximum (FWHM) of XRC for InN/LT-InN on GaN/A1203
(0001) at N/In = 50,000, T = 530 C and TLT-InN = 450 C with different
inlet tubes. .................... ................ .................................148

5-60. Cross-sectional SEM for InN/LT-InN on GaN/A1203 (0001) at N/In = 50,000, T
= 530 C and TLT-InN = 400 oC with the horizontal and the vertical inlet
tu b e s. ..................................................................... 14 8

5-61. Grazing Angle Incident X-ray Diffraction (GIXD) for InN grown on A1203
(0002) with the incident angle of 1 degree when the horizontal inlet tube
w as u sed ................................................................... 14 9

5-62. Grazing Angle Incident X-ray Diffraction (GIXD) for InN grown on GaN/A1203
(0002) with the incident angle of 1 degree when the horizontal inlet tube
w as u sed ................................................................... 150

5-63. Grazing Angle Incident X-ray Diffraction (GIXD) for InN grown on (a) A1203
(0002) and (b) GaN/A1203 (0002) with the incident angle of 1 degree
w hen the vertical inlet tube w as used....................................................... 151

5-64. X-ray Diffraction (XRD) 0-20 scan of InN/LT-InN/GaN/LT-GaN on Si (111) at
N/In = 50,000, T = 530 C and TLT-InN = 400 C for both horizontal and
vertical inlet tubes. ............................. .......... .................. ..... .......... 15 1

5-65. Full Width Half Maximum (FWHM) of XRC of InN/LT-InN on A1203 (0001)
at N/In = 50,000, T = 530 C for both horizontal and vertical inlet tubes.
The annealing test is performed at T = 450 C for 30 min.........................152

5-66. Full Width Half Maximum (FWHM) of XRC of InN/LT-InN on GaN/A1203
(0001) at N/In = 50,000, T = 530 C for both horizontal and vertical inlet
tubes. The annealing test is performed at T = 450 C for 30 min. .............152















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

INDIUM NITRIDE GROWTH BY METAL-ORGANIC VAPOR PHASE EPITAXY

By

Taewoong Kim

August 2006

Chair: Timothy J. Anderson
Major Department: Chemical Engineering

InN and In-rich compositions of InxGal-xN, have potential for a variety of device

applications including solar cells. This work addresses the growth of high quality InN by

metalorganic vapor phase epitaxy. To better understand the material a thermodynamic

assessment of the In-N-C-H system was performed to yield the In-N P-T diagram. In

addition, the InN critical thickness was calculated for several candidate substrates to

guide substrate selection. Furthermore, computational fluid dynamics was used to design

an improved reactor. A vertical NH3 tube design produced the lowest reported Q-28

rocking curve FWHM value of (574 arcsec) for InN grown on GaN/A1203 (0001). The

film surface was also mirror-like as judged by AFM (RMS roughness = 4.2 nm). The PL

peak energy of 0.82 eV was obtained for InN grown on Si, consistent with recent reports

of a considerably lower of bandgap energy.














CHAPTER 1
INTRODUCTION

During the last few years the interest in indium nitride (InN) has brought

remarkable attention due to its highly attractive inherent properties, such as high mobility

and high saturation velocity [Yam02 and Yam04b].

Epitaxial growth of InN films by metal-organic vapor phase epitaxy (MOVPE),

was first reported by Matsuoka et al. and Wakahara et al. in 1989 independently [Mat89,

Wak89]. In the 1990s, epitaxial growth of InN films was performed by several scientists

[Wak90, Yam94a, Yam94b, Guo95a, Guo95b, Uch96, Che97, Yam97a, Sat97a, Yam98a,

Yan99, Tsu99, Pan99, Yam99a]. These studies included the growth by MOVPE and

MBE on different substrates such as Si, GaAs, GaAsB, A1203 and GaP over a wide range

of growth conditions but had not shown any good results, however, no high quality films

were produced.

Because of the low decomposition temperature of InN (- 650 C), poor lattice

matched substrate, high equilibrium pressure of nitrogen, and the low cracking efficiency

of NH3 at the growth temperature, the growth of high quality InN film is challenging.

Since the bandgap energy of InN has recently been discovered to be 0.7

eV,[Wu02] the use of InN with GaN and A1N make it possible to extend the emission of

nitrided-based LEDs from ultraviolet to infrared regions. The bandgap energies of the

semiconductor materials are shown in Fig. 1.1. Alloying InN with GaN creates an InxGal_

xN active layer that is suitable for light emitting devices because InxGal-xN considerably









increases luminescence efficiency due to the localized energy states formed by alloy

composition fluctuations of the InxGal-xN [Nak92].


0 direct bandgap
6.0 II indirect bandgap

3.05.0 MgS

1.0 ZnS MgSe
GaN
P 3.0


C-i-
1.0 InP


3.0 4.0 5.0 6.0
Lattice Constant (A)


Figure 1-1. Bandgap energies Eg of the semiconductor materials.

InN was predicted to have the lowest effective wou mass for electrons among all III-

nitride semiconductors, which leads to high mobility and high saturation (drift) velocity.

The theoretical maximum mobility calculated in InN at 300K is 4400 cm V155

(GaN 1000 cm2 VSi ) [Chi94]. It was found that InN exhibits an extremely high peak

drift velocity of 4.2x107 cm/s [Be199]. Thus, InN is promising as a highly potential

material for the fabrication of high-speed high-performance heterojunction field-effect

transistors (FETs). The use of wurtzite InN would permit photonic devices in infrared and

much faster electronic devices, because it could induce higher mobility and high peak

(saturation) velocity than most other III-nitride based materials.

Single crystalline epitaxial InN films by MOVPE were first reported in [Mat98 and

Wak89]. The typical FWHM of single crystalline InN grown by MOVPE is 4000 5500

arcsec [Che97]. Significant improvement in the growth of InN film has been made by









MOVPE during the last few years. The average reported data on FWHM of the X-ray

Rocking Curve (XRC) is -2000 arcsec for single crystal InN [Yam01b, Yan02a, and

Yam04a] which may indicate that the highest crystalline quality of this material has not

been achieved yet. The high crystalline quality InN can reduce the leakage current and

extend the lifetime of the laser diodes (LDs) due to reduced dislocations. Therefore, the

growth of high quality crystalline InN is essential to obtain high performance devices.

The highest mobility and lowest background carrier concentration of InN by

MOVPE are reported to be 900 cm2/Vs and 5x101 cm-3, [Yam04b]. Better results were

achieved using molecular beam epitaxy (MBE), and the highest reported data on mobility

and lowest background carrier concentration of InN are 2050 cm2/Vs and 3.49x 107 cm3,

respectively [Lu02a].

In addition, InN has potential for highly efficient low cost solar cell. Yamamoto et

al. proposed InN for a top cell material of a two-junction tandem solar cell [Yam94a].

In summary, InN is a very attractive material for semiconductor device applications

and high structural quality, low defect density material with high mobility and low carrier

concentration has not been achieved to date. Therefore, more research is needed for the

improvement of crystalline quality, transport and optical properties of InN epitaxial films

by MOVPE.














CHAPTER 2
LITERATURE REVIEW

2.1 Indium Nitride (InN) and Indium Gallium Nitride (InxGai-xN) Properties

In this section, the fundamental properties of InN and InxGal-xN such as structural,

physical, electrical, and optical properties will be discussed. The understandings of these

properties are important when selecting a suitable substrate and buffer layer for obtaining

high quality InN and InxGal-xN films. Also, possible applications of InN and InxGal-xN

will be assessed.

2.1.1 Structural Properties

Lattice parameters of the wurtzite crystalline structure of InN was first reported as

a = 3.53 A and c = 5.69 A [Juz38]. However, the lattice parameter in the rf-sputtered

InN film measured by Tansley and Folsey, a = 3.548 A and c = 5.760 A [Tan86a],

showing a slight increase in the lattice parameter values, which also differs from the

lattice parameter measured in the rf-sputtered InN film by Kubota et al., a = 3.540 A and

c = 5.705 A [Kub89]. The crystalline quality of InN obtained by Kubota et al. was higher

than the other previously reported InN films and the lattice parameter is much closer with

the lattice parameter measured in the single crystalline InN film. Based on the recent

results reported by Davydov, the lattice parameter in the high quality single crystal

hexagonal InN film was reported to be a = 3.5365 A and c = 5.7039 A [Dav02a]. The

lattice parameters for polycrystalline and single crystalline InN reported by several

groups are plotted in Fig. 2.1 [Bhu03b]. Probable reasons for the variation in lattice

parameter are different crystalline quality and oxygen incorporation [Yam03].









pattern [Lim99]. At 2002, Bhattacharya et al. reported the observation of zincblende

phase in InN thin film grown by pulsed laser deposition (PLD) and measured a lattice

constant of 5.09+0.04A [Bha02]. The different lattice constants of InN were summarized

in Table 2-1.


3.58

3.56 Single crystal
S- Kubota X
3.54 S" X x
; Osamnura Tansley
SX etaL etal
4 3.52 Davydov
taL
at.
3.5 i i a t k
5.68 5.7 5.72 5.74 5.76 5.78
c axis (A)
Figure 2-1. Lattice parameter for polycrystalline and single crystalline InN reported by
different groups.

Table 2-1. Lattice constants of InN.
Structure a (A) c (A) References
Wurtzite 3.53 5.69 Zuda and Hahn
[Juz38]
Wurtzite 3.548 5.760 Tansley and Foley
[Tan86a]
Wurtzite 3.540 5.705 Kubota [Kub89]
Wurtzite 3.5365 5.7039 Davydov [Dav02a]
Zincblende 4.98 Strite [Str93]
Zincblende 4.98-5.04 Lima [Lim99]
Zincblende 5.09+0.04 Bhattacharya
[Bha02]
InxGaxN films were usually deposited on GaN buffer layers, because the lattice

constant of InxGaxN is closer to that of GaN than sapphire when the mole fraction of

indium (x) in InxGaxN is less than 0.3.

The indium mole fraction of InxGal-xN films is estimated from the lattice constant

along with the c axis measured by x-ray diffraction, assuming that the lattice constant









changes linearly with the indium mole fraction given (Eq. 2-1). In the calculation, aGaN=

3.189 A, CGaN= 5.178 A, ainN = 3.548 A and cInN = 5.7034 A [Nak02, Qia02, Yos91]. The

fundamental properties of GaN and InN are listed below in Table 2-2.



alnxGal-xN = alInN + (-x)aGaN


C In xGa l xN = X CInN + (1- X ) CGaN


(2-1)


Table 2-2. Properties of GaN and InN.
Fundamental properties of GaN


Wurtzite type:
Band gap energy

Temperature coeff.
Pressure coefficient
eV/kbar
Lattice constants
Thermal expansion

Thermal conductivity
Index of refraction

Zincblende polytype:
Band gap energy
Lattice constants
Index of refraction


Eg(300K)= 3.39 eV
Eg(1.6K)= 3.50 eV
dEg/dT=-6.0x 10-4eV/K
dEg/dT= 4.2x 103

a=3.189 A c=5.185 A
Aa/a= 5.59x 10-6 K
Ac/c= 3.17x 10-6 K
K=1.3 W/cmK
n (1 eV)= 2.33
n (3.38 eV)= 2.67

Eg(300K)= 3.2-3.3 eV
a= 4.52 A
n(3 eV ?)= 2.9


Fundamental properties of InN


Wurtzite type:
Band gap energy
Temperature coeff.
Lattice constantsa
Thermal expansion

Thermal conductivity
Index of refraction

Zincblende polytype:
Band gap energy
Lattice constant


Eg(300K)= 0.6-0.9 eV
dEg/dT=-1.8 x104 eV/K
a=3.537 A c=5.704 A
Aa/a&4 x 10-6K
Ac/c3 x 10-6K
K=0.8+ 0.2 W/cm K
n=2.9-3.05


Eg(300K)= 2.2 eV
a=5.09 A


[bBha02, aDav02a, Mor94]

In summary, the different lattice constants of InN obtained by several scientists

were discussed. The difference of lattice constants is thought to be caused by the

difference in the crystalline quality of InN. The lattice constant of InxGaxN can be

calculated by using the Vegard's law with the lattice constants of InN and GaN.









2.1.2 Physical Properties

Directly measured density of wurtzite InN is 6.89x103 kg m-3 at 25 C [Hah40]. A

comparable value of 6.81x103 kg m-3 has been estimated from X-ray data [Pea67]. The

cell volume, taken in conjunction with a molar mass of 128.827 g mol-1, yields densities

of (6.81+0.05)x103 kg m-3 and 6.97x10 kg m-3 for the wurtzite and zinc blende polytypes,

respectively. Bulk modulus has been calculated from first principles by a local-density

approximation [Cam90] and by a linear muffin-tin orbital method [Kub89], suggesting a

value ofB = 165 GPa.

The five distinguishable second-order elastic moduli in a hexagonal crystal are c11,

c12, c13, c33 and c44. Other researchers have utilized empirical and theoretical approaches

to calculate the thermoelastic properties of the wurtzite structure InN [She91, Kim96a,

Wri97, Mar98, Chi99]. Table 2-3 summarizes the room-temperature elastic constants

from both experimental and theoretical results. Estimates of the principal transverse and

longitudinal elastic constants ct and ci are given in Table 2-4.

Table 2-3. Elastic constants of wurtzite InN at room temperature.
Elastic Sheleg and Kim et al. Wright Marmalyuk Chisholm
constants Savastenko [Kim96a] [Wri97] et al. et al.
[She79] [Mar98] [Chi99]
C1 (GPa) 190 271 223 257 297.5
C12 (GPa) 104 124 115 92 107.4
C13 (GPa) 121 94 92 70 108.7
C33 (GPa) 182 200 224 278 25.05
C44 (GPa) 9.9 46 48 68 89.4
[WanOl]

Table 2-4. Physical properties of InN.
Property Value Ref Comments










Density (wurtzite)

Density (zinc blende)
Molar mass
Mol. Vol. (wurtzite)
Mol. Vol. (zinc
blende)
ct
Cl
Deformation potential
hcoTo

hiwLo


6.89x103 kg m-3
(6.81+0.05)x103 kg m-3
6.97x103 kg m3
128.827g mol-1
31.2 A3
30.9 A3
-2
4.42x1011 dyn cm2
12-2
2.65x1012 dyn cm2
7.1 eV
59.3 meV (478 cm-1)
57.1 meV (460 cm-1)
86.2 meV (694 cm-1)
89.2 meV (719 cm-1)


H. Hahn

S. Strite



V. W. Chin
V. W. Chin
V. W. Chin
K. Osamura
T. L. Tansley
K. Osamura
T. L. Tansley


Meas. by displacement
Various X-ray data
X-ray data

From lattice constants
From lattice constants
Estimate
Estimate
Estimate
Reflectance meas.
Transmission meas.
Est.-Brout sum rule
Est.-Brout sum rule


[Edg94], (reprinted from the Institute of Electrical Engineers with the permission of
INSPEC)

The piezoelectric constant has not been reported, but its dependence on the dielectric

constants Sr and e14 [Wol89] allows values of about 50 % of those found in AIN to be

inferred [Chi94].

Indium nitride has twelve phonon modes at the zone centre (symmetry group C6v),

three acoustic and nine optical with the acoustic branches essentially zero at k = 0. The

IR active modes are E1 (LO), Ei(TO), Ai(LO) and Ai(TO). A transverse optical mode has

been identified at 478 cm1(59.3 meV) by reflectance and 460 cm-1 (57.1 meV) by

transmission [Tan88]. In both reports the location of a longitudinal optical mode is

inferred from the Brout sum rule, giving respective values of 694cm-1 (86.1 meV) and

719cm-1 (89.2 meV).

In summary, the physical properties of InN films were briefly discussed, especially

the elastic constants used to calculate the strain energy and thus estimate the critical

thickness of InN film.


I









2.1.3 Electrical Properties of InN

2.1.3.1 Background Defects

As-grown InN is always n-type with a very high background carrier concentration.

There has been much speculation as to what species is responsible for the high

background donor concentration in InN. Potential candidates for such high background

donors are native defects, such as N vacancy or nitrogen antisite, and impurities, such as

ON, Siin, and possibly interstitial H.

According to the oldest and most common view, the nitrogen vacancy is the most

probable reason for n-type conductivity of InN. Tansley and Foley [Tan84b] had

speculated that the n-type behavior is caused by an antisite defect: N on an In site (Nin),

which they had suggested might be a double donor. Jenkins and Dow [Jen89] showed

that the native defect responsible for naturally occurring n -type InN is a nitrogen

vacancy. Another defect possibly responsible for the n-type character of InN is oxygen on

an N site, which is not a native defect but is nevertheless likely to be present in

significant concentration. It is most likely that every nitrogen vacancy donates a single

donor but possibly donates three electrons to the conduction band [Jen89]. Tansley and

Egan [Tan92a, Tan92b] have also speculated that the N vacancy might be the defect

responsible for natural n-type character of InN. There is a simple approach to how the

nitrogen vacancy contributes a donor in the as-grown InN film. The donor nature of the N

vacancy is constructed as a missing N atom surrounded by four indium atoms that

provide three valence electrons to complete the bonding octet with the five missing

electrons of nitrogen. Two of these three electrons would be donated to the conduction

band. Therefore, it has been believed that nitrogen vacancy is the dominant donor in the

as-grown InN film [YamOl, Yam02].









In contrast with the above views, there are also some theoretical and experimental

evidence, which argues against the nitrogen vacancy being responsible for the

background n-type conductivity. Stampfl et al. [StaOO] performed first-principles density-

functional calculation to investigate the electronic and atomic structure and formation

energies of native defects and selected impurities (O, Si, and Mg) in InN. Their

calculation showed that oxygen and silicon impurities act as donors and that they can

easily be incorporated during growth.

At 2002, Look et al. [Loo02] presented a rule to determine donor and acceptor

concentrations in degenerate InN. From a comparison with glow discharge mass

spectroscopy measurement and the developed theory, they suggested that a potential

candidate for the dominant donor in InN is H. However, the native defects also cannot be

completely ruled out.

As discussed above both theoretical calculation and experimental result give

conflicting views and opinions regarding the major reasons responsible for high n-type

conductivity of as-grown InN film. However, on the basis of the data available in the

literature, two major reasons can be concluded. One is native defects, mainly nitrogen

vacancy, and one is impurities, mainly oxygen.

2.1.3.2 Hall mobility and Electron Concentration in Undoped InN

The carrier concentrations and Hall mobilities reported for undoped InN films

grown by a variety of techniques are plotted against the calendar year in Fig. 2.2

[Bhu03b].

The growth methods are divided into five categories: molecular beam epitaxy

(MBE), metal-organic chemical vapor phase epitaxy (MOVPE), hydride vapor phase

epitaxy (HVPE), sputtering, and others, including electron beam plasma method, reactive









evaporation and pulsed laser deposition. Until the 1980s most of the InN films were

deposited using sputtering. The grown films were polycrystalline with a carrier

concentration scattered from 101 to 1021 cm-3 and Hall mobility from 20 to 250 cm2/Vs

with the exception of the results obtained by Tansley and Foley [Tan84a].

Tansley and Foley [Tan84a] attained a dramatic reduction of the carrier

concentration with very high electron mobility. A room temperature electron mobility of

2700 cm2/V s, which reached a maximum value of 5000 cm2/V s at 150 K, was

measured. These are the best electrical properties ever reported in InN. It should be

noted that the InN was a polycrystalline. Unfortunately, the InN film prepared by reactive

sputtering in other laboratories has not met these results of Tansley and Foley and has

universally high carrier concentration near 1020 cm-3 and constantly low electron mobility

of less than 100 cm2/Vs. The InN film grown by different techniques also showed the

high carrier concentrations and low electron mobility.

Sato [Sat97b] achieved a carrier concentration of 4x1019 cm-3 in the InN epitaxial

layer grown on sapphire substrate by plasma-assisted MOVPE in 1997. However, there is

no further improvement or report on the electrical properties of InN by plasma-assisted

MOVPE.

The significant improvements in conventional MOVPE grown InN films started

with the work of Yamamoto et al. and Pan et al. [Yam98a, Pan99] in which they reported

an electron concentration of 5x 1019 cm-3 with a Hall mobility of about 300 cm2/V s in the

InN film grown on sapphire substrate.












1021 L .




S10 v 102

E 101
10"17

i016 1 I 10


0: MBE 0 MOVPE
: HVPE V: Sputter k Others
I I h I






V
b
t 6 r


1970 1980 1990 2000 1970 1980 1990 2000
Calendar year Calendar year

Figure 2-2. Carrier concentration and hall mobility reported for undoped InN film grown
in a variety of technique is plotted against the calendar year.

Yamaguchi et al. [Yam99a] showed that using a GaN underlying layer increased

InN film thickness and significantly improve the Hall mobility. A Hall mobility of about

700 cm2/V s was obtained in the InN film grown on GaN even at an electron

19 3
concentration of 5x 1019 cm3. Yamamoto et al. [Yam98a, YamOl, Yam02] showed a

high NH3/TMI molar ratio and enhanced NH3 decomposition (by growth temperature,

atmospheric pressure growth, reduced flow velocity, etc.) significantly improved the

electrical properties of MOVPE grown InN film.

As a result, a carrier concentration in the order of 1018 cm3 and the electron

mobility of 730 cm2/Vs were reported. Recently, Yamamoto et al. [Yam04b] also

reported a carrier concentration of 5x10 1 cm-3 and the electron mobility of 900 cm2/V s

for the MOVPE grown InN film.

Laser-assisted MOVPE has the potential to decompose NH3 photolytically

independent of the substrate temperature [Bhu02a].










Lu et al. [LuOO] have obtained an electron concentration of 3 x1018 cm-3 with a Hall

mobility of 542 cm2/V s in the InN film grown by MEE (Migration Enhanced Epitaxy).

They also showed that the Hall mobility for both growth methods, MEE and MBE,

increases with film thickness. Similar thickness dependence in Hall mobility was also

observed in the MOVPE grown InN film [Yam99a]. The thickness dependence of the

Hall mobility is presumed to be caused by the reduced defect density away from the

lattice-mismatched substrate. Higashiwaki and Matsui [Hig02] found that there was an

immediate sharp increase in mobility up to a film thickness of 150 nm, beyond which it

almost leveled out. The room-temperature Hall mobility as a function of InN thickness in

the InN film grown by MBE, MPVPE, and MEE is shown in Fig. 2.3 [Hig02a].

Lu et al. [Lu02a] have achieved a carrier concentration in the order of 1017 cm-3 and

a mobility of more than 2000 cm2/V s for the thick InN film grown on HVPE grown on

bulk GaN template. The use of a buffer layer of AIN, GaN or InN seems to contribute to

the improvement of structural and electrical properties of MBE grown InN. The better

electrical properties in the MBE InN film compared with the MOVPE are believed to be

because the active nitrogen can be supplied independently of the growth temperature and

reduced impurity incorporation in the MBE growth.


200






1S 129)

10 W 10 150 2W 2" 3= 3so 4oo
S. "
ao


M U U
e .1

~ 01 o 1(Rf. f177)MOVPEi
0 6 100 1..i 200 2150 300 350 400
khN Ihckneaa (nn)


Figure 2-3. Room-temperature Hall mobility as a function of InN thickness in InN films
grown by MBE, MOVPE, and MEE.









Table 2-5. Carrier concentration and Hall mobility for the different growth methods.
Growth methods Carrier concentration Hall mobility References
(cm-3) (cm2/V s)
MOVPE 5x1018 -900 Yamamoto [Yam04b]
PA-MOVPE 4x1019 -Sato [Sat97b]
HVPE 1017 2000 Lu [Lu02a]
MBE 10171020 600-1200 Bhuiyan [Bhu02a]
MEE 3x1018 -542 Lu [LuOO]
Sputtering 10 1-1021 20-250 Bhuiyan [Bhu02a]


The typical range of carrier concentrations and mobilities for the different growth

methods including MOVPE, PA-MOVPE, HVPE, MBE, MEE, and sputtering was

discussed in detail and summarized in Table 2-5.

2.1.4 Optical Properties of InN

Until 2001, the measured bandgap of 1.89 eV has been commonly accepted for InN

[Tan86a]. However, a few groups recently showed by PL measurements that the band

gap energy of InN is in between 0.65 and 0.90 eV, [Dav02a, Dav02b, Dav02c, Wu02,

Tat02, Hor02, Sai02, Miy02] which is much smaller than 1.89 eV.

Evidence of a narrower band gap for InN was reported in 2001. Inushima et al. insisted

that the fundamental absorption edge of MBE grown InN layer lies around 1.1 eV, which

is much lower than the previously reported values [InuOl]. Davydov et al. reported a

band gap value of 0.9 eV for high quality MBE grown InN, studied by means of optical

absorption, PL, photoluminescence excitation (PLE) spectroscopy, as well as by ab initio

calculation [Dav02a]. Figure 2-4 shows photoluminescence spectra for MBE grown InN

sample which showed that the band gap of InN was much less than the previously

reported value (around 1.9 eV) [Dav02a]. They further studied in detail with different

high quality hexagonal InN films grown by different epitaxy methods. Analysis of optical

absorption, PL, PLE, and photoreflectivity data obtained on single crystalline hexagonal









InN film leads to the conclusion that the true band gap of InN is Eg ~ 0.7 eV [Dav02b,

Dav02c].

The larger band gap (-1.89 eV) cited in the literature may be due to the formation of

oxynitrides, which have much larger band gaps than that of InN. As can be seen in Fig.

2.5, the energy gap data less than 1 eV were obtained for single crystalline InN film with

a relatively low carrier concentration, while the larger values were mostly for

polycrystalline InN film [Bhu03a]. It should also be pointed out that the band gap

obtained from epitaxial films shows a remarkable dependence of carrier concentration,

which is different from the larger one obtained from polycrystalline films. Polycrystalline

films show a similar band gap (~ 2 eV) in spite of the wide range variation of carrier

concentration 1016-1021 cm-3


InN (b) TfK













Figure 2-4. Photoluminescence spectra for MBE grown InN.

As Motlan et al. [Mol02] reported, oxygen incorporation is one of the causes for the

large band gap energy. Therefore, the larger values may be related to oxygen

incorporation into grown InN because polycrystalline films can contain a high density of

oxygen atoms at their grain boundaries.
f \ WIt I* 1i


06 GA 1-D 1.1

Figure 2-4. Photoluminescence spectra for MBE grown InN.

As Motlan et al. [Mol02] reported, oxygen incorporation is one of the causes for the

large band gap energy. Therefore, the larger values may be related to oxygen

incorporation into grown InN because polycrystalline films can contain a high density of

oxygen atoms at their grain boundaries.













Tar reLey amd Folev a
Eg- 1.89+ 2.1 K)6f n' AL








IDaydo et al.

1016 1017 iOI8 1019 1021 1021 IJ22
Carrier comncetratica (cma )
Figure 2-5. Band gap energy for InN films as a function of carrier concentration.

Davydov et al. [Dav02c] showed that the sample with band gap in the region of 1.8-

2.1 eV contained up to 20 % of oxygen, much higher than for samples with narrow band

gap. It can be assumed that oxygen is responsible for a high concentration of defects.

Therefore, this increase of the band gap energy can be caused by formation of

oxynitrides, which have a much larger band gap than that of InN.

2.1.5. Indium Nitride (InN) andlindium Gallium Nitride (InxGal-xN) Applications

The latest progress in improving the InN film quality indicates that the InN film

almost meets the requirements for application to practical devices. Nowadays, the

bandgap energy of InN is known as 0.7 eV and thus InxGal-xN layers can be used as

absorber layers in tandem solar cells where the mole fraction of indium (x) is varied from

0 to 1 which tunes the bandgap from 0.7 to 3.4 eV. This energy range covers the

majority of the solar spectrum, therefore improving efficiency.

In addition to the tandem solar cells, InN can also be applied to LED and LD

similar to other III-V nitride compounds. Because the reported band gap value of InN is

about 0.7 eV, which is compatible with the wavelength of the optical fiber, another very









important potential application of InN, fabrication of high-speed LD and PD in the

optical communication system, is expected.

It is expected to be a highly promising material for the fabrication of high

performance high electron mobility transistor (HEMT). InN as a HEMT channel requires

a larger band gap barrier to induce and confine electrons. The significant lattice

mismatch between InN and GaN or AIN can result in a large piezo-electric charge, which

is very advantageous for HEMT applications. The strained InxGal-xN or InxAll-xN is also

a good choice as a barrier layer.

InxGal-xN is a very important compound semiconductor among III-V nitride

compounds because the InxGal-xN active layer emits light by the recombination of the

injected electrons and holes into this active layer. The addition of a small amount of

indium into the GaN was very important in obtaining a strong band-to-band emission

because GaN without the indium could not emit a strong band-to-band emission at RT.

This reason is considered to be related to deep localized energy states.

Currently, InxGal-xN is usually applied for the active layer in LEDs and LDs for

this characteristic of the deep localized energy states, which can facilitate the efficiency

of the band-to-band emission. For InxGal-xN-based LDs, however, the TDs (threading

dislocations) density had to be decreased to lengthen the lifetime by using the ELOG

(Epitaxial Lateral Overgrowth). For InxGal-xN-based LEDs, the lifetime of the LEDs is

more than 100,000 hours in spite of the large number of dislocations. This difference in

lifetime-behavior between LDs and LEDs is probably caused by the difference in the

operating current density in the two devices. The operating current density of LDs is

about one order higher than that of LEDs. Numerous studies have investigated the origin









of these defects, and their effects on the structural, optical, electronic, and morphological

properties of heteroepitaxial InxGal-xN layers [Chen06, Cho04, Jin06, Lil06].

2.2 Thermodynamic Analysis and Phase Separation in the InxGal-xN System

Thermodynamics give the guideline for the epitaxial growth process for all

techniques, including MOVPE, since epitaxial growth is simply a highly controlled phase

transition. A thermodynamic understanding of epitaxy allows the determination of alloy

composition as well as the solid stoichiometry.

The thermodynamics of mixing of semiconductor alloys (III/V, II/VI, and IV/IV)

determines many characteristics of the growth process as well as the properties of the

resultant materials. For example, Thermodynamic factors may limit the mutual solubility

of the two (or more) components of an alloy. When the sizes of the constituent atoms are

sufficiently different, miscibility gap exist. In addition to solid-phase immiscibility in

important alloys systems such as GaInAsP and InxGal-xN, this size difference also leads

to microscopic structures far different than the random, totally disordered state normally

expected for alloys. Both miscibility gaps and deviations from a random distribution of

the atoms constituting the lattice affect the electrical and optical properties of

semiconductor alloys in ways that are extremely important for many types of devices.

The thermodynamics of the surface must also be considered in any effort to

understand the growth processes as well as the characteristics of the materials produced

epitaxially.

The basic goal of thermodynamics, as applied to epitaxy, is to define the

relationship between the compositions of the various phases in an equilibrium system at

constant temperature and pressure.









2.2.1 Thermodynamic Models in Solid Solution

2.2.1.1 Regular Solution Model

The term regular solution was first used by Hiderbrand to describe a class of

solutions that are nonideal but consist of a random arrangement of the constituents. The

term has since come to designate a more restricted, semiquantative model for the

calculation of the free energy of mixing of multicomponent systems. Two additional

assumptions are (1) interactions between the constituent atoms occur only pair-wise- that

is, only between nearest neighbor pairs, and (2) the atoms reside on a lattice with each

atom surrounded by Z neighbors. The bond energies are commonly thought of as being

the sum of "chemical" energies, frequently related to charge transfer due to differences in

electronegativity, and "strain" energies related to distortions in the lattice due to

differences in the sizes of the constituent atoms. The enthalpy of mixing is obtained by

summing nearest-neighbor bond energies

AHM = x(1 x). Q (2-2)

where interaction parameter (Q) is


S= ZN H .(H + Hc) (2-3)


where N is Avogadro's number.

2.2.1.2 Bonding in Semiconductor Solid Solutions Model

Traditionally, semiconductor alloys have been described in terms of the virtual

crystal approximation (VCA), where the lattice on which the atoms are situated is

uniform; that is, the individual bonds are dislocated to form a microscopically uniform

solid solution. This was believed to be dictated by the accuracy with which Vegard's law

describes the linear dependence of lattice constant on solid solution. However, it has been










recently found that the virtual crystal model for semiconductor solid solution is in fact not

a good description of the solid. The bond lengths in the alloy more nearly resemble the

bond lengths in the pure binary compounds than the average values anticipated from the

virtual crystal model [Ega02]. The valence force field (VFF) model can be used to

explain this behavior [Pos02]. The interactions between atoms are considered to be due to

entirely to strain (i.e., the stretching and bending of the bonds). The simplest form of the

VFF calculation for an alloy AC-BC assumes that the lattice is composed of five types of

hetrahedra shown in Fig.2.6 [Ich86, Kea66]. It is known that the bonding in a

semiconductor is due to long-range effects, particularly the distributed electron energy

stated in the solid. The same valence electrons that determine the optical and electrical

properties of the semiconductor also determine the bonding, as well as the elastic

constant. This is contrary to the basic assumptions of the regular solution model, which

cannot be expected to provide a physically accurate, predictive description of the

enthalpy of mixing in semiconductor alloys.




Ai B
I I
-A A--- ----

TYPE 0 TYPE 1 TYPE 2






TYPE 3 TYPE 4

Figure 2-6. Tetrahedral cells in a ternary III-V alloy semiconductor.









2.2.1.3 Delta Lattice Parameter (DLP) Model for Enthalpy of Mixing

Important information needed for the calculation of solid-solid, solid-liquid, and

solid-vapor phase equilibriums is the heat of mixing in the solid, AH'. This coupled with

the assumption of a random distribution of constituents on their respective sublattices

allows the calculation of the free energy of mixing of the solid alloy. Several researchers

have suggested that the bonding energy in semiconductors is linearly related to the

bandgap [May90, Pan98, Sun94, and Ho96]. The work of Phillips and Van Vechten

suggested that the average band gap should be used in this relationship. Since it varies as

a2 5 in semiconductors that are nearly covalent such as the III-V compounds, JH"t, which

is used as a measure of bonding energy, might be written



AHt = K a 25 (2-4)



Considering the zero of enthalpy to be infinitely separated atoms, the interaction

parameter can be calculated from the enthalpy of mixing at x = 1/2, yielding




Q" = 4K a + B 25 a5) a99K (aA (2-5)
2 2 (+a B )

Using Vegard's law to obtain the lattice constant at x=0.5. The value of K was

obtained by making a least-square fit of Equation (Eqn.2-5) to available experimental

values of Q' that are listed in Table 2-6. The DLP calculation also appears to be quite

accurate for the III/V nitride alloys. A striking feature of the DLP model is that the

interaction parameter, hence the enthalpy of mixing, is always positive.









Table 2-6. Comparison of interaction parameters calculated using various models with
experimental data.
Alloy QS Qs Qs Qs Phase
(exp)a (DLP) (VFF) (Mod VFF) Separation
(kcal/mol) (kcal/mol) (kcal/mol) (kcal/mol)
AlGaN 1.19 1.34 0.87
AlInN 17.45 18.10 11.44
GaInN 9.60 9.62 5.98 Yes
A1PN 19.68 60.79 36.56
A1AsN 57.93 85.33 53.42
GaPN 23 28.90 42.43 27.38 Yes
GaAsN 42.78 59.09 36.84 Yes
InPN 19.68 29.09 16.33 Yes
InAsN 26.71 39.14 21.87 Yes
[Str99]

2.2.1.4 Strain Energy Model

In the traditional regular solution model, the uniformly positive values of enthalpy

of mixing strongly suggest that the enthalpy of mixing is due to strain, rather than

chemical factors. The mixing enthalpy can also be estimated using the simplified VFF

model. The solid is considered to be made up to identical tetrahedra (Fig.2.6) with the

position of the central atoms, located on the sub-lattice with no mixing, allowed to relax

to the position giving the lowest strain energy, considering both stretching and bending

distortions. The strain energy due to the stretching and bending of the bonds in each type

of tetrahedron is summed over the five types of tetrahedral weighted by the distribution

probability (random arrangement was not assumed in reference [Ich86]). The two terms

are coupled and must be solved simultaneously [Ich86]. This approach allows a

calculation of the free energy of mixing. There are two major drawbacks to the simple

forms of the VFF model described here. First, when the lattice is assumed to be made up

of tetrahedral where the corner atoms take the VCA positions, one of the sublattices is not

relaxed. This causes a significant overestimation of the total strain energy. Second, the









difference in energy between the several tetrahedral types is much greater than kT for

many types of tetrahedrals. Taking into account the effects of the resulting short-range

order (SRO) makes the calculation of the mixing enthalpy difficult, since it couples the

two factors [Ich86]. These problems can be surmounted by considering a large ensemble

of several hundred atoms with the positions of each allowed to relax while maintaining a

relatively simple calculation by considering only the dilute limit, where the effect of the

SRO is negligible [Sch91, Ho96]. This approach was developed specifically for dealing

with systems with very low solubility limits, in particular for the solubility of the very

small N atom in conventional III/V semiconductors such as GaAs, InP, GaP, and so forth.

2.2.1.5 First-Principal Models

Advances in fundamental insight for the energy of a semiconductor lattice and the

methodology of solving mathematical problems of large matrices have been achieved

recently due to the availability of high-powered computers.

These achievements can make possible the first-principle local density self-

consistent total energy minimization calculations in semiconductor alloy systems

[Zun94]. Using these quantum mechanical calculations, the thermodynamics of

semiconductor solid solutions can be calculated without any of the extreme simplifying

approximations necessary to obtain simple analytic models.

The total energy minimization calculations are based on the entire complex band

structures. The results from such calculations are included in Table 2-4. The mixing

enthalpies have also been calculated for InxGal-xN, InAIN, and AlGaN alloys using a

pseudopotential perturbation approach [Ito97].









2.2.2 Thermodynamic Analysis of InN

Koukitu and Seki have performed a thermodynamic analysis of the MBE growth of

III-nitrides [Kou97a]. The equilibrium partial pressure and the growth rate were

calculated for input V/III ratio, input partial pressure of group III elements and growth

temperature. A summary of their calculation results as a phase diagram for the

deposition, indicating etching, droplet formation and growth regions are shown in the

Fig. 2.6. The chemical reaction, which connects all species at the substrate surface, is

In(g) + N(g) = InN(s). (2-6)

The equilibrium equation for the reaction is as follows:

K 1=I/(PI+PN). (2-7)

From the conservation constraint we have

P -PI= PN PN, (2-8)

where PO, and PO are the input partial pressures, which are obtained from the

incident beam flux, and Pin and PN are the equilibrium partial pressures. Equation (2-8)

expresses that the deposition occurs in the ratio of 1: 1 for In and N. The equilibrium

partial pressures at the substrate surface can be obtained from the solution of the above

simultaneous equations. The value of the equilibrium constant was obtained from the

literature [Kou97a]. The corresponding free energy to the chemical reaction (Eq.2-6)

used in the analysis is as follows:

AG(kcal/mol)= (-1.764 x 102)+ 3.067 x 102/T

+ (-1.451 x 10-3) T x In(T)+7.909 x 10-2 XT

+ 3.883 x 10-" xTx T. (2-9)









The calculation for the MBE growth technique using an NH3 source was performed

in a similar manner, using atomic nitrogen. The chemical reaction is

In(g)+ NH3(g)= InN(s) + (3/2)H2(g). (2-10)

In NH3 case, they introduce a, the molar fraction of decomposed NH3, into the

calculation as follows:

NH3(g) z (1-a)NH3(g) + (1/2)N2(g) + 3(1/2)H2(g) (2-11)

The value of a is assumed appropriately as that of MOVPE growth [Kou96],

because it is difficult to know the exact value. The equilibrium partial pressure and the

growth rate were calculated for input V/III ratio, input partial pressure of In, and growth

temperature. In the growth of InN, they conclude that three deposition modes, i.e.,

etching, droplet formation and growth regions, appear in the temperature range from 500

to 900 OC. The temperature suitable for the InN growth is predicted to be from 600 to 700

C with V/III > 1, which is essential in the MBE growth. However, the experimental

growth temperature is much lower than this theoretical prediction, and almost

experiments have been done in the temperature range from 450 to 550 C. They also

reported that there is a difference between the atomic nitrogen and the NH3 source as

shown in the corer of diagram (Fig. 2.7) where the etching region appears [Kou97a]. In

the case of the atomic nitrogen source, the etching region appears constantly at the region

where the input V/III ratio and the input P,0 are low value. On the other hand, in the case

of the NH3 source, it appears at the region where the V/III ratio is high and the input Pn

is low. They concluded that this is due to the decomposition of NH3: when NH3 is

decomposed, H2 gas is produced, and the produced H2 drives Eqn. (2-10) to the left hand.

Consequently, the deposition moves into the etching mode due to the increase in H2.










Atom ic N rIVHE

10-E







N H 10-2 e .. .........t...... e_
,Soo
-- -- -



10-2 "0-




1, ( T-orr > V k'wr )
10wt Dmp~et Et, 1r i
OirXvtt zup'laL HB IFt4hilnR


Figure 2-7. Calculated phase diagram for the MBE deposition of InN using atomic N and
NH3 gases. There are three deposition modes: etching, droplet and growth.

2.2.3 Phase Separation in InxGal-xN

The large positive enthalpy of mixing for systems with a large lattice mismatch can

overwhelm the negative entropy of mixing for temperatures below the critical

temperature. This results in a free energy versus composition curve shown schematically

in Fig. 2.8, with an upward bowing in the center [Str99]. This dictates that at equilibrium,

a random alloy with composition between points A and B will decompose into a mixture

of two phases. Two other important points in the G versus energy curve shown in Fig. 2.8

are the inflection points lying between A and B. Between these two points the solid

solution is unstable against an infinitesimal fluctuation of composition. The spinodal

appears on the T-x phase diagram, as indicated in Fig. 2.9 [Str99]. In the pseudobinary

phase diagram, the boundary of the unstable region is defined by the locus of (d2G/dx2),p

= 0 [25], called the spinode. Inside this region, the solid can decompose "spoinodally,"

with no energy barrier.




















Solid Composition (x)

Figure 2-8. Free energy versus solid composition for a hypothetical semiconductor alloy
having a large positive enthalpy of mixing. Point A and B are the bimodal
points, and points C and D represent the spinodal points.

The growth of InxGal-xN alloys has proven to be extremely challenging, mostly due

to the trade-off between the epilayer quality and the amount of InN incorporation into the

alloy as the growth temperature is changed. Growth using high temperatures of

approximately 800C, typically results in high crystalline quality but the amount of InN

in the solid is limited to low values because of the high volatibility of N over InN. Ho and

Stringfellow performed a theoretical calculation of the enthalpy of mixing, the solid

phase interaction parameter, and the extent of the miscibility gap for InxGal-xN alloy

system using a modified valence-force-field (VFF) model calculation where the lattice is

allowed to relax beyond the first nearest neighbor [Str97].


Figure 2-9. Schematic liquid-solid pseudobinary phase diagram.










The VFF model, itself, is found to overestimate the total strain energy of a ternary

system due to the constraint that only one of the two sublattices is allowed to relax. The

calculation of the enthalpy of mixing or the interaction parameter in III-V system has

been a topic of interest for nearly twenty-five years. In 1972 Stringfellow developed the

semi-empirical delta-lattice-parameter (DLP) model, which is found to yield surprisingly

accurate interaction parameters for a wide range of III-V alloys knowing only the lattice

constants of the binary constituents. The temperature dependence of the bimodal and

spinodal lines in the InxGal-xN system was calculated using a modified VFF model. The

strain energy is found to decrease until approximately the sixth nearest neighbor, but this

approximation is suitable only in the dilute limit. Assuming a symmetric, regular

solution-like composition dependence of the enthalpy of mixing yields an interaction

parameter of 5.98 kcal/mole and a critical temperature for the phase separation of 1250

C (Fig. 2.10) [Ho96]. At a typical growth temperature of 800 C, the solubility of indium

in GaN calculated to be less than 6 %. The miscibility gap is expected to represent a

significant problem for the epitaxial growth of these alloys [Ho96].


1200 -
1000-
P 800-

E 400 o
I-
200
0 I -I-*- -
0 0.2 0.4 0-6 0.8 1
GaN X1 InN


Figure 2-10. Binodal (solid) and spinodal (dashed) curves for the InxGal-xN system,
calculated assuming a constant average value for the solid phase interaction
parameter.









Singh et al. reported the growth of InGaN thick (0.3-0.4 [tm) films and InxGal-xN

/GaN double heterostructures by MBE at the substrate temperatures 700-800C. X-ray

diffraction and optical absorption studies showed that the phase separation of InN of

InxGal-xN thick films occurred with x > 0.3. On the other hand, InxGal-xN/GaN double

heterostructures showed no evidence of phase separation. These observations were

accounted for using Stringfellow's model (DLP model) on phase separation, which gives

a critical temperature for miscibility of the GaN-InGaN system equal to 2457 K. The

maximum value of the critical temperature (Tc) above which the InN-GaN system is

completely miscible can be computed from Stringfellow's equation for a binary system

(Eq.2-12) [Sin97].


T = 8.75K (Aa)2 (2-12)
4R 4


Where Aa is the difference in the lattice constants of GaN and InN, a is the

average lattice of GaN and InN, and R is the gas constant. K is the proportionality

constant between atomization enthalpy (bonding energy). Phase separation in any alloy

requires long-range diffusion and thus a correlation should exist between phase

separation and a time required for the growth of the film. They believed this is one of the

reasons for the non-observable phase separation in GaN/ InxGal-xN /GaN double

heterostructures with thin InxGal-xN layers. Strain associated with thin InxGal-xN

quantum wells could also stabilize alloys against phase separation.

Wakahara et al. calculated the compositional imhomogeneity in InxGal-xN by using

a theoretical estimation of the interaction parameter based on DLP model [Wak97]. Table

2-7 summarize the lattice mismatching, the interaction parameter (a) and the critical









temperature of spinodal decomposition, which is denoted as Tc = a /2R, for the III-V

ternary alloy system.

It can clearly be seen that the nitride alloys including the InN have a very large

interaction parameter thus, the critical temperature of the spinodal decomposition also

becomes very high. It is expected that the immiscibility of the InxGal-xN alloy is very

strong. The critical temperature of the spinodal decomposition defined at the composition

x = 0.5 is much higher than the typically used growth temperature of 800 C. The

evidence of the phase separation in the InN containing nitride alloys was resulted in

[Mor94]. Recently, Koukitsu and Seki [Zun94] reported compositional inhomogeneity

based on a thermodynamic analysis of the vapor-solid interface. They predicted that the

compositional inhomogeneity of InxGal-xN increases with an increase of the growth

temperature and in the partial pressure of the hydrogen but decreases with V/III ratio.

Table 2-7. Interaction parameters for various III-V ternary alloy systems.
III-V ternary Lattice Interaction parameter Critical temperature
alloy system mismatch a (DLP model) Tc (K)
Aa/a (%) (cal/mol)
A1As-GaAs 0.159 0 0
GaAs-InAs 6.92 2815 709
AlP-GaP 0.239 0 0
GaP-InP 7.39 3630 914
GaP-GaN 18.9 28900 7276
AIN-GaN 2.93 931 233
AIN-InN 13.38 17300 4338
GaN-InN 10.46 10300 2583
[Wak97]

The resulting strain in the layers could lead to deviations from homogeneity of the

sublattice. Zunger and Mahajan have reviewed several observations, which indicate that

when the tetrahedral radii are different, two types of structural variations are observed:

phase separation and atomic ordering. The difficulties in InxGal-xN growth are mainly









due to (a) very high equilibrium vapor pressure (EVPs) of nitrogen over InN and (b) a

large lattice mismatch (11 %) between InN and GaN. The lattice mismatch between InN

and GaN (due to the very different tetrahedral radii) results in highly strained InxGal-xN

alloys. Therefore, at relatively low growth temperature (650-800 C), phase separation is

a major concern. The majorities of the III-V ternary and quaternary alloys are predicted

to be thermodynamically unstable and show a tendency towards clustering and phase

separation. The phase separation and ordering phenomena in InxGal-xN alloys with MBE

was studied by Doppalapudi et al. [Kel98]. The ordering parameter was found to increase

with the growth rates of the films, a result which is consistent with the notion that

ordering is induced at the surface of the growing films where it is thermodynamically

stable and is then subsequently "frozen in" during further growth. Phase separation was

found to be essential for films with high indium content (> 25 %), while ordering was

noticeable for films with small indium content (< 10 %). This competition between the

two phenomena is consistent with the proposal that lattice strain is the driving force for

both. The effect of elastic strain in epitaxial InxGal-xN layers coherently grown on GaN

wafers on spinodal decomposition of the ternary compound was examined. The effect

results in considerable suppression of phase separation in the strained InxGal-xN layers.

The elastic strain effect is predicted to lower the critical temperature. This effect does

work only if the relaxation of strain in the epitaxial layer is not yet started.

In summary, the results obtained by the several models used in the thermodynamic

analysis about InN and InxGal-xN were reviewed. For InN, the thermodynamic result

showed that the growth region is and the stable InN growth can be achieved at high V/III

ratio. For InxGal-xN, it was found that phase separation commonly occurs and that the









critical temperatures could be calculated through the interaction parameter and these

critical values depend on the chosen model. The maximum mole fraction of indium

incorporated into InxGal-xN, which depends on the value of elastic strain of InxGal-xN

was studied. From this result, it can be suggested that the maximum mole fraction of

indium incorporated into InxGal-xN can be increased by decreasing the strain energy of

InxGal-xN.

2.3 Indium Nitride (InN) and Indium Gallium Nitride (InxGal-xN) Growth
Challenges

There are several problems to be overcome for high crystalline InN and InxGal-xN

film growth. These problems are narrow region of growth temperature due to the low

decomposition temperature, low cracking efficiency of NH3, no suitable nitrogen

precursors to improve the decomposition efficiency of NH3, and carrier gas. These

problems which occur during the growth of InN and InxGal-xN are briefly discussed in

this part.

2.3.1 Growth Temperature and V/III Ratio

The growth of InN is the most difficult among the III-nitrides because the

equilibrium vapor pressure of nitrogen over the InN is several orders higher than A1N and

GaN [Amb96]. Because of the low InN dissociation temperature and high equilibrium N2

vapor pressure over the InN film [McC70], the growth of InN requires a low growth

temperature. Due to the low (-550 C) growth temperature, the MOVPE growth of InN is

thought to be restricted by a low decomposition rate of NH3. Although a higher growth

temperature is expected to result in a higher decomposition rate of NH3, it can also result

in thermal decomposition (thermal etching) of the grown InN. On the other hand, the

growth at a low temperature (lower than 400 C) is dominated by the formation of









metallic indium droplets due to the shortage of reactive nitrogen. Epitaxial growth at low

temperature becomes impossible due to reduced migration of the deposited materials.

Therefore, the region suitable for the deposition of InN is very narrow.

Koukitu et al. carried out a thermodynamic study on the MOVPE growth of III-

nitrides [Kou97b]. They pointed out that a high input V/III ratio, the use of an inert

carrier gas, and a low mole fraction of the decomposed NH3 are required for the growth

of InN. Experimental results also match well with the first two points (high input V/III

ratio and the use of inert carrier gas) but are not clear on the last point (a low mole

fraction of the decomposed NH3).

High input V/III provides sufficient amount of reactive nitrogen, since the NH3

decomposition rate is low at low growth temperature. NH3 is decomposed

thermodynamically into H2 and N2 with low decomposition efficiency at temperature

higher than 300 C, which results in the increase of H2 partial pressure [this sentence does

not make sense, there is low decomposition efficiency at T < 300oC, and at higher

temperatures the H2 partial pressure increases]. Thus too high ratio of V/III significantly

prevents the growth of InN and leads to the etching of InN due to the increase of H2

partial pressure (Eq. (2-10)). A suitable region of V/III ratio and growth temperature is

required for the high quality InN growth without indium droplets formation during the

growth.

The main problem in growing InxGal-xN has been the phase separation at high

indium incorporation (x (In) > 0.3) due to the very high equilibrium vapor pressure of

nitrogen over InN. The compositional control has only been achieved for relatively low

growth temperature, up to 650 C. However, crystal quality is not good because the









decomposition rate of NH3 is very low below 1000 C. Yoshimoto et al. [Yos91]

obtained the relatively high-quality InxGal-xN using a high temperature (800 C) and a

high indium flow rate. In that case of the temperature growth (650-800 oC), the phase

separation may occur because of the large mismatch (11 %) between InN and GaN. The

lattice mismatch between InN and GaN (due to the very different tetrahedral radii) results

in highly strained InxGal-xN. Therefore at the growth temperature of 650-850 oC, phase

separation is still a major concern.

2.3.2 Nitrogen Source

InN and InxGal-xN are typically grown by MOVPE using conventional group III

precursors such as tri-methyl indium and tri-ethyl indium with NH3 as the active nitrogen

source. However, InN and InxGal-xN are relatively difficult to produce with the high

quality required for minority carrier devices due to high equilibrium N2 vapor pressure

and the low decomposition efficiency of NH3 as discussed earlier.

NH3 is almost stable even at 1000 C and decomposes only 15 % at 950 oC, even

when catalyzed by GaN [Che91]. The combination of high growth temperatures and high

nitrogen volatility leads to high concentrations of N vacancies in GaN and InN. This is

often cited as the reason that the GaN and InN epitaxial layers are n-type.

Solution of this problem will probably require N precursor to give the high active

nitrogen reduction at the low growth temperature. Hydrazine (N2H4) is an attractive N

precursor because it contains no carbon atoms to be incorporated into the solid, and the

hydrogen atoms are potentially beneficial for removal of the alkyl radicals (from the

group III precursors) from the surface. It decomposes at temperatures as low as 4000 C,

considerably lower than temperatures required for NH3 because of the weaker N-N bond

[Gas86], thus making it suitable for growth at temperatures well at the current growth









temperature (-550 C) and at low V/III ratio as low as 10 which prevents the waste of N

precursor. It also has a favorable vapor pressure of approximately 10 Torr at 180C, as

indicated in Table 2-8. However, hydrazine is a toxic material, a rocket fuel, and

explosive. When hydrazine is used in III-V nitride growth with the trimethyl-group III

alkyls, the adduct formation between hydrazine and the group III precursors was

observed. The layers were found to be contaminated with both oxygen and carbon. The

hazard associated with the toxicity and explosiveness of hydrazine makes its use in a

production environment unlikely.

Unsymmetrical dimethylhydrazine (H2N2 (CH3)2, 1,1 DMHy) is a considerably

safer alternative to hydrazine. It has a vapor pressure of 157 Torr at 25 C and pyrolyzes

at temperatures considerable lower than for NH3. However, relatively high (>1019 cm-3)

levels of oxygen and carbon were observed, both of which are associated with the use of

DMHy.

A potentially less hazardous precursor, phenyl hydrazine, has also been explored

[Jon95]. However, the vapor pressure of 0.03 Torr at room temperature is far too low to

be acceptable.

Hydrogen azide, or hydrazoic acid (HN3) has also been successfully used for

MOVPE growth in a low-pressure reactor [Cht97].This precursor is attractive because it

has a high vapor pressure (the boiling temperature is 37 C) and decomposes at

approximately 300 C to yield HN radicals with two dangling bonds, a potentially good

source of atomic nitrogen, and N. However, it is highly toxic and potentially explosive.

The N precursor tertiary-butylamine ((C4H9)NH2 or TBAm) has a convenient vapor

pressure of 340 Torr at 25 C, a low toxicity, and is stable. However, the use of TBAm









for the growth of GaN has proven unsuccessful [Rus96, Bea97]. They observed no GaN

deposition but rather deposition of a layer consisting mostly of carbon [Bea97].

In summary, several candidates for nitrogen sources were reviewed and all of them

have some problems such as toxicity, explosion, low decomposition efficiency, and

contamination. NH3 has been still widely used as a nitrogen source in spite of low

decomposition efficiency. The design of the optimum nitrogen source for the growth of

III/V nitrides is still required even though it is tricky.

Table 2-8. Properties of nitrogen precursors for MOVPE.
Melting Boiling Vapor Pressure
Precursor Point Point a b, K P (Torr)/T(C)
(C) (C)
NH3 -77.7 9.9974 31.211
N2H4 1.3 10/18
MMHy 113 49.7/25
1,1 DMHy 157/25
N2H3(C6H5) (phenyl 8.749 3.014 0.03/23
hydrazine) 37
HN3 (hydrogen azide) -67 45.2 7.61 1,509.8 288/20
TBAm


[Str99]

2.3.3 Carrier Gas

Carrier gas is used as the medium to give the uniform flow pattern of precursors in

MOVPE reactor. H2 and N2 carrier gas have been usually used in InN growth. Koukitu et

al. carried out a detailed thermodynamic study on the role of hydrogen during the

MOVPE growth of III-nitrides [Kou99a]. They showed that increase of hydrogen in the

growth system results in a decrease of InN deposition rate (called etching), which they

suggested was due to the decrease of driving force for the deposition.

Thus the reaction for the growth of InN proceeds more effectively in the inert gas

system and is prevented with the increase of H2. Therefore it is necessary to use inert









carrier gas in the growth of InN. These theoretical and experimental results confirm that

using a N2 carrier gas is preferred (and widely used) for successful InN growth.

2.4 Indium Nitride (InN) Growth Techniques

In this part, we review the several growth techniques commonly used for InN

growth each of which deals briefly with the characteristics of the reaction system, the

precursors, the chemistry, the applications, the disadvantages and advantage of each

growth technique.

2.4.1 Chemical Vapor Deposition (CVD)

CVD involves the dissociation and/or chemical reactions of gaseous reactants in an

activated (heated, plasma etc.) environment, followed by the formation of a solid film.

The deposition involves homogeneous gas phase reactions, which occur in the gas phase,

and heterogeneous chemical reactions which occur on a heated surface leading to the

formation of epitaxial films.

In general, the CVD process involves the following key steps as shown in Fig. 2.11

[ChoOOb, Cho03].

(1) Generation of active gaseous reactant species.

(2) Transport of the gaseous species into the reaction chamber.

(3) Gaseous reactants undergo gas phase reactions forming intermediate species:

(4) Absorption of gaseous reactants onto the heated substrate, and the

heterogeneous reaction occurs at the gas-solid interface (i.e. heated substrate) which

produces the deposit and by-product species.

(5) The deposits will diffuse along the heated substrate surface forming the

crystallization centre and growth of the film.










(6) Gaseous by-products are removed from the boundary layer through diffusion or

convection,

(7) The unreacted gaseous precursors and by-products will be transported away

from the deposition chamber.



homnogeous gas
phase reaction




AB,(1) I cystauizah
B----- ~~ trg---..... ..... -. .
Boundary reaction diffusion
layer
Heated substrate
Vapor precursor Efluent gas
y orp~ Effluentgas
feed system Deposilion chamber/ractor treatment system



Figure 2-11. Schematic illustration of the key CVD steps during deposition.

2.4.1.1 Metal-Organic Vapor Phase Epitaxy (MOVPE)

Metalorganic Vapor Phase Epitaxy (MOVPE) is one growth method among CVD,

which has been classified according to the use of metalorganics as precursors.

Compounds containing metal atoms bonded to organic radicals are known as

"Metalorganics". MOVPE can be used to deposit a wide range of materials in the form of

amorphous, epitaxial, and polycrystalline films.

The schematic of MOVPE was shown in Fig. 2.12 where TMI is delivered by N2

carrier gas and NH3 is also delivered directly into MOVPE reactor and the thermal

environment for the decomposition and/or deposition reaction of the precursors can be

supplied using resistance heating, radio-frequency or infrared lamp heating. MOVPE tend

to involve endothermic reactions, thus cold-wall reactors with a single temperature zone

can be used.









@Re@a




@@@
1 RF-Coil

Vacuum
I PFump

N, Dilubicn
Gas VJ 1 I-ent






Figure 2-12. Schematic of horizontal cold-wall MOVPE system.

The metalorganic precursors generally undergo decomposition or pyrolysis

reactions. In general, metalorganics precursors have lower decomposition or pyrolysis

temperatures than halides, hydrides or halohydrides. Thus, metalorganic precursors

enable MOVPE process to perform at a lower deposition temperature than conventional

CVD, which generally uses halides or hydrides.

The source materials generally used for the MOVPE growth of InN, are

trimethylindium (TMI) as In source, and ammonia (NH3) as N source.

The pyrolysis of TMI in MOVPE was first studied by Jacko and Price who founded

that the decomposition occurred in three steps as each of the In-CH3 bonds were broken

at the temperature above 400 C [ Jac64]. The methyl radicals thus formed were then

found to recombine to yield ethane (C2H6). This mechanism is given by Eq. (2-13) to (2-

16).


In(CH3)3 In(CH3)2 + CH3.


(2-13)









In(CH3)2 In(CH3) + CH3. (2-14)

In(CH3) In + CH3. (2-15)

CH3 + CH3. C2H6 (2-16)

The indium reacts with NH3 at the substrate surface in high temperature (> 500C)

to form InN (Eq. (2-17)).

In (g) + NH3 (g) = InN (s) + 3/2 H2 (g) (2-17)

The other method was suggested by Koukitu and the reaction procedure is given as

follows [Kou97b]. First, thermodynamically, almost all the NH3 is decomposed into N2

and H2 at temperatures higher than 300 C. However, it is well known that the

decomposition rate of NH3 under typical growth conditions is slow without a catalyst and

the extent of the decomposition strongly depends on the growth conditions. The mole

fraction of decomposed NH3, into the calculation as follows (Eq. (2-18)).

NH3 (g) -- (1-c) NH3 (g) + a/2 N2 (g) + 3a/2 H2 (g) (2-18)

The metal-organic precursors TMI are decomposed irreversibly, according to the

following homogeneous reaction, near the vapor-solid interface (Eq. (2-19)).

(CH3)3In (g) + 3/2 H2 (g) In (g) + 3CH4 (2-19)

The chemical reaction which occurs at the substrate surface to form InN is the same

as the former method given in Eq.(2-17).

MOVPE can be performed at atmospheric pressure and low pressure (about 2.7-

26.7 kPa). For a typical MOVPE process, the deposition is entirely kinetically controlled

at very low deposition pressure (< 1 kPa), even though the deposition temperature is

relatively high. At pressures above 1 kPa, the growth rate is predominantly controlled by

diffusion-rate limited mechanism [Dup95].









Despite the high cost of precursors, MOVPE have been developed especially for

the growth of epitaxy of III-V as well as II-VI and IV-VI semiconducting material for

optoelectronic applications (e.g. light-emitting diode, heterojunction bipolar transistor,

solar cells, photocathode advanced laser designs such as quantum well and double

heterostructures, etc.).

2.4.1.2 Hydride Vapor Phase Epitaxy (HVPE)

Hydride vapor phase epitaxy (HVPE) has been important in the development of a

variety of semiconductors including the III-arsenides, the III-phosphides and the III

nitrides such as InN. HVPE occurs usually at atmospheric pressure (horizontal or

vertical).

Generally, HVPE using chloride sources provides a high growth rate (> 30tm)

compared with that of MOVPE and MBE. Because of the group III element is transported

to the substrate as a volatile compound (usually a chloride), this technique is often

referred to as chloride-transported vapor phase epitaxy.

The source material generally used for the HVPE growth of InN is liquid indium as

indium source, which will react with HC1 and form indium monochloride (InCl) or

indium trichloride (InC13) and ammonia (NH3) or monomethylhydrazine (MMHy) as

nitrogen source. The source of InCl is formed by the reaction between metallic In and

HC1 at 780 C and the InC13 is presynthesized and is evaporated from the source boat in

the temperature range from 325 to 375 C (Fig. 2.13) [Tak97a]. The reaction chemistry

for InN growth was given

InC (g) + NH3 (g) = InN (s) + HC1 + H2 (2-20)










Deposition Source
Zone Zone

SNH3


HCI



In (I)


Figure 2-13. Schematics of horizontal hot-wall hydride vapor phase epitaxy chamber.

Theoretically, the chlorine in the HVPE chemistry should reduce the amount of

impurities from the system due to the formation of highly volatile species, producing

films with low carrier concentrations. However, Si and O impurities from the quartz tube

can cause highly n-type films.

2.4.1.3 Plasma Enhanced Chemical Vapor Deposition (PECVD)

Plasma Enhanced Chemical Vapor Deposition (PECVD) is also known as glow

discharge chemical vapor deposition. It uses electron energy (plasma) as the activation

method to enable deposition to occur at a low temperature and at a reasonable rate.

Supplying electrical power at a sufficiently high voltage to a gas at reduced pressures

(<1.3 kPa) results in the breaking down of the gas and generates a glow discharge plasma

consisting of electrons, ions and electronically excited species.

Uncracked trimethylindium (TEI) carried by Ar or H2 as In source and uncracked

N2 as N source are ionized and dissociated by electron impact, and hence generating

chemically active ions (radicals). These radicals undergo the heterogeneous chemical

reaction at or near the heated substrate surface and deposit the thin film (Fig.2.14)

[Cho03].










MFC

Ar --



r25 Pa
Substrats
Heater

TEI VCV
MB+RP


Figure 2-14. Schematics of PECVD.

The disadvantage of PECVD is that it requires the use of a vacuum system to

generate the plasma, and a more sophisticated reactor to contain the plasma. PECVD is

often more expensive and in general has difficulty in depositing high purity films. This is

mostly due to the incomplete desorption of by-product and unreacted precursor at low

temperatures, especially hydrogen which remains incorporated into the films.

However, PECVD can find applications where technology will balance the cost of

fabrication and also where low deposition temperatures are required on temperature

sensitive substrates, which can not be met by the conventional CVD.

The main advantage of PECVD over other CVD methods is that the deposition can

occur at relatively low temperatures on large areas. It also offers flexibility for the

microstructure of the film and deposition to be controlled separately. The ion

bombardment can be substituted for deposition temperature to obtain the required film

density. Such low temperature deposition is important for applications that involve the

use of temperature sensitive substrates.









2.4.2 Molecular Beam Epitaxy (MBE) and Metalorganic Molecular Beam Epitaxy
(MOMBE)

MBE is a common technique for thin epitaxial growth of semiconductors, metals

and insulators. MBE has the capability of growing device quality layers of

semiconductors with atomic resolution. The low growth temperature (300-600 C)

ensures negligible dopant diffusion. The apparatus consists of an UHV cold-wall

chamber, independently controlled thermal or e-beam cells to supply the sources, in situ

heating and cleaning, and in-situ monitors of growth and chemical analysis.

The simplest method for generating molecular beams is from heated Knudsen cells

containing Ga, In, Al or dopant material. Shutters open to allow the molecular beams to

leave the cells and the beams are directed at a heated substrate. Thermal beams of atoms

or molecules react on a clean substrate surface to grow an epitaxial film under UHV

conditions. The atomic species undergo adsorption and migration on the surface. MBE

can be also equipped with a number of in-situ probes that monitor the growth real time

including RHEED where high-energy electrons are diffracted off the growing surface and

imaged to describe the nature of the epitaxy.

In the MBE growth of III-nitrides, the solid sources of the group III elements such

as Ga, In, and Al are used in general, but the nitrogen is supplied by the gas source such

as N2 and NH3. In general, this type of MBE system is called gas source MBE [Dav97].

When organometallic sources replace the group-III elemental source, it is called

metalorganic molecular beam epitaxy (MOMBE) or chemical beam epitaxy [Abe97]. In

both types of MBE, the key issue in the growth is the nitrogen source. The dissociation

energy of N2 molecules is as high as 9.5 eV, therefore, the supply of N2 gas to the

substrate surface with the group-III elemental beams can not induce any growth of the









nitrides. For obtaining atomic reactive nitrogen, the N2 molecules are dissociated by the

radio-frequency (rf)-plasma or the electron cyclotron resonance (ECR) method. Although

the rf-plasma source is the most popular and the rf-radical source produces considerably

fewer ions than an ECR source due to the higher plasma pressures [Hug95], it is well

known that ion damage can still induce during the epitaxy [Pow93]. Some techniques to

avoid such ion damage by an ion tapping system using the static electric field, have been

tried [Bot95, Mol94, Iwa96]. The other serious problems induced by the plasma may be

some contamination such as oxygen or carbon dioxide.

MOMBE is also one of the potential growth techniques where the advantages of

both MOVPE and MBE can be utilized. Film can be grown relatively at low temperature

by MOMBE and premature reaction of the precursors, a serious problem in MOVPE, is

minimized due to the large mean free path of gaseous molecules.

2.4.3 Atomic Layer Deposition (ALD)

Atomic Layer Deposition (ALD) can be considered as a special mode of CVD. It is

a surface deposition process that can be used for the controlled growth of epitaxial films,

and the fabrication of tailored molecular structures on the surfaces of solid substrates.

'Monatomic layers' can be grown in sequence which is a characteristic feature of ALD.

Therefore, the desired coating thickness can be produced simply by counting the number

of reaction sequences in the process. The surface reconstruction of the monolayer formed

in the reaction sequence will influence the saturation mechanism and the saturation

density of the precursor.

The ALD reaction sequences are normally perform in an 'effective overdosing'

condition to ensure a complete saturation of the surface reaction to form the monoatomic

layer. Furthermore, such effective overdosing' condition also provides good conformal









coverage that allows uniform coatings onto complex shaped substrates. The sequencing

in ALD also eliminates the gas phase reactions, and enables a wider choice of reactants

(e.g. halides, metalorganics, elemental metal, etc.).

The ALD process has the potential to be scaled up for the deposition of high quality

thin films with excellent uniformity and reproducibility onto large area substrate [Nii96,

Lau98].

The ALD process can be performed at atmosphere pressure or in a vacuum system

as in molecular beam epitaxy. The use of vacuum enables a variety of in-situ surface

analysis methods to be incorporated into the ALD equipment for the in-situ analysis of

the growth mechanism and the deposited surface structures [Bac97, Kou97c, Her99].

The distinctive sequencing feature in ALD makes it an attractive method for the

precise growth of crystalline compound layers, complex layered structures [Cha98]

superlattices [TorOO, Har98] and layered alloys with precise interfaces.

Currently, a wide range of thin films have been synthesized using ALD methods.

These include semiconductor III-V, II-VI, oxides, nitrides, phosphides, and metallic films

[Gup98, Hsu98, Utr99, Mar99, Ish97b, UtrOO].

The ALD process can produce films with good conformal coverage and it has the

ability to control film thickness accurately at the sub-nanometer level. Such distinctive

advantages have made it a potentially valuable tool for nanotechnology.

2.5 Substrate Materials

The important properties of substrate are the lattice constant which causes the

lattice mismatch for the epitaxial InN film and thermal expansion coefficient which also

can create dislocations during cooling when there is a large thermal coefficient mismatch









between the film and substrate. The properties of the substrate commonly used in III-V

nitrided film growth are summarized based on InN film in Table 2-9.

Table 2-9. Structural properties of substrates.
Substrates Lattice Lattice Lattice TEC (10-6 K-1) TEC
constant constant mismatch mismatch
a (A) c (A) (%) a c (%)
InN-Wurtzite 0.537b 5.704b 5.70b 3.70b 0
GaN-Wurtzite 3.189a (5.524 5.18a -10.9 5.59a 3.17a -30.9
when rotated
30 C)
AIN wurtzite 3.112b 4.98b -13.7 4.2b 5.3b -8.1
6H-SiC wurtzite 3.08b 15.1b 14.8 4.2b 4.7b -8.1
A1203 4.758b 12.991b 25.7 7.5b 8.5b -48.5
rhombohedral
ZnO wurtzite 3.252b 5.213 -8.8b 2.9b 4.8b 33.1
LiA102 5.17b 5.1687b -12.9 7.1b 7.5b -45.6
tetragonal (3.134 as a (wurtzite)
wurzite)

LiGaO2 5.40b 5.007b -11 6b 7b -35.7
orthorhombic (3.186 as a (wurtzite)
wurtzite)
GaN zincblende 4.53b 21.9 5.2b -25.8
Si (111)- cubic 5.43b( 3.84 as 7.9 6.2b -37.7
a hexagonal) (hexagonal)
3C-SiC 4.36b 18.9 2.7b 43
zincblende
GaAs zincblende 5.65b 37.4 6.0b -35.7
[aMor94, bDav02a]

2.5.1 Sapphire Substrate (A1203) (0001)

Sapphire is the most extensively used substrate material for the epitaxial growth of

III-V nitride materials. Large area good quality crystals of sapphire are easily available at

relatively low cost. They are transparent and stable at high temperature. The large lattice

mismatch (25.7%) and thermal expansion coefficient difference (48.5 %) for InN can

result in an extremely high density of structural defects of InN film. However,

researchers have revealed that the substrate surface pretreatment and insert of an









intermediate buffer layer between the substrate and epilayer can significantly improve the

film quality. Nitridation of the sapphire substrate surface significantly improves the

crystalline quality of III-V nitride growth as a result of the formation of A1N which

reduce the lattice mismatch from 25.7 % for InN/A1203 to 13.7 % for InN/AlN.

The single crystal can be described by both rhombohedral unit cells with volume

84.929 A3 and hexagonal unit cell with volume 254.792 A3. The unreconstructed basal c-

plane perspective views for both unit cells are shown in Fig. 2.15 [Edg02]. The faceting

of sapphire crystal is shown in Fig. 2.16 [Amb98].






%V '^ _' 0^ .0 j 40





(a) (b)


Figure 2-15. Perspective views in (2 x 2 x 1) unit cell: (a) along [0001] direction in a
rhombohedral unit cell; (b) along the (0001) direction in hexagonal unit cell.

(e)
(0001)
(iIn) (0

(1120) 305- (1010) 7n) \ "(r)


(10) n,, f t/


(a) "-i (b) ,'____

Figure 2-16. Common facets of sapphire crystals: (a) view down c-axis; (b) surface
planes.









2.5.2 Silicon (Si) Substrate

Si (111) substrate is usually spotlighted as an attractive substrate because of the

high quality and low cost of Si. The availability of the either n or p-type substrate is

advantageous. The doped substrates can significantly simplify device structures.

The bulk Si crystal is a diamond structure and has lattice constant a = 5.43 A at

room temperature. However, Si (111) surface has hexagonal surface and lattice parameter

of a = 3.84 A. Therefore, Si has the small lattice mismatch (7.9 %) for InN.

The unit cell is outlined as a diamond shape with seven atoms along each edge, for

two different orientations. Si has a diamond-lattice structure with the space group of


Fd3 m (no.227), which belongs to the cubic-crystal family. It can be represented as two

interpenetrating fcc sublattices with one sublattice displaced from the other by one

quarter of the distance along a body diagonal of the cube (i.e. the displacement of a

V3 a/4. where a = 0.543 nm). Each atom in the lattice is surrounded by foul equidistant

nearest neighbors that lie at the comers of a tetrahedron. Figure 2-17 illustrates the

perspective views along the [001], [011] and [111] directions of a Si unit cell [Liu02].

There are several methods for Si substrate preparation (Table 4) [Yan96, Gru91, DadOla,

Wat93].









(aM HII) (cL

Figure 2-17. Perspective views of Si along various directions: (a) [001]; (b) [011]; (c)
[111].









2.5.3 Gallium Nitride (GaN) and Aluminium Nitride (AIN) Substrate

Gallium nitride substrate has a small lattice mismatch of 10.9 % with InN compared

with sapphire substrate (25.7 %) and A1N substrate (13.7 %) even if it is greater than that

of silicon of 7.9 %. Gallium nitride normally has a wurtzite structure, with the space

group of P63mc (no. 186). The wurtzite structure consists of alternating biatomic close-

packed (0001) planes of Ga and N pairs stacked in an ABABAB sequence. Atoms in the

first and third layers are directly aligned with each other. Figure 2-18 displays the


perspective views of wurtzite GaN along [0001], [1120] and [1010] directions, where the

large circles represent gallium atoms and the small circles nitrogen [Liu02]. The close-

packed planes are the (0001) planes. The group III nitrides lack an inversion plane

perpendicular to the c-axis, thus, crystals surfaces have either a group III element (Al, Ga,


or In) polarity (designated (0001) or (0001)A) or a N-polarity (designated (0001) or

(0001)B). An excellent review on crystal polarity is given by Hellman [Hel98]. The


zincblende structure (space group F4 3m) of GaN can be stabilized in epitaxial films. The

stacking sequence for the (111) close-packed planes in this structure is ABCABC.

Perspective views of the zincblende structure are shown in Fig. 2.19 [Liu02].










ll) (b) (c)


Figure 2-18. Perspective views of wurtzite GaN along various directions: (a) [0001]; (b)
[1120]; (c) [1010].






















Figure 2-19. Perspective views of zincblende GaN along various directions: (a) [100]
(1x1x1 unit); (b) [110] (2x 2 x2 units); (c) [111] (2 x 2 x 2 units).

A1N normally has the wurtzite structure, although epitaxial layers of zincblende

structure A1N have been made [Oku98a, Oku98b]. Wurtzite A1N has the space group of

P63mc (no. 186) as same as wurtzite GaN. The (0001) surfaces of A1N are polar, which

has an important effect on its etching, bulk crystal growth and GaN epitaxy. A1N has the

properties such as high thermal conductivity, low thermal expansion coefficient, high

electrical resistivity, good dielectric properties, and excellent oxidation resistance.

2.5.4 Other Substrates

GaAs has the same structure as zincblende GaN. GaAs is less stable than SiC or

sapphire. Above 800 C its decomposition rate to liquid gallium and arsenic vapor is

considerable. GaAs has the large lattice mismatch of 37.4 % for InN film.

Zinc oxide (ZnO) has a wurtzite structure and its stacking order match with lattice

constants closely matched to GaN (a=3.249 A, c= 5.205 A). The small lattice mismatch

of 8.8 % for InN makes ZnO attractive substrate for InN growth. Lithium gallate

(LiGaO2) also has the small lattice mismatch of 11 % for InN film. Therefore, LiGaO2 is

another candidate for the suitable substrate for InN growth.









2.5.5 Buffer Layer

There is no lattice matched substrate available for InN so far. For example, the InN

has a lattice mismatch of 25 % with sapphire, 8% with Si (111), 37.4 % with GaAs, and

11 % with GaN. High quality single crystalline InN is very difficult to be obtained

because of these problems.

The two-step growth method or growth using buffer layer has now become a

standard method for the heteroepitaxial growth of thin films. This method is commonly

used to alleviate lattice mismatch and thermal expansion coefficient difference the

substrate and epilayer. In this method, a thin buffer layer is grown at a low temperature in

the first step. The main epilayer is grown in the second step at a high temperature. The

buffer layer provides the high density of nucleation centers and promotes the lateral

growth of the main epilayer. The two-step growth of InN is not well studied, especially in

the MOVPE growth.

There are very few studies about the MOVPE growth of InN using buffer layer

such as GaN, A1N, and InN. There is no significant report that use of low temperature

InN buffer layers in the growth InN gives improvement. Pan et al. studied two-step

growth of InN using conventional MOVPE [Pan99]. Based on their findings, they

concluded that the two-step growth is not adequate for InN, which may correlate to the

unstable nature of the InN film. Guo et al. reported that if a single crystalline InN film is

heated above 550 C in a N2 flow, the surface undergoes a considerable change, owing to

the decomposition and desorption of nitrogen [Guo93].









2.6 Summary for Growth of InN on Different Substrate

2.6.1 Growth on Sapphire (A1203) Substrate

The growth of InN in horizontal MOVPE reactor has been studied using a-A1203

(0001) substrate by Yamamoto [Yam94b]. A single-crystalline InN film was obtained on

a-A1203 substrate at 500 oC, in spite of the larger lattice mismatch for InN (0001)/a-A1203

(0001) by the nitridation of the A1203 (0001) substrate prior to the growth. Nitridation of

a-A1203 surface occurs at the temperature region from 800 C to 1000 C. A1N is formed

during the nitridation and the lattice mismatch is reduced from 25 % for InN/ a-A1203 to

about 13 % for InN/AIN [Yam94b, Pan99].

Chen found that the InN film quality is strongly dependent on the growth

temperature and V/III ratio [Che97]. He reported the best quality of InN film was grown

at 375 C under a high V/III ratio of 146000 and the flow rate of NH3 of 2000 sccm. InN

film growth was carried out in the atmospheric-pressure horizontal MOVPE reactor with

a cross-section of 30x14 mm2. The FWHM of the best quality of InN (0002) was 96

arcsec with InN (10-11) existing while the typical FWHM of XRC of MOVPE-grown

InN is from 4000 to 5500 arcsec [Che97].

Surface morphology study of InN grown in MOVPE was carried out by AFM with

different growth condition by Yamamoto [YamOla]. A continuous InN film with

enhanced two-dimensional growth was obtained at 630-650 C. It was reported that

growth rate was increased with increasing growth temperature in the range of 500-630

C, while it is independent of growth temperature at a temperature higher than 630 C. It

was suggested that when the growth is performed at 630-650 C, growth rate is

proportional to TMI supply. The increase in growth rate with increasing growth

temperature at a temperature less than 630 C can be explained by taking account that









growth rate is limited by NH3 decomposition rate. Yamamoto studied the effect of GaN

buffer layer on InN and found that uniformity for grown InN film is markedly improved

by employing a GaN buffer layer and this improvement is due to the uniform nucleation

of InN [Yam04a].

The growth of InN in vertical resistive heated MOVPE reactor was performed by

Hwang [Hwa0l] where InN was grown at 360-540 C; high V/III ratio was used to

prevent indium droplet formation. The best InN was obtained at 540 C and there was no

reported about the value of FWHM.

Takahashi carried out the growth of InN by HVPE at V/III = 10-100 using InCl and

InC13 as In sources and NH3 and MMHy as N sources where the source of InCl was

formed by the reaction between metallic In and HC1 at 780 C and InC13 was evaporated

from the source boat in temperature range 325-375 C. The InC13-NH3 system showed an

appreciable growth rate of InN (-0.3 [tm/hr) and the growth rate initially increases with

increasing growth temperature up to 550 C and then gradually decreases to 700 C. The

other systems such as InC13-MMHy, InCl-NH3, and InCl-MMHy showed the very small

growth rate (< 0.05 [tm) [Tak97a].

The hydride vapor phase epitaxy growth of InN was performed by Yuichi Sato

where HC1 (diluted with N2 to 1 %) gas was passed over the indium metal source, which

was kept in a quartz boat and the indium source was maintained at 900 C in order to

form InCl. The growth rate of the film gradually increases with increasing growth

temperature and reaches the maximum growth rate of 0.3 [tm/h at 510 C [Sat94a].









In addition to MOVPE and HVPE, atomic layer deposition (ALD) and molecular

beam epitaxy (MBE) were also used for InN growth on sapphire substrate by several

researchers [Bed 97, Hig20, and Mam99].

In summary, the growth conditions of InN on sapphire substrate for the different

growth methods such as MOVPE, HVPE, and ALD were discussed. The MOVPE is the

commonly used growth method for InN and the high quality single crystalline InN

growth by MOVPE is still required because the typical range of FWHM of XRC is higher

than 1000 arcsec.

2.6.2 Growth on Silicon (Si) Substrate

The growth of InN on Si in horizontal MOVPE reactor was carried out by

Yamamoto [Yam94b]. For Si substrate, relatively well oriented InN films are grown at

about 400 C. Polycrystalline InN films are grown both at 350 C and at 500 C on Si

substrate. Polycrystalline InN growth below 350 C is believed to be due to reduced

migration of deposited materials on Si or decomposition rate of raw materials. The

growth at a temperature higher than 450 C results in serious deterioration of InN films

grown on Si substrates. It was shown that the nitridation occurs at a temperature as low as

500 C by exposing to NH3 [Yam94b]. The cause for poorly-oriented or polycrystalline

InN film growth on Si at a temperature above 400 C was due to the formation of

amorphous silicon nitride (SiNx) on Si substrates before the growth.

He suggested that epitaxial growth of InN on Si without a buffer layer is found to

be difficult owing to the nitridation of Si substrate. The application of InN on Si to a

tandem solar cell was suggested by Yamamoto [Yam94a, Yam94b].

Yang et al. improved the growth rate of InN on Si with a double-zone MOVPE

system consisting of a high temperature NH3 pre-cracking zone and a low temperature









deposition zone [Yan02c]. A maximum growth rate of 6 tmm /h was achieved due to the

high cracking efficiency of NH3. In this experiment, he used N2 as a carrier gas, the flow

rate of NH3 at 800-1600 sccm, and V/III ratio of several hundreds. The optimal growth

temperature was 450 C [Yan02c].

In summary, single crystalline InN growth on Si was obtained but no reports on

crystalline quality (FWHM of XRC) have been reported. Therefore, the growth

conditions for high quality single crystalline InN should to be studied and optimized.

2.6.3 Growth on Gallium Arsenide (GaAs) Substrate

InN films was obtained on GaAs(1 11) at 500 C, 1.3 Torr, and N2 flow rate of 200

sccm, using microwave-excited MOVPE by Guo et al.[Guo95b]. Yamamoto et al.

studied thermal nitridation of GaAs (111) in flowing NH3 and horizontal MOVPE growth

of InN on the nitrided GaAs (111) as a result of the thermal nitridation [Yam97a].

In the case of GaAs(1 11) substrates, crystal structure of a GaN layer formed by the

nitridation before the InN growth was found to be dependent on nitridation temperature

TN; zincblende structure for TN< 700 C and wurtzite for TN> 800 oC.

For an InN film grown on a GaAs (111) substrate with a zincblende GaN layer, its

crystalline structure is changed from zincblende to wurtzite when the thickness exceeds

about 0.2 itm. On a GaAs (111) with a wurtzite GaN layer, on the other hand, growth of

zincblende InN is not found [Yam98a].

Using an atmospheric HVPE system, InN growth was carried out on a GaN layer

which was formed on a GaAs (100) substrate inclined 10 o toward the <111>B direction

of GaAs substrate. An important requirement for growth was to keep low temperatures of

less than 750 C in the upstream region of the reactor to raise the amount of InC13, where









indium chloride is formed at the temperature higher than 750 C. Furthermore, it was

necessary to exclude H2 from the reaction system for deposition to occur because the high

partial pressure of H2 increases the amount of InCl. These results indicate that the

effective chemical substance of indium chlorides for the growth is InC13. Growth rate of

1.5 [tm/h was obtained at 570 C and single crystalline InN growth was confirmed by X-

ray diffraction measurement [Sun96].

The growth of InN using MOVPE and HVPE was discussed in terms of growth

conditions. For MOVPE, the structure of InN depends on the nitridation temperature for

GaAs(111)B substrates. For HVPE, InC13 forms InN film more effectively than InCl

does.

2.6.4 Growth on Gallium Phosphorus (GaP) Substrate

Guo et al. reported that InN films had been grown on GaP (111) substrate at 500

C using microwave-excited MOVPE and TMI and nitrogen were used as the source

materials. The epitaxial InN film was obtained on GaP (111) by exposing the substrate to

the nitrogen plasma for 60min before growth [Guo95b]. InN films have a wurtzite

structure [Guo95b].

Bhuiyan et al. obtained InN on GaP(ll1)B by the horizontal MOVPE reactor

where single crystalline InN films can be obtained on GaP(ll1)B only when the

nitridation of the substrate is not made intentionally. InN films grown on nitrided

GaP(111)B are found to be polycrystalline. XPS analysis shows the formation of PNx as

well as GaN after the nitridation of GaP (111)B substrate surfaces by flowing NH3 above

500 C. Formation of PNx is responsible for the poor crystalline structure for InN. A

single crystalline InN film with an excellent surface morphology can be grown on









GaP(111)B at high temperature (600-650 C) using a low temperature InN buffer layer

[Bhu00a,Bhu01,Bhu02b].

The growth of InN on GaP substrate using MOVPE was briefly discussed. When

the growth of InN is performed on GaP substrate, the nitridation step should not be

required in order to obtain the single crystalline InN.

2.6.5 Growth on Gallium Nitride (GaN) and Alumimum Nitride (AIN) Substrate

Yamaguchi et al. presented the result of the InN film grown on GaN substrate with

A1N buffer layer using atmospheric MOVPE. Growth temperature was 450 C and V/III

ratio was 105. The FWHM of XRC decreases with increasing the thickness of InN film

[Yam99a].

The effects of reactant-gas velocity on the growth of InN on GaN/sapphire by

MOVPE were studied by Yang et al. With a high-speed reactant gas, the thickness of the

stagnant layer is reduced so that the reactant species can reach the surface effectively. A

layer like growth of InN was achieved, resulting in a significant improvement of the film

quality. In addition, significant enhancement of the growth rate up to 2 [tm/h was

obtained. The FWHM of XRC decreased with increasing gas velocity. FWHM of XRC

for InN (0002) with 476 arcsec was reported but there was no report about whether the

InN is single or poly crystalline [Yan02a].

The possibility that high quality single crystalline InN can be grown on

GaN/sapphire substrate using MOVPE is studied and it is found that the flow pattern of

source materials can have an effect on the InN film quality.









2.7 Overview

The latest lattice constant of single crystal InN with wurtzite was reported to be a =

3.537 A and c = 5.704 A [Dav02a]. The band-gap energy of InN is nowadays accepted to

~ 0.7 eV instead of 1.89 eV.

Thermodynamic analysis helped us to understand at which growth condition the

growth and etching happen and therefore help us to predict where InN film can be grown

before the epitaxial growth is performed.

The possible candidates as N precursor were reviewed due to low decomposition

efficiency of NH3 at low growth temperature of InN (- 550 C). Because all of other

candidates for nitrogen source have the several problems such as toxicity, explosion and

contamination, NH3 is still widely used N precursor and N2 carrier gas is better than H2 as

carrier gas.

MOVPE is still the most widely used growth technique for InN for the industry and

academy to date.

Based on available published data, the typical range of FWHM of XRC for single

crystalline InN grown by MOVPE is higher than 2000 arcsec [Yam04a]. For Si substrate,

it is still very difficult to have the high crystalline InN because of the bad coverage of

InN on Si substrate despite of a small lattice mismatch [this topic was not reviewed in the

InN growth on Si substrates section]. Therefore the study for the growth high quality

single crystalline InN has been still required.

It is found that some factors such as growth temperature, V/III ratio, substrate,

nitridation treatment, buffer layer, and flow pattern of source gases can have an effect on

the structural quality of InN film. Based on the results of this review, these factors will be

analyzed in detail to conduct our experiments of InN growth by MOVPE.














CHAPTER 3
THERMODYNAMIC ANALYSIS OF InN AND InxGal-xN MOVPE GROWTH

3.1 Thermodynamic Analysis of InN and InxGal-xN

The results of a study on the effect of pressure and temperature on the equilibrium

growth of InN and InxGal-xN are presented in this chapter. Specifically, equilibrium in

IN-Ga-N-C-H system is studied to clarify the impact of process variables on film

composition and to estimate a suitable growth condition for InN and InxGal-xN to support

the experimental studies. For example, it is interesting to know the maximum content of

indium that can be incorporated into the InxGal-xN phase without the phase separation.

The Gibbs energy functions for InN and GaN from the Scientific Group

Thermodata Europe (SGTE) and the ThermoCalc software package were used for these

calculations.

3.1.1 Reaction Mechanism and Kinetics of InN MOVPE

Growth of InN by MOVPE typically uses trimethylindium (TMI) and NH3

precursors in a N2 carrier gas. The pyrolysis of TMI was studied by Jacko et al. [Jac64,

Tra78, and Lar85] and they proposed the sequential hemolytic fission of the In-C bond

along methyl radical recombination described by Eq. (3-la) to (3-1d).

In(CH3)3 In(CH3)2 + CH3 (3-la)

In(CH3)2 In(CH3) + CH3 (3-1b)

In(CH3) In + CH3* (3-1c)

CH3. + CH3 C2H6 (3-1d)









It is reported that reactions 3.1a and 3.1c are slow steps, thus producing In(CH3)

into the vapor phase.

Stepwise hemolytic fission of the In-C bond in TMIn was first proposed as in Eq.

3-1 [Jac64] and recently mono-methyl indium (MMIn) and atomic indiums were

experimentally observed in the gas phase [Par02]. New reaction intermediates are

proposed based on experimental evidence using in situ Raman and computational

chemistry supports by Hwang [Hwa04]. It has been suggested from Hwang's

experimental results [Hwa04] that MMIn and/or dimethyl indium (DMIn) seem to hide

from Raman detection by forming another intermediate, presence of which in contrast to

MMIn is evident. A new intermediate (HInCH3) was found to exist during TMIn

decomposition in a nitrogen carrier. It has some of its characteristic vibrations at 416

[v(H-In-C)], 464 [v(In-C)], and 1560 cm-1 [v(H-In)]. The new intermediate was

experimentally observed to decompose very quickly in a high temperature region of the

reactor. In addition, it was considered highly probable that (DMIn)2 and DMIn-MMIn

would form during TMIn decomposition, as shown in Eq. (3-2) [Hwa04].

2DMIn > (DMIn)2 (3-2a)

DMIn + MMIn > DMIn-MMIn (3-2b)

DMIn-MMIn < CH3InCH2 + HInCH3 (3-2c)

TMIn + MMIn > (DMIn)2 (3-2d)

In terms of the kinetics of TMI decomposition (Eq. (3-1)), several experimental

results are summarized in Table 3-1. Reaction rate constant and activation energies of

TMI decomposition are shown in Table 3-1.









Table 3-1. Reported reaction rate constants for TMIn decomposition.
ko (s-1) Ea (kcal/mol) Carrier
Hwang [Hwa04] 1017.9 56.1 N2
Jacko & Price 10157 47.2 Toluene
[jac64]
Larsen & 1012.6 40.5 N2
Stringfellow 1012.0 35.9 H2
[Lar86]
Buchan et at. 1017.9 54.0 He
[Buc88] 1013.4 39.8 D2
1015.0 42.6 H2

Ammonia is the most widespread precursor for III-Nitrides growth by MOVPE.

Complex chemical equilibrium analysis by Koukitu suggested that most of the NH3

should be decomposed into N2 and H2 at temperatures greater than 300 C [Kou97b].

It is well known, however, that the decomposition rate of NH3 under typical growth

conditions is slow without a catalyst and the extent of the decomposition strongly

depends on the growth conditions. The decomposition reactions of NH3 are presented by

Eq. (3a-3c).

NHI3 > NI2 + H. (3-3a)
NH2 2 NH + H. (3-3b)

NH->N+H. (3-3c)

N. + N. N2 (3-3d)

H.+ H. H2 (3-3e)

Based on the aforementioned consideration, the several possible reactions for InN

formation are suggested (Eq. (3-4)).

In + N. InN (3-4a)

In(CH3) + NH. -> InN + CH4 (3-4b)

In(CH3)2 + NH2 -> InN + 2CH4 (3-4c)









(DMIn)2 + 2H. -> 2InN + 2CH4 (3-4d)

This complex chemical equilibrium analysis of the growth of InN requires

computing the equilibrium state in the In-N-C-H system. The growth conditions of InN

were calculated as a function of deposition temperature, pressure, and composition. It is

assumed that the vapor phase follows an ideal gas mixture and the vapor species whose

equilibrium mole fractions are negligible (below 10-30) are not taken into account in this

calculation because the same result for P-T diagram was obtained in either case when the

species with the mole fraction less than 10-30 (C4H1o, C4H2, C4, and N3 etc.) were

included or excluded in the calculation. Therefore, the species with the mole fraction less

than 10-30 are thought to be insignificant in the calculation. The equilibrium mole

fractions of each component can be obtained from the equilibrium result data of

ThermoCalc. With this assumption, species, phase, and thermodynamic properties

included in this analysis are summarized in Table 3-2. Diamond was not considered in

the calculation as the phase of graphite is taken as a stable one.

The equilibrium state for the growth of InN without indium formation of 2-phase

(In (1) + InN) was computed in the range of P = 1 to 101.3 kPa (7.5 to 760 Torr), T = 400

to 1000 C, and V/III (NH3/TMI) = 50,000.

The P-T diagram is shown in Fig. 3.1 and the results indicate that the etching

temperature (decomposition temperature) is 810 C at P = 13.3 kPa (100 Torr) and V/III

= 50,000. Pressure and V/III ratio were chosen based on our current operation conditions

of the MOVPE system used in this study.










Table 3-2. Species, phases, and thermodynamic properties included in the analysis of
MOVPE of InN.
Phase Species Parameter (J/mol)


C, InN, In


In, N


C, C1H4, H,
H2, NH3,
In, In2, N,
N2


Solid (s)


Gcs) = -17368.4408+170.730318T-24.3TLN(T)-4.723x 104 T2+
2562600T--2.643 x 10T2+1.2x 101T-3
GSN(s) -149963.181+215.110609T-38.0744TLN(T) -0.0060668T2
G,(,)= -6978.89011+92.3381153T-21.8386TLN(T) -0.00572566T2
-2.12032167x10-6T3-22906T1
Gi() = -6978.89011+92.3381153T-21.8386TLN(T) -0.00572566T2
-2.12032167x10-6T3-22906T-1 + 3283.7-7.6402121T
GN(1)= -4461.675+60.74575T-12.7819TLN(T)-0.00176686T2
+2.680735x10-9T3-32374T1

Gcg)= 710430.933-17.7062915T-20.97529TLN(T)+1.998237x10-4T2
-3.34617167x 10-T3+1680.6515T-+RTLN(105P)
Gc1H4(g) = -77295.5632-147.095196T-2.234656TLN(T) -0.048463265T2
+4.33754333x10-6T3-305431.45T-+RTLN(105P)
GH(g) = 211801.621+24.4989821T-20.78611TLN(T)+ RTLN(10-5P)
GH2(g)= 9522.9741+78.5273879T-31.35707TLN(T)+0.0027589925T2
-7.46390667x 10-7T3+56582.3T1+ RTLN(105P)
GN3(g) = -53688.8736-38.3667407T-21.21774TLN(T) -0.022871695T2
+1.80809167x10-6T3-76698.65T l+RTLN(10-5P)
G,,g) = 236267.082-68.7705731*T-15.35206TLN(T)-0.00527185T2
-3.98269833x 10-T3-94519.9T -+RTLN(10-5P)
GIn2() = 407673.852-41.5349376T-35.82134TLN(T) -0.00654889T2
+2.03928167x10-8T3-20133.605T +RTLN(105P)
GN(g) = 466446.153-13.3752574T-20.89393TLN(T)+8.45521x10- T
-1.0018685x10-T3+2788.7865T-+RTLN(10-P)
GN2g) = -8000.12556-8.81620364T-27.22332TLN(T)-0.0012599175T2
-5.39381x10-T3-38326.695T +RTLN(10-5P)


Liquid (1)


Gas (g)









101
V/III=50,000
91
81
71 (InN(s) + Gas) (Gas)

61
51 Growth Etching
I 41
2 31 Experimental Growth T
21 -100 Torr
11


400 500 600 700 800 900 1000
Temperature (oC)
Figure 3-1. Calculated P-T phase diagram for InN at X(In) = 5.31212x10-6, X(N) =
0.24998, X(H) = 0.75000, X(C) = 1.59364x10-5 and V/III = X(N)/X(In)=
50,000.

In summary, the InN P-T diagram was computed and the maximum growth

temperature was estimated at the condition used in this study (P =13.3 kPa (100

Torr) for V/III = 50,000) using ThermoCalc software. It is known that the typical

MOVPE growth temperature of InN is in the range of 400 to 7000C depending on V/III

ratio, pressure, residence time, and probing methods [Bhu03b].

In terms of the growth temperature region of InN, the difference between

thermodynamic calculation and the experimental result is thought to be caused by the fact

that the MOPVE growth of InN is non-equilibrium reaction.

3.1.2 Pressur-Temperature (P-T) Phase Diagram of InxGal-xN and Phase Separation
in InxGal-xN

For MOVPE growth of InxGal-xN in this study, trimethylindium (TMI) was used as

In precursor, triethylgallium (TEG) as Ga precursor, NH3 as the reactive N source, and









N2 as a carrier gas. Thus chemical equilibrium state was computed in the In-Ga-N-C-H

system.

The growth condition for the growth of InN was calculated as a function of

deposition temperature, pressure, and composition. In the thermodynamic analysis, the

calculation procedures for InxGal-xN are the same as that of InN, which was already

discussed 3.1.1. The species, phases and thermodynamic properties included in this

analysis are summarized in Table 3-3. The Redlich-Kister equation was used and the

interaction parameters (Lo and Li) of liquid indium and liquid gallium are presented in

Table 3-3.

The input mole fraction ratio of TMI to TEG which leads to the growth of In0.3Gao.7N

was calculated to show the relation of the input ratio of indium with the indium content in

InxGal-xN, which is obtained experimentally (Fig. 3.2) [Mat92]. The indium content of

0.3 in In0.3Gao.7N is chosen in this calculation because the maximum indium mole

fraction is reported to be about 0.3 [Elm98, Sin97, Nak94, Shi94, Nak93, Mat92]. The

flow rate of TEG of 0.44 sccm, the flow rate of H2 of 4 slm, the flow rate of TMI of 0.27

sccm and the flow rate ofNH3 of 1.6 slm were chosen in this study. This condition

corresponds to X(In) = 1.87328x10-5, X(Ga) = 3.05276x10-5, X(N) = 0.111, X(H) =

0.8887, X(C) = 2.39364x10-4. The P-T phase diagram (Fig. 3.3) shows that the stable

In0.3Gao.7N can be obtained below the growth temperature of 730 C and at P = 13.3 kPa

(100 Torr), which is the operating pressure of our MOVPE system [Mat92]. From theses

results, it is also clear that In0.3Gao.7N is decomposed at T > 800 C. Experimentally, the

growth of In0.3Gao.7N is very difficult at the temperature above 800 C [Mat92]. These

calculated results are in the good agreement with the data obtained by Matsuoka [Mat92].










Table 3-3. Phases and species included in the analysis of MOVPE of InxGal-xN.
Phase Species Parameter (J/mol)


C, InN, Gc(s)= -17368.4408+170.730318T-24.3TLN(T)-4.723x 104 T2+
GaN, 2562600T'-2.643x 10T 2+1.2x 10T -3
InGaN GnN(s)= -149963.181+215.110609T-38.0744TLN(T) -0.0060668T
GaN() = -137112+272.388T-44.3769TLN(T)-0.0063011T2
+586387T1
LIlaN(s) = 40320
In, Ga, N GIn(l)= -6978.89011+92.3381153T-21.8386TLN(T) -0.00572566T2
-2.12032167x 106T3-22906T-1 + 3283.7-7.6402121T
GGa() = 5491.298-18.073995T-7.017x 10-1T7 -7055.643+132.73019T
-26.0692906TLN(T)+1.506x10-4T2 -4.0173x10-8T3-118332T1
+1.645x1023T9
GN(1)= -4461.675+60.74575T-12.7819TLN(T)-0.00176686T2
+2.680735x10-9T3-32374T1
Lo(Ga, In) = 4450+1.19185T
L1 (Ga, In) = 0.25943 T


C, C1H4, H,
H2, N1H3,
Gal, Ga2,
In, In2, N,
N2


Gcg)= 710430.933-17.7062915T-20.97529TLN(T)+1.998237x 10-4T2
-3.34617167x10-8T3+1680.6515T-+RTLN(10-P)
Gc1H4(g) = -77295.5632-147.095196T-2.234656TLN(T) -0.048463265T2
+4.33754333x 106T3-305431.45T l+RTLN(10-5P)
GH(g) = 211801.621+24.4989821T-20.78611TLN(T)+ RTLN(10 5P)
GH2(g)= 9522.9741+78.5273879T-31.35707TLN(T)+0.0027589925T2
-7.46390667x 10 -T3+56582.3T1+ RTLN(105P)
GNH3(g)= -53688.8736-38.3667407T-21.21774TLN(T) -0.022871695T2
+1.80809167x10-6T3-76698.65T -+RTLN(105P)
GGa(g) = 259072.279+88.0130701T-38.71057TLN(T)+0.01053784T2
-9.86907833x10-T3 (298.13 K 263812.519+33.4871429T-30.75007TLN(T)+0.00537745T2
-5.46534x 10-T3-150942.65T-1 (600 K GGa2(g) = 422882.385-36.0787973T-33.72863TLN(T)-0.009368525T2
+7.62775167x10-T3-19520.385T1
GIng) = 236267.082-68.7705731*T-15.35206TLN(T)-0.00527185T2
-3.98269833x 10-T3-94519.9T +RTLN(10-5P)
Gn2(g) = 407673.852-41.5349376T-35.82134TLN(T) -0.00654889T2
+2.03928167x10-8T3-20133.605T +RTLN(105P)
GN(g) = 466446.153-13.3752574T-20.89393TLN(T)+8.45521x10 T
-1.0018685x10-T3+2788.7865T +RTLN(10-P)
GN2(g) = -8000.12556-8.81620364T-27.22332TLN(T)-0.0012599175T2
-5.39381x 10-T3-38326.695T -+RTLN(105P)


Solid (s)


Liquid (1)


Gas (g)









K 1.i


0.2 0.4 0.6


Figure 3-2. Relation between indium mole fraction (x)
ratio of the sum of group III source of TMI


of InxGai-xN and the flow rate
and TEG.


100 Torr

Jnt3Gaa7N+
7gas)

- Growth


InQ3GaQ7N+alloy(Ga(1)+In(l)))
+(gas)

Growth


700 720 740 760 780
Temperature (C)


800 820 840


Figure 3-3. Calculated P-T phase diagram for In0.3Gao.7N at X(In)=1.87328x 105,
X(Ga)=3.05276x10-5, X(N)=0.111, X(H)=0.8887, X(C)=2.39364x10-4 and the
data points (0) are from the measurements observed by Matsuoka.


D.B 1.0


TMI
TM1+TEG


28
25
22
019
16
13
10
7
4
1


Ga(I)+ln(I)
+ (gas)

Etching









The main problem in growing InxGal-xN has been the phase separation at high

indium content (X(In) > 0.3) due to the very high equilibrium vapor pressure of nitrogen

over InN. This phase separation leads to the miscibility gap in InxGal-xN [Ho96].

In the calculation of the miscibility gap in InxGal-xN, the 2-sublattice regular

solution model was used for solid phase which consists of InN and GaN sub-lattices and

the interaction parameter (L) of 5.98 kcal/mol [Ho96] and the Redlich-Kister model was

used for liquid phase where the interaction parameter of Lo and L1 are given in Table 3-2.

The regular solution model for the Gibbs excess energy of mixing for InxGal-xN is given

by

AGx (InxGa _xN) = x xGaNL (3-5)

The Redlich-Kister model for the Gibbs excess energy of mixing for In (1) and Ga

(/) is given by

AG (In(1) Ga()) = Xn(IGa() (L + Ll (xn-xGa)) (3-6)

The existence of a miscibility gap and phase separation of InxGal-xN grown by

MOPVE was confirmed by using ThermoCalc Software (Fig.3.4) [Pin98].

The maximum calculated indium content in InxGal-xN grown by MOVPE is 0.1 at

780 C, and the experimental data were used from the paper of E.L. Piner [Pin99]. Our

calculations of the maximum indium content in InxGal-xN and the critical temperature are

in the good agreement with results presented in [Str99]. However, based on experimental

data the maximum indium content in InxGal-xN is 0.3 [Elm98].










1500
1400 Phase separation
1300
1200
0 1100
1000 -
900
CL 3aN Metastable InN
S800 -
700 -
600 A: Single Phase
500 : Double Phase
400 -
0 0,2 0.4 0.6 0.8 1
GaN X(n) InN


Figure 3-4. Thermodynamically calculated miscibility gap of InxGal-xN grown by
MOVPE and the data points (AM) are from the measurements observed by
Piner.

This gap between the experimental result and theoretical one is caused by the fact that

MOVPE growth of InxGal-xN is non-equilibrium reaction and this thermodynamic results

show the binodal curve which corresponds to the stable region but not show the spinodal

curve corresponding to the meta-stable region which can be achieved experimentally.

3.2 Quantum Calculation of Phase Separation in InxGal-xN

3.2.1 Boundary Passivation Method with Hydrogen

The phase separation in InxGal-xN was studied using Quantum calculation method

(also called the first-principle method or ab-initio method). For this calculation, the

Hatree-Fock Self-Consistent-Field (HF-SCF) method was used and the total energy of the

system was calculated by the Schrodinger equation (Eq.3-7).


H H = EH (3-7)

A
where H is the Hamiltonian operator, E is the energy of system, and 'PHP is the

Hatree product (the wave function of the system);









,HP = 12 ..... VN (3-8)

hv=s, = V/, (3-9)

Therefore,

-HTHp = hfi Hp =f hVi V2 ..... ZN = E1 (3-10)
1 =1


h, = 1V Z-k + (3-11)
2 k=1 k


V {j} Idr=Y Vf j (3-12)


where V,{/} represents an interaction potential with all of the other electrons

occupying orbitals {j} and i, j represent the electron and k the nucleus.

In the first step of the SCF, one guesses the wave function i/ for all of the

occupied molecular orbitals (MOs) and uses these to construct the necessary one-electron

operator, h. Solution of each differential equation (Eq. 3-11) provides a new set of y,

presumably different from the initial guess. The one-electron Hamiltonians are formed

anew using these presumably more accurate V/. At some point, the difference between a

newly determined set and the immediately preceding set falls some threshold criterion

and the final set of i/ is referred to as the 'converged' SCF orbitals. The computational

process is shown in Fig. 3.5.










Guess at initial v

--

0_F Compute W1, = ONc .



Compute h 1 =-v-z +V}
2



Solve for hA = g



Not yet Yes
Converge?


Figure 3-5. Flow chart of the HF-SCF procedure.

For the calculation of phase separation in InxGal-xN, the structure of the unit cell

was set up, which contains In, Ga, N, and H with different indium composition and all

nitrogens at the wall sides were passivated by hydrogen to calculate the total energy (see

Fig. 3.6). As the indium content increases, the site of Ga is exchanged with In atom.

Three sets of bond lengths for In-H, Ga-H, and N-H were used for HF-SCF/3-21G

calculation. The first one is derived from the calculation of the software of Molden, the

second one from that of Hiraoka [Hir94], and the third one is obtained using PM3 method,

one of semi-empirical methods with Hyperchem software (Table 3-4). The other bond

lengths were obtained from the data of Inaba, who calculated these bond lengths using

the CAmbridge serial Total Energy Package (CASTEP) code [Ina01]. The energies were








calculated using three different calculation methods with Gaussian software (Table 3-5)
where Hatree energy is equal to 627.51kcal/mol.


K-


GaN


Ino.25Gao.7sN


Ino.38Gao.62N


Ino.7Gao.25N


InN


Figure 3-6. Structures used to compute the total energy for the InxGal-xN vs. indium mole
fraction.









Table 3-4. Bond lengths for the calculation using HF-SCF.
In-H Ga-H N-H Ga-N In025Gao 75N In.. :Ga.. N Ino75Ga. N InN
(x=0) In-N Ga-N In-N Ga-N In-N Ga-N (x=l)
1 1.813 1.813 1.008 1.96 2.17 1.968 2.174 1.969 2.187 1.971 2.188
2 1.588 1.588 0.991 1.96 2.17 1.968 2.174 1.969 2.187 1.971 2.188
3 1.694 1.624 0.991 1.96 2.17 1.968 2.174 1.969 2.187 1.971 2.188


Table 3-5. Calculated total energy for three types of different bond length.
Energy HF/3-21G(1) HF/3-21G (2) HF/3-21G (3)
(Hatree)
GaN -15758 -15763 -15762
Ino.25Gao.75N -23365 -23366 -23366
In0.38Gao.62N -27166 -27167 -27167
Ino.75Gao.62N -38558 -38570 -38570
InN -46173 -46173 -46173

In the method of HF/3-21G, the 3 signifies that three core expanded Gaussian basis

function are used as the core function and the 21G indicates that the valence functions are

split into one basis function with two Gaussian type orbitals and one with only one

Gaussian type orbital.

When these energies were compared, all three showed a similar value for each mole

fraction of In. After the first one was selected, the energy in the state without phase-

separation and the energy in the state with phase separation were compared. The energy

in the state with phase separation was calculated from the summation of the energy of

GaN and the energy of InN according to the lever rule. It is found out that the energy

with phase separation is less than the energy without phase separation when the indium

mole fraction (X(In)) is changed from 0.25 to 0.38 (Table 3-6).









Table 3-6. Calculated energies of InxGal-xN with the phase separation and without phase
separation.
Energy Energy for InxGal-xN Energy for (1-x)GaN + x InN
(Hatree) (HF/3-21G (1)) (Phase separation)
GaN -15758 -15758
Ino.25Gao.75N -23365 -23361
In0.38Ga0.62N -27166 -27315
Ino.75Gao.25N -38558 -38569
InN -46173 -46173



Based on presented data, the phase separation in InxGal-xN started to occur at 0.25<

X(In) < 0.38. This result is in good agreement with the reported value of X(In) = 0.28

[Elm98].

In summary, the phase separation in InxGal-xN was studied using quantum

calculation method and 2-sublattice regular solution model. Calculated values of the

maximum indium mole fraction, X (In), in InxGal-xN by quantum calculation method are

in a good agreement with the reported experimental data.














CHAPTER 4
CALCULATION OF THE CRITICAL THICKNESS OF InN ON GaN, A1N, Si, AND
A1203

As the lattice mismatch between an epitaxial film and substrate increases, high-

quality epitaxial growth can not continue indefinitely because the strain energy of the

layer is eventually completely or partially relieved (or relaxed) by the generation of

dislocations at the interface (misfit dislocations).

When a film with an unstrained lattice constant, af is deposited on a substrate with

a different lattice constant, a, (Fig. 4.la), it initially grows with a lattice constant equal to

that of the substrate. The mismatch is accommodated by strain in the layer. This is known

as a pseudomorphic film. This continues until the film reaches some critical thickness he

(Fig. 4. b). The critical thickness is the thickness of the overgrown film at which misfit

dislocation begins to occur. When the film thickness exceeds he, the misfit is

accommodated by the formation of misfit dislocations which emanate from the interface

between the film and substrate, and the lattice constant of the film relaxes toward the

unstrained value (Fig. 4. c) [Woo83].

The critical thickness for pseudomorphous growth is very important from a

technological point of view. First, the misfit dislocations deteriorate the performance of

the heterostructure devices due to the increased leakage current. Second, in uniformly

strained epilayers, the interatomic spacing differs from that in the unstrained (relaxed)

ones, thus changing the bandgap energy. Sapphire traditionally has been the most

commonly used substrate for III-Nitride growth. Unfortunately, in case of InN the lattice










mismatch is significant 25.7%. Several alternative substrates are considered, such as

GaN, A1N, and Si with the lattice mismatch of 10.9 %, 13.7 %, and 7.9 % respectively.

These values are much smaller compare to In/A1203 case. The values of critical thickness

of InN on grown on GaN, A1N, A1203, and Si substrates were calculated using different

models and results are presented in this chapter.









--- ----------

Mismatch
accommodated
ili by strain
(pseudomorphic)


I--- <----C(



Mismatch
(i accommodated
I by misfit
dislocation


Figure 4-1. Schematic representation of the formation of misfit dislocations: (a)
unstrained lattice; (b) thickness of the film is less than h,; (c) thickness of the
film is greater than he misfit dislocations are generated.

4.1 Calculation of Critical Thickness of InN by Matthews' Method.

In the simple model of Matthews (Mat75, Tu92), the critical thickness (he) for

pseudomorphic epitaxy is derived by minimizing the total energy, as the sum of the

stored energy as strain and the net energy which film releases from dislocation formation.

It is assumed that film and substrate are cubic and prepared from elastically

isotropic material. The elastic constants of the two crystals are assumed equal. The misfit









dislocations are considered to be arranged in a square network, to be in edge orientation,

and to have Burgers vectors in the interface plane. The energy of a square grid made up

of two perpendicular and non-interacting arrays of edge dislocation is therefore given by

Eq. (4-1). The energy associated with the elastic strain in the film is given by Eq. (4-2).

Gb h (4-1)
Edislocaon o-(f g) n +1 (4-1)
2r(1 v) b

(1+ v)
Es =r2 2G V)h (4-2)
(1 v)

(1t+ v) G Cb _h
Etotaz = E22Gf h( + V) -f (f- ) In +l1 (4-3)
(1- v) 2(1 v)b

where the in-plane strain e is defined as

aJI1 a f
8 = (4-4)
af

af as
as (4-5)
a1

2(as xa,)
b = 2(ax) (4-6)
af + a

where a. is the parallel-to- the-interface or in-plane lattice constant of the

deposited film material, afis the lattice constant of film material in the bulk or unstrained

state, as is the lattice constant of substrate, fis the misfit between film and substrate, Gf is

the shear modulus of the film, b is the Burger vector of the dislocation, and v is Poisson's

ratio for the film.

The strain that minimizes the total energy is obtained by setting dEtota de = 0 to

give the critical strain (s*).









8 b vh In- + 1 (4-7)
8;r(l +v)h b) I

The critical thickness (hc) is obtained by iteration with e* =f


h = 8 b( In )fL+1 (4-8)
8(1 + v)Of L b )

According to Eq. (4-7), the critical thickness of the overgrown film depends on the

misfitf Burger vector b and the Poisson's ratio v for the film.

For the calculation of the critical thickness of InN film on GaN and A1N substrates,

selected physical properties of these materials are needed including the lattice constants

of InN, GaN, and A1N. These property values are summarized in Table 4-1, where the

Burger vector is calculated using Eq. (4-6) and Misfit dislocation is calculated using Eq.

(4-5).

Table 4-1. Physical properties required for the calculation of the critical thickness of InN
on GaN, A1N, A1203, and Si substrates.
GaN (0001) A1N (0001) A203 (0001) Si (111)
substrate substrate substrate substrate
Burger vector (b) 3.354A 3.311 A 4.758A 3.840A
Poisson's ratio of InN (u) 0.3a (300K) 0.3a (300K) 0.3b (300K) 0.3b
(300K)
Misfit dislocation (f) -0.098 -0.120 0.345 0.086
[aBe04, bLiu02]

Table 4-2. Calculated critical thickness of InN on GaN, A1N, A1203, and Si substrates
using Mattews' method.
GaN (0001) A1N (0001) Al203 (0001) Si (111)
substrate substrate substrate substrate
Critical thickness ( A) 0.658 0.599 0.440 0.760









All these calculated values of critical thickness of InN lead to the conclusion that

the misfit dislocations are formed during the growth of the first monolayer on all

considered substrates.

4.2 Calculation of Critical Thickness of InN by van der Merwe's Method.

When the square atomic meshes of the adjoining crystal planes are considered, the

energy of the interface due to the lattice misfit will be equal to the energy of the

homogeneous strain Ehs given by


Eh, = 2Gfhf2 b (4-9)
1- vb

Beyond the critical thickness, misfit dislocations are introduced at the interface so

that initially homogeneous strain and misfit dislocation energy coexist. According to the

theory of van der Merwe, the energy of the misfit dislocations (the homogeneous strain is

absent) is naturally divided into two parts. The first is the energy of intersection between

the atoms of the two crystal halves (Eq. 4-10) [Mar03].


4, = (1+A +A (4-10)


where d b is the separation of the atoms of the adjoining crystal planes.

The second is the energy of the periodic elastic strain energy which is distributed in

the two crystal halves. This second energy is the total strain energy per atom of the misfit

dislocations (Eq. 4-11).


E, = ln(2A 2 2A2 (4-11)


The misfit dislocation energy Ed is then given by van der Merwe (Eq. (4-10))

through the sum of(Eq. 4-10) and (Eq. 4-11) [Mar03].










Ed = -G 1 +- 1 ln(2Zl+A2 -2A2) (4-12)


A =2G'b
G, p
1 1- V 1 -vb
G' G G,
G=VGG

2afas
b-
a +as)
aa
aP a
af -ad

where G is the shear modulus at the interface, Gf is the shear modulus of the film,

Gs is the shear modulus of the substrate, b is Burger vector, and p is the vernier of misfit

or the dislocation spacing as shown in Fig. 4.2 [Mar03]. The dashed lines located at a

distance p/2 from the contact plane show the boundary beyond which the periodic strains

originating from the dislocations practically vanish. The physical properties for the

calculation of critical thickness of InN on GaN, A1N, A1203, and Si substrates are

summarized in Table 4-3.




p Film


Substrate







Figure 4-2. Model of epitaxial interface between two semi-infinite crystals resolved in a
sequence of misfit dislocations spaced at an average distance.









Table 4-3. Physical properties required for the calculation of critical thickness of InN on
GaN, A1N, A1203, and Si substrates.
InN GaN (0001) A1N (0001) A1203 (0001) Si (111)
substrate substrate substrate substrate

Poisson's ratio (u) d0.3 d0.23 d0.2 a0.275 b0.218
Bulk Modulus (GPa) C165 '236 c248 a338 b98
[aBel80, bGeo99, Chi99, dBel04]

In the isotropic materials, the shear modulus (G) can be calculated from a bulk

modulus (B) or tensile (Young's) modulus (E) by


3(1- 2v)
2(1 + v)


1
2( +v)E
2(1 + v)


(4-13)


The modulus (G) calculated from the given bulk modulus (B) for each material

using equation (4-13) is shown in Table 4-4.

Table 4-4. Calculated shear moduli for InN, GaN, A1N, A1203, and Si materials.
InN GaN (0001) A1N (0001) A203 (0001) Si (111)
substrate substrate substrate substrate

Shear modulus 76 155 186 179 68
(GPa)


Equating the 2 Ed of a square grid of two perpendicular and noninteracting arrays of

misfit dislocations and the energy Eh, of the homogeneous strain gives ()a= uIb = ))


G, ( -v) f(A)
4 r 2G (1 +v) f2


f af -as
af


f(A) = 1+ l + 1 2Aln(2AZI +2 22)


(4-14)


(4-15)










To illustrate the minimum energy considerations, the homogeneous strain energy

Ehs is plotted for the number of InN epilayers and the dislocation energy Ed against the

misfit for InN on a GaN substrate (see Fig. 4.3).


18
816 InN on GaN substrate
14 Ehs(n=1)
12
10
E(J/m2) 8 2E
6
4
2
0
0 0.02 0.04 0.06 0.08 0.1
0.047

Misfit ((af-as)/af)


Figure 4-3. Homogeneous strain energy, Eh, for the 1st InN epilayer and dislocation
energy, Ed vs. the misfit, f on GaN substrate.

Based on results presented in Fig.4.3, if the misfitf > 0.047 (the critical value), Eh,

is greater than Ed even for a monolayer, which means that the misfit dislocation occurs

during the growth of the 1st InN epilayer since the misfit f between the InN film and the

GaN substrate is 0.098. The calculated critical thickness (using Eq. 4-14) of InN on GaN

substrate is 3.927 A.

Identical calculation procedure was performed for the growth of InN on an A1N

substrate with the matching result that the misfit dislocation also occurs during the

growth of the 1st InN epilayer since the misfit (f) between InN film and A1N substrate is

0.120, which is larger than 0.048 of the critical value of (Fig.4.4). The critical thickness

of InN on A1N substrate is 3.044 A (using Eq. 4-14).